CLINICAL ARTICLE

J Neurosurg 135:671–682, 2021

Recurrent brainstem cavernous malformations following

primary resection: blind spots, fine lines, and the
right-angle method
Roxanna M. Garcia, MD, MS, MPH,1 Taemin Oh, MD,2 Tyler S. Cole, MD,1
Benjamin K. Hendricks, MD,1 and Michael T. Lawton, MD1
1
Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona; and
2
Department of Neurological Surgery, University of California, San Francisco, California

OBJECTIVE  Proximity of brainstem cavernous malformations (BSCMs) to tracts and cranial nerve nuclei make it costly
to transgress normal tissue in accessing the lesion or disrupting normal tissue adjacent to the lesion in the separation
plane. This interplay between tissue sensitivity and extreme eloquence makes it difficult to avoid leaving a remnant on
occasion. Recurrences require operative intervention, which may increase morbidity, lengthen recovery, and add to over-
all costs. An approximately 20-year experience with patients with recurrent BSCM lesions following primary microsurgi-
cal resection was reviewed.
METHODS  A prospectively maintained database of 802 patients who underwent microsurgical resection of cerebral
cavernous malformations during 1997–2018 was queried to identify 213 patients with BSCMs. A retrospective chart re-
view was conducted for patients with recurrent BSCM after primary resection who required a second surgery.
RESULTS  Fourteen of 213 patients (6.6%) underwent repeat resection for recurrent BSCM. Thirty-four hemorrhagic
events were observed among these 14 patients over 576 patient-years (recurrent hemorrhage rate, 5.9% per year; me-
dian discrete hemorrhagic events, 2; median time to rehemorrhage, 897 days). BSCM occurred in the pons in 10 cases,
midbrain in 2 cases, and medulla in 2 cases. A blind spot in the operative corridor was the most common cause of re-
sidual BSCM (9 patients). All recurrent BSCMs were removed completely, although 2 patients each required 2 operations
to treat recurrence. Twelve patients had unchanged or improved modified Rankin Scale scores at last clinical evaluation
compared with admission, and 2 patients had worse scores. Recurrence was more common among patients who were
operated on in the first versus the second half of the series (8.5% vs 4.7%).
CONCLUSIONS  The 6.6% rate of BSCM recurrence requiring reoperation reflects the fine lines between complete
resection and recurrence and between safe and harmful surgery. The detection of remnants is difficult postoperatively
and remains so even at 6 months when the resection bed has healed. The 5.9% annual hemorrhage risk associated with
recurrent BSCM in this experience is consistent with that reported for unoperated BSCMs. The right-angle method helps
to anticipate blind spots and meticulously inspect the resection cavity for residual BSCM during surgery. A low percent-
age of recurrent BSCM (5%–10%) ensures ongoing effort toward an acceptable balance of safety and completeness.
https://thejns.org/doi/abs/10.3171/2020.6.JNS201555
KEYWORDS  brainstem cavernous malformation; cavernoma; recurrent; remnant; reoperation; residual; vascular
disorders

E volving surgical perspectives have transformed the

brainstem cavernous malformation (BSCM) from
an inoperable lesion in a surgical no man’s land to
a lesion that is best managed microsurgically using skull
atic reviews of BSCMs describing the natural history and
management of these lesions1–3 have established microsur-
gical resection as the standard of care following symptom-
atic hemorrhage within the context of the patient’s clinical
base approaches and the lesion’s presentation on pial or status and life expectancy, the lesion’s accessibility within
ependymal surfaces to access it safely while avoiding neu- the brainstem, and the estimated risk of rehemorrhage.4,5
ral tracts and cranial nerve (CN) nuclei. Several system- Gross et al.2 reported an overall 91% complete resection

ABBREVIATIONS  BSCM = brainstem cavernous malformation; CN = cranial nerve; mRS = modified Rankin Scale.
SUBMITTED  April 30, 2020.  ACCEPTED  June 30, 2020.
INCLUDE WHEN CITING  Published online November 20, 2020; DOI: 10.3171/2020.6.JNS201555.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  671
Garcia et al.

rate among 1390 patients aggregated from the literature, neurosurgeon were included because each had a second
with rates ranging from 85% to 100% in larger case se- recurrence after surgery that was treated by the senior au-
ries.4–7 When resecting cavernous malformations in any thor (3 BSCM operations in total were performed for each
location, success is measured first by the complete remov- patient). Lesions located entirely within the basal ganglia
al of the symptomatic lesion and second by the absence or thalamus were excluded as well.
of new permanent neurological deficits. Curative resection
at the time of index surgery rids the patient of the offend- Data Collection
ing lesion and eliminates the risk of future deterioration All data were retrospectively collected from clinical as-
caused by rehemorrhage, at least in the sporadic form of sessments (at admission, during the immediate postoper-
the disease. Surgical success is common with most cav- ative period, and at last follow-up), radiographic imaging,
ernous malformations in the cerebral and cerebellar hemi- operative notes, intraoperative photographs and videos,
spheres. Still, resection of cavernous malformations in postoperative imaging, and final pathology reports. The
the brainstem is challenging. The proximity of BSCMs to senior author dictated all operative notes. An independent
the descending and ascending tracts and CN nuclei make neuroradiologist reviewed all neuroimaging data. All pa-
it costly to transgress any normal tissue in accessing the tients included in the data set had pathologically proven
lesion or disrupting any normal tissue adjacent to the le- cavernous malformations. Operative notes and radio-
sion in the separation plane. This interplay between tissue graphic images were used to determine the presence of
sensitivity and extreme eloquence makes it difficult, if not developmental venous anomalies. The modified Rankin
impossible, to avoid leaving a remnant on occasion. The Scale (mRS) score was utilized to assess and quantify pa-
largest institutional surgical case series by a single neuro- tient outcomes.11 The outcome was considered favorable if
surgeon reported a rate of residual or recurrent BSCM of the patient had an mRS score ≤ 2 at the time of the final
11% (29 of 260 patients).4 clinic evaluation; an mRS score > 2 was considered an un-
The natural history and management of residual or re- favorable score. The timing of rehemorrhage was defined
current BSCMs are poorly described in the literature.4,8–10 from the date of surgical intervention to the date of clini-
Current findings suggest that completely resected BSCMs cal rehemorrhage or radiographic evidence of hemorrhage
do not recur de novo in the same location, which implies or finding of a residual lesion on postoperative imaging,
that some residual BSCM or remnants are likely pres- after radiographic evidence of rehemorrhage.
ent but poorly visualized immediately after resection.4,9
BSCM recurrence is associated with many difficulties. Statistical Analysis
For the patient, it is a clinical setback that can generate Summary statistics were reported using the median
frustration, disappointment, and additional morbidity. For and IQR due to small sample bias. Descriptive statisti-
the neurosurgeon, recurrence can rattle surgical self-con- cal analysis was conducted using the Wilcoxon rank-sum
fidence, cause embarrassment, and diminish the trust of test and Fisher’s exact test for continuous and categorical
patients and colleagues. Symptomatic recurrences require variables, respectively, given small cell counts. Statistical
operative intervention, which may increase morbidity, tests were considered significant if p < 0.05. All statistical
lengthen recovery, and add to overall hospitalization and analyses were performed using Stata Intercooled version
rehabilitation costs. Although rare, this problem deserves 14 (StataCorp).
special attention. We therefore reviewed an approximately
20-year experience with BSCM among patients with re-
current lesions following primary microsurgical resection. Results
Patient Characteristics and Presentation
Methods A total of 14 of 213 patients (6.6%) underwent a sec-
ond resection for recurrent BSCM after primary resection
Study Population (Table 1). The median age of patients who underwent a
This study was approved by the Committees on Human second resection for recurrent BSCM was 32 years (IQR
Research at the University of California, San Francisco, 32–41 years), with a sex preponderance toward female pa-
and the Barrow Neurological Institute. The prospectively tients (n = 13).
maintained database of the senior author (M.T.L.) was Nine patients were clinically symptomatic at the time
queried to identify 802 patients who underwent microsur- of recurrent hemorrhage and presented with worsening
gical resection of a cavernous malformation in the brain neurological deficits. Five patients had asymptomatic re-
between October 1, 1997, and November 30, 2018. From currences detected on surveillance MRI at follow-up im-
this cohort, a retrospective chart review was conducted for aging (cases 1, 2, 3, 10, and 13). Two patients (cases 2 and
213 patients who underwent microsurgical resection of 3) had suspicious immediate postoperative imaging and
BSCMs during this period. Patients with recurrent BSCM underwent multiple postoperative scanning sessions until
after primary resection who required a second surgery the surgical remnant became evident during hospitaliza-
were included in the study. Patients were included only tion. Three patients (cases 1, 10, and 13) did not have any
if the primary and secondary surgeries were performed neurological deterioration or remnants found on MRI im-
by the senior author; 9 patients with symptomatic hemor- mediately after surgery, but hemorrhagic blood products
rhage from recurrent BSCM after resection performed by (cases 1 and 10) or a remnant (case 13) was ultimately
another neurosurgeon were excluded. Two patients with identified. The median time from surgical intervention to
recurrent BSCM after resection performed by another a remnant or asymptomatic rehemorrhage without neu-

672 J Neurosurg Volume 135 • September 2021

 
* | * 04:30 *
 TABLE 1. Characteristics among patients with recurrent BSCM requiring reoperation
Days to Clinical Vol (mm3) Max Diameter (mm) Cross
Case Age (yrs), No. of Total No. Yrs Btwn Rebleed or Timing Presentation Depth Axial BSCM
No. Sex Events of Ops Ops Remnant* to Op at Rebleed Location Primary Secondary Primary Secondary to Pia Surface Midline Grade†
1 41, F 4 2 2.1 707 Subacute No Pons 4826 30.1 Superficial Dorsal Yes 4
2 37, F 2 2 0.1 7 Acute No Pons 7107 9472 25.4 29.5 Deep Dorsal Yes 2
3 22, F 2 2 0.0 1 Acute No Medulla 2045 2033 17.3 17.2 Superficial Ventral Yes 1
4 56, M 2 2 0.1 21 Acute Yes Pons 13,694 15,497 30.5 33.5 Deep Ventral Yes 5
5 38, F 2 2 5.7 2082 Subacute Yes Pons 1541 2098 16.4 16.9 Superficial Dorsal Yes 3
6 67, F 3 2 2.3 838 Acute Yes Pons 1145 2077 16.3 16.9 Deep Dorsal No 3
7 39, F 3 3‡ 8.9 3249 Acute Yes Pons 3710 27.4 Superficial Ventral Yes 2
8 26, F 2 2 5.9 2159 Acute Yes Midbrain 1782 3093 17.4 29.3 Deep Ventral No 2
9 32, F 2 2 2.8 1005 Acute Yes Medulla 786 1320 12.0 15.2 Deep Ventral Yes 2
10 26, F 3 2 0.4 127 Acute No Pons 1121 271 15.4 8.5 Exophytic Dorsal Yes 1
11 37, F 2 2 4.5 1620 Acute Yes Pons 523 3316 12.2 19.3 Deep Dorsal No 1
12 39, F 2 2 1.2 430 Acute Yes Pons 1192 546 7.4 10.7 Exophytic Ventral No 1
13 43, F 3 2 3.2 1113 Acute No Midbrain 405 1116 9.9 11.7 Superficial Dorsal No 3

* | * 04:30 *
14 40, F 2 3§ 0.7 957 Subacute Yes Pons 5502 383 24.5 10.2 Superficial Ventral Yes 3
* Days to rebleed or residual represents the time from last surgery to clinical presentation, radiographic evidence of rehemorrhage, or residual lesion without clinical deterioration.
† BSCM grade was measured using 5 factors: size of cavernoma (0–2 cm, 0 points; > 2 cm, 1 point), crossing axial midpoint (no, 0 points; yes, 1 point), presence of developmental venous anomaly (no, 0 points; yes, 1
point), age (0–40 years, 0 points; > 40 years, 2 points), and hemorrhage timing (acute [0–3 weeks], 0 points; subacute [3–8 weeks], 1 point; or chronic [> 8 weeks], 2 points).
‡ Patient underwent 3 operations.
 
§ Patient had slow progressive evidence of deterioration and opted for a third surgery.

J Neurosurg Volume 135 • September 2021

673
Garcia et al.
Garcia et al.

rological deterioration was 127 days (IQR 4–906 days). Patient Outcomes
Overall, there were 34 hemorrhagic events among these The mean hospitalization time was 8 days (IQR 6–11
14 patients over 576 patient-years, resulting in a retrospec- days), and the median follow-up time was 15 months (IQR
tive recurrent hemorrhage rate of 5.9%, or 5.9 events per 6–39 months). There were no intraoperative complications
patient-year. The median number of discrete hemorrhagic or deaths among the patient cohort. Four patients devel-
events was 2 (IQR 1–3 events). The median time from the oped medical complications or required additional pro-
last surgery to rehemorrhage was 1005 days (IQR 634– cedures: ventriculoperitoneal shunt (case 2), gastrostomy
2211 days) among those without evidence of remnant on tube placement (case 4), pseudomeningocele (case 10), and
postoperative neuroimaging. chemical meningitis (case 13).
Overall, patients improved neurologically after treat-
BSCM Characteristics ment. All patients initially presented with at least one fo-
Three patients had multiple cerebral cavernomas, and cal neurological deficit (1 patient presented with obtun-
11 had single sporadic BSCMs, with 23 total lesions dation, 7 patients had motor deficits, 11 had at least one
among the 14 patients. Recurrent BSCM was located in cranial nerve deficit, 7 had sensory dysfunction, and 3 had
the pons in 10 cases, in the midbrain in 2 cases, and in the cerebellar dysfunction or ataxia). In the immediate post-
medulla in 2 cases. Seven lesions had ventral locations. operative period, 2 patients had worsening cranial neu-
The median maximum diameter was 16.9 mm (IQR 11.7– ropathies, and 1 had new mild hemiparesis (4+/5, case 14)
27.4 mm). The median distance to the pial surface was after the third resection; otherwise, all patients remained
0.25 mm (IQR 0–2.0 mm). Two recurrent BSCMs were at neurological baseline. At the time of the last clinical
exophytic, 6 abutted the pial surface, and 6 were deep. evaluation, 2 patients had motor deficits, 5 patients had
Eight patients had a concurrent developmental venous sensory dysfunction, and 1 patient had cerebellar dysfunc-
anomaly that drained the index BSCM. BSCM grades are tion. Four patients had made a significant recovery in their
shown in Table 1. cranial nerve deficits. Overall, 13 patients had at least one
CN deficit at the time of the last follow-up. Twelve patients
Surgical Results had mRS scores that were unchanged or improved at the
last clinical evaluation compared with mRS scores at ad-
The extended retrosigmoid (n = 6) and suboccipital- mission, and 2 patients had worse mRS scores (Table 2).
transventricular (n = 5) surgical approaches were the most
common. Two far-lateral approaches, 1 orbitozygomatic- Neurosurgeon Experience
pterional approach, 1 lateral supracerebellar-infratentorial
approach, and 1 translabyrinthine approach were also per- Although the overall recurrence rate was 6.6%, recur-
formed (Table 2). rence was more common among patients who were oper-
Eleven of the 14 patients underwent primary surgical ated on in the first half of the series than those operated on
treatment in an acute time frame (less than 3 weeks), and in the second half. Overall, 9 recurrences were observed
3 patients underwent primary surgical treatment in a sub- among the first 106 patients (8.5%), and 5 recurrences
acute time frame (3–8 weeks) (Table 1). Two patients un- were observed among the last 107 patients (4.7%), which
derwent a second operation for recurrence (third surgery suggests that increased surgeon experience may be associ-
overall). The first patient (case 7) experienced rebleeding ated with lower recurrence rates (Fig. 5).
8.5 years after the second operation. The other patient
(case 14) underwent resection 8.5 months after her second Discussion
resection when a residual lesion was found on her 6-month In this consecutive series of patients with BSCMs treat-
surveillance MRI. In these 2 cases of pontine BSCMs, the ed microsurgically over 20 years, the rate of recurrence
initial incomplete resection was performed by another requiring reoperation was 6.6% (14 of 213 patients), with
neurosurgeon, and the exposure in both cases was deemed all but 2 of these patients presenting with acute clinical
by the senior author to be inadequate. In one patient (case deterioration from new hemorrhage. Patients experienced
7), the prior retrosigmoid approach was expanded to a rebleeding an average of 2.5 years after their primary
translabyrinthine approach. In the other patient (case 14), resection, which is comparable to the timing reported in
the prior “mini” retrosigmoid craniotomy was expanded other large institutional case series12 and which demon-
to an extended retrosigmoid craniotomy. strates the difficulty of detecting remnants. With our man-
A review of intraoperative photographs and videos de- agement protocol, postoperative MRI is performed within
termined that a blind spot in the operative corridor was the 48 hours after resection to identify obvious remnants, and
most common cause of residual BSCM, which was found patients with positive imaging findings are returned to the
in 9 patients (Figs. 1 and 2). An anatomical obstacle hid operating room to complete the resection. There was no
the remnant in 2 patients (a developmental venous anomaly immediate return to the operating room in this experi-
in 1 patient and CN IX and X in 1 patient). In 2 patients, ence. However, 2 patients (cases 2 and 3) remained hos-
the BSCM was in clear view, but the tissue at the depth of pitalized due to inconclusive imaging findings and ulti-
the cavity was misinterpreted as normal brainstem rather mately underwent resection. If no residual BSCM is found
than a remnant of the lesion (Fig. 3). Recurrent BSCM was in the immediate postoperative period, a follow-up MRI is
found at reoperation at this exact location in these 2 pa- done at 6 months to evaluate the cavity after the blood has
tients. In one patient, the recurrence was felt to be a satellite been resorbed and the tissues have healed. Surveillance
lesion immediately adjacent to the main BSCM (Fig. 4). imaging is recommended again at 5 years. There were an

674 J Neurosurg Volume 135 • September 2021

 
* | * 04:30 *
 TABLE 2. Outcomes among patients with BSCMs requiring reoperation
Primary Preop Secondary Postop Last Clinical
Neuro Deficits Neuro Deficits Evaluation Deficits
No. No. No. Preop Last
Case Secondary of CN Motor of CN Motor of CN Motor Op mRS mRS FU LOS
No. Preop Symptoms Deficits Deficits Other Deficits Deficits Other Deficits Deficits Other Approach DVA Score Score (mos) (days)
1 Motor weakness, paresthe- 1 MRC 4/5 None 3 4+/5 HemiA 3 None None Suboccipital transventricular No 3 2 66 6
sias, imbalance, slurred
speech, dizziness
2 HA, nausea, vomiting, 3 None Ataxia 3 4+/5 HemiA 1 None None Suboccipital transventricular No 4 2 10 13
dysarthria, weakness
3 Vertigo, diplopia, numbness, 4 MRC 3/5 Dysmetria 2 4+/5 HemiA, 2 None Ataxia Rt far lat No 2 2 5 6
dysphagia ataxia
4 Obtunded 1 MRC 4/5 None 5 0/5 None 2 MRC 2/5 HemiA Lt extended retrosigmoid Yes 4 5 93 19
5 Motor weakness, facial 2 None None 1 3/5 HemiA 1 None None Suboccipital transventricular Yes 4 2 13 6
numbness
6 Motor weakness 1 None HemiA 2 3/5 HemiA 5 None HemiA Suboccipital transventricular Yes 3 3 31 8
7 Paresthesias, diplopia, 3 None HemiA 2 None HemiA 1 None HemiA Translabyrinthine; rt ex- No 1 1 39 10
facial numbness & droop, tended retrosigmoid
vertigo
8 Diplopia, motor weakness 5 MRC 1/5 UA 1 4+/5 None 1 None None
Lt orbitozygomatic/pteronial Yes 2 1 17 12

*30 *
9 HA, diplopia 1 None HemiA 1 None HemiA 1 None None
Rt extended retrosigmoid Yes 2 1 6 7
10 HA, diplopia 2 None None 0 None None 1 None None
Suboccipital transventricular No 1 1 6 5
11 Facial numbness, spastic leg 1 MRC 4/5 None 2 None None 1 None Ataxia
Rt extended retrosigmoid Yes 3 2 95 8
12 Diplopia, blurry vision 1 None None 0 None None 1 None None
Lt far lat Yes 1 1 38 11
13 HA, paresthesias, facial 1 None Ataxia 1 None HemiA 1 None HemiA
Lt lat suboccipital, supracer- Yes 1 1 52 3
 
numbness, diplopia ebellar/infratentorial
14 Motor weakness, paresthe- 0 MRC 4/5 None 0 None None 0 MRC 4+/5 HemiA Rt extended retrosigmoid (2) Yes 1 3 41 1
sias
DVA = developmental venous anomaly; FU = follow-up; HA = headache; HemiA = hemianesthesia; LOS = length of stay; MRC = Medical Research Council; UA = unable to assess due to patient’s poor mental status.

J Neurosurg Volume 135 • September 2021

675
Garcia et al.
Garcia et al.

FIG. 1. Case 12. A residual pontine cavernous malformation hid in the inferior blind spot of a left far-lateral/transpontomedullary
sulcus approach, behind the olivary nucleus. This 39-year-old woman presented with diplopia and blurred vision. A and B: Axial
(A) and sagittal (B) T2-weighted FLAIR MR images demonstrating the lesion. C: A far-lateral craniotomy exposed the lesion in the
vago-accessory triangle between the vagus nerve (CN X), spinal accessory nerve (CN XI), and olivary nucleus (Olive). D: This
exophytic lesion was resected through a transpontomedullary sulcus approach, but the residual lesion remained inferiorly behind
the olivary nucleus (asterisk). E: Sagittal T2-weighted FLAIR MR image showing no residual lesion postoperatively, but the patient
presented 14 months later with symptomatic rehemorrhage. F and G: Axial (F) and sagittal (G) T2-weighted FLAIR MR images
showing evidence of hemorrhage. H: At reoperation through the same left far-lateral craniotomy, the cavernous malformation (as-
terisk) was found in the previous resection cavity inferiorly, behind the olivary nucleus. I: Postoperative axial T2-weighted FLAIR
MR image confirming complete removal of the lesion. The patient recovered with no new neurological deficits. AICA = anterior in-
ferior cerebellar artery; Cbl = cerebellum; ChPlx = choroid plexus; CN IX = glossopharyngeal nerve. Panels C, D, and H: copyright
Barrow Neurological Institute, Phoenix, Arizona. Published with permission.

additional 3 returns (cases 1, 10, and 13) to the operating associated with a significant risk of additional rehemor-
room in this experience after positive surveillance imag- rhage and morbidity, and therefore surgical intervention is
ing. This vigilant protocol failed to identify residual or re- recommended even when the patient is clinically asymp-
current lesions in 9 of the 14 patients, demonstrating the tomatic. Routine surveillance should continue with MRI 5
difficulty of detecting tiny remnants in the perioperative years after resection of a recurrent BSCM to ensure that a
period and also suggesting that new hemorrhage may be complete resection has been achieved. Conservative man-
needed to bring them to detection. Tissue edema, hemo- agement of asymptomatic remnants with hemorrhagic
static agents, and pooled blood in the resection cavity can activity is not usually recommended but may be appropri-
hide a remnant, which is why we pursue hemostasis me- ate when the reoperative risk is high or the patient prefers
ticulously with bipolar cautery, avoid the use of oxidized observation.
cellulose products inside the cavity or to line its walls, and A 93.4% complete resection rate is consistent with oth-
fill the resection bed with irrigant before closing. If sur- er reports in the literature, but some of the 199 patients
veillance imaging at 6 months or later identifies a remnant who were not operated on whose BSCM was deemed to
lesion, the reappearance often results from bleeding ac- have been cured may have asymptomatic recurrences that
tivity, as in both of our cases. Hemorrhagic remnants are have not yet bled or been diagnosed. Therefore, the 6.6%
676 J Neurosurg Volume 135 • September 2021
 
*30 *
 Garcia et al.

FIG. 2. Case 8. A residual midbrain cavernous malformation hid in the medial blind spot under the cerebral peduncle after a left
orbitozygomatic-pterional/transsylvian approach. This 26-year-old woman presented with diplopia and hemiparesis. A: Axial
T2-weighted FLAIR MR image demonstrating the lesion. B: Through a left orbitozygomatic-pterional craniotomy and transsylvian
approach, the cerebral peduncle (CPed) was exposed and mapped to identify the descending motor tract (1). C: The lesion was
resected medial to the motor tract through the hemosiderin stain on the CPed. The lesion was accessed between the thalamoper-
forators (Perf), but residual lesion remained medially (asterisk). D: Axial T2-weighted FLAIR MR image showing no residual lesion
postoperatively, but the patient presented 6 years later with symptomatic rehemorrhage. E: Axial T1-weighted MR image showing
evidence of hemorrhage. F: At reoperation through the same left orbitozygomatic-pterional craniotomy/transsylvian approach, the
recurrent cavernous malformation was found medial to the previous resection cavity, underneath a more medial thalamoperfora-
tor trunk (asterisk). G: Axial T2-weighted MR image confirming complete removal of the lesion. The patient recovered with no new
neurological deficits. CN III = oculomotor nerve; ICA = internal carotid artery; PCA = posterior cerebral artery; Tent = tentorium.
Panels B, C, and F: copyright Barrow Neurological Institute, Phoenix, Arizona. Published with permission.

recurrence rate in our study may be an underestimate. In Surgical Blind Spots

addition, patients with recurrent BSCM who opted to fore- Blind spots, or areas in the surgical corridor with a
go repeat surgery are difficult to identify retrospectively. compromised or inadequate view, were identified as the
Although we do not know of any such patients in this se- most common cause of incomplete BSCM resection in this
ries, they would not have been included in the study. The experience. The inherent depth of lesions in the brainstem
bleeding behavior of recurrent BSCM in this experience, and the length of the corridor make blind spots inevitable.
namely, the 5.9% annual hemorrhage risk, is consistent The surgical corridor comprises 1) the craniotomy, 2) a
with that reported for BSCMs not operated on and appears subarachnoid or ventricular segment, 3) a parenchymal
to be associated with a similar clinical course. segment, and 4) an intralesional segment (Fig. 6). Each
J Neurosurg Volume 135 • September 2021 677
 
*30 *
Garcia et al.

FIG. 3. Case 10. Recurrence resulted from misinterpretation of the deep margin of the cavernous malformation. This 26-year-old
woman presented with headache and diplopia. A and B: Axial (A) and sagittal (B) T2-weighted MR images demonstrating the
lesion. C: A suboccipital craniotomy and transventricular approach easily exposed the lesion on the floor of the fourth ventricle.
D: After what was thought to be a complete resection, the deep wall of the resection cavity was interpreted as gliotic pons (aster-
isk). E and F: Postoperative axial T1-weighted (E) and sagittal T2-weighted (F) FLAIR MR images do not appear to suggest re-
sidual BSCM, but when the patient presented for routine surveillance imaging without evidence of neurological deterioration, there
was radiographic evidence of rehemorrhage. G: Axial T1-weighted MR image showing evidence of hemorrhage. H: At reoperation
through the same suboccipital craniotomy, recurrent cavernous malformation was found at the depths of the previous resection
cavity (asterisk). Deeper dissection into the cavity exposed the residual lesion, and it was removed completely. I and J: Axial (I)
and sagittal (J) T2-weighted FLAIR MR images demonstrating complete removal of the lesion. The patient recovered with no new
neurological deficits. Panels C, D, and H: copyright Barrow Neurological Institute, Phoenix, Arizona. Published with permission.

component adds length and limits visualization. For ex- keyhole is displaced to the end of an already long corridor
ample, the trans–middle cerebellar peduncle approach to without that range of viewing angles, which decreases vi-
a pontine BSCM has a corridor made up of the retrosig- sualization and requires that the lesion be mobilized into
moid craniotomy, subarachnoid dissection into the cer- the viewing axis. The blind spots created by this geom-
ebellopontine angle, a tunnel through the middle cerebel- etry force possible blind dissection and hide the pathology.
lar peduncle, and entrance into the lesion. The view of the One must conceptualize the blind spots and deliberately
lesion through this 6-cm corridor can be obscured by the inspect these spaces at the end of the resection, even after
edges of the craniotomy, a prominent flocculus, the col- the resection seems complete.
lapsing white matter of the middle cerebellar peduncle, or This difficult geometry is compounded by darkness,
the corners at the interface between the middle cerebel- swollen brain tissue, and poor approach selection. Lighted
lar peduncle and the lesion laterally. Corridors to BSCMs instruments (bipolar forceps and suction) in long corridors
have small openings, long lengths, and limited viewing are invaluable. Brain edema can narrow the corridor and
angles. This situation is the opposite of the keyhole con- make the walls more susceptible to injury with manipula-
cept in which a small opening plus a wide range of views tion and retraction. Delaying surgery for at least 2 weeks
still enable a good survey of the lesion. With BSCMs, the after a hemorrhage allows time for edema to resolve and
678 J Neurosurg Volume 135 • September 2021
 
*30 *
 Garcia et al.

FIG. 4. Case 11. This 37-year-old woman presented with right facial numbness. A and B: Axial T2-weighted (A) and T1-weighted
Gd-enhanced (B) MR images demonstrating a cavernous malformation in the right lateral pons at the level of the trigeminal
root entry zone. Note the satellite cavernous malformation in the middle cerebellar peduncle, lateral to the pontine lesion. C: An
extended retrosigmoid craniotomy and dissection into the cerebellopontine angle exposed the greenish hemosiderin stain on the
lateral pontine surface, superior to the facial (CN VII) and vestibular (CN VIII) nerves. D: The lesion was identified in the lateral
pons and removed. E: Postoperative CT scan showing no findings suggestive of a residual lesion. MRI also lacked findings sug-
gestive of a residual lesion (not shown). F and G: Axial T2-weighted FLAIR (F) and T1-weighted (G) MR images showing new
hemorrhage at the site of the prior surgery 4.5 years later (at 37 years of age). The hematoma was contiguous with the unresected
satellite lesions seen previously in the middle cerebellar peduncle. H: At reoperation through an extended retrosigmoid craniotomy,
the satellite lesion was resected through a transfloccular/trans–middle cerebellar peduncle approach (double asterisks), and a
small recurrence was resected at the previous resection cavity in the lateral pons (asterisk). I: Postoperative axial T1-weighted MR
image confirming complete resection. The patient recovered with no new neurological deficits. Cbl = cerebellum; CN VII = facial
nerve; CN VIII = vestibulocochlear nerve; CN IX = glossopharyngeal nerve; CN X = vagus nerve; Floc = flocculus. Panels C, D,
and H: copyright Barrow Neurological Institute, Phoenix, Arizona. Published with permission.

associated hematoma to liquefy, both of which make it is placed where the BSCM comes to a pial or ependymal
easier to dissect the BSCM separation planes. Dynamic surface. The second point is placed at the center of the le-
retraction is important within deep corridors because it sion, and the line connecting these 2 points defines the le-
shifts the tissues, viewing axis, and illumination slightly sion’s axis. The 2-point method is a heuristic that matches
with each move to gain a glimpse into blind spots. Fixed the lesion’s axis with the surgical axis, which is the trajec-
retraction is rarely helpful in seeing within the resection tory of the operative corridor. When these axes align, the
cavity and often obstructs the dissection maneuvers. lesion is well visualized within the resection cavity; when
these axes do not align, visualization is impaired.
One way to conceptualize blind spots is with another
Right-Angle Method method: the “right-angle method.” The surgical axis is de-
The “2-point method” has proven to be a useful guide fined by 2 points: one at the superficial entrance into the
to select the correct approach to BSCMs.13 The first point operative corridor (corridor entry, point 1) and one at the
J Neurosurg Volume 135 • September 2021 679
 
*30 *
Garcia et al.

FIG. 5. Effect of neurosurgeon experience on the incidence of residual or recurrent BSCMs. The cumulative case volume of all
cerebral cavernous malformations resected by the senior author during an approximately 20-year period is plotted. Patients with
residual or recurrent BSCM are superimposed chronologically as horizontal lines, with the line beginning at the time of the first op-
eration and ending at the time of the last operation. Initial operations are indicated with a white dot, and operations to treat residual
or recurrent BSCM are indicated with a black dot. If the second operation occurred less than 6 months after the initial operation,
both operations are indicated with a single white dot. Two patients underwent 3 operations; for one of these patients, the initial
operation was performed by another surgeon in 1995.

entry into the BSCM (front wall, point 2; Fig. 7). The sur- tious dissection define these differences. Sometimes color,
gical axis is drawn by connecting these points and extend- contour, and texture make this distinction obvious, with
ing this line through the lesion to its back wall (point 3). mulberry-like lesions embedded in hemosiderin-stained
With the right-angle method, a perpendicular line is drawn parenchyma or a thick capsule that peels away cleanly
from the surgical axis to the point on the lesion’s capsule from the brainstem. Other times, hemorrhagic parenchy-
farthest from the axis. Lobulations and compartments of- ma blends indistinctly with the pathology, as in 2 cases in
ten make these lesions irregular and asymmetrical, rather which the remnant was not in a blind spot but left behind
than simple spheres. Perpendicular lines drawn on axial, in plain sight. The brainstem’s supreme eloquence can
sagittal, and coronal MR images identify multiple poten- make neurosurgeons cautious. Conservatism keeps opera-
tial blind spots, and any perpendicular line exceeding 1 tions safe and patients intact, whereas probing deeper and
cm in length identifies a potential blind spot. The surgical pushing further can be devastating. At these moments of
axis approximates the line of sight into the resection bed, uncertain interpretation, completeness and morbidity pull
which makes spots in the near corners more difficult to at each other like a tug-of-war in one’s mind. This percep-
see than those in the far corners. Sight down long corri- tion is the fine line that separates complete resection from
dors cannot make right turns, which transforms ledges of recurrence and good outcomes from new deficits.
eloquent tissue in the near field into blind spots. The right- The capsule is key to safe and complete lesionectomy.
angle method helps anticipate the blind spots, which drives A small opening into the capsule is necessary to drain
meticulous inspection for residual BSCM during surgery the liquefied hematoma and reduce the lesion’s size. Cav-
(Fig. 7). erns are opened and interconnected internally to explore
the different portions of the lesion. Capsular preservation
keeps the lesion whole and protects the brainstem from the
The Fine Line neurosurgeon’s maneuvers. Gradually, the capsule pulls
After blind spots, misinterpretation was the next most away from the parenchyma. The dissection then switches
common cause of residual or recurrent lesions. Every re- from intracapsular to extracapsular, freeing adhesions cir-
section requires the painstaking evaluation of the plane that cumferentially and collapsing the capsule inward to find
separates the cavernous malformation from the brainstem remaining points of adhesion. Ideally, circumferential dis-
itself. In other words, overcoming blind spots and obtain- section liberates the BSCM with its capsule intact, except
ing the proper views brings the greater challenge of dis- for the initial entry site. Sharp dissection is performed
secting at the border between the lesion and parenchyma, with round knives of varying sizes and angles and rarely
between morbidity and cure. Direct visualization and cau- with microscissors. Delicate traction with a grasping in-
680 J Neurosurg Volume 135 • September 2021
 
*30 *
 Garcia et al.

FIG. 7. Illustration of the right-angle method. The surgical axis is defined

by 2 points: one at the superficial entrance into the operative corridor
(corridor entry, point 1) and one at the entry into the BSCM (front wall,
point 2). The surgical axis is drawn by connecting these points and
extending this line through the lesion to its back wall (point 3). The right-
angle method draws a perpendicular line from the surgical axis to the
point on the lesion’s capsule farthest from the axis. Any perpendicular
line exceeding 1 cm in length is identified as a potential blind spot. This
illustration shows two pontine cavernous malformations, each exposed
through an extended retrosigmoid/trans–middle cerebellar peduncle
approach. The long axis of lesion A aligns with the surgical axis, and
the right-angle method identifies no blind spot. In contrast, the long axis
of lesion B does not align with the surgical axis, and the right-angle
method identifies a posterolateral blind spot hiding a remnant (red) of
the cavernous malformation. Copyright Barrow Neurological Institute,
Phoenix, Arizona. Published with permission.

this operation but requires a steeled mindset to press on-

ward when neurophysiological signals change, represent-
FIG. 6. Illustration of the anatomy of the blind spot. The inherent depth ing another fine line to be managed.
of a BSCM and the length of the corridor required to reach it make blind The fine lines between complete resection and recur-
spots inevitable. The surgical corridor comprises 1) the craniotomy; 2) a rence, between safe and harmful resection, and between
subarachnoid, cisternal, or ventricular segment; 3) a parenchymal seg- heeding and ignoring neuromonitoring make this opera-
ment; and 4) an intralesional segment. With a BSCM, the small keyhole tion one of the most challenging in neurosurgery. It lacks
opening into the lesion is displaced to the end of a very long corridor the drama of intraoperative rupture or the elegance of
that limits the viewing angles, decreases visualization, and creates
blind spots in which a remnant can hide. Copyright Barrow Neurological
deep bypasses, but it demands circumspection at every
Institute, Phoenix, Arizona. Published with permission. step, from the approach selection to the exposure to each
maneuver during the resection. It demands constant cal-
culation regarding how hard to pull, how deep to probe,
and how thorough to search. While these other challeng-
strument and countertraction with the suction finally de- ing operations are in subarachnoid spaces or deep cisterns,
liver the lesion. the resection of a cavernous malformation is within the
Neurophysiological monitoring identifies areas of high- brainstem, which raises the stakes to the highest level.
est eloquence, like the facial colliculus in the fourth ven- Even extensive neurosurgeon experience does not seem to
tricular floor or the pyramidal tract in the cerebral pedun- eliminate the problem of incomplete resection. Residual
cle of the midbrain. Neuromonitoring can also warn the lesions were observed less frequently in the second half
neurosurgeon when the dissection is too rough or strays of this experience, which reflects the maturation of tech-
into monitored tracts and nerves. Although these alarms nique, comfort working within the brainstem, awareness
help redirect or modulate the dissection, they can also of blind spots, and a feel for separation planes. However,
derail the neurosurgeon. A sudden signal change elicits recurrences were still observed late in this surgical experi-
worry and hesitation, often at the end of a resection, when ence of more than 800 cerebral cavernous malformations.
the borders of the lesion have been reached. Too much cau- One of the greatest benefits of experience is that it lessens
tion at the finish line can easily lead to leaving a BSCM the uncomfortable dissonance that comes from walking
remnant. Small diminutions of signals should not deter the these fine lines. The enduring problem of residual BSCM,
pursuit of curative resection because they often translate to despite extensive experience, highlights the need for a neu-
transient deficits postoperatively. The information must be rosurgeon to constantly modulate between an aggressive
processed without impeding progress to complete resec- approach, associated with undue operative morbidity, and
tion. Neuromonitoring has become a standard of care for a conservative approach, associated with high rates of re-
J Neurosurg Volume 135 • September 2021 681
 
* | * 04:30 *
Garcia et al.

currence. Recurrent BSCM occurs in 5%–10% of patients,   5. Garcia RM, Ivan ME, Lawton MT. Brainstem cavernous mal-
and this possibility ensures that we constantly work to- formations:​surgical results in 104 patients and a proposed
ward an acceptable balance of safety and completeness. grading system to predict neurological outcomes. Neurosur-
gery. 2015;​76(3):​265–278.
  6. Wang CC, Liu A, Zhang JT, et al. Surgical management of
Conclusions brain-stem cavernous malformations:​report of 137 cases.
Surg Neurol. 2003;​59(6):​444–454.
In this large series of patients with BSCM treated mi-   7. Pandey P, Westbroek EM, Gooderham PA, Steinberg GK.
crosurgically over more than 20 years, the recurrence rate Cavernous malformation of brainstem, thalamus, and basal
in patients requiring reoperation was 6.6% (14 of 213 pa- ganglia:​a series of 176 patients. Neurosurgery. 2013;​72(4):​
tients), with all but 2 presenting with acute clinical dete- 573–589.
rioration from a new hemorrhage. Detecting remnants is   8. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera-
difficult during the immediate postoperative period and tive remnants of brainstem cavernomas:​incidence, risk fac-
even 6 months after the operation, when the resection bed tors and management. Acta Neurochir (Wien). 2008;​150(9):​
has healed. The 5.9% annual hemorrhage risk of recurrent 879–887.
  9. Abla AA, Lekovic GP, Garrett M, et al. Cavernous malforma-
BSCM in this experience is consistent with that reported tions of the brainstem presenting in childhood:​surgical expe-
for unoperated BSCMs. Blind spots were the most com- rience in 40 patients. Neurosurgery. 2010;​67(6):​1589–1599.
mon cause of incomplete BSCM resection in this expe- 10. Bertalanffy H, Gilsbach JM, Eggert HR, Seeger W. Micro-
rience. The right-angle method helps to anticipate blind surgery of deep-seated cavernous angiomas:​report of 26
spots and meticulously inspect the resection cavity for re- cases. Acta Neurochir (Wien). 1991;​108(3-4):​91–99.
sidual BSCM during surgery. The percentage of patients 11. van Swieten JC, Koudstaal PJ, Visser MC, et al. Interobserver
with BSCM who experience recurrence is low (5%–10%), agreement for the assessment of handicap in stroke patients.
Stroke. 1988;​19(5):​604–607.
which ensures that we constantly work toward an accept- 12. Steinberg GK, Chang SD, Gewirtz RJ, Lopez JR. Microsur-
able balance of safety and completeness. gical resection of brainstem, thalamic, and basal ganglia an-
giographically occult vascular malformations. Neurosurgery.
Acknowledgments 2000;​46(2):​260–271.
13. Brown AP, Thompson BG, Spetzler RF. The two-point meth-
Dr. Roxanna Garcia served as the StrokeNet research fellow od:​evaluating brain stem lesions. BNI Q. 1996;​12(1):​20–24.
from 2018 to 2019 and is a Fogarty Global Health Trainee from
2019 to 2020, Department of Neurological Surgery, Northwestern
University, Feinberg School of Medicine. Research reported in Disclosures
this publication was supported under the StrokeNet award num-
ber U24 NS107233-01 and the Fogarty International Center and The authors report no conflict of interest concerning the materi-
National Institute of Mental Health, of the National Institutes als or methods used in this study or the findings specified in this
of Health, under award number D43 TW010543. The content is paper.
solely the responsibility of the authors and does not necessarily
represent the official views of the National Institutes of Health. Author Contributions
We thank the staff of Neuroscience Publications at Barrow Conception and design: Lawton, Garcia. Acquisition of data:
Neurological Institute for assistance with manuscript and illustra- Garcia, Oh, Cole, Hendricks. Analysis and interpretation of data:
tion preparation. Lawton, Garcia. Drafting the article: Lawton, Garcia. Critically
revising the article: Lawton, Garcia. Reviewed submitted version
of manuscript: all authors. Statistical analysis: Garcia. Admin-
References istrative/technical/material support: Lawton. Study supervision:
  1. Gross BA, Batjer HH, Awad IA, Bendok BR. Brainstem Lawton.
cavernous malformations. Neurosurgery. 2009;​64(5):​E805–
E818. Supplemental Information
  2. Gross BA, Batjer HH, Awad IA, et al. Brainstem cavernous Current Affiliations
malformations:​1390 surgical cases from the literature. World
Dr. Garcia: Department of Neurological Surgery, Northwestern
Neurosurg. 2013;​80(1-2):​89–93.
University, Chicago, IL.
  3. Gross BA, Du R. Hemorrhage from cerebral cavernous mal-
formations:​a systematic pooled analysis. J Neurosurg. 2017;​
126(4):​1079–1087. Correspondence
  4. Abla AA, Lekovic GP, Turner JD, et al. Advances in the Michael T. Lawton: c/o Neuroscience Publications, Barrow
treatment and outcome of brainstem cavernous malformation Neurological Institute, St. Joseph’s Hospital and Medical Center,
surgery:​a single-center case series of 300 surgically treated Phoenix, AZ. [email protected].
patients. Neurosurgery. 2011;​68(2):​403–415.

682 J Neurosurg Volume 135 • September 2021

 
* | * 04:30 *
 CLINICAL ARTICLE
J Neurosurg 135:683–692, 2021

Optimal access route for pontine cavernous malformation

resection with preservation of abducens and facial nerve
function
Helmut Bertalanffy, MD,1 Shinya Ichimura, MD,1,2 Souvik Kar, PhD,1 Yoshihito Tsuji, MD,1,3 and
Caiquan Huang, MD1
International Neuroscience Institute, Hannover, Germany; 2Department of Neurosurgery, Kawasaki Chuo Clinic, Kawasaki; and
1

Department of Neurosurgery, Matsubara Tokushukai Hospital, Matsubara, Japan

3

OBJECTIVE  The aim of this study was to analyze the differences between posterolateral and posteromedial approach-
es to pontine cavernous malformations (PCMs) in order to verify the hypothesis that a posterolateral approach is more
favorable with regard to preservation of abducens and facial nerve function.
METHODS  The authors conducted a retrospective analysis of 135 consecutive patients who underwent microsurgical
resection of a PCM. The vascular lesions were first classified in a blinded fashion into 4 categories according to the
possible or only reasonable surgical access route. In a second step, the lesions were assessed according to which ap-
proach was performed and different patient groups and subgroups were determined. In a third step, the modified Rankin
Scale score and the rates of permanent postoperative abducens and facial nerve palsies were assessed.
RESULTS  The largest group in this series comprised 77 patients. Their pontine lesion was eligible for resection from ei-
ther a posterolateral or posteromedial approach, in contrast to the remaining 3 patient groups in which the lesion location
already had dictated a specific surgical approach. Fifty-four of these 77 individuals underwent surgery via a posterolater-
al approach and 23 via a posteromedial approach. When comparing these 2 patient subgroups, there was a statistically
significant difference between postoperative rates of permanent abducens (3.7% vs 21.7%) and facial (1.9% vs 21.7%)
nerve palsies. In the entire patient population, the abducens and facial nerve deficit rates were 5.9% and 5.2%, respec-
tively, and the modified Rankin Scale score significantly decreased from 1.6 ± 1.1 preoperatively to 1.0 ± 1.1 at follow-up.
CONCLUSIONS  The authors’ results suggest favoring a posterolateral over a posteromedial access route to PCMs in
patients in whom a lesion is encountered that can be removed via either surgical approach. In the present series, the
authors have found such a constellation in 57% of all patients. This retrospective analysis confirms their hypothesis in a
large patient cohort. Additionally, the authors demonstrated that 4 types of PCMs can be distinguished by preoperatively
evaluating whether only one reasonable or two alternative surgical approaches are available to access a specific lesion.
The rates of postoperative sixth and seventh nerve palsies in this series are substantially lower than those in the majority
of other published reports.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201023
KEYWORDS  cavernous malformation; pons; abducens nerve; facial nerve; cranial nerve function; surgical approach;
vascular disorders

C
avernous malformations of the brain are rare le- medullary lesions.7 Due to the risk of symptomatic bleed-
sions that account for 8%–15% of all intracranial ing, sometimes with devastating consequences, PCMs are
vascular malformations.1–3 Fewer than 35% of ce- a potential threat to the patient. The main goal of surgery
rebral cavernous malformations involve the brainstem, is to eliminate the risk of bleeding. Moreover, surgery
and these are more frequently located in the pons.2,4–6 Con- is also aimed at preserving cranial nerve (CN) function,
sidering the complex and unique segmental organization particularly that of CN VI (abducens nerve) and CN VII
of the brainstem, it is preferable to study pontine cavern- (facial nerve).
ous malformations (PCMs) separately from midbrain and As reflected in the pertinent literature, neurosurgeons

ABBREVIATIONS  CL = type C lesion approached posterolaterally; CM = type C lesion approached posteromedially; CN = cranial nerve; CPA = cerebellopontine angle;
EMG = electromyography; mRS = modified Rankin Scale; PCM = pontine cavernous malformation.
SUBMITTED  April 3, 2020.  ACCEPTED  July 8, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS201023.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  683

* | * 04:31 *
Bertalanffy et al.

have utilized various surgical approaches to expose and sible or only reasonable surgical approach to access each
access brainstem cavernous malformation.4,5,8–16 The specific lesion. To obtain an ideally unbiased estimation,
following standard or skull base approaches have been cutouts of preoperative axial, coronal, and sagittal MRI
commonly used to remove PCMs: retrosigmoid, lateral slices focused on the brainstem that contained neither the
supracerebellar infratentorial, median suboccipital (with patient’s name nor the date of examination were analyzed
or without telovelar exposure), far-lateral, subtemporal, in a blinded fashion. Thus, the examiners remained un-
anterior petrosal, and orbitozygomatic approaches. Ex- aware about each patient’s identity, the year of surgery,
cept for the last 3, which are anterolateral approaches, and which access route was taken in surgery.
the remaining can broadly be divided into posterolateral To assess the suitability of a certain surgical approach,
and posteromedial access routes. Comprehensibly, and we based our evaluation on the following criteria: lesion
as found in many previous publications, neurosurgeons size, shape, exact intrinsic location, and extent; relation-
tended to expose PCMs via the posterior midline since ship between lesion and brainstem surface; lesion axis in
this standard approach offers direct access to the rhom- oval-shaped PCMs; expected accessibility of “hidden cav-
boid fossa.5,10,11,15,17,18 Less frequently, PCMs have also ernous malformation portions”; the main composition of
been exposed by a limited incision of the lateral or an- the PCM (e.g., solid cavernous malformation vs predomi-
terolateral pontine surface at the cerebellopontine angle nantly fresh hematoma); expected distortion or displace-
(CPA).5,12,​19–21 ment of intrinsic fibers and nuclei; expected entry point
The present study compares the results obtained with into the pons; expected vascular anatomy on the surface
the posterolateral access route with those of the tradition- of the cerebellum and brainstem; and shape of the skull
al, more frequently used posteromedial transrhomboid base. Accordingly, we distinguished the following 4 lesion
fossa approach used for PCM resection. Our main purpose categories.
was to verify the hypothesis that posterolateral access to
PCMs is more likely to ensure morphological and func- Type A Lesion
tional preservation of CNs VI and VII. Only a posterolateral route seemed reasonable and
suitable to access PCM type A lesions because the lesion
Methods had abutted the lateral surface of the pons and was locat-
ed rather remotely from the rhomboid fossa (Fig. 1). Ac-
We retrospectively analyzed clinical and operative re- cess from a medial approach would have been associated
cords, radiological findings, and follow-up data of 135 pa- with significant and unjustifiable brainstem parenchymal
tients with PCMs. The senior author (H.B.) has surgically damage.
treated all individuals during the past 24 years, mainly at
three different institutions. Type B Lesion
All patients underwent preoperative T1- and T2- Only a posteromedial route seemed reasonable and
weighted MRI in addition to other routine MR sequences. suitable to access a type B PCM because the lesion was
The maximum diameter of the lesions was measured on located rather far from the lateral pontine surface, but suf-
T1-weighted images either in the axial or in the sagittal ficiently close to, or even bulging through, the rhomboid
plane. In cases in which a cavernous malformation extend- fossa into the fourth ventricle (Fig. 1). Access from a lat-
ed from the pons into the midbrain or medulla, the lesion eral approach would have been associated with significant
was defined as PCM when its sagittal width in the pons and unjustifiable brainstem parenchymal damage.
exceeded its sagittal width in the midbrain or medulla.
A surgical indication was established in patients who Type C Lesion
had experienced a clinically pertinent singular or repeated Either a posterolateral or a posteromedial approach was
hemorrhage, or in individuals harboring cavernous mal- regarded as possible and both are equally valid options to
formations with mass effect causing significant neurologi- be chosen as an access route to type C PCM lesions. Since
cal deficits, regardless of whether the lesion had reached the lesion was located between the lateral pontine surface
the pontine surface. and rhomboid fossa, it may well have been accessed from
Bilateral motor, somatosensory, and brainstem auditory either side, and expectedly each approach may have yield-
evoked potentials as well as unilateral or bilateral electro- ed a comparable surgical outcome (Figs. 1 and 2); neither
myography (EMG) of CNs VI and VII were continuously approach appeared to be clearly superior over the other.
monitored in all patients during surgery, although EMG
was not performed in the few patients who already had Type D Lesion
lost one or both of these CN functions. Within 48 hours Only an anterolateral pontine exposure appeared rea-
after surgery, each patient underwent a control MRI ex- sonable for type D PCMs because the major part of the
amination to verify the brainstem morphology and the PCM or the entire lesion lay anterior to the pontine coro-
completeness of the PCM resection. Analysis of surgical nal midline (Fig. 1).
approaches in this patient series was carried out in 3 con- The posterolateral approaches to PCMs in this series
secutive steps. comprised the lateral supracerebellar infratentorial (Fig.
3), the retrosigmoid, the far-lateral transcondylar (Fig. 4),
Step 1 and the subtemporal transtentorial approaches. The pos-
In the first step, the PCMs were retrospectively classi- teromedial approach to PCMs was a median suboccipital
fied into 4 categories (types A–D) according to the pos- craniotomy with telovelar exposure of the rhomboid fossa

684 J Neurosurg Volume 135 • September 2021

 
* | * 04:31 *
 Bertalanffy et al.

FIG. 1. The 4 different lesion types (types A–D) defined according to the most suitable or possible surgical approach to access the
PCM. Each column shows axial (upper) and sagittal (lower) MR images of one patient harboring a representative PCM type. For
a detailed explanation, see Methods.

(Fig. 5). The anterolateral approaches used to expose type A permanent postoperative CN VI or VII deficit was
D lesions included a retrosigmoid craniectomy and in only considered either in patients with preoperatively intact CN
one instance the anterior transpetrosal access route. function who had abducens or facial nerve palsy of any
degree by the time of the latest postoperative follow-up, or
Step 2 in individuals in whom these deficits were already present
In a second step, this time nonblinded, only patients before surgery but were found to have clearly worsened by
harboring a type C lesion were dichotomized on their part the time of the latest follow-up examination.
into two subgroups according to which approach has in
fact been utilized to remove their PCM microsurgically. Statistical Analysis
Consequently, the patient subgroup CL included all pa- Data analysis was performed using IBM SPSS (version
tients in group C who harbored a PCM that was actually 22.0, IBM Corp.) on a personal computer. We used the
exposed by a posterolateral approach (namely, a CL-sub- Wilcoxon signed-rank test to compare pre- and postop-
type lesion); likewise, the patient subgroup CM comprised erative mRS scores and Fisher’s exact probability test to
all group C individuals whose PCMs (CM-subtype le- compare CN function between patients; p < 0.05 was con-
sions) were accessed by the posteromedial approach. sidered statistically significant.
Hereinafter, type A–D lesions refer to specific PCMs as The local research ethics committee of each institution
defined above and as shown in Fig. 1, while groups A–D approved the use of anonymized patient data for this ret-
refer to the 4 patient groups that harbored one of these 4 rospective study.
lesion types.

Step 3 Results
The third step aimed at assessing the overall outcome; Table 1 shows the main clinical characteristics, lesion
for this reason, we determined the modified Rankin Scale size, composition of each patient group and subgroup, and
(mRS) score in each patient group (A–D) and subgroup clinical outcome expressed as the mRS score. There were
(CL and CM) before surgery and by the time of the latest 50 males and 85 females who ranged in age from 3 to 69
postoperative follow-up examination to assess the overall years (mean age [± SD] 35.4 ± 13.9 years). Each patient
surgical outcome. Additionally, we documented the ab- had experienced at least one clinically significant bleed
ducens and facial nerve function in each individual and from the underlying PCM by the time of admission; the
compared the permanent CN deficit rates between patients number of hemorrhagic episodes varied between 1 and 4
who underwent surgery via a posterolateral approach (mean 1.6 ± 0.7). The lesion size ranged from 5 to 30 mm
(group A + subgroup CL) and those who were operated on (mean 15.2 ± 6.0 mm). The largest group in this series was
via a posteromedial approach (group B + subgroup CM). group C (n = 77), with 54 patients belonging to subgroup
Separately, functioning of the two cranial nerves was also CL and 23 belonging to subgroup CM.
assessed in the CL and CM subgroups. A posterolateral surgical approach was performed in 67

J Neurosurg Volume 135 • September 2021 685

 
* | * 04:31 *
Bertalanffy et al.

FIG. 2. Artist’s illustration exemplifying a transverse section through the middle pons with an underlying PCM. The vascular lesion
does not abut the pontine surface, but it displaces, here unilaterally, the surrounding intrinsic structures, including the nuclei and
fibers of the abducens and facial nerves. According to anatomical studies, the facial nucleus (VII. nucleus) is located at an average
distance of 5.1 mm from the floor of the fourth ventricle, while the abducens nucleus (VI. nucleus) lies at a mean distance of only
0.2 mm from the surface of the rhomboid fossa. The green arrows demonstrate the 2 possible access routes, either medial or
lateral, to this left-sided lesion. p-laterally = posterolaterally; p-medially = posteromedially. Copyright Caiquan Huang. Published
with permission. Figure is available in color online only.

patients who belonged to either group A or subgroup CL proach (groups A + CL; 3.0%) and those who underwent
(A + CL), while a posteromedial approach was used in 53 surgery via the posterior midline (groups B + CM; 9.4%),
individuals who belonged to either group B or subgroup this difference did not reach statistical significance (Table
CM (B + CM). We opted for an anterolateral approach for 2). However, when comparing these rates only between the
a type D PCM exposure in only 15 patients. patient subgroups CL and CM, both abducens and facial
Postoperative MRI confirmed gross-total resection of palsy rates were significantly higher in patients who were
the intrinsic PCM in all patients except for 2 individuals operated on via the rhomboid fossa than in those who un-
in whom we intentionally left a small portion of the PCM derwent PCM removal from laterally (21.7% vs 3.7% and
behind to avoid excessive manipulation in the depth of the 21.7% vs 1.9%, respectively).
pons. Neither patient had additional neurological deficits Table 3 lists the mean mRS scores by the time of fol-
nor recurrent hemorrhage from the residual PCM until the low-up in patients who sustained a permanent postoper-
one in this study. ative CN deficit compared with the mean mRS scores in
The overall mean mRS score decreased from baseline individuals without such a deficit. No statistically signifi-
by the time of the last follow-up examination, and it im- cant difference was found in this context between the two
proved in the majority of individuals in patient groups and evidently unequal patient populations. PCM size did not
subgroups. There was a statistically significant difference influence the occurrence of postoperative CN VI or CN
between postoperative and preoperative mRS scores in all VII deficits.
patients and, with few exceptions, in the subpopulations.
The rates of permanent abducens and facial nerve pal-
sies after surgery in the entire patient series were 5.9% Discussion
and 5.2%, respectively. Despite an obvious difference be- Background and Aim of This Study
tween these rates in patients operated on by a lateral ap- Since the early 1990s when neurosurgeons began to op-

686 J Neurosurg Volume 135 • September 2021

 
* | * 04:31 *
 Bertalanffy et al.

FIG. 3. This 46-year-old female had sensory deficits involving the left side of her body, including her face. A and B: Preoperative
axial (A) and sagittal (B) T2-weighted images showing evidence of a large hemorrhagic type C (subtype CL) intraaxial PCM. Due
to the lesion’s compressive effect, the dorsal structures of the pons are severely compressed (red arrows), suggesting that there
is only a thin parenchymal layer between the lesion and rhomboid fossa surface. C: The patient underwent surgery in the left
lateral park-bench position via a lateral retrosigmoid craniotomy that also exposed the distal transverse sinus. D and E: Total PCM
removal is documented on postoperative contrast-enhanced axial (D) and sagittal (E) T1-weighted MR images. After decompres-
sion, the pontine parenchyma dorsal to the resection cavity has clearly unfolded and has regained a significant width (red arrows).
F and G: Intraoperatively, bulging and slight discoloration of the pontine surface (F) is seen between the trigeminal (CN V) and
cochlear (CN VIII) nerves; this area was chosen as the entry point into the pons (G, white arrow), and total PCM removal was
achieved without complication. The patient had an uneventful postoperative course, and there were no additional neurological
deficits. Figure is available in color online only.

erate within the brainstem with increasing frequency, they pothesis that a posterolateral entry into the pons is more
have sought so-called “safe entry zones” into the brain- favorable than a transrhomboid fossa approach in terms
stem11,13,16,22,23 and particularly into the pons. The main of preserving abducens and facial nerve function in PCM
purpose of defining such safe entry zones into the pons surgery. With the present study, we intended to verify this
was avoiding additional morphological damage to the hypothesis in a relevant patient population. Theoretically,
brainstem and preventing permanent neurological deficits a prospective randomized controlled study would appear
by surgical manipulation in a highly vulnerable brain pa- as the scientifically most suitable tool for such a com-
renchyma. The use of electrophysiological mapping of the parative analysis. Practically, however, attempting such a
rhomboid fossa9,11,24 and detailed anatomical studies of the prospective study appeared inadequate because of diffi-
floor of the fourth ventricle15,25,26 added important knowl- cult valid randomization, numerous anatomical variations
edge about brainstem morphology and function. This im- and heterogeneous morphological appearances of PCMs,
proved understanding contributed significantly to facili- a relatively low number of cases available per year, and,
tating surgery within the pons in addition to augmenting consequently, an expected overlong duration of such a
the safety of the procedure. kind of study. During our blinded evaluation process, we
Already 3 decades ago, some authors observed that the first focused on the possibility of accessing the PCM ei-
pons may tolerate a small lateral incision quite well, and ther posterolaterally or posteromedially. In cases in which
evacuation of an intrinsic pontine hematoma or vascular we came to the conclusion that both approaches were pos-
malformation can be carried out via a lateral exposure of sible and reasonable alternatives, the divergent advantages
the pons without producing permanent neurological defi- and drawbacks of these two access routes seemed to bal-
cits.19,21 Considering these early observations and also our ance each other. Despite minor differences, we could not
own experience in brainstem surgery gained over more identify features that would have rendered one approach
than 2 decades, we postulated a few years ago the hy- clearly superior to the other. Surprisingly, more than half

J Neurosurg Volume 135 • September 2021 687

 
* | * 04:31 *
Bertalanffy et al.

FIG. 4. This 45-year-old female harbored a voluminous hemorrhagic type C PCM (subtype CL) occupying the central pontine area.
The patient presented preoperatively in a bedridden condition and with an altered level of consciousness. Neurologically, left-sided
hemiparesis and hemihypesthesia, as well as right-sided facial paralysis, sixth nerve palsy, and severe dysphagia, were noted.
A and B: Preoperative axial (A) and sagittal (B) T2-weighted MR images. The red arrows indicate the surface of the rhomboid
fossa, which seems extremely thinned out by the underlying compressive vascular lesion. C: The patient underwent surgery in the
left lateral park-bench position. The procedure was carried out via a left-sided far-lateral transcondylar approach. D and E: Post-
operative axial (D) and sagittal (E) T2-weighted images showing total PCM removal. The red arrows indicate a rather significant
parenchymal width of the dorsal pons. This area contains the nuclei of CN VI and VII among other structures that may have been
seriously injured had a transrhomboid fossa exposure been chosen. F: Exposure of the lateral pons revealed the slightly hemor-
rhagic facial nerve at the level where it exits the brainstem (white arrow), visible above the rootlets of CNs IX–XI. G: A lesionecto-
my was performed by entering the pons through the pontomedullary sulcus. The patient’s postoperative course was uneventful,
and there were no complications. One year after surgery, the patient has fully recovered, and only minimal residual right facial
weakness and mild hemihypesthesia were noted. Figure is available in color online only.

of our entire patient population, namely, 77 of 135 indi- a posteromedial surgical approach to PCMs and postop-
viduals (57%), belonged to this patient group C. erative CN VI and VII deficits. One might argue that a
Following our first blinded evaluation, we then assessed certain distinction can already empirically be expected
the number of group C patients who in fact underwent sur- regarding the impact of the surgical approach on the final
gery from a posteromedial approach (subgroup CM) and, outcome. Indeed, due to anatomical reasons (as shown in
likewise from a posterolateral approach (subgroup CL). Fig. 2), the surgical removal of a PCM through the rhom-
boid fossa may appear more likely to cause CN VI and/or
Interpretation of Results VII damage than resection from a posterolateral approach.
Yet, these nerve fibers and even their nuclei might still
Sex distribution (63% of patients were female), mean be injured when approaching the lesion laterally, either
age, preoperative bleeding episodes, and lesion size in this deeply within the pons or in the CPA. In clinical practice,
patient series compare favorably to those in other stud- surgeons must decide to choose either the posterolateral or
ies.4,27,28 the posteromedial approach when both access routes ap-
For the first time, we have demonstrated that the 4 types pear as equally valid options in the same patient. In such
of PCMs can be distinguished according to the most suit- a situation, our results suggest favoring the posterolateral
able or possible surgical access route. Such an allocation is approach as being the more favorable one pertaining to the
an important prerequisite for selecting the optimal surgi- preservation of abducens and facial nerve function.
cal route to access PCMs. Another interesting result of this study is the relative-
The most important result of this study, however, is the ly low overall rates of postoperative abducens and facial
obvious difference between the rates of permanent postop- nerve palsies (5.9% and 5.2%, respectively). This may be
erative CN VI and VII deficits in the CL and CM patient attributed at least in part to the aforementioned bias in se-
groups, and the statistically significant correlation between lecting more frequently a lateral access route to the PCM.
688 J Neurosurg Volume 135 • September 2021
 
* | * 04:31 *
 Bertalanffy et al.

FIG. 5. A and B: Preoperative axial (A) and sagittal (B) T2-weighted images obtained in a 16-year-old male who presented with
a left-sided sensorimotor syndrome and internuclear ophthalmoplegia. Despite the lesion’s considerable size and its vicinity to
the facial nucleus, the patient had not developed facial palsy. C and D: The patient underwent microsurgical removal of this type
B PCM in a semisitting position (C) via a longitudinal straight-line skin incision for a midline suboccipital craniotomy (D). E and F:
Postoperative axial (E) and sagittal contrast-enhanced (F) T1-weighted MR images demonstrating total removal of the vascular le-
sion. Surgical access to this PCM was obtained by the dorsal midline approach. G: Intraoperative inspection of the rhomboid fossa
after telovelar exposure. This hemorrhagic PCM had already broken through the floor of the fourth ventricle over a distance of 5–6
mm. Using the electrophysiological stimulation probe (white arrow) the location and extent of the facial colliculus were determined
bilaterally. H: Total PCM removal is documented on this intraoperative photograph that shows the empty resection cavity contain-
ing a millimeter scale and more superiorly the sylvian aqueduct (white arrow). Electrical stimulation of the facial colliculi (blue
arrows) yielded a satisfactory response by the end of the intraaxial procedure. Accordingly, there was no postoperative facial palsy
or, fortunately, additional deficits. This young patient rapidly recovered after an uneventful postoperative course. Figure is available
in color online only.

TABLE 1. Clinical characteristics and comparisons of pre- and postoperative mRS scores in patient groups and subgroups
No. of M/F Mean Age Mean No. of Preop Mean Lesion Mean Preop Mean Postop
Patients Sex (yrs) Bleeding Episodes Size (mm) mRS Score mRS Score p Value
Total 135 50/85 35.4 ± 13.9 1.6 ± 0.7 15.2 ± 6.0 1.6 ± 1.1 1.0 ± 1.1 0.0001*
Group A 13 6/7 36.9 ± 8.9 2.0 ± 0.7 13.9 ± 5.0 1.0 ± 0.7 0.6 ± 0.7 0.1
Group B 30 13/17 36.4 ± 15.2 1.7 ± 0.8 11.7 ± 4.6 1.9 ± 1.2 1.3 ± 1.3 0.006*
Group C 77 25/52 35.8 ± 14.1 1.5 ± 0.7 16.7 ± 6.3 1.7 ± 1.1 1.0 ± 1.0 0.0005*
  Subgroup CL 54 14/40 34.8 ± 13.7 1.5 ± 0.7 17.2 ± 6.4 1.5 ± 1.0 0.8 ± 0.8 0.002*
  Subgroup CM 23 11/12 38.2 ± 15.1 1.7 ± 0.5 15.5 ± 6.0 2.2 ± 1.0 1.5 ± 1.4 0.08
Group D 15 6/9 30.0 ± 13.8 1.1 ± 0.6 15.8 ± 5.7 1.1 ± 1.1 0.5 ± 0.5 0.24
Mean values are presented as the mean ± SD. Refer to Methods for a detailed explanation of the groups and subgroups.
* Statistically significant improvement between preoperative and postoperative mRS scores.

J Neurosurg Volume 135 • September 2021 689

 
*31 *
Bertalanffy et al.

TABLE 2. Distribution of patients with permanent CN VI and VII TABLE 3. Long-term outcomes and lesion sizes in patients with
deficits and those without CN VI and VII deficits at follow-up
No. of Patients CND VI p Value CND VII p Value CN Status at No. of mRS Score p PCM Size p
Follow-Up Patients at Follow-Up Value (mm) Value
Total 135 8 (5.9) 7 (5.2)
A + CL 67 2 (3.0) 1 (1.5) CN VI deficit 8 1.4 ± 0.3 15.1 ± 4.1
0.13 0.02* 0.1 0.9
B + CM 53 5 (9.4) 6 (11.3) CN VI intact 127 1.0 ± 1.1 15.0 ± 6.1
CL 54 2 (3.7) 1 (1.9) CN VII deficit 7 1.5 ± 1.1 15.3 ± 4.1
0.02† 0.008† 0.04* 0.88
CM 23 5 (21.7) 5 (21.7) CN VII intact 128 1.0 ± 1.1 15.0 ± 6.1
D 15 1 (6.7) 0 (0.0) Mean values are presented as the mean ± SD.
* Statistically significant.
CND = CN deficit.
Values represent the number of patients (%) unless stated otherwise.
* Significantly higher deterioration rate compared with the A + CL patient
subgroup.
† Significantly higher deterioration rate compared with the CL patient sub-
fashion in quite a number of individuals harboring deep-
group.
seated lesions and also in many cases in which the PCM
did not reach the lateral surface of the pons, as exempli-
fied in Fig. 3. To improve the viewing trajectory, we have
modified this posterolateral route in many other instances
In many previous publications, the rates of CN VI and VII into a far-lateral paracondylar or transcondylar exposure
palsy were rather close to 20%, ranging between 11.5% by extending the craniectomy inferiorly. This allowed for
and 24.1%.4,14,29 We found only one report in which these entering the pons from a more favorable lateral-to-medial
rates were only 4.3% and thus lower than those in the pres- and caudal-to-cranial direction,8,32 as demonstrated in the
ent study.30 representative case shown in Fig. 4.
Contrary to the conjecture that lesion size may influ-
ence the postoperative result in the treatment of brainstem Posterior Midline Approach
cavernous malformations,4,7 we found no significant corre- The posteromedial approach is the traditional—and
lation between PCM size and the occurrence of postoper- still widely utilized—method of accessing the cerebel-
ative CN VI and VII deficits. This can be attributed to the lum, the fourth ventricle, and the posterior aspect of the
fact that even large, hemorrhagic PCMs usually expand brainstem. Exposure of PCMs in this series via the floor
smoothly and thus may only distort or displace the intrin- of the fourth ventricle was usually a telovelar exposure as
sic pontine structures, including the CN nuclei and fibers, exemplified in Fig. 5. When using this tailored transrhom-
but do not injure them irreparably in the majority of cases. boid fossa access to the PCM, we relied on knowledge of
Apart from lesion size, we have verified other factors, the topographical and intrinsic anatomy of the pons and
such as dorsal bilateral extension of the PCM or bulging rhomboid fossa.33 In particular, we took into account the
through the rhomboid fossa, observed in the CN deficit lesion’s vicinity to the abducens and facial nerve fibers
groups; however, none of these factors clearly correlated and their nuclei, among other intrinsic pontine structures
with complication risk. such as the posterior longitudinal fascicle and lateral, spi-
The highly significant decrease in the mean mRS score nal, and trigeminal lemnisci. It has long been known that
after surgery (1.0 ± 1.1) compared with the preopera- the exact location and extent of the facial colliculus can-
tive value (1.6 ± 1.1) clearly reflects the favorable overall not precisely be determined by inspection of the rhomboid
outcome and the efficiency of surgery in this quite large fossa alone.18,26 As illustrated in Fig. 2, the facial collicu-
patient series. Understandably, CN deficits seem to have lus consists of the facial nerve fibers (the facial loop) that
deteriorated at the final overall outcome, which can be rec- cross over the underlying sixth nerve nucleus.25,26 In a pre-
ognized as a higher postoperative mRS score, as shown in vious study, we showed that the location and extent of the
Table 3. It can be speculated, however, that the observed facial colliculus can vary widely from patient to patient
lack of statistical significance may be attributed to the and even in one individual between right and left side.31
small number of patients in the 2 deficit groups. Even though EMG of CNs VI and VII was most important
for identifying the facial colliculus in the rhomboid fossa,
Posterolateral Access Route the value of this tool in preventing permanent CN deficit
Microsurgical removal of PCMs from a posterolateral in quite large lesions (such as in the case shown in Fig.
approach has been practiced with increasing frequency 5) was rather limited in the present series. In 6 patients
during the past 3 decades. Access to the lesion from a dor- with posterolateral access to the PCM who had postoper-
solateral approach is usually obtained via a standard ret- ative permanent facial palsy, we lost the EMG response in
rosigmoid craniectomy and exposure through the CPA by 4 instances, while it remained intact in 2 individuals. We
using a viewing trajectory toward the exit of the trigeminal speculate that in these 2 cases we were able to stimulate
root. Many PCMs can thus be reached by the transbra- the peripheral portion of the facial loop in the rhomboid
chium pontis approach. Several authors have described fossa but did not detect an injury in its proximal portion.
their method and have mentioned that less eloquent paren- However, in smaller PCMs or if the lesion was not in the
chyma is encountered with this technique.2,4,​5,​10,​12,​17,​19,​21,31 direct vicinity of the facial colliculus, CN monitoring was
We have utilized this posterolateral approach in a tailored a highly valuable and indispensable tool.
690 J Neurosurg Volume 135 • September 2021
 
*31 *
 Bertalanffy et al.

Selection of the Optimal Access Route is more likely to ensure the morphological and functional
In clinical practice, we recommend the following policy. preservation of CNs VI and VII in a lesion that may be
First, determine the PCM type. If a type A, B, or D lesion exposed equally well and with otherwise comparable out-
is encountered, decision-making on the optimal surgical comes from either posterolateral or posteromedial access.
approach is straightforward. However, the situation is dif- In the present series, we have found such a constellation in
ferent in the presence of a type C lesion, particularly when 57% of patients. The rates of permanent sixth and seventh
the PCM does not abut the lateral pontine surface, as ex- nerve palsies after surgery in this series are substantially
emplified in Fig. 4. Many neurosurgeons may tend to favor lower than those in the majority of other published reports,
the transrhomboid fossa approach in such cases because which can be attributed to our former empirical preference
the lesion appears closer to the floor of the fourth ventricle of the posterolateral access route.
than to the lateral pontine surface. Choosing the shortest
distance to the lesion was also propagated many years ago References
by the so-called “two-point method,”34 which may perhaps   1. Del Curling O Jr, Kelly DL Jr, Elster AD, Craven TE. An
be helpful in some mesencephalic cavernous malforma- analysis of the natural history of cavernous angiomas. J Neu-
tions. In a type C PCM, however, this method may rather rosurg. 1991;​75(5):​702–708.
suggest using a transrhomboid fossa approach. We empha-   2. Hebb MO, Spetzler RF. Lateral transpeduncular approach to
size that the thickness of the dorsal pontine parenchymal intrinsic lesions of the rostral pons. Neurosurgery. 2010;​66(3)
layer should not be underestimated when it is severely (Suppl Operative):​26–29.
compressed by a voluminous intrinsic hemorrhagic le-   3. Robinson JR, Awad IA, Little JR. Natural history of the cav-
ernous angioma. J Neurosurg. 1991;​75(5):​709–714.
sion. The postoperative MR images in Figs. 3 and 4 clearly   4. Abla AA, Lekovic GP, Turner JD, et al. Advances in the
show that this dorsal pontine area, which contains impor- treatment and outcome of brainstem cavernous malformation
tant anatomical structures, still has a substantial width that surgery:​a single-center case series of 300 surgically treated
unfolds after lesionectomy from a posterolateral approach. patients. Neurosurgery. 2011;​68(2):​403–4145.
Evacuating the PCM by traversing this unexpectedly thick   5. Bertalanffy H, Benes L, Miyazawa T, et al. Cerebral caverno-
parenchymal layer may undoubtedly endanger the nuclei mas in the adult. Review of the literature and analysis of 72
and fibers of CNs VI and VII. We therefore do not advo- surgically treated patients. Neurosurg Rev. 2002;​25(1-2):​1–55.
  6. Kondziolka D, Lunsford LD, Kestle JR. The natural history
cate such a strategy in a type C lesion. Likewise, we do of cerebral cavernous malformations. J Neurosurg. 1995;​
not recommend using a laborious transpetrosal exposure 83(5):​820–824.
to access PCMs from a lateral approach as propagated by   7. Tsuji Y, Kar S, Bertalanffy H. Microsurgical management of
others27,35 because of the limited advantage of this skull midbrain cavernous malformations:​predictors of outcome
base approach in PCMs. and lesion classification in 72 patients. Oper Neurosurg (Ha-
gerstown). 2019;​17(6):​562–572.
Study Limitations   8. Bertalanffy H, Burkhardt J-K, Kockro RA, et al. Resection of
cavernous malformations of the brainstem. In:​Rigamonti D,
We recognize several limitations of this study. First, this ed. Cavernous Malformations of the Nervous System. Cam-
is a retrospective analysis. Patients were selected arbitrari- bridge University Press;​2011:​143–160.
ly, according to the senior author’s experience in brainstem   9. Bricolo A, Turazzi S. Surgery for gliomas and other mass
surgery and personal preference (biased selection). Only lesions of the brainstem. Adv Tech Stand Neurosurg. 1995;​22:​
one neurosurgeon has carried out all microsurgical pro- 261–341.
cedures. Therefore, the favorable surgical outcome of this 10. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​
long-term results of microsurgical resection in 52 patients.
study cannot be generalized. Certainly, the choice of surgi- Neurosurgery. 2005;​56(6):​1203–1214.
cal approaches in the patients in this series was not random- 11. Kyoshima K, Kobayashi S, Gibo H, Kuroyanagi T. A study of
ized but also biased, having been influenced by a certain safe entry zones via the floor of the fourth ventricle for brain-
empirical preference for the posterolateral access route. stem lesions. Report of three cases. J Neurosurg. 1993;​78(6):​
Despite our endeavor for impartiality, the classification of 987–993.
PCMs into 4 distinct types may also not entirely have been 12. Ohue S, Fukushima T, Kumon Y, et al. Surgical management
objective because selection of surgical approaches for PCM of brainstem cavernomas:​selection of approaches and micro-
surgical techniques. Neurosurg Rev. 2010;​33(3):​315–324.
removal ultimately remained subject to the senior author’s 13. Recalde RJ, Figueiredo EG, de Oliveira E. Microsurgical
estimation based on his longstanding experience. Neverthe- anatomy of the safe entry zones on the anterolateral brain-
less, we are convinced that these limitations did not cru- stem related to surgical approaches to cavernous malforma-
cially attenuate the inferences drawn from this study. tions. Neurosurgery. 2008;​62(3)(suppl 1):​9 –17.
14. Samii M, Eghbal R, Carvalho GA, Matthies C. Surgical man-
agement of brainstem cavernomas. J Neurosurg. 2001;​95(5):​
Conclusions 825–832.
We have demonstrated that 4 different types of PCMs 15. Strauss C, Lütjen-Drecoll E, Fahlbusch R. Pericollicular sur-
can be distinguished by evaluating preoperatively whether gical approaches to the rhomboid fossa. Part I. Anatomical
only one reasonable or two alternative surgical approaches basis. J Neurosurg. 1997;​87(6):​893–899.
are available to access a specific lesion. This categorization 16. Yagmurlu K, Rhoton AL Jr, Tanriover N, Bennett JA. Three-
dimensional microsurgical anatomy and the safe entry zones
may facilitate selecting the optimal access route for the of the brainstem. Neurosurgery. 2014;​10(suppl 4):​602–620.
microsurgical resection of PCMs. The present retrospec- 17. Abla AA, Turner JD, Mitha AP, et al. Surgical approaches to
tive analysis confirms our previous hypothesis with statis- brainstem cavernous malformations. Neurosurg Focus. 2010;​
tical methods, namely that posterolateral access to PCMs 29(3):​E8.

J Neurosurg Volume 135 • September 2021 691

 
*31 *
Bertalanffy et al.

18. Bogucki J, Gielecki J, Czernicki Z. The anatomical aspects of 31. Bertalanffy H, Tissira N, Krayenbühl N, et al. Inter- and
a surgical approach through the floor of the fourth ventricle. intrapatient variability of facial nerve response areas in the
Acta Neurochir (Wien). 1997;​139(11):​1014–1019. floor of the fourth ventricle. Neurosurgery. 2011;​68(1)(Suppl
19. Konovalov AN, Spallone A, Makhmudov UB, et al. Surgical Operative):​23–31.
management of hematomas of the brain stem. J Neurosurg. 32. Bertalanffy H, Bozinov O, Sürücü O, et al. Dorsolateral ap-
1990;​73(2):​181–186. proach to the craniocervical junction. In:​Cappabianca P,
20. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- Califano L, Iaconetta G, eds. Cranial, Craniofacial and Skull
formations of the brainstem:​experience with 100 patients. J Base Surgery. Springer-Verlag;​ 2010:​175–196.
Neurosurg. 1999;​90(1):​50–58. 33. Mussi AC, Rhoton AL Jr. Telovelar approach to the fourth
21. Symon L, Jackowski A, Bills D. Surgical treatment of ponto- ventricle:​microsurgical anatomy. J Neurosurg. 2000;​92(5):​
medullary cavernomas. Br J Neurosurg. 1991;​5(4):​339–347. 812–823.
22. Cavalcanti DD, Preul MC, Kalani MY, Spetzler RF. Micro- 34. Brown AP, Thompson BG, Spetzler RF. The two-point meth-
surgical anatomy of safe entry zones to the brainstem. J Neu- od:​evaluating brain stem lesions. BNI Q. 1996;​12:​20–24.
rosurg. 2016;​124(5):​1359–1376. 35. Gross BA, Dunn IF, Du R, Al-Mefty O. Petrosal approaches
23. Zenonos GA, Fernandes-Cabral D, Nunez M, et al. The to brainstem cavernous malformations. Neurosurg Focus.
epitrigeminal approach to the brainstem. J Neurosurg. 2018;​ 2012;​33(2):​E10.
128(5):​1512–1521.
24. Morota N, Deletis V, Epstein FJ, et al. Brain stem mapping:​
neurophysiological localization of motor nuclei on the floor Disclosures
of the fourth ventricle. Neurosurgery. 1995;​37(5):​922–930. The authors report no conflict of interest concerning the materi-
25. Lang J Jr, Ohmachi N, Lang J Sr. Anatomical landmarks of als or methods used in this study or the findings specified in this
the rhomboid fossa (floor of the 4th ventricle), its length and paper.
its width. Acta Neurochir (Wien). 1991;​113(1-2):​84–90.
26. Strauss C, Romstöck J, Nimsky C, Fahlbusch R. Intraopera- Author Contributions
tive identification of motor areas of the rhomboid fossa using
direct stimulation. J Neurosurg. 1993;​79(3):​393–399. Conception and design: Bertalanffy. Acquisition of data: all
27. Garcia RM, Ivan ME, Lawton MT. Brainstem cavernous mal- authors. Analysis and interpretation of data: all authors. Drafting
formations:​surgical results in 104 patients and a proposed the article: Bertalanffy, Ichimura, Kar. Critically revising the
grading system to predict neurological outcomes. Neurosur- article: all authors. Approved the final version of the manuscript
gery. 2015;​76(3):​265–278. on behalf of all authors: Bertalanffy. Statistical analysis:
28. Zaidi HA, Mooney MA, Levitt MR, et al. Impact of timing Ichimura, Bertalanffy, Tsuji. Study supervision: Bertalanffy.
of intervention among 397 consecutively treated brainstem
cavernous malformations. Neurosurgery. 2017;​81(4):​620–626. Correspondence
29. Huang AP, Chen JS, Yang CC, et al. Brain stem cavernous Helmut Bertalanffy: International Neuroscience Institute,
malformations. J Clin Neurosci. 2010;​17(1):​74–79. Hannover, Germany. [email protected].
30. Chen LH, Zhang HT, Chen L, et al. Minimally invasive
resection of brainstem cavernous malformations:​surgical
approaches and clinical experiences with 38 patients. Clin
Neurol Neurosurg. 2014;​116:​72–79.

692 J Neurosurg Volume 135 • September 2021

 
*31 *
 » This article has been updated from its originally published version
to correct the Fig. 6 legend. See the corresponding
erratum notice in this issue, p 977. «

CLINICAL ARTICLE
J Neurosurg 135:693–703, 2021

Holmes tremor: a delayed complication after resection of

brainstem cavernomas
*Marco Cenzato, MD,1 Davide Colistra, MD,1 Giorgia Iacopino, MD,2
Christian Raftopoulos, MD, PhD,3 Ulrich Sure, MD, PhD,4 Marcos Tatagiba, MD, PhD,5
Robert F. Spetzler, MD, PhD,6 Alexander N. Konovalov, MD, PhD,7 Andriy Smolanka, MD, PhD,8
Volodymir Smolanka, MD, PhD,8 Roberto Stefini, MD,9 Carlo Bortolotti, MD,10 Paolo Ferroli, MD,11
Giampietro Pinna, MD,12 Angelo Franzini, MD,11 Philipp Dammann, MD,4 Georgios Naros, MD,5
Davide Boeris, MD, PhD,1 Paolo Mantovani, MD,10 Domenico Lizio, PhD,13 Mariangela Piano, MD,14
and Enrica Fava, MD, PhD1,15
1
Neurosurgery, Great Metropolitan Hospital Niguarda, Milan; 2Neurosurgery, Department of Human Neurosciences, Sapienza
University, Rome, Italy; 3Department of Neurosurgery, Saint-Luc University Clinic, Catholic University of Louvain, Brussels,
Belgium; 4Department of Neurosurgery, University Hospital Essen, University of Duisburg-Essen; 5Department of Neurosurgery,
University Hospital Tübingen, Germany; 6Neurosurgery, Barrow Neurological Institute, Phoenix, Arizona; 7Neurosurgery,
Burdenko Neurosurgical Institute, Moscow, Russia; 8Neurosurgery, University Hospital Uzhhorod, Ukraine; 9Department of
Neurosurgery, Legnano Hospital, Milan; 10Department of Neurologic Surgery, Bellaria Hospital, Institute of Neurological Sciences
of Bologna; 11Department of Neurosurgery, Carlo Besta Neurological Institute, Milan; 12Neurosurgery Department, University
Hospital Verona; Departments of 13Medical Physics and 14Neuroradiology, Great Metropolitan Hospital Niguarda, Milan; and
15
Department of Medical Biotechnologies and Translational Medicine, University of Milan, Italy

OBJECTIVE  In this paper, the authors aimed to illustrate how Holmes tremor (HT) can occur as a delayed complication
after brainstem cavernoma resection despite strict adherence to the safe entry zones (SEZs).
METHODS  After operating on 2 patients with brainstem cavernoma at the Great Metropolitan Hospital Niguarda in
Milan and noticing a similar pathological pattern postoperatively, the authors asked 10 different neurosurgery centers
around the world to identify similar cases, and a total of 20 were gathered from among 1274 cases of brainstem caver-
nomas. They evaluated the tremor, cavernoma location, surgical approach, and SEZ for every case. For the 2 cases at
their center, they also performed electromyographic and accelerometric recordings of the tremor and evaluated the post-
operative tractographic representation of the neuronal pathways involved in the tremorigenesis. After gathering data on
all 1274 brainstem cavernomas, they performed a statistical analysis to determine if the location of the cavernoma is a
potential predicting factor for the onset of HT.
RESULTS  From the analysis of all 20 cases with HT, it emerged that this highly debilitating tremor can occur as a
delayed complication in patients whose postoperative clinical course has been excellent and in whom surgical access
has strictly adhered to the SEZs. Three of the patients were subsequently effectively treated with deep brain stimula-
tion (DBS), which resulted in complete or almost complete tremor regression. From the statistical analysis of all 1274
brainstem cavernomas, it was determined that a cavernoma location in the midbrain was significantly associated with the
onset of HT (p < 0.0005).
CONCLUSIONS  Despite strict adherence to SEZs, the use of intraoperative neurophysiological monitoring, and the
immediate success of a resective surgery, HT, a severe neurological disorder, can occur as a delayed complication after
resection of brainstem cavernomas. A cavernoma location in the midbrain is a significant predictive factor for the onset
of HT. Further anatomical and neurophysiological studies will be necessary to find clues to prevent this complication.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201352
KEYWORDS  Holmes tremor; cavernoma; brainstem; safe entry zone; vascular disorders

ABBREVIATIONS  DBS = deep brain stimulation; DTT = dentatothalamic tract; EMG = electromyographic; FSL = Functional MRI of the Brain Software Library; FTOZ =
frontotemporoorbitozygomatic; HT = Holmes tremor; Max = maximum intensity; NST = nigrostriatal tract; ROI = region of interest; ROT = rubroolivary tract; SEZ = safe entry
zone; VIM = ventral intermediate nucleus.
SUBMITTED  April 20, 2020.  ACCEPTED  July 14, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS201352.
* M.C. and D.C. contributed equally to this work.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  693

*32 *
Cenzato et al.

C avernomas account for 5%–10% of all vascular treatment with DBS). We also evaluated the postoperative
malformations,1–3 and brainstem cavernomas ac- tractographic representation of the neuronal pathways in-
count for about 20% of all intracranial cavernous volved in the tremorigenesis.
angiomas.4 Surgery should be considered in selected pa-
tients with brainstem cavernomas, particularly in cases Electromyography
of a second symptomatic bleeding or in cases of a single EMG analysis of the most involved limb was per-
debilitating bleed.5 formed in the resting, postural, and intention positions (in
Accurate knowledge of the main safe entry zones illustrative case 3, even after treatment with DBS). Surface
(SEZs) to the brainstem is fundamental to minimize adhesive electrodes were placed in pairs on the following
surgical morbidity. A rare complication of surgery for muscles: biceps brachii, triceps brachii, flexor communis
brainstem cavernoma is the Holmes tremor (HT). In the digitorum, and extensor radialis carpi. The frequency and
medical literature, this syndrome is described in four case average amplitude of the muscle jerks during the tremor
reports6–9 and cited in two series.10,11 HT is characterized were evaluated. The examination was performed at the
by a resting, postural, and intentional tremor that occurs same time (11:00 am) for all patients.15 Patients were asked
sometime after a damaging event. It is characterized by not to use drugs known to affect tremor, such as alcohol or
low-frequency muscle contractions (< 5 Hz)12 and is ex- caffeinated drinks, before the examination.16
plained in the majority of cases by the interruption of the
nigrostriatal, dentatothalamic, and/or rubroolivary contra- Accelerometry
lateral bundles.13 The most effective treatment for HT is
surgery, which entails either lesioning or deep brain stim- We recorded the tremor by using an iPhone acceler-
ulation (DBS) of the ventral intermediate nucleus (VIM) ometer (in illustrative case 3, before and after treatment
of the thalamus.14 with DBS), accessible through a free application (app)
called VibSensor 2.0.0 (Now Instruments and Software
Our current study focuses on a series of 20 patients out
Inc.). VibSensor was built to measure the frequency of
of 1274 cases that were operated on for brainstem cav-
seismographic activity on all three axes, one for each di-
ernoma in 11 different neurosurgery centers all over the
rection of space.17 When testing the 2 patients operated on
world in order to illustrate how HT can occur as a delayed
in Milan, we recorded the most involved (upper) limb in
complication after resective surgery despite strict adher-
all positions. The phone was placed on the dorsal surface
ence to the SEZs, the use of intraoperative neurophysi-
of the hand at the wrist level and was secured with a rub-
ological monitoring, and the immediate clinical success
ber band. The spectrum of signal strength in the different
of the surgery.
frequencies is displayed in the graph produced by the app.
Methods MRI and Fiber Tracking
After identifying the delayed onset of HT in 2 patients Patients underwent a 1.5-T magnetic field tomography
operated on at the Great Metropolitan Hospital Niguarda (Achieva 1.5T, Philips) diagnostic MRI examination. Im-
in Milan, we discussed our findings during international ages with contrast in T1, T2, and diffusion tensor imaging
clinical meetings, noting that this complication was also were obtained. The processing for reconstruction of the
reported by colleagues with extensive experience in brain- white matter tracts of interest was performed using the
stem surgery. We then contacted 10 different neurosurgery Oxford Centre for Functional MRI of the Brain Software
centers, gathering a total of 20 similar cases. Library (FSL).18 The nigrostriatal tract (NST) was recon-
The patient was assessed in the sitting position with the structed by selecting fibers that pass through two regions
limbs fully supported against gravity (rest tremor) and in of interest (ROIs): the substantia nigra and the homolat-
the upright position with the hands along the body (pos- eral striatum (Fig. 1A and B). The dentatothalamic tract
tural tremor). The intention tremor was assessed using the (DTT) was reconstructed by the selection of fibers passing
index-nose test. through four ROIs: the dentate nucleus of the cerebellum
For each patient, the following variables were recorded: beneath the floor of the fourth ventricle,19 the superior cer-
sex, age, location of the cavernoma, preoperative symp- ebellar peduncle between the upper part of the pons and
toms, surgical approach, SEZ, immediate and postoper- the cerebellum,20 and the contralateral red nucleus21 (Fig.
ative symptoms, symptoms at follow-up, time of HT pre- 1C–E), in addition to the contralateral thalamus, the lat-
sentation, type of treatment for HT, outcome, and time of ter obtained by automatic segmentation of the FSL. The
follow-up. rubroolivary tract (ROT) was obtained by selecting fibers
After gathering data on all 1274 brainstem cavernomas that pass through two ROIs: the red nucleus and the ipsi-
from all centers, we employed Fisher’s exact test to deter- lateral inferior olivary nucleus (Fig. 1F and G).
mine if the location of the cavernoma (medulla, pons-me- The maximum intensity (Max) value was measured
dulla, pons, midbrain) is a potential predicting factor for from the reconstructions of all the fiber bundles. This val-
the onset of HT. Statistical analysis was performed using ue represents the maximum number of streamlines in the
the software Stata/SE 16.1 (StataCorp.). voxel that has the highest probability density to contain
In our illustrative cases 3 and 4, the clinical appearance the reconstructed tract. This value, compared between the
of tremor was evaluated by measuring the electromyo- same bundles of different hemispheres, may be indicative
graphic (EMG) and accelerometric characteristics of the of differences within the white matter due to, for example,
tremor itself (in illustrative case 3, both before and after an injured area.

694 J Neurosurg Volume 135 • September 2021

 
*32 *
 Cenzato et al.

Results
The mean age of the 20 patients with HT was 43.5
years (range 24–68 years). Ten patients were males and
10 were females. There were 15 cavernomas within the
midbrain, 3 at the pontomedullary junction, and 2 within
the superior cerebellar peduncle (Table 1).
The surgical approach was selected on a case-by-case
basis depending on the anatomical location of the lesion.
The following approaches were used for midbrain lesions:
supracerebellar infratentorial (10 cases), pterional (1 ipsi-
lateral and 1 contralateral to the lesion), subtemporal (2
cases), frontotemporoorbitozygomatic (FTOZ; 1 ipsilat-
eral and 1 contralateral to the lesion), and transfrontal
transchoroidal (1 case). A midline suboccipital telovelar
approach was chosen for the 3 cavernomas at the ponto-
medullary junction (Table 1).
The SEZs used to access the midbrain were lateral mes-
encephalic (9 cases) or anterior mesencephalic (5 cases) or
infracollicular (3 cases), while the subfacial triangle was
chosen in all 3 pontomedullary lesions (Table 1).
The signs and symptoms of presentation were different,
including diplopia in 13 patients, hemiparesis in 6, ataxia
in 5, hemihypesthesia in 2, and dysmetria and dysarthria
in 2. Three patients (cases 1, 2, and 11) underwent surgery
because of a second symptomatic bleed, but on admission
they had been asymptomatic. After surgery, 15 patients
were clinically stable, 4 clinically worsened, and 1 neuro-
logically improved.
HT developed between 3 days and 24 months after
surgery, contralaterally to the site of the cavernoma in 18
cases and ipsilaterally in 2 cases. The tremor was associ-
ated with palatal tremor and inferior olive hypertrophic
degeneration in 2 cases. HT spontaneously disappeared in
1 case and improved in 6 cases (1 spontaneously, 2 after
medical treatment, and 3 after DBS), while it did not im-
prove in 12 cases and worsened in 1 case. The follow-up
was between 5 months and 7 years.
The total number of brainstem cavernomas operated
on in all centers was 1274 (Table 2). One hundred thirty-
three (10.4%) cavernomas were located in the medulla,
206 (16.2%) in the pons-medulla, 594 (46.6%) in the pons, FIG. 1. Axial T1-weighted MR images showing ROIs for reconstruction
and 341 (26.8%) in the midbrain. So, based on these data, of the NST (blue areas), located at the level of the substantia nigra
(A) and the ipsilateral striatum (B); of the DTT (red areas), located at
1.57% (20/1274) of patients developed HT after resection the level of the dentate nucleus (C), the superior cerebellar peduncle
of brainstem cavernomas, 1.46% (3/206) of patients after (D), and the contralateral red nucleus (E); and the ROT (green areas),
resection of cavernomas of the pons-medulla, and 4.99% located at the level of the red nucleus (F) and the ipsilateral inferior
(17/341) of patients after resection of cavernomas of the olivary nucleus (G). Figure is available in color online only.
midbrain. In the statistical analysis of all 1274 cases, a
cavernoma location in the midbrain was significantly as-
sociated with the onset of HT (p < 0.0005; Supplemental
Table). showed total resection of the cavernoma (Fig. 2E and F).
The patient was discharged 1 week after surgery with no
adjunctive deficits in a good clinical condition.
Illustrative Cases One month later she developed a tremor, suspected to
Case 3 be an HT, on the left side and involving mainly the upper
A 50-year-old woman was admitted to our hospi- limb. The tremor included a resting and postural compo-
tal for the onset of diplopia and left hemiparesis. Brain nent and worsened on the index-nose test. The EMG and
MRI showed a suspected hemorrhagic cavernoma within accelerometric recordings showed a tremor frequency of 3
the right cerebral peduncle (Fig. 2A and B). The patient Hz in the three positions (Fig. 3A, C, E, and G). Electro-
underwent resection of the cavernous malformation via myography showed a synchronous bursting of hand flexor
an ipsilateral subtemporal approach through the lateral and extensor muscles, with a higher burst amplitude of
mesencephalic SEZ (Fig. 2C and D). Postoperative MRI the flexors. Postoperative fiber tracking showed a reduced

J Neurosurg Volume 135 • September 2021 695

 
*32 *
Cenzato et al.

TABLE 1. Summary of data from collected cases with HT

Case Age Cavernoma Surgical
No. Surgeon Hospital Sex (yrs) Location Approach SEZ
1 R. F. Spetzler Barrow Neurological Institute, Phoenix M 47 Superior cerebel- Supracerebellar infratentorial Infracollicular
lar peduncle
2 R. F. Spetzler Barrow Neurological Institute, Phoenix F 31 Superior cerebel- Supracerebellar infratentorial Infracollicular
lar peduncle
3 M. Cenzato Great Metropolitan Hospital Niguarda, Milan F 50 Rt mesencephalon Subtemporal Lat mesencephalic
4 M. Cenzato Great Metropolitan Hospital Niguarda, Milan F 64 Rt pontomedullary Midline suboccipital Subfacial
telovelar
5 R. Stefini Legnano Hospital, Milan M 28 Rt mesencephalon Contralateral FTOZ Ant mesencephalic
6* C. Raftopoulos Saint-Luc University Clinic, Brussels F 29 Rt mesencephalon Ipsilat pterional Ant mesencephalic
transsylvian
7 C. Raftopoulos Saint-Luc University Clinic, Brussels M 56 Lt mesencephalon Transfrontal transchoroidal Ant mesencephalic
8 C. Raftopoulos Saint-Luc University Clinic, Brussels F 59 Lt mesencephalon Ipsilat FTOZ Ant mesencephalic
9 C. Raftopoulos Saint-Luc University Clinic, Brussels F 31 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
10 C. Bortolotti Bellaria Hospital, Bologna M 55 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
11 P. Ferroli C. Besta Neurological Institute, Milan F 48 Rt pontomedullary Midline suboccipital Subfacial
telovelar
12 A. N. Konovalov Burdenko Neurosurgical Institute, Moscow F 67 Rt mesencephalon Supracerebellar infratentorial Infracollicular
13 A. N. Konovalov Burdenko Neurosurgical Institute, Moscow M 24 Lt mesencephalon Subtemporal Lat mesencephalic
14 A. Smolanka University Hospital Uzhhorod, Uzhhorod M 25 Rt mesencephalon Contralat pterional Ant mesencephalic
transsylvian
15 A. Smolanka University Hospital Uzhhorod, Uzhhorod M 25 Rt mesencephalon Supracerebellar infratentorial Lat mesencephalic
16 M. Tatagiba University Hospital Tübingen, Tübingen F 68 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
17 U. Sure University Hospital Essen, Essen M 52 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
18 U. Sure University Hospital Essen, Essen M 37 Lt pontomedullary Midline suboccipital Subfacial
telovelar
19 U. Sure University Hospital Essen, Essen F 38 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
20 G. Pinna University Hospital Verona, Verona M 35 Lt mesencephalon Supracerebellar infratentorial Lat mesencephalic
Ant = anterior.
* This case has been inserted courtesy of Delaunois et al., 2018.9

TABLE 2. Location of 1274 brainstem cavernomas

Hospital Midbrain Pons Pons-Medulla Medulla Total Cases
Barrow Neurological Institute, Phoenix 151 (30.2%) 216 (43.2%) 77 (15.4%) 56 (11.2%) 500
Great Metropolitan Hospital Niguarda, Milan 15 (23.1%) 36 (55.4%) 9 (13.8%) 5 (7.7%) 65
Legnano Hospital, Milan 1 (20.0%) 2 (40.0%) 1 (20.0%) 1 (20.0%) 5
Saint-Luc University Clinic, Brussels 10 (33.3%) 12 (40.0%) 5 (16.7%) 3 (10.0%) 30
Bellaria Hospital, Bologna 8 (26.7%) 15 (50.0%) 3 (10.0%) 4 (13.3%) 30
C. Besta Neurological Institute, Milan 16 (35.6%) 18 (40.0%) 6 (13.3%) 5 (11.1%) 45
Burdenko Neurosurgical Institute, Moscow 67 (19.1%) 182 (51.9%) 70 (19.9%) 32 (9.1%) 351
University Hospital Uzhhorod, Uzhhorod 7 (29.2%) 13 (54.2%) 3 (12.5%) 1 (4.2%) 24
University Hospital Tübingen, Tübingen 20 (29.9%) 30 (44.8%) 7 (10.4%) 10 (14.9%) 67
University Hospital Essen, Essen 28 (29.8%) 38 (40.4%) 19 (20.2%) 9 (9.6%) 94
University Hospital Verona, Verona 18 (28.5%) 32 (50.8%) 6 (9.5%) 7 (11.1%) 63
Total cases 341 (26.8%) 594 (46.6%) 206 (16.2%) 133 (10.4%) 1274

696 J Neurosurg Volume 135 • September 2021

 
*32 *
 Cenzato et al.

representation of the NST and ROT through the right red

nucleus, homolateral to the lesion, and no representation
of the right DTT. The data were confirmed by Max values.
The tremor did not improve spontaneously or after le-
vodopa treatment; thus, after 9 months, the patient under-
went surgery for DBS of the VIM thalamic nucleus. After
surgical treatment, the patient clinically improved, and the
EMG and accelerometric (Fig. 3B, D, F, and H) recordings
showed outstanding improvement in the tremor.

Case 4
A 64-year-old woman was admitted to our hospital
for the onset of progressively worsening diplopia and un-
steady gait. Brain MRI showed a suspected hemorrhagic
cavernoma at the right pontomedullary junction, under
the floor of the fourth ventricle (Fig. 4A–C). The patient
underwent resection of the cavernous malformation via a
midline suboccipital telovelar approach through the subfa-
cial SEZ (Fig. 4D and E). The patient’s neurological status
was unchanged postoperatively.
Three months later, the patient developed a tremor, sus-
pected to be an HT, in the left upper limb. The tremor had
resting and postural components and appeared to be worse
during the index-nose test. On EMG and accelerometric
recordings, the tremor showed a frequency of around 1.5–
2 Hz in the three assessment positions. Electromyography
showed a synchronous bursting of hand flexor and exten-
sor muscles, with a higher burst amplitude of the exten-
sors. The postoperative fiber tracking showed a reduced
representation of the ROT through the right red nucleus,
ipsilateral to the lesion, and surprisingly a reduced rep-
resentation of the contralateral NST and DTT. The data
were confirmed by Max values.
The patient did not respond to levodopa and is currently
making a decision about DBS.

Discussion FIG. 2. Case 3. Preoperative axial (A) and coronal (B) T2-weighted
Brainstem cavernomas represent a relatively rare vas- MR images showing a brainstem cavernoma at the level of the right
cular malformation that can lead to severe neurological cerebral peduncle. Intraoperative images showing the surgical access
symptoms by direct compression or following hemor- through the lateral mesencephalic sulcus (C) and the dyschromic area
rhage. Today, it is crucial that brainstem cavernomas are corresponding to the cavernoma (D) inside the brainstem. Postoperative
axial T1-weighted MR images with gadolinium enhancement show in
exclusively treated in highly specialized centers, with all more caudal (E) and more cranial (F) sections the surgical entry point at
the available new technologies. Strict adherence to the the level of the right lateral mesencephalic sulcus and complete removal
SEZs to the brainstem is also required to minimize surgi- of the cavernoma. Figure is available in color online only.
cal morbidity.
HT is one of many possible postsurgery complications.
It is very unusual and is generally found quite some time cerebellar dentate nucleus to the thalamus presumably
after surgery. HT can be highly debilitating for the affect- leads to thalamic functional changes. In other words, this
ed individuals. The syndrome is characterized by a rest- suggests that the approximately 4-Hz rhythmic discharges
ing, postural, and intentional tremor that ensues sometime observed in the thalamus during functional neurosurgery,
after the damaging event with a low-frequency muscle which correspond to the HT rhythm, contribute to the de-
contraction (< 5 Hz).12 To date, it is generally accepted that velopment of the tremor. So, the coagulation of thalamic
HT occurs as a result of lesions in the brainstem/cerebel- neurons through VIM lesioning or, similarly, the func-
lum and in the thalamus.13 According to anatomopatho- tional changes obtained through VIM DBS represent two
logical22 and PET data,23 both the nigrostriatal dopami- different techniques to control the tremor.
nergic and dentatothalamic pathways must be damaged to HT can also be caused by damage to the Guillain-Mol-
generate a contralateral HT.13 A recent tractographic study laret triangle. A recent paper has described a reduction of
has also shown a reduction in the NST and DTT from the the fibers of the central tegmental tract on the side of the
injured mesencephalic side.7 Damage to the DTT due to damaged brainstem, according to tractographic examina-
disruption of afferent pathways (deafferentation) from the tion.24

J Neurosurg Volume 135 • September 2021 697

 
*32 *
Cenzato et al.

FIG. 3. Case 3. EMG results showing in all positions a tremor frequency around 3 Hz but a notable difference in the amplitude of
the tremor between the pre-DBS (A, C, E) and post-DBS (B, D, F) recordings. Power spectrum graphs obtained with VibSensor,
showing a stable tremor frequency around 3.3 Hz but a considerable reduction in the power spectral density (PSD) of the tremor
between the pre-DBS (G) and post-DBS (H) recordings. BB = biceps brachii muscle; ERC = extensor radialis carpi muscle; FCD =
flexor communis digitorum muscle; TB = triceps brachii muscle. Figure is available in color online only.

Review of the Literature and Analysis of Our Series noma who, 3 months after partial excision, presented with
In the literature, authors have reported a case of HT a progressively increasing contralateral HT of the hand,
due to the presence of a mesencephalon cavernoma, with which almost completely resolved after thalamic DBS.6
a worsening of symptoms after surgery.25 There are also Seidel et al. reported the case of a 16-year-old girl with a
4 well-described cases of HT following the resection of left midbrain cavernoma who, 9 months after surgery, had
brainstem cavernomas. Pahwa et al. described the case of a highly debilitating contralateral HT, which almost com-
a 45-year-old woman with a right mesencephalon caver- pletely resolved after 3 months of combined pramipexole
698 J Neurosurg Volume 135 • September 2021
 
*32 *
 Cenzato et al.

and levodopa treatment.7 Aydin et al. described the case

of a 30-year-old woman with a right pontomesencephalic
cavernoma who, 6 months after an excision performed via
a telovelar suboccipital approach, presented with an HT
in the contralateral upper limb, which almost completely
recovered after DBS of the VIM nucleus of the thalamus
and the inner globus pallidus.8 Delaunois et al. reported
the case of a 29-year-old woman with a right mesencepha-
lon cavernoma who, 10 days after a homolateral transsyl-
vian transuncal surgery performed via the anterior mesen-
cephalic SEZ, developed an HT in the contralateral upper
limb, which disappeared after 1 year of treatment with
levodopa.9 This case is included as case 6 in our series
(Table 1).
We compared the literature to our analysis of the 20
cases of HT, finding that HT after the resection of brain-
stem cavernomas is a rare morbidity: 1.57% (20/1274) of
all patients in our series developed HT. In the statistical
analysis of 1274 cases, a cavernoma location in the mid-
brain was significantly associated with the onset of HT (p
< 0.0005). This strong association can be easily explained
by the fact that the majority of anatomical structures in-
volved in the genesis of HT are densely packed in the mid-
brain (Fig. 5).
In our series HT occurred equally in both sexes, where-
as in the largest study of HT ever published (29 cases),
the authors found a slight female predominance.14 The rest
of the literature reports a male prevalence. In the 4 cases
of HT occurring after brainstem cavernoma surgery that
have been reported so far, the patients were all women.6–9
The average age in our study was 43.5 years (range 24–68
years), while in those other 4 cases the average age was FIG. 4. Case 4. Preoperative MRI showing a cavernoma at the
30 years (range 16–45 years). In the literature on HT, the pontomedullary junction in axial Gd-enhanced T1-weighted (A), coronal
age at clinical presentation has varied. As described above, T2-weighted (B), and sagittal Gd-enhanced T1-weighted (C) sections.
the onset delay has ranged between 1 and 24 months after Postoperative CT scan showing in more caudal (D) and more cranial
surgery; in our series the delay was between 3 days and 24 (E) sections the surgical entry point at the level of the floor of the fourth
months, while in those previously mentioned 4 cases the ventricle, reached through the midline suboccipital telovelar approach.
time span ranged from 10 days to 9 months. Also in line
with what has been reported in the literature, there was
a clear prevalence of cavernomas at the mesencephalic cases, via the supracerebellar infratentorial approach in 7
level (15 cases) versus the pontomedullary level (3 cases) cases and the subtemporal approach in 2 cases. When the
in our series. In the previously mentioned 4 cases of HT point closest to the surface was anterior to the mesenceph-
following brainstem cavernoma resection, the localization alon, all 5 cases were operated on via the anterior mesen-
was exclusively mesencephalic. In only 2 of our cases was cephalic SEZ, using the transsylvian pterional approach
the cavernoma located in the superior cerebellar peduncle, (2 cases, 1 ipsilateral and 1 contralateral), FTOZ approach
causing a homolateral HT, probably as a result of damage (2 cases, 1 ipsilateral and 1 contralateral), or transfrontal
to the dentate-rubral fibers and the ascending arm of the transchoroidal approach (1 case). When the cavernoma
dentate-olivary fibers before the decussation. In the litera- emerged near the floor of the fourth ventricle (3 cases), the
ture, there are very few cases in which the tremor homo- subfacial SEZ was chosen with a telovelar midline suboc-
lateral to the lesion of the superior cerebellar peduncle is cipital approach. The approach was specified in only 2 of
described as “intentional”26 or “reminiscent of moderate the 4 cases of HT following brainstem cavernoma resec-
essential tremor.”27,28 tion in the literature: the pterional transsylvian approach
The location and characteristics of the cavernoma, in with the anterior mesencephalic SEZ in the case published
particular the outcrop area or the area closest to the epen- by Delaunois et al., in which the cavernoma reached the
dymal surface, determine the choice of the SEZ. Among anterior surface of the midbrain, and the telovelar suboc-
the 20 cases we examined, the infracollicular SEZ—with cipital approach in the case published by Aydin et al., in
an infratentorial supracerebellar approach—was chosen in which the cavernoma reached the pontine surface of the
all 3 cases in which the cavernoma had the point closest floor of the fourth ventricle.8,9
to the surface located near the quadrigeminal plate. When Overall, the delayed onset of the HT occurred with all
this point was located laterally with respect to the mid- the SEZs, regardless of the surgeon.
brain, the lateral mesencephalic SEZ was chosen in all 9 In line with the literature, the onset symptoms of caver-
J Neurosurg Volume 135 • September 2021 699

* | D*2 *32 *
Cenzato et al.

FIG. 5. Anatomical relationships between HT pathways and the lateral mesencephalic SEZ. Drawing of the nigrostriatal (blue),
dentatothalamic (green), and rubro-olivo-dentato-rubric (red) pathways. The NST, DTT, and ROT at the level of the cerebral
peduncle run close to the lateral mesencephalic sulcus (yellow), which is the SEZ used in case 3. Copyright Davide Colistra.
Published with permission. Figure is available in color online only.

700 J Neurosurg Volume 135 • September 2021

 
* | D*2 *32 *
 Cenzato et al.

nomas varied.14 It is important to note that 15 of the 20 pa-

tients remained stable after surgery and 1 even improved,
while in 4 cases there were additional neurological deficits.
In 2 cases, after some time from the acute event, an infe-
rior olive hypertrophic degeneration with palatal tremor
was associated with the HT, a rare but well-reported event
in the literature.22,29
DBS proved to be the most effective treatment in our
study, in line with findings in the literature.

Analysis of the Illustrative Cases

In case 3 the tremor was predominant in the flexor mus-
cles; in case 4 it was predominant in the extensors. The
frequency of the tremor was different, settling at around
3 Hz in case 3 and at 1.5–2 Hz in case 4. Tractographic
data were different in the 2 cases as well: in case 3 the FIG. 6. Anatomical relationships between HT pathways and the subfa-
NST, DTT, and ROT were less represented on the lesion cial SEZ. Freehand drawing showing how the central tegmental tract
side (mesencephalic), but in case 4 only the ROT was less (red), within which the ROT runs, is at the level of the floor of the fourth
represented on the lesion side (pontomedullary) and the ventricle just below the subfacial area (yellow), which is the SEZ used in
NST and DTT were reduced on the opposite side. It is case 4. Copyright for drawing: Davide Colistra; published with permis-
possible that these data, especially the tractographic data, sion. The drawing is based on a photograph of a cadaveric dissection
reflect the different lesion sites and therefore the different prepared by Kaan Yağmurlu, which appears in The Rhoton Collection;
with permission (CC BY-NC-SA 4.0 [http://creativecommons.org/licens-
pathways of the fibers involved in the damage. In fact, in es/by-nc-sa/4.0]). Figure is available in color online only.
the HT with the mesencephalic lesion the studied bundles
would all seem to be involved homolaterally, while in the
HT with the pontomedullary lesion only the ROT would
seem to be involved homolaterally. Moreover, the fact that
the fiber tracts theoretically responsible for tremorigenesis In cases of mesencephalic lesions in the area of the
are reduced on the lesion side, specifically those passing tegmentum, consider the following. In the anterior mesen-
through the SEZ, is further proof that the current SEZs to cephalic surgical SEZ, which is directed through the ce-
the brainstem are insufficient to prevent the delayed onset rebral peduncle between the corticospinal and corticobul-
of HT. Thus, in case 3, in which the lateral mesencephalic bar tracts laterally and the exit point of the oculomotor
SEZ was used, the affected pathways of the NST, DTT, nerve medially, care should be taken in the deepest part of
and ROT at the level of the cerebral peduncle run close to the working space to avoid the red nucleus with the NST,
this SEZ (Fig. 5). DTT, and ROT. In the lateral mesencephalic SEZ, which
In case 4, in which the subfacial SEZ was used, the is located between the medial lemniscus posteriorly and
affected tract was the ROT within the central tegmental the substantia nigra anterolaterally, care should be taken
tract, which at the level of the floor of the fourth ventricle to avoid the substantia nigra and, at an average of 3.4 mm
runs just below this SEZ (Fig. 6). deep to the surface of the midbrain,31 the red nucleus with
In case 3, the treatment with DBS reduced the ampli- the NST, DTT, and ROT. In the infracollicular SEZ, which
tude but not the frequency of the tremor. is located between the trochlear nerve inferiorly and the
The synchronous pattern of the tremor observed in inferior margin of the inferior colliculus superiorly, care
the 2 illustrative cases contrasts with the definition of the should be taken as the incision deepens from dorsal to ven-
classic HT as an alternating-pattern tremor.16 According tral to avoid the ROT within the central tegmental tract
to Milanov,16 HT is present even at rest and has the low- and the DTT within the decussation of the superior cer-
est frequency (less than 4 Hz) but an alternating pattern ebellar peduncles.
of EMG bursting. The tremors observed in our illustrative In the case of lesions at the pontomedullary junction,
cases have mixed features, being present at rest and show- near the floor of the fourth ventricle, if the surgeon choos-
ing a low frequency (as in HT) but a synchronous pattern. es the subfacial SEZ, care should be taken to avoid the
Myorhythmia, as described by Baizabal-Carvallo et al.,30 ROT within the central tegmental tract, which runs deep
shares some of the features of the tremors observed in our inside the subfacial triangle.
illustrative cases. The pattern is reported to be alternat- This study represents the first series of postsurgical
ing or synchronous, with one agonist muscle contracting HT cases reported in the literature, as well as the first se-
more intensely than the antagonist, resulting in oscillatory ries on HT after the resection of brainstem cavernomas.
movement. Its major limitation is its inclusion of retrospective cases.
Our data demonstrate that the HT complication is more
Surgical Tips and Other Important Considerations than just episodic. The sample was collected from among
From the results of this study, we can establish that a cav- patients operated on by neurosurgeons from 11 different
ernoma location in the midbrain is a significant predictive centers, all of whom had similar experiences in surgery
factor for the onset of HT, and we can hypothesize which ar- for brainstem cavernomas. Unfortunately, at the moment
eas, during the resection, are most at risk for the onset of HT. there is no evidence that intraoperative neurophysiological
J Neurosurg Volume 135 • September 2021 701

* | D*2 *32 *
Cenzato et al.

monitoring can help in preventing this complication, but surgery:​a single-center case series of 300 surgically treated
in the present study we could link the surgical SEZ to the patients. Neurosurgery. 2011;​68(2):​403–415.
anatomical structure involved in HT. 11. Wang CC, Liu A, Zhang JT, et al. Surgical management of
brain-stem cavernous malformations:​report of 137 cases.
Surg Neurol. 2003;​59(6):​444–454.
Conclusions 12. Bhatia KP, Bain P, Bajaj N, et al. Consensus statement on the
classification of tremors. From the task force on tremor of
Brainstem cavernomas are a relatively rare vascular pa- the International Parkinson and Movement Disorder Society.
thology. Surgery, although complex and accompanied by Mov Disord. 2018;​33(1):​75–87.
a high risk of serious complications, should be advised as 13. Deuschl G, Bergman H. Pathophysiology of nonparkinsonian
the treatment of choice in selected cases. Surgery should tremors. Mov Disord. 2002;​17(suppl 3):​S41–S48.
be performed using SEZs to the brainstem. HT is one of 14. Raina GB, Cersosimo MG, Folgar SS, et al. Holmes tremor:​
the possible, albeit rare, complications of this type of sur- clinical description, lesion localization, and treatment in a
gery and is in most cases secondary to damage to the ni- series of 29 cases. Neurology. 2016;​86(10):​931–938.
15. van Hilten JJ, van Dijk JG, Dunnewold RJW, et al. Diurnal
grostriatal, dentatothalamic, and rubroolivary pathways. variation of essential and physiological tremor. J Neurol Neu-
From our study it has emerged that, despite strict adher- rosurg Psychiatry. 1991;​54(6):​516–519.
ence to SEZs, the use of intraoperative neurophysiological 16. Milanov I. Electromyographic differentiation of tremors. Clin
monitoring, and the immediate success of the surgery, HT Neurophysiol. 2001;​112(9):​1626–1632.
can occur as a delayed complication following resection of 17. Bhatti D, Thompson R, Hellman A, et al. Smartphone apps
brainstem cavernoma. A cavernoma location in the mid- provide a simple, accurate bedside screening tool for orthostatic
brain is a significant predictive factor for the onset of HT. tremor. Mov Disord Clin Pract (Hoboken). 2017;​4(6):​852–857.
18. Kwon HG, Hong JH, Hong CP, et al. Dentatorubrothalamic
We listed the anatomical pathways involved in the genesis tract in human brain:​diffusion tensor tractography study.
of this highly debilitating complication, but further ana- Neuroradiology. 2011;​53(10):​787–791.
tomical and neurophysiological studies will be necessary 19. Salamon N, Sicotte N, Drain A, et al. White matter fiber tractog-
to find a means of prevention. raphy and color mapping of the normal human cerebellum with
diffusion tensor imaging. J Neuroradiol. 2007;​34(2):​115–128.
20. Hong JH, Kim OL, Kim SH, et al. Cerebellar peduncle injury
Acknowledgments in patients with ataxia following diffuse axonal injury. Brain
We thank the illustrator, Lorenzo Di Santo, for his special con- Res Bull. 2009;​80(1-2):​30–35.
tribution to this article. 21. Habas C, Cabanis EA. Cortical projection to the human red
nucleus:​complementary results with probabilistic tractogra-
phy at 3 T. Neuroradiology. 2007;​49(9):​777–784.
References 22. Masucci EF, Kurtzke JF, Saini N. Myorhythmia:​a wide-
  1. McCormick WF, Hardman JM, Boulter TR. Vascular mal- spread movement disorder. Clinicopathological correlations.
formations (“angiomas”) of the brain, with special reference Brain. 1984;​107(Pt 1):​53–79.
to those occurring in the posterior fossa. J Neurosurg. 1968;​ 23. Remy P, de Recondo A, Defer G, et al. Peduncular ‘rubral’
28(3):​241–251. tremor and dopaminergic denervation:​a PET study. Neurol-
  2. Metellus P, Kharkar S, Kapoor S, et al. Cerebral cavernous ogy. 1995;​45(3 Pt 1):​472–477.
malformations. Neurosurg Q. 2008;​18:​223–229. 24. He JM, He J, Lin HX, et al. Holmes tremor with impairment
  3. Giombini S, Morello G. Cavernous angiomas of the brain. of the Guillain-Mollaret triangle following medullar hemor-
Account of fourteen personal cases and review of the litera- rhage. Neurol Sci. 2018;​39(7):​1305–1306.
ture. Acta Neurochir (Wien). 1978;​40(1-2):​61–82. 25. Samadani U, Umemura A, Jaggi JL, et al. Thalamic deep brain
  4. Moriarity JL, Clatterbuck RE, Rigamonti D. The natural stimulation for disabling tremor after excision of a midbrain cav-
history of cavernous malformations. Neurosurg Clin N Am. ernous angioma. Case report. J Neurosurg. 2003;​98(4):​888–890.
1999;​10(3):​411–417. 26. Fukui T, Ichikawa H, Sugita K, Tsukagoshi H. Intention
  5. Akers A, Al-Shahi Salman R, Awad IA, et al. Synopsis of tremor and olivary enlargement:​clinico-radiological study.
guidelines for the clinical management of cerebral cavernous Intern Med. 1995;​34(11):​1120–1125.
malformations:​consensus recommendations based on systemat- 27. Albin RL. Cerebellar input tremor. Neurology. 1998;​50(1):​
ic literature review by the Angioma Alliance scientific advisory 307–308.
board clinical experts panel. Neurosurgery. 2017;​80(5):​665–680. 28. Savoiardo M. Cerebellar input tremor:​inferior or superior
  6. Pahwa R, Lyons KE, Kempf L, et al. Thalamic stimulation cerebellar peduncle lesion? Neurology. 1998;​51(6):​1777–1778.
for midbrain tremor after partial hemangioma resection. Mov 29. Rieder CRM, Rebouças RG, Ferreira MP. Holmes tremor in
Disord. 2002;​17(2):​404–407. association with bilateral hypertrophic olivary degeneration
  7. Seidel S, Kasprian G, Leutmezer F, et al. Disruption of ni- and palatal tremor:​chronological considerations. Case report.
grostriatal and cerebellothalamic pathways in dopamine Arq Neuropsiquiatr. 2003;​61(2B):​473–477.
responsive Holmes’ tremor. J Neurol Neurosurg Psychiatry. 30. Baizabal-Carvallo JF, Cardoso F, Jankovic J. Myorhythmia:​
2009;​80(8):​921–923. phenomenology, etiology, and treatment. Mov Disord. 2015;​
  8. Aydin S, Abuzayed B, Kiziltan G, et al. Unilateral thalamic Vim 30(2):​171–179.
and GPi stimulation for the treatment of Holmes’ tremor caused 31. Yagmurlu K, Rhoton AL Jr, Tanriover N, Bennett JA. Three-
by midbrain cavernoma:​case report and review of the literature. dimensional microsurgical anatomy and the safe entry zones
J Neurol Surg A Cent Eur Neurosurg. 2013;​74(4):​271–276. of the brainstem. Neurosurgery. 2014;​10(suppl 4):​602–620.
  9. Delaunois J, Vaz G, Raftopoulos C. Transsylvian transuncal
approach for an anterior midbrain cavernous malformation
resection:​a case report. Oper Neurosurg (Hagerstown). 2018;​ Disclosures
14(3):​E38–E43. The authors report no conflict of interest concerning the materi-
10. Abla AA, Lekovic GP, Turner JD, et al. Advances in the als or methods used in this study or the findings specified in this
treatment and outcome of brainstem cavernous malformation paper.

702 J Neurosurg Volume 135 • September 2021

 
* | D*2 *32 *
 Cenzato et al.

Author Contributions Supplemental Information

Conception and design: Colistra, Cenzato. Acquisition of data: Online-Only Content
Colistra, Raftopoulos, Sure, Tatagiba, Spetzler, Konovalov, Supplemental material is available with the online version of the
A Smolanka, V Smolanka, Stefini, Bortolotti, Ferroli, Pinna, article.
Franzini, Dammann, Naros, Boeris, Mantovani, Lizio, Piano, Supplemental Table. https://thejns.org/doi/suppl/​10.3171/​
Fava. Analysis and interpretation of data: Colistra, Iacopino, 2020.7.JNS201352.
Lizio, Fava. Drafting the article: Colistra, Iacopino, Fava.
Critically revising the article: Cenzato, Raftopoulos, Sure, Correspondence
Tatagiba, Spetzler, Konovalov, A Smolanka, V Smolanka, Fava.
Reviewed submitted version of manuscript: Raftopoulos, Sure, Davide Colistra: Great Metropolitan Hospital Niguarda, Milan,
Tatagiba, Spetzler, Konovalov. Approved the final version of the Italy. [email protected].
manuscript on behalf of all authors: Colistra. Administrative/
technical/material support: Lizio. Study supervision: Cenzato.

J Neurosurg Volume 135 • September 2021 703

* | D*2 *32 *
 CLINICAL ARTICLE
J Neurosurg 135:704–711, 2021

Association between flat-panel computed tomography

hyperattenuation and clinical outcome after successful
recanalization by endovascular treatment
Jang-Hyun Baek, MD,1,2 Byung Moon Kim, MD, PhD,3 Ji Hoe Heo, MD, PhD,2
Dong Joon Kim, MD, PhD,3 Hyo Suk Nam, MD, PhD,2 Young Dae Kim, MD, PhD,2
Hyun Seok Choi, MD,3 Jun-Hwee Kim, MD,3 and Jin Woo Kim, MD4
1
Department of Neurology, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul; 2Department
of Neurology, Severance Stroke Center, Severance Hospital, Yonsei University College of Medicine, Seoul; 3Interventional
Neuroradiology, Severance Stroke Center, Severance Hospital, and Department of Radiology, Yonsei University College of
Medicine, Seoul; and 4Department of Radiology, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul,
Republic of Korea

OBJECTIVE  Hyperattenuation on CT scanning performed immediately after endovascular treatment (EVT) is known to
be associated with the final infarct. As flat-panel CT (FPCT) scanning is readily accessible within their angiography suite,
the authors evaluated the ability of the extent of hyperattenuation on FPCT to predict clinical outcomes after EVT.
METHODS  Patients with successful recanalization (modified Thrombolysis in Cerebral Infarction grade 2b or 3) were
reviewed retrospectively. The extent of hyperattenuation was assessed by the Alberta Stroke Program Early CT Score
on FPCT (FPCT-ASPECTS). FPCT-ASPECTS findings were compared according to functional outcome and malignant
infarction. The predictive power of the FPCT-ASPECTS with initial CT images before EVT (CT-ASPECTS) and follow-up
diffusion-weighted images (MR-ASPECTS) was also compared.
RESULTS  A total of 235 patients were included. All patients were treated with mechanical thrombectomy, and 45.5% of
the patients received intravenous tissue plasminogen activator. The mean (± SD) time from stroke onset to recanaliza-
tion was 383 ± 290 minutes. The FPCT-ASPECTS was significantly different between patients with a favorable outcome
and those without (mean 9.3 ± 0.9 vs 6.7 ± 2.6) and between patients with malignant infarction and those without (3.4 ±
2.9 vs 8.8 ± 1.4). The FPCT-ASPECTS was an independent factor for a favorable outcome (adjusted OR 3.28, 95% CI
2.12–5.01) and malignant infarction (adjusted OR 0.42, 95% CI 0.31–0.57). The area under the curve (AUC) of the FPCT-
ASPECTS for a favorable outcome (0.862, cutoff ≥ 8) was significantly greater than that of the CT-ASPECTS (0.637) (p
< 0.001) and comparable to that of the MR-ASPECTS (0.853) (p = 0.983). For malignant infarction, the FPCT-ASPECTS
was also more predictive than the CT-ASPECTS (AUC 0.906 vs 0.552; p = 0.001) with a cutoff of ≤ 5.
CONCLUSIONS  The FPCT-ASPECTS was highly predictive of clinical outcomes in patients with successful recanaliza-
tion. FPCT could be a practical method to immediately predict clinical outcomes and thereby aid in acute management
after EVT.
https://thejns.org/doi/abs/10.3171/2020.7.JNS193214

C
KEYWORDS  endovascular treatment; stroke; prognosis; flat-panel computed tomography; vascular disorders

onventional multidetector CT is often performed so-called contrast staining.1–5 This contrast material–re-
after endovascular treatment (EVT) of acute stroke, lated hyperattenuation is thought to result from leakage of
as CT imaging can show periprocedural hemor- contrast medium into the parenchymal extracellular space
rhagic conditions, including procedure-related intracranial due to blood-brain barrier disruption caused by focal is-
hemorrhage and immediate hemorrhagic transformation chemic injury, chemotoxicity, or reperfusion injury.6 There-
due to reperfusion. However, most hyperattenuation on CT fore, hyperattenuation due to contrast staining on postpro-
scanning is generally contributed by the contrast medium, cedural CT can be presumed to reflect the infarct core.1,2

ABBREVIATIONS  ASPECTS = Alberta Stroke Program Early CT Score; AUC = area under the ROC curve; EVT = endovascular treatment; FPCT = flat-panel CT; ICA =
internal carotid artery; mRS = modified Rankin Scale; NIHSS = National Institutes of Health Stroke Scale; ROC = receiver operating characteristic; tPA = tissue plasminogen
activator.
SUBMITTED  November 27, 2019.  ACCEPTED  July 15, 2020.
INCLUDE WHEN CITING  Published online December 25, 2020; DOI: 10.3171/2020.7.JNS193214.

704 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *33 *
 Baek et al.

CT-like images can be obtained directly within the Acquisition of FPCT

angiography suite with the use of modern angiographic FPCT was routinely performed at the end of the EVT
systems equipped with a flat-panel detector (flat-panel CT procedure to evaluate if procedure-related complications
[FPCT]). FPCT is often performed during or at the end of (e.g., subarachnoid hemorrhage) or hemorrhagic transfor-
the EVT procedure, primarily to observe periprocedural mation had occurred. In patients who underwent intracra-
hemorrhagic conditions, due to the ease with which it can nial stenting or received intraarterial glycoprotein IIb/IIIa
be performed and its recently improved image quality. inhibitor, FPCT was performed before the introduction of
Given that hyperattenuation on postprocedural CT is as- rescue modalities to evaluate the hemorrhagic condition.
sociated with the infarct core, we reasoned that the extent FPCT images were acquired on a biplane angiography
of hyperattenuation on FPCT performed at the end of the system (Allura Xper FD20/20 or Allura Clarity FD20/20
EVT procedure might be closely associated with patients’ Interventional Angiographic Systems, Philips Healthcare).
clinical outcomes. In particular, we hypothesized that the This imaging system allows motorized rotational move-
greater the extent of hyperattenuation on FPCT, the poor- ment of the C-arm in the plane perpendicular to the longi-
er the patients’ clinical outcomes. To test this hypothesis, tudinal table axis around the head of the patient’s position
we evaluated clinical outcomes according to the extent of at the system isocenter. A total of 620 fluoroscopic frames
hyperattenuation of FPCT and compared the ability of in a circular motion were acquired with the following
FPCT to predict clinical outcome with that of other imag- parameters: voltage 120 kV, tube current 250 mA, frame
ing modalities. speed 30 frames per second, acquisition time 5 msec per
frame, total angle 240°, rotation speed 22° per second, ex-
Methods posure time 20 seconds, field of view 48 cm, and 1024
Participants × 1024–pixel matrix detector. Image reconstruction was
performed using a dedicated workstation. Image postpro-
We retrospectively reviewed consecutive acute stroke cessing to correct for beam hardening and scatter was per-
patients with intracranial large-vessel occlusion of the formed to achieve maximum spatial and contrast resolu-
anterior circulation who underwent EVT between Janu- tion. For evaluation, volume-rendering reconstruction was
ary 2010 and December 2018. Patients were selected from performed with 5-mm-thick slices and no gap, although
a prospectively maintained stroke registry at a tertiary the thickness can be customized from 1 to 100 mm.
stroke center. EVT was generally considered for patients
who met the following criteria: 1) a CT angiography–de-
Determination of FPCT-ASPECTS, CT-ASPECTS, and
termined endovascularly accessible intracranial large-
MR-ASPECTS
vessel occlusion associated with neurological symptoms;
2) presentation to the hospital within 8 hours from stroke The extent of FPCT hyperattenuation was semiquanti-
onset (in the later study period, patients within 8 to 12 tatively assessed using the ASPECTS (FPCT- ASPECTS).
hours were also considered if they had an Alberta Stroke To compare the predictive power of the FPCT-ASPECTS
Program Early CT Score [ASPECTS] ≥ 7); and 3) base- with other periprocedural imaging modalities, the
line National Institutes of Health Stroke Scale (NIHSS) ASPECTS on noncontrast CT images obtained before the
score ≥ 4. We also generally performed EVT in patients EVT procedure (CT-ASPECTS) and diffusion-weighted
with premorbid modified Rankin Scale (mRS) scores ≤ 3. images obtained after the EVT procedure (MR-ASPECTS)
The intracranial internal carotid artery (ICA) and M1 or were also assessed. The CT-ASPECTS was calculated ac-
proximal M2 segment of the middle cerebral artery were cording to the original method.7 Similarly, the ASPECTS
defined as intracranial large vessels. In the present study, was calculated by subtracting the number of hyperattenu-
patients 1) whose occlusion was successfully recanalized ated regions from 10 to calculate the FPCT-ASPECTS
and 2) who underwent FPCT at the end of the EVT proce- and MR-ASPECTS. Locations of interest for determining
dure were included. the FPCT-ASPECTS and MR-ASPECTS were the same
All patients eligible for intravenous tissue plasmino- as those for the CT-ASPECTS. The CT-ASPECTS, FPCT-
gen activator (tPA) or EVT were evaluated by noncon- ASPECTS, and MR-ASPECTS were assessed by a total of
trast CT and CT angiography. Patients eligible for intra- six independent neuroradiologists and interventional neu-
venous tPA treatment were treated with 0.9 mg/kg tPA. roradiologists. The ASPECTS on each imaging modality
EVT procedures were performed according to common was assessed independently by two raters. All raters were
recommendations. The frontline EVT modality was stent blinded to clinical and procedural outcomes. Interobserver
retriever thrombectomy, contact aspiration thrombectomy, agreements of the CT-ASPECTS, FPCT-ASPECTS, and
or a combination of these techniques. A balloon-guiding MR-ASPECTS were 0.872, 0.936, and 0.957, respectively.
catheter was routinely used. Successful recanalization in Cases with discrepancies were resolved by consensus.
EVT was defined as a modified Thrombolysis in Cerebral
Infarction grade of 2b or 3. Follow-up MRI, including Statistical Analysis
diffusion-weighted imaging and MR angiography, were To determine if there was an association between the
performed routinely at around 24 hours after the EVT FPCT-ASPECTS and clinical outcome after EVT, pa-
procedure. tients were grouped according to clinical outcome. In the
The institutional review board approved this study and present study, clinical outcome was evaluated based on
waived the requirement of informed consent for study in- functional outcome and malignant infarction.
clusion based on its retrospective design. Functional outcome was considered favorable if the

J Neurosurg Volume 135 • September 2021 705

 
* | D*2 *33 *
Baek et al.

mRS score at 3 months after EVT was ≤ 2. Based on this

threshold, patients were assigned to the favorable outcome
or the unfavorable outcome group. First, demographic
factors, common risk factors for stroke, clinical and pro-
cedural information, and the FPCT-ASPECTS were com-
pared between groups. The Student t-test, Mann-Whitney
U-test, chi-square test, and Fisher’s exact test were used for
comparison. To identify independent variables associated
with a favorable outcome, multivariable binary logistic re-
gression analysis was performed. Variables with p < 0.10
in univariable analysis were entered into the multivariable
model. Second, to determine the predictive power of the
FPCT-ASPECTS for a favorable outcome, we performed
receiver operating characteristic (ROC) curve analysis
and calculated the area under the ROC curve (AUC) of
the FPCT-ASPECTS for a favorable outcome. An opti-
mal cutoff point of the FPCT-ASPECTS for a favorable
outcome was determined by the Youden index. Then, us-
ing DeLong’s test, the AUC of the FPCT-ASPECTS for a
favorable outcome was compared with those of the CT-
ASPECTS and MR-ASPECTS.
The association of the FPCT-ASPECTS with malig-
nant infarction was assessed using the same statistical
approach as described above. For malignant infarction,
patients were assigned to the malignant infarction group
or the no malignant infarction group. Malignant infarc-
tion was defined as brain swelling that was treated with
decompression craniectomy or that caused death without
intracerebral hemorrhage. Brain swelling was assessed
on MR images routinely obtained around 24 hours af-
ter the EVT procedure. If a patient’s neurological status
worsened, follow-up CT or MRI was performed imme-
diately. Additional brain imaging was often performed
serially for high-risk patients to observe the evolution of
brain swelling. All statistical analyses were performed
using SPSS (version 23, IBM Corp.) and MedCalc (ver-
sion 16.8.4) software; p < 0.05 was considered statisti-
cally significant.

Results
A total of 279 patients had both a successful recanali-
zation and FPCT during the study period. FPCT images
could be assessed in 235 patients (mean age 68.1 ± 13.6
years; 50.2% male) (Fig. 1).

FPCT-ASPECTS and a Favorable Outcome

After excluding 16 patients whose 3-month mRS scores FIG. 1. Patient selection flowchart.
were not available, 219 patients were analyzed (Fig. 1).
Among them, 157 (71.7%) were classified as having a fa-
vorable outcome. The mean FPCT-ASPECTS was signifi- point of the FPCT-ASPECTS for a favorable outcome was
cantly higher in patients with a favorable outcome (9.3 ± ≥ 8 (sensitivity 96.8%, specificity 54.8%).
0.9) than in those with an unfavorable outcome (6.7 ± 2.6)
(p < 0.001; Table 1 and Fig. 2). Along with a lower NIHSS FPCT-ASPECTS and Malignant Infarction
score, the FPCT-ASPECTS was an independent factor The mean FPCT-ASPECTS was significantly lower in
for a favorable outcome (OR 3.28, 95% CI 2.15–5.01; p patients with malignant infarction (3.4 ± 2.9) than in pa-
< 0.001). The AUC of the FPCT-ASPECTS for a favor- tients without malignant infarction (8.8 ± 1.4) (p < 0.001;
able outcome was 0.862 (95% CI 0.809–0.905; p < 0.001), Table 2). Furthermore, the FPCT-ASPECTS was an inde-
which was significantly greater than that of the CT-AS- pendent factor for malignant infarction (OR 0.42, 95% CI
PECTS (0.637; p < 0.001) and similar to that of the MR- 0.31–0.57; p < 0.001). The AUC of the FPCT-ASPECTS
ASPECTS (0.853; p = 0.983; Fig. 3A). An optimal cutoff for malignant infarction was 0.906 (95% CI 0.861–0.940;

706 J Neurosurg Volume 135 • September 2021

 
* | D*2 *33 *
 Baek et al.

TABLE 1. Comparison of variables and FPCT-ASPECTS between patients with and those without a favorable outcome
Univariable Analysis Multivariable Analysis
All Patients Favorable Outcome Unfavorable Outcome OR
(n = 219) (n = 157) (n = 62) p Value (95% CI)* p Value
Age, yrs 68.1 ± 13.7 66.0 ± 13.9 73.1 ± 11.7 <0.001 0.96 (0.92–1.00) 0.057
Male sex 112 (51.1) 84 (53.5) 28 (45.2) 0.295
Hypertension 121 (55.3) 88 (56.1) 33 (53.2) 0.764
Diabetes 39 (17.8) 25 (15.9) 14 (22.6) 0.246
Smoking 26 (11.9) 21 (13.4) 5 (8.1) 0.356
Atrial fibrillation 86 (39.3) 56 (35.7) 30 (48.4) 0.092 0.67 (0.28–1.59) 0.363
Coronary artery disease 26 (11.9) 19 (12.1) 7 (11.3) 0.999
Initial NIHSS score 15.0 [7.0, 23.0] 14.0 [6.7, 21.3] 18.0 [12.0, 24.0] <0.001 0.83 (0.75–0.93) 0.001
Intravenous tPA 102 (46.6) 74 (47.1) 28 (45.2) 0.881
ICA occlusion 85 (38.8) 54 (34.4) 31 (50.0) 0.045 0.92 (0.36–2.32) 0.851
Time profile, mins
 Onset-to-image 238 ± 284 232 ± 465 251 ± 288 0.664
 Image-to-puncture 89 ± 67 90 ± 73 87 ± 43 0.807
 Puncture-to-recanalization 52 ± 38 46 ± 34 67 ± 42 <0.001 0.99 (0.98–1.01) 0.533
 Onset-to-recanalization 380 ± 298 368 ± 299 406 ± 294 0.399
CT-ASPECTS 9.3 ± 1.1 9.5 ± 0.9 8.9 ± 1.4 0.001 1.33 (0.90–1.97) 0.157
FPCT-ASPECTS 9.0 ± 2.0 9.3 ± 0.9 6.7 ± 2.6 <0.001 3.28 (2.15–5.01) <0.001
Univariate results are expressed as mean ± SD, the number of patients (%), or median [IQR].
* Odds ratio for a favorable outcome.

p < 0.001) with an optimal cutoff point of ≤ 5 (sensitiv- recanalization, and a patient’s collateral status, to mention
ity 82.4%, specificity 97.3%). The AUC of the FPCT- a few factors.6,8–11 Although many factors are well known
ASPECTS was significantly greater than that of the CT- to be associated with clinical outcomes after EVT, it is
ASPECTS (0.552; p = 0.001; Fig. 3B). not easy to draw a conclusion by integrating these factors.
Successful recanalization is also a strong determinant of a
Discussion patient’s clinical outcome. Nowadays, most patients have
a successful recanalization with advances in endovascular
In this study, we found that the FPCT-ASPECTS was technique. However, despite the successful recanalization,
significantly associated with favorable outcome and ma- a favorable outcome cannot be achieved in a considerable
lignant infarction in patients with successful recanaliza- number of patients. Thus, something more direct and prac-
tion by EVT. More importantly, the FPCT-ASPECTS was tical is required to immediately predict clinical prognosis
significantly more predictive of a favorable outcome and in patients with a successful recanalization.
malignant infarction than the CT-ASPECTS. The FPCT- Postprocedural CT findings may be useful for pre-
ASPECTS was comparable to the MR-ASPECTS with dicting a patient’s clinical prognosis. Conventional CT is
regard to prediction of a favorable outcome. To the best primarily performed after an EVT procedure to observe
of our knowledge, this is the first study of hyperattenua- hemorrhagic conditions such as procedure-related intra-
tion on FPCT focusing on patients’ clinical outcomes and cranial hemorrhage or hemorrhagic transformation. Fur-
the predictive ability of FPCT; most previous studies have thermore, hyperattenuation on postprocedural conven-
focused on the association between hyperattenuation on tional CT images is a known risk factor for subsequent
imaging and hemorrhagic risk. hemorrhagic transformation.3,12–14 According to previous
studies, a substantial portion of hyperattenuation is due
Hyperattenuation on Postprocedural CT and Clinical to contrast staining, not hemorrhage. Contrast staining on
Prognosis postprocedural conventional CT results from blood-brain
Early prediction of a patient’s clinical prognosis is im- barrier disruption, which is in turn caused mostly by is-
portant in acute stroke management. Physicians can make chemic injury of the affected region. In other words, re-
short- or long-term management plans and monitor possible gions of contrast staining on postprocedural CT indicate
detrimental clinical events based on the prediction results. the extent of blood-brain barrier disruption and may be
In routine clinical practice, malignant infarctions develop a surrogate of the final infarct core volume or infarction
in the acute period, and later functional outcomes require area. In support of this, the extent of hyperattenuation on
prediction. Clinical outcomes after EVT have been pre- postprocedural conventional CT images has been shown
dicted using common prognostic factors such as stroke ex- to match the final infarct volume well.1,2 FPCT is read-
tent based on initial brain images, time from stroke onset to ily assessable and is widely performed during or at the

J Neurosurg Volume 135 • September 2021 707

 
* | D*2 *33 *
Baek et al.

FIG. 2. Distribution of functional outcome according to the ASPECTS on FPCT (FPCT-ASPECTS).

end of endovascular procedures for the same purposes as We think that the postprocedural findings from FPCT
postprocedural conventional CT. FPCT can detect post- could be also applied to findings from a conventional or
procedural hemorrhagic conditions equally as well as con- portable CT scanner.
ventional CT.15,16 The extent of hyperattenuation on FPCT FPCT can be practically useful in a few ways. First, by
is also associated with subsequent hemorrhagic transfor- performing FPCT at the end of the EVT procedure, pro-
mation and final infarct volume.17,18 If hyperattenuation on cedure-related intracranial hemorrhage can be evaluated.
conventional CT or FPCT is significantly associated with Hemorrhagic transformation that develops during or as a
final infarct volume, the extent of hyperattenuation may be result of EVT can also be observed. In cases in which cer-
predictive of a patient’s clinical prognosis. vical carotid stenting, intracranial stenting, or glycoprotein
IIb/IIIa infusion is required, hemorrhagic conditions can
Advantages of Using FPCT After EVT easily be detected by performing FPCT during the EVT
procedure. Although clinical or neurological examination
In addition to its clinical relevance, FPCT has several findings could also hint at intracranial hemorrhage, neuro-
advantages over conventional CT. FPCT is easier to per- logical deterioration might possibly be due to other condi-
form; it can be performed on the same angiography table tions, such as the progression of ischemic stroke, metabolic
without transferring the patient to the CT room. Although derangement, and unstable vital signs. In addition, unrec-
the burden in transporting a patient to the CT room has ognized intracranial hemorrhage should be found in pa-
been reduced by the recent availability of a portable CT tients who need immediate antithrombotic medications af-
scanner, FPCT is still more convenient. Furthermore, ter or during EVT. Second, by evaluating hyperattenuation
FPCT can be repeated at any time during the endovascu- on FPCT at the end of the EVT procedure, a patient’s func-
lar procedure. For this reason, FPCT performed at the end tional outcome, in addition to the possibility of malignant
of the EVT procedure is starting to replace conventional infarction, can be predicted at an earlier stage. This could
CT performed after the EVT procedure. In addition, by help physicians monitor and prepare for further detrimen-
performing FPCT at the end of the EVT procedure, FPCT tal clinical events. With this clinical utilization, our center
might show hyperattenuation more optimally. Because hy- has performed FPCT in cases of EVT. After every proce-
perattenuation on CT might attenuate even over a short pe- dure, we perform FPCT scanning rather than CT scanning.
riod of time, the timing of CT scanning is important. Even Although the AUC of the FPCT-ASPECTS in predict-
on conventional CT scanning performed approximately 1 ing patient’s functional outcome is similar to that of the
hour after the EVT procedure, a considerable area of the MR-ASPECTS, we do not think it directly means that MR
infarction can become iso- or hypoattenuated.19 However, images obtained after EVT can be replaced with FPCT.
our findings in this study do not mean that FPCT is the Generally, MR images have their own role in acute stroke
only CT modality to show the predictability of clinical management: to define the exact extent of acute stroke
outcomes after EVT. If postprocedural CT findings need and involved anatomical structures, to detect hemorrhagic
to be evaluated, any type of CT scanner can be considered. transformation more sensitively, and to evaluate the follow-

708 J Neurosurg Volume 135 • September 2021

 
* | D*2 *33 *
 Baek et al.

FIG. 3. ROC curves for a favorable outcome and malignant infarction. A: ROC curves of the FPCT-ASPECTS, MR-ASPECTS, and
initial noncontrast CT (CT-ASPECTS) for a favorable outcome. The AUC of the FPCT-ASPECTS was 0.862, which was signifi-
cantly greater than that of the CT-ASPECTS (0.637) (p < 0.001). The AUC of the FPCT-ASPECTS was comparable to that of the
MR-ASPECTS (0.853; p = 0.983). B: ROC curves of the FPCT-ASPECTS and CT-ASPECTS for malignant infarction. The AUC of
the FPCT-ASPECTS (0.906) was significantly greater than that of the CT-ASPECTS (0.552) (p = 0.001).

up patency of the recanalized artery. In addition to the infor- core progresses until recanalization is achieved, the final
mation about intra- or postprocedural hemorrhage, FPCT infarct core might be significantly different from the ini-
might be clinically referred to predict a patient’s clinical tial infarct core, depending on the status of the collateral
outcome in advance to obtain MR images after EVT. vessels and the elapsed time from initial CT to recana-
Although the ASPECTS on initial CT images (CT- lization. For a patient with an initially large infarct core,
ASPECTS) is an important prognostic factor for a pa- the CT-ASPECTS might be predictive of a worse clinical
tient’s functional outcome, the CT-ASPECTS was less outcome. However, the CT-ASPECTS might not predict
predictive than the FPCT-ASPECTS in our study. First, a good clinical outcome even if a patient has an initially
the CT-ASPECTS was originally devised to aid in de- small infarct core. On the contrary, the FPCT-ASPECTS
termining treatment eligibility, not for assessing clinical appears to reflect the volume of the final infarct core, as
outcome. Second, the CT-ASPECTS only reflects the ex- demonstrated by an AUC that is comparable to that of the
tent of the infarct core before EVT. Because the infarct follow-up MR-ASPECTS.

TABLE 2. Comparison of variables and FPCT-ASPECTS between patients with and those without malignant infarction
Univariable Analysis Multivariable Analysis
All Patients (n = 235) Malignant Infarction (n = 17) No Malignant Infarction (n = 218) p Value OR (95% CI)* p Value
Age, yrs 68.1 ± 13.6 66.7 ± 12.3 68.2 ± 13.7 0.675
Male sex 118 (50.2) 10 (58.8) 108 (49.5) 0.616
Hypertension 131 (55.7) 8 (47.1) 123 (56.4) 0.461
Diabetes 43 (18.3) 1 (5.9) 42 (19.3) 0.324
Smoking 28 (11.9) 2 (11.8) 26 (11.9) 0.999
Atrial fibrillation 95 (40.4) 6 (35.3) 89 (40.8) 0.799
Coronary artery disease 28 (11.9) 2 (11.8) 26 (11.9) 0.999
Initial NIHSS score 15.0 [7.0, 23.0] 14.0 [4.0, 24.0] 15.0 [7.0, 23.0] 0.019 1.03 (0.88–1.21) 0.695
Intravenous tPA 107 (45.5) 9 (52.9) 98 (45.0) 0.616
ICA occlusion 91 (38.7) 11 (64.7) 80 (36.7) 0.036 2.07 (0.41–10.5) 0.379
CT-ASPECTS 8.9 ± 1.3 3.4 ± 2.9 9.3 ± 1.1 0.028 1.02 (0.65–1.59) 0.932
FPCT-ASPECTS 9.0 ± 2.0 3.4 ± 2.9 8.8 ± 1.4 <0.001 0.42 (0.31–0.57) <0.001
Univariate results are expressed as mean ± SD, the number of patients (%), or median [IQR].
* Odds ratio for malignant infarction.

J Neurosurg Volume 135 • September 2021 709

 
* | D*2 *33 *
Baek et al.

Limitations mon use of FPCT to evaluate hemorrhagic conditions im-

Contrast material–related hyperattenuation (contrast mediately after EVT, FPCT can help immediately predict
staining) can be categorized as contrast extravasation or clinical outcomes and thereby aid in acute management
contrast enhancement.5 Compared with contrast extrava- after EVT.
sation, contrast enhancement on postprocedural CT is
known to be less hyperattenuating, is confined to an ana- Acknowledgments
tomical border without mass effect, and disappears on This research was supported by a grant from the Korea Health
follow-up images around 24 hours. Contrast enhancement Technology R&D Project through the Korea Health Industry
was not associated with hemorrhagic lesions, whereas Development Institute (KHIDI), funded by the Ministry of Health
contrast extravasation was a negative prognostic sign and and Welfare, Republic of Korea (HC15C1056).
highly associated with intracerebral hemorrhage. Howev-
er, we did not discriminate between contrast extravasation References
and contrast enhancement for this study. First, for earlier   1. Amans MR, Cooke DL, Vella M, et al. Contrast staining on
prediction of a patient’s clinical outcomes, we wanted to CT after DSA in ischemic stroke patients progresses to in-
establish the ability of FPCT to predict outcomes right af- farction and rarely hemorrhages. Interv Neuroradiol. 2014;​
ter the EVT procedure, when contrast extravasation and ​
20(1):106–115.
contrast enhancement cannot be distinguished reliably.   2. Antonucci MU, Mocco J, Bennett JA. New insight into tran-
Second, as our hypothesis was based on the pathomecha- sient contrast enhancement on computed tomography after
nism of blood-brain barrier disruption by ischemic injury, endovascular treatment of stroke. Interv Neuroradiol. 2012;​
​
18(3):303–308.
the principal observation in this study should include both   3. Nakano S, Iseda T, Kawano H, et al. Parenchymal hyperden-
contrast extravasation and contrast enhancement. The two sity on computed tomography after intra-arterial reperfusion
types of contrast staining might simply reflect the sever- therapy for acute middle cerebral artery occlusion:​incidence
ity of blood-brain barrier damage rather than a completely ​ ​
and clinical significance. Stroke. 2001;32(9):2042–2048.
different pathomechanism.5,20 Considering that iatrogenic   4. Wildenhain SL, Jungreis CA, Barr J, et al. CT after intracra-
intracranial hemorrhage due to vessel rupture is very rare nial intraarterial thrombolysis for acute stroke. AJNR Am J
during the EVT procedure, hyperattenuation on FPCT, re- ​ ​
Neuroradiol. 1994;15(3):487–492.
  5. Yoon W, Seo JJ, Kim JK, et al. Contrast enhancement and
gardless of its severity, should develop in the area of ische- contrast extravasation on computed tomography after intra-
mic injury to the blood-brain barrier. In fact, mingled (or arterial thrombolysis in patients with acute ischemic stroke.
undiscriminating) hyperattenuation on postprocedural CT ​ ​
Stroke. 2004;35(4):876–881.
was shown to be significantly associated with final infarct   6. Kim BM, Baek JH, Heo JH, et al. Collateral status affects the
volume in previous studies.2,13,​17,18 Therefore, as our and onset-to-reperfusion time window for good outcome. J Neu-
previous studies have shown, hyperattenuation on FPCT ​ ​
rol Neurosurg Psychiatry. 2018;89(9):903–909.
appears to reflect the final infarct volume.   7. Pexman JH, Barber PA, Hill MD, et al. Use of the Alberta
Stroke Program Early CT Score (ASPECTS) for assessing
This study was retrospectively conducted in a single CT scans in patients with acute stroke. AJNR Am J Neurora-
stroke center. Although FPCT has been recommended ​ ​
diol. 2001;22(8):1534–1542.
for patients who undergo EVT at our institution, its per-   8. Kim BJ, Chung JW, Park HK, et al. CT angiography of col-
formance was not forced under a specific protocol. Thus, lateral vessels and outcomes in endovascular-treated acute
FPCT was not performed in all patients who underwent ​ ​
ischemic stroke patients. J Clin Neurol. 2017;13(2):121–128.
EVT. However, the omission of FPCT was mostly acci-   9. Saver JL, Goyal M, van der Lugt A, et al. Time to treatment
dental rather than intentional. Fortunately, basic demo- with endovascular thrombectomy and outcomes from ische-
mic stroke:​a meta-analysis. JAMA. 2016;316(12):1279–1288.
​ ​
graphics and clinical outcomes were not significantly 10. Sheth SA, Sanossian N, Hao Q, et al. Collateral flow as caus-
different between patients with FPCT and those without ative of good outcomes in endovascular stroke therapy. J
FPCT (Supplemental Table). Furthermore, our goal was to ​
Neurointerv Surg. 2016;8(1):2–7. ​
evaluate the ability of FPCT to predict patients’ clinical 11. Yoo AJ, Zaidat OO, Chaudhry ZA, et al. Impact of pretreat-
outcomes, not to determine clinical outcomes according ment noncontrast CT Alberta Stroke Program Early CT
to FPCT’s performance. Thus, the fact that FPCT was not Score on clinical outcome after intra-arterial stroke therapy.
performed in all patients is unlikely to have significantly ​ ​
Stroke. 2014;45(3):746–751.
12. Kim JT, Heo SH, Cho BH, et al. Hyperdensity on non-con-
affected our findings. trast CT immediately after intra-arterial revascularization. J
In addition, the clinical significance of cutoff points ​ ​
Neurol. 2012;259(5):936–943.
of the FPCT-ASPECTS for clinical outcomes might be 13. Lummel N, Schulte-Altedorneburg G, Bernau C, et al. Hyper-
a little bit limited. Although these points were calculated attenuated intracerebral lesions after mechanical recanalization
under statistical significance, an internal or external vali- ​ ​
in acute stroke. AJNR Am J Neuroradiol. 2014;35(2):345–351.
dation process was not performed in this study. Further 14. Parrilla G, García-Villalba B, Espinosa de Rueda M, et al.
validation study is necessary in a prospective or multi- Hemorrhage/contrast staining areas after mechanical intra-
arterial thrombectomy in acute ischemic stroke:​imaging
center-based cohort. findings and clinical significance. AJNR Am J Neuroradiol.
​ ​
2012;33(9):1791–1796.
Conclusions 15. Payabvash S, Khan AA, Qureshi MH, et al. Detection of
intraparenchymal hemorrhage after endovascular therapy in
The FPCT-ASPECTS was highly predictive of a favor- patients with acute ischemic stroke using immediate postpro-
able outcome and malignant infarction in patients with a cedural flat-panel computed tomography scan. J Neuroimag-
successful recanalization by EVT. In addition to the com- ​ ​
ing. 2016;26(2):213–218.

710 J Neurosurg Volume 135 • September 2021

 
* | D*2 *33 *
 Baek et al.

16. Psychogios MN, Buhk JH, Schramm P, et al. Feasibility of Author Contributions
angiographic CT in peri-interventional diagnostic imaging:​a Conception and design: BM Kim. Acquisition of data: all authors.
comparative study with multidetector CT. AJNR Am J Neuro- Analysis and interpretation of data: all authors. Drafting the
radiol. 2010;​31(7):​1226–1231. article: BM Kim, Baek. Critically revising the article: all authors.
17. Rouchaud A, Pistocchi S, Blanc R, et al. Predictive value of Reviewed submitted version of manuscript: all authors. Approved
flat-panel CT for haemorrhagic transformations in patients the final version of the manuscript on behalf of all authors: BM
with acute stroke treated with thrombectomy. J Neurointerv Kim. Statistical analysis: BM Kim, Baek. Study supervision: BM
Surg. 2014;​6(2):​139–143. Kim.
18. Schneider T, Mahraun T, Schroeder J, et al. Intraparenchymal
hyperattenuations on flat-panel CT directly after mechani- Supplemental Information
cal thrombectomy are restricted to the initial infarct core on
diffusion-weighted imaging. Clin Neuroradiol. 2018;​28(1):​ Online-Only Content
91–97. Supplemental material is available with the online version of the
19. Dekeyzer S, Reich A, Othman AE, et al. Infarct fogging on article.
immediate postinterventional CT-a not infrequent occur- Supplemental Table. https://thejns.org/doi/suppl/​10.3171/​
rence. Neuroradiology. 2017;​59(9):​853–859. 2020.7.JNS193214.
20. Hamann GF, Okada Y, del Zoppo GJ. Hemorrhagic trans-
formation and microvascular integrity during focal cerebral Correspondence
ischemia/reperfusion. J Cereb Blood Flow Metab. 1996;​16(6):​ Byung Moon Kim: Yonsei University College of Medicine, Seoul,
1373–1378. Republic of Korea. [email protected].

Disclosures
The authors report no conflict of interest concerning the materi-
als or methods used in this study or the findings specified in this
paper.

J Neurosurg Volume 135 • September 2021 711

 
* | D*2 *33 *
 CLINICAL ARTICLE
J Neurosurg 135:712–721, 2021

Functional DTI tractography in brainstem cavernoma

surgery
Anna Zdunczyk, MD, Fabia Roth, Thomas Picht, MD, and Peter Vajkoczy, MD

Department of Neurosurgery, Charité–Universitätsmedizin Berlin, Germany

OBJECTIVE  Surgical resection of brainstem cavernomas is associated with high postoperative morbidity due to the
density of local vulnerable structures. Classical mapping of pathways by diffusion tensor imaging (DTI) has proven to be
unspecific and confusing in many cases. In the current study, the authors aimed to establish a more reliable, specific,
and objective method for somatotopic visualization of the descending motor pathways with navigated transcranial mag-
netic stimulation (nTMS)–based DTI fiber tracking.
METHODS  Twenty-one patients with brainstem cavernomas were examined with nTMS prior to surgery. The resting
motor threshold (RMT) and cortical representation areas of hand, leg, and facial function were determined on both
hemispheres. Motor evoked potential (MEP)–positive stimulation spots were then set as seed points for tractography.
Somatotopic fiber tracking was performed at a fractional anisotropy (FA) value of 75% of the individual FA threshold.
RESULTS  Mapping of the motor cortex and tract reconstruction for hand, leg, and facial function was successful in all
patients. The somatotopy of corticospinal and corticonuclear tracts was also clearly depicted on the brainstem level.
Higher preoperative RMT values were associated with a postoperative motor deficit (p < 0.05) and correlated with a
lower FA threshold (p < 0.05), revealing structural impairment of the corticospinal tract (CST) prior to surgery. In patients
with a new deficit, the distance between the lesion and CST was below 1 mm.
CONCLUSIONS  nTMS-based fiber tracking enables objective somatotopic tract visualization on the brainstem level
and provides a valuable instrument for preoperative planning, intraoperative orientation, and individual risk stratification.
nTMS may thus increase the safety of surgical resection of brainstem cavernomas.
https://thejns.org/doi/abs/10.3171/2020.7.JNS20403
KEYWORDS  brainstem cavernoma; diffusion tensor tractography; diffusion tensor imaging; navigated transcranial
magnetic stimulation; vascular disorders

C
avernous malformations are rare vascular lesions of In patients with lesions that are deep seated or inaccessible
the central nervous system with an estimated preva- from safe entry zones, a conservative treatment is advocat-
lence of 0.4%–0.8%,1–3 accounting for 10%–20% of ed due to the high risk of perioperative morbidity.5,8
all vascular malformations. Up to one-third4,5 are brain- Modern neuroimaging techniques based on diffusion
stem cavernous malformations (BSCMs), which arise in MRI delineate the course of white matter pathways based
the brainstem and are associated with higher hemorrhage on water molecule diffusion along neuronal axons.9,10 By
(2.3%–13.6%1,6) and repeat hemorrhage (15%–60.9%1,6) exploration of each white matter tract voxel by voxel, a
rates. Surgery is the primary treatment method, particu- predictable pattern of diffusion and thus tract orientation
larly in symptomatic patients presenting with recurrent in a 3D space can be displayed.11,12 Diffusion tensor trac-
bleeding. The complex anatomy of vulnerable structures tography (DTT) is traditionally performed through ana-
on the brainstem level, however, makes surgical resections tomically predefined regions of interest (ROIs). These are
technically demanding and leads to a high rate of postoper- used as starting points to track down specific fiber tracts at
ative morbidity. Symptomatic patients with lesions abutting a specific fractional anisotropy threshold (FAT). The FA
the pial or ependymal surface or present with anatomically value is a parameter of mean diffusivity, i.e., the direction
safe entry zones are the primary candidates for surgery.5,7 dependence of a diffusion process.11,13 This can therefore

ABBREVIATIONS  BSCM = brainstem cavernous malformation; CBT = corticobulbar tract; CST = corticospinal tract; DTI = diffusion tensor imaging; DTT = diffusion tensor
tractography; FA = fractional anisotropy; FAT = FA threshold; FDI = first digital interosseus muscle; MEP = motor evoked potential; MRC = Medical Research Council; mRS
= modified Rankin Scale; nTMS = navigated transcranial magnetic stimulation; RMT = resting motor threshold; ROI = region of interest.
SUBMITTED  February 9, 2020.  ACCEPTED  July 13, 2020.
INCLUDE WHEN CITING  Published online December 25, 2020; DOI: 10.3171/2020.7.JNS20403.

712 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *34 *
 Zdunczyk et al.

also be used as a measure of diffusion along neuronal ax- section thickness T1 inversion recovery (TR/TE/TI/α 7.8
ons, indicating tract integrity. msec/3.1 msec/500 msec/16°) 3D gradient-echo sequence
Recent studies have pointed out the benefit of diffusion (IR 3D-FSPGR). A diffusion-weighted single-shot echo-
tensor imaging (DTI) in determining the association be- planar sequence along 23 different geometrical directions
tween the lesion and the white matter tracts in order to at a b value of 1000 sec/mm2 was repeated to sample a
establish the best neurosurgical approach8,14–18 in deep- diffusion tensor. Scan parameters were TR/TE 11,000
seated lesions. However, no clear correlation between DTI msec/83 msec, matrix size 128 × 128, and FOV 240 × 240
parameters and postoperative outcomes has been demon- mm.
strated.16,17,19
Navigated transcranial magnetic stimulation (nTMS) is Navigated Transcranial Magnetic Stimulation
a noninvasive tool that creates a map of functional motor All patients underwent preoperative nTMS brain map-
areas and allows the evaluation of cortical and corticospinal ping (eXimia, Nexstim) using a high-precision, figure-
excitability.20,21 During the last decade, nTMS has proven to eight stimulation coil, combined with MRI-based neu-
be a valuable tool for preoperative motor mapping in brain ronavigation and analytical software to deliver biphasic
tumors located in eloquent motor and speech areas.22–25 magnetic stimulation to spots on the motor cortex. The
nTMS-based fiber tracking was recently introduced muscle output was recorded on the system’s integrated
in order to establish an objective algorithm to visualize electromyogram using surface electrodes (Neuroline 720,
motor pathways derived from functional data.26 Here, the Ambu) attached to the mentalis muscle, first digital inter-
“motor-positive” nTMS points are used as seed points to osseus muscle (FDI), and tibialis anterior muscle bilater-
track functionally relevant motor pathways for hand, leg, ally.
and facial function. Compared with the traditional ap-
proach, this protocol allows for a specific tractography of Resting Motor Threshold
each cortical region. In a cohort of patients with tumors The resting motor threshold (RMT) serves as a mea-
located in or adjacent to the primary motor cortex, this sure of cortical excitability. The hotspots for the FDI and
method led to a change in operative strategy in nearly half respective RMT were determined as previously described
of the patients.26 for both hemispheres.27,28 The RMT was defined by an au-
The aim of the present study was to evaluate whether tomated algorithm approximating the lowest output inten-
nTMS-based fiber tracking is also beneficial for preopera- sity producing reliable motor evoked potentials (MEPs) ≥
tive planning and risk stratification in the surgery of brain- 50 µV with a 95% confidence interval of ± 2.0.
stem cavernomas.
Diffusion Tensor Tractography
Methods Visualization of tracts was performed according to our
The study proposal was approved by the local ethics previously published algorithm on nTMS-based tractogra-
commission. Patients provided written informed consent phy11,13,29 using a commercial software package (Elements,
for all medical evaluations and treatments. Brainlab; Fig. 1). Briefly, nTMS spots that elicited an MEP
on the cortical surface for hand, face, and leg muscles
Patient Selection were imported into the DTI software and enlarged to a
diameter of 3 mm to create a continuous seed point area.
Twenty-one patients (mean age 44 years, range 25–67 Then another ROI was set according to the color-coded
years; 11 female) with BSCMs were enrolled in this study. conventional FA map on the level of the pons, including
The radiological diagnosis was assessed by an experienced the corticospinal or corticobulbar projections.30 The FAT
neuroradiologist. Exclusion criteria were other pathologies was determined for each ROI by stepwise increases of the
in the vicinity of the corticospinal tract (CST) above the FA value until the complete disappearance of tracts. Fiber
lesion site (i.e., tumor or infarction), high-grade paresis of tracking was then performed at 75% FAT using an antero-
the upper extremity (grade < 3, British Medical Research grade direction with a vector step length of 1.6 mm, an
Council [MRC] scale for muscle strength), neuroinflam- angular threshold of 30°, and a minimum tract length of
matory or neurodegenerative disease, and pregnancy. 110 mm. The resulting tracts were then classified depend-
ing on their radiological appearance as 1) normal, 2) dis-
Assessment of Clinical Neurological Status placed but intact, 3) deformed, or 4) interrupted.17,31 Then,
At the time of enrollment, all patients underwent neu- the shortest distance between the lesion and the CST/cor-
rological examination. The severity of paresis was as- ticobulbar tract (CBT) was measured. After surgical re-
sessed based on the MRC scale (grade range 0–5). Patient section, the preoperative nTMS-DTI plan was fused with
disability in daily life practice was categorized according the postoperative MRI in order to analyze the condition of
to the modified Rankin Scale (mRS; score range 0–6). tracts beneath the resection cavity.
Neurological recovery was reassessed after 3 months in
the outpatient clinic. Surgical Management
The decision for surgical treatment was individually
MRI Protocol tailored to every patient depending on lesion localization
Conventional MRI was performed on a 1.5-T or 3-T and associated anatomical structures. The nTMS-DTI re-
imaging unit (Siemens 3 T, GE Healthcare) with a 1.0-mm sults were used as additional information to evaluate the

J Neurosurg Volume 135 • September 2021 713

 
* | D*2 *34 *
Zdunczyk et al.

FIG. 1. Workflow for nTMS-based DTI fiber tracking. A: nTMS determination of primary motor area (hand, leg, face). B: Transfer
of nTMS-based seed points. C: Individual FAT determination. D: Somatotopy of tracts on brainstem level (orange, CST arm; blue,
CST leg; pink, CBT). Figure is available in color online only.

risk for perioperative morbidity and determine the surgi- patients with complete remission of symptoms (6/6) after
cal approach that would spare the CST. During surgery, thorough consultation about the surgical risks and natural
the tractography results were presented via a heads-up history of the disease. Further criteria were close anatomi-
display outline. Additional neuromonitoring procedures cal relation to functional tracts (3/6 patients), deep-seat-
in the form of motor and sensory evoked potentials and ed location (2/6), pregnancy (1/6), and personal decision
electrophysiological monitoring of the cranial nerves were (against surgeon’s advice, 1/6).
performed to guide the operative approach and microsur- In the surgical group, 53% of the lesions presented
gical resection. with a pial surface, the other 47% were deep seated. A
gross-total resection was achieved in 14 patients (93%).
Statistical Analysis Postoperatively, only 1 patient (7%) encountered a new
The data are expressed as mean ± SEM. Comparison motor deficit, and 3 patients (20%) experienced worsening
of groups was done with t-testing. While for dependent of preoperative paresis. At the 3-month follow-up, all 6
variables a paired design was considered, in nonparamet- patients reported a partial regression of motor weakness.
ric data the Wilcoxon signed-rank test was applied. Rela- A new cranial nerve deficit was seen in 2 patients (13%),
tionships between FAT and RMT, as well as motor status but no new sensory deficit was reported (Table 1). At the
and FAT and RMT, were tested with Pearson correlation follow-up examination, most patients presented with an
analysis. excellent clinical status (mRS score 0–2 in 50% of the pa-
To assess predictive values for a new motor deficit, we tients, average mRS score 2.2; Table 1). In 28% of the pa-
performed a multiple logistic regression analysis with new tients the preoperative mRS score improved, and in 43%
motor deficit as the dependent variable and RMT, FAT, it remained stable.
distance to CST > 5 mm, and tract disruption as the in-
dependent variables. Differences were considered statisti- Navigated Transcranial Stimulation
cally significant at p < 0.05 (GraphPad Prism 8.0.2). Resting Motor Threshold
We compared the RMTs of patients with lateralized le-
Results sions divided into two groups, patients with and those with-
Clinical Results out a preoperative deficit. Laterality was determined by the
A total of 21 patients (age range 24–66 years, mean 42.8 distance between the lesion and the CST. From previous
years) were enrolled in this study. At admission, most of studies on supratentorial lesions, it is known that a higher
the patients presented in a good clinical condition (average RMT correlates with a worse postoperative outcome.30,32
mRS score 1.9), mainly complaining about cranial nerve In this study, patients presenting with a preoperative mo-
deficits (12/21 patients), followed by sensory disturbances tor weakness showed significantly higher RMT values
(7/21) and motor weakness of the limbs (6/21). Microsurgi- and thus lower cortical excitability than those without (p <
cal resection was performed in 15 patients (Table 1). All of 0.05). This was also true for all patients with a postopera-
the surgical patients were symptomatic prior to admission tive motor deficit (p < 0.05; Fig. 2A). In the subgroup of pa-
and had signs of bleeding around the cavernous lesion on tients who developed a new or worsened motor deficit, the
cranial MRI. Based on the additional information gained preoperatively assessed RMT was also higher than in the
from the nTMS-DTI results, operative resection was per- patients who remained stable (51.3 vs 37.7, p = 0.06). This
formed in 5 patients and conservative treatment in 2 pa- comparison, however, did not reach significance, which is
tients. Six patients received conservative treatment. The probably mainly due to the small sample size of patients
mutual decision to withdraw from surgery was made in with postoperative motor deterioration (n = 4).

714 J Neurosurg Volume 135 • September 2021

 
* | D*2 *34 *
 TABLE 1. Clinical data for the surgical group patients

Patient Age Distance to Motor Status (MRC grade) mRS Score CST Dist, Tract Integrity
No. (yrs)/Sex Level, Position Surface, mm Op Approach Preop Postop 3-Mo FU Preop 3-Mo FU RMT, %* FAT* mm† (CST)
1 44/F Pons, dorsal 1 Kawase 4+ 4+ 5 3 1 53 0.15 3 Displaced
2 32/F Mes, ventral 0 Transcall 5 4 4+ 4 2 35 0.34 1 Displaced
3 44/F Mes/thal, ventral 2 Transcall 3 3 4 3 3 50 0.08 0 Displaced
4 31/M Pons, ventral 0 Ant petros 5 5 5 1 0 35 0.38 2.6 Normal
5 25/M Pons, dorsal 0 Med subocc 5 5 5 2 3 26 0.33 2.5 Normal
6 59/M Mo, dorsal 0 Med subocc 5 5 5 1 1 31 0.22 6.7 Normal
7 35/M Pons, dorsal 0 Med subocc 4 4− 4 3 3 64 0.28 0 Interrupted
8 41/F Mes, dorsal 2 Subtemp 5 5 5 2 1 34 0.31 4 Displaced
9‡ 52/F Pons, ventral 0 Retrosigmoid 5 5 5 1 2 33 0.39 1.5 Interrupted
10 42/F Mes/pons, dorsal 0 Supracereb 5 5 5 1 3 47 0.13 3 Displaced
11 40/M Mo, vent-dors 2 Med subocc 5 5 5 2 2 39 0.35 2.1 Displaced
12 35/M Pons/mo, vent-dors 2 Retrosig 4+ 3 4− 2 3 53 0.12 0 Interrupted
13 30/F Pons, ventral 0 Med subocc 4 3 4 2 2 70 0.32 0 Interrupted
14 56/F Mo, dorsal 1.5 Med subocc 5 5 5 1 1 39 0.35 5.3 Displaced

* | D*2 *34 *
15 48/M Mes, ventral 0 Pterional 5 5 5 1 1 43 0.38 4.2 Displaced
Ant = anterior; dist = distance; FU = follow-up; Kawase = approach through the middle fossa; med = medial; mes = mesial; mo = motor cortex; petros = petrosal; subocc = suboccipital; subtemp = subtemporal; supracereb
= supracerebellar; thal = thalamic; transcall = transcallosal; vent-dors = ventrodorsal.
* Affected hemisphere.
 
† Distance between CST and cavernoma.
‡ Partial resection due to intraoperative reduction of MEP amplitude.

J Neurosurg Volume 135 • September 2021

715
Zdunczyk et al.
Zdunczyk et al.

cavity was observed to enclose the tracts on the postoper-

ative MRI, 75% developed a new deficit (see Figs. 4 and 5;
patients 7, 9, 12, and 13 in Table 1). On the contrary, only
25% of the patients showed an aggravation of neurological
function if the tracts were just beneath the lines of resec-
tion (see Figs. 3 and 6; patients 1, 2, 3, 4, and 15 in Table 1).
Multiple logistic regression analysis was then per-
formed to display the association of the four crucial pa-
rameters (RMT, FAT, distance to CST, and tract disrup-
tion) with the dependent variable, a new motor deficit.
This analysis resulted in ORs for the RMT of 0.9 (95%
CI 0.6021–1.097); for the FAT, 0.003 (95% CI 5.646e-
018 to 21,540,537); for the distance to CST, 23.81(95%
CI 0.5818–9085); and for tract integrity, 1.927 (95% CI
0.006588–216.2), with a negative predictive power of 75%
and positive predictive power of 81.8% (p = 0.007, pseudo
R2 = 0.53). According to this model of risk stratification, a
patient with an elevated RMT, low FAT, distance to tracts
< 1 mm, and disruption of tracts would carry a risk of 84%
for a new motor deficit.
Comparisons of the nTMS tractography results with
conventional DTI fiber tracking (predefined anatomical
FIG. 2. A: Comparison of RMT in patients with a new motor deficit (MRC ROI, FA 0.2) are illustrated in Figs. 3–6. Intraoperative
grade < 5) and those without (MRC grade 5). *p < 0.05. B: Correlation neuromonitoring was also used to confirm tractography
of FAT and rRMT, p < 0.05, r 2 = −0.57. C: Comparison of FAT in patients data (Fig. 5). The traditional algorithm led to a higher
with (MRC grade 1) and without (MRC grade 5) a preoperative motor
deficit. *p < 0.05. D: Distance to CST in patients with (MRC grade < 5)
number of fibers. However, in most cases this finding did
and without (MRC grade 5) a postoperative motor deficit. not show any correlation with intraoperative neuromoni-
toring (Fig. 5) or postoperative outcome (Figs. 3–6).

Case Illustrations
Fractional Anisotropy
Case 1
Applying our proposed algorithm, we defined an indi-
vidual FAT for each cortical region separately (mean ± SD This 56-year-old female patient (Fig. 3) presented with
FAT: CST 0.3 ± 0.1, CBT 0.23 ± 0.08). There was a nega- a 2-week history of double vision and vertigo (patient 14
tive correlation between the FAT and RMT values (p < in Table 1). Upon clinical examination, there was no fur-
0.05, r2 = −0.57; Fig. 2B), confirming that alterations of ther deficit. The MRI revealed a hemorrhagic mass on the
FAT reflect disturbance of CST integrity. We furthermore midportion of the medulla oblongata (Fig. 3). The nTMS-
encountered lower FA values in patients presenting with a based tractography depicted the CBT and the CST for the
preoperative motor deficit (0.16 vs 0.32, p < 0.05; Fig. 2C). lower extremity abutting the lesion laterally (Fig. 3B) but
without any fiber disruption (Fig. 3C). The RMT (39%)
nTMS-Based Tractography and FAT (0.35) did not show any aberration (Table 1). The
Motor mapping with definition of cortical areas for the traditional deterministic algorithm displayed the CST (leg)
hand, leg, and face regions was feasible in all patients. The and CBT in the ventrolateral portion of the lesion, which
nTMS coordinates were used as seed points to perform a might suggest a perioperative impairment (Fig. 3E). Based
selective tractography for each of the 3 cortical regions on the nTMS DTI results, a suboccipital craniotomy and
and display a somatotopy of tracts on the brainstem level, dorsomedian brainstem entry point was chosen, sparing
and then the distance between the lesion and tracts was the left CBT (Fig. 3D). After surgery, the patient presented
determined (mean distances: CBT 4.1 mm, CST 4.1 mm). without a new neurological deficit, and her preoperative
Fiber disruption was observed in 13% of the cases and a diplopia resolved gradually.
clear fiber displacement in 67% of the cases. In all of the
patients who presented with a new postoperative motor Case 2
deficit, the distance between the tracts and the lesion was This 35-year-old male patient presented with a history
below 1 mm (sensitivity 100%, specificity 90%, positive of right-sided hemiparesis (MRC grade 4) and facial palsy
predictive value [PPV] 80%; Fig. 2D; also see Figs. 4 and (House-Brackmann grade II, patient 12 in Table 1). There
6). Similarly, in those with the development of a new fa- was a cavernous lesion on the ventral-left part of the pons
cial deficit, the distance to the CBT was 2 mm (sensitivity with signs of subacute bleeding on the preoperative MRI
100%, specificity 67%, PPV 66%). In 75% of the patients (Fig. 4A). nTMS DTI revealed a close anatomical correla-
with interruption of tracts, a new postoperative deficit oc- tion to tracts of both hemispheres (Fig. 4B), in particular
curred, whereas in patients without tract interruption, 54% the left CST/CBT, which also showed partial disruption
of the patients showed intact fibers and 36% showed dis- (Fig. 4C). The elevated RMT value (53%) and low FAT
placed fibers (Table 1). In patients for whom the resection (0.12) suggested an alteration of the left CST (Table 1).

716 J Neurosurg Volume 135 • September 2021

 
* | D*2 *34 *
 Zdunczyk et al.

FIG. 3. Illustrative case 1. A 56-year-old female patient (patient 14 in Table 1) with preoperative double vision and vertigo. A: Cav-
ernous malformation in the central part of the medulla oblongata. B: Distance > 2 mm to all tracts. C: No effect on tracts visible on
preoperative DTI. D: Patient experienced no postoperative deficits. E: Conventional tractography does not display CBT but does
show closeness to leg fibers. The patient did not show any disturbance in leg function in the postoperative period. Figure is avail-
able in color online only.

FIG. 4. Illustrative case 2. A 35-year-old male patient (patient 12 in Table 1) with preoperative hemiparesis (MRC grade 4) and
facial palsy (House-Brackmann grade II). A: Cavernous malformation in the left part of the pons. B: Close anatomical correlation
to the right CBT/CST with partial disruption; distance to CST, 0 mm (yellow arrow). C: Disruption of left CST and CBT. D: Postop-
erative MRI display tracts at the rim of the resection cavity (red star); the patient experienced an aggravation of motor deficit (MRC
grade 3) and facial palsy (House-Brackmann grade III). E: Mainly CST for hand function was found in the vicinity of the lesion.
Nevertheless, the patient’s facial function deteriorated. Figure is available in color online only.

J Neurosurg Volume 135 • September 2021 717

 
* | D*2 *34 *
Zdunczyk et al.

FIG. 5. Illustrative case 3. A 52-year-old female patient (patient 9 in Table 1) with preoperative double vision. A: Cavernous
malformation in the central part of the pons. B: Right and left CSTs adjacent to the lesion; distance of CST from lesion, 1 mm.
Intraoperatively, only a partial resection (red star) was performed due to loss of MEP amplitude for the right hand during surgery
(yellow arrow). C: Disruption and dislocation of CST. D: No new neurological deficit after partial resection (red star). E: Leg CST
within and closest to the lesion, although there were no alterations in leg MEP amplitude during surgery. Figure is available in color
online only.

Deterministic DTI showed closeness mainly to the hand closeness to the CBT and right CST (Fig. 6E). Postopera-
region CST (Fig. 4E). Postoperatively, the patient suffered tively, the patient suffered from severe dysphagia but no
from aggravation of right-sided hemiparesis (MRC grade motor deficit in the extremities, as indicated by the nTMS
3) and facial palsy (House-Brackmann grade III), which DTI data (Fig. 6B and D).
confirmed the nTMS DTI data (Fig. 4B and D).
Case 3
Discussion
This 52-year-old female patient presented with double Due to their critical neuroanatomical location, BSCMs
vision (patient 9 in Table 1). The preoperative MRI dis- that present with hemorrhage are highly associated with
plays a cavernous malformation in the central part of the severe neurological deficits. Bleeding rates vary from 2.4%
pons (Fig. 5A). The preoperative nTMS DTI plan depicts to 4.6% per patient-year and increase up to 5%–34.7% for
both CST adjacent (< 1 mm) to the lesion (Fig. 5B) with rebleeding once bleeding has occurred.2,3,6 Surgical resec-
partial disruption and dislocation of the left CST. Both the tion represents the best treatment method to prevent fur-
RMT (33%) and FAT (0.39) were unremarkable (Table 1). ther hemorrhagic events. Naturally, the highly eloquent lo-
Traditional DTI depicts CST fiber for leg function clos- calization of BSCM is associated with significant periop-
est to the lesion (Fig. 5E). Intraoperatively, only a partial erative morbidity and mortality (5%–27.7% and 0%–6.3%,
resection was possible due to a loss of MEP amplitude of respectively4,33,34), with transient postoperative deficits re-
the right hand region (Fig. 5B, yellow arrow), as previously ported in up to 86% of cases.35,36
indicated in the nTMS DTI plan. After surgery, the patient
remained neurologically intact. DTI in Brainstem Cavernoma
Technical advances in DTI have made this a promis-
Case 4 ing preoperative tool to visualize white matter tracts and
This 42-year-old female patient had a history of double thereby enhance surgical planning and postoperative out-
vision and vertigo (patient 10 in Table 1). Further MRI come.8,16,18 Due to the space-occupying effects of BSCMs,
confirmed a cavernoma in the right part of the mesenceph- high anatomical variability may occur by displacement or
alon. The nTMS DTI plan showed displacement of the left disruption of tracts.16,17 Traditionally, DTI is carried out
CBT (Fig. 6C) and abutting the lesion (Fig. 6B). The RMT through predefined anatomical landmarks used as seed
(49%) appears elevated, and the FAT (0.13) was decreased points and custom FA values,37 which highlights the main
(Table 1). The conventional DTI fiber tracking suggested limitation of this technique in a situation of obscured
718 J Neurosurg Volume 135 • September 2021
 
* | D*2 *34 *
 Zdunczyk et al.

FIG. 6. Illustrative case 4. A 42-year-old female patient (patient 10 in Table 1) with preoperative double vision and vertigo. A: Cav-
ernous malformation in the right part of the mesencephalon. B: Distance to CBT, 0 mm (yellow arrow); distance to CST, 3 mm.
C: Displacement of left CBT. D: Patient experienced postoperative dysphagia, left CBT bordering (yellow arrow) resection cavity
(red star). E: Left CST (hand/leg) and CBT abutting the lesion. However, the patient did not suffer from postoperative paresis of the
extremities. Figure is available in color online only.

anatomy and signal alterations that might result from the were defined and used to track functionally relevant fibers
lesion itself, surrounding edema, or hemosiderin depos- and successfully reconstruct a somatotopy of tracts on the
its.38 In such cases, an objective definition of seed points brainstem level.
and individual FAT would help to account for pathologi-
cal anatomy and disturbed tract integrity. By localization Preoperative Risk Stratification
of essential tracts, nTMS-based DTI can deliver crucial Due to the density of highly eloquent anatomical struc-
information to define safe entry zones and guide intraop- tures surrounding BSCMs on the one hand and the natural
erative orientation and monitoring.26 Furthermore, the sole course of BSCMs on the other hand, defining the individu-
visualization of white matter tracts does not carry infor- al risk profile of surgical morbidity is a crucial step in the
mation about their functionality and vulnerability. While decision-making process for surgical treatment. Of equal
these features crucially depend on the FA profile of the importance is the ability to counsel the patient about the
tracts and the FAT set for tractography, determining an in- risks of a postoperative deficit before deciding on a spe-
dividual threshold level by using supplemental functional cific treatment option based on objective data.
data might overcome this methodological weakness.26 nTMS-based DTI is a noninvasive method that not only
allows for a preoperative visualization of essential tracts,
Seed Region Definition Based on nTMS but also combines radiographic and electrophysiological
nTMS noninvasively establishes a link between stimu- features. Estimating the excitability and conductivity of
lation of a cortical area and registration of motor output, the tracts adds important information to the topographical
a process analogous to the gold standard of direct corti- data, i.e., the minimal distance between tract and lesion,
cal stimulation (DCS).20,21,23 The resulting functional maps by providing information on the vulnerability of the tracts.
obtained with nTMS have been demonstrated to provide In our series, we were able to show a negative correlation
high reliability and precision of results compared with between the FAT and RMT, i.e., measure of anisotropy or
DCS findings.20,21 Using individually defined seed points tract integrity and cortical excitability. A disturbance in
derived from the nTMS measurement helps to define es- fiber conductivity was thus associated with a low excit-
sentially functional tracts. A customized FAT determina- ability. Both parameters also correlate to a pre- and post-
tion has furthermore proven to be a superior method for operative motor deficit. Patients with disturbed fiber integ-
reducing aberrant fibers and interexaminer variability, as rity and thus low FAT also presented with a lower cortical
recently shown in supratentorial lesions.26,30 In the pres- excitability displayed by a high RMT. Also, a distance of
ent study, seed regions for hand, leg, and facial function 1 mm between lesion and tracts appears to be the critical
J Neurosurg Volume 135 • September 2021 719
 
* | D*2 *34 *
Zdunczyk et al.

lower limit: all patients who developed a new motor deficit Conclusions
or an increased paresis had a minimal distance between nTMS-based tractography allows for an objective vi-
lesion and tract of 1 mm or less. Of equal importance is sualization of relevant motor pathways on the brainstem
the development of a new cranial nerve deficit, particu- level. By combining functional and radiographical data,
larly dysarthria or dysphagia, since it significantly restricts an evaluation of fiber vulnerability and lesion-to-tract
quality of life and might result in the need for further sur- distance can be achieved. We feel that with the use of
gical therapy, i.e., tracheostomy. In our study, the critical nTMS-DTI complex brainstem cavernomas become more
distance to CBT for a new facial deficit was 2 mm. amenable to surgical resection and neurological postoper-
ative deterioration can be prevented. This method provides
Surgical Planning and Intraoperative Orientation valuable information that can positively impact patient
Choosing the right surgical approach is a crucial step in consultation, preoperative planning, and intraoperative
the surgery of brainstem lesions.7,35,36 Visualization of the orientation.
CST and CBT in relation to the tumor is essential to define
a safe surgical corridor and brainstem entry point. In the
present study, the tractography results were used in all of Acknowledgments
the surgical procedures to set up an individually tailored We thank our study nurse H. Schneider (Department of Neu-
surgical plan with the lowest risk of affecting the CST or rosurgery, Charité–Universitätsmedizin, Berlin) for research
assistance.
CBT. By intraoperative navigation, the nTMS DTI results
can be made readably available during surgery through a
heads-up display in the surgical microscope and serve as a References
guide for microsurgical dissection as well as direct electri-   1. Gross BA, Lin N, Du R, Day AL. The natural history of in-
cal stimulation. tracranial cavernous malformations. Neurosurg Focus. 2011;​
30(6):E24.
​
Clinical Outcome   2. Kim PY, Park YG, Choi JU, et al. An analysis of the natural
history of cavernous malformations. Surg Neurol. 1997;48(1):​ ​
In our series of 15 patients undergoing brainstem cav- 9–18.
ernoma resection, we observed a new motor deficit in only   3. Robinson JR Jr, Awad IA, Little JR. Natural history of the
1 patient (7%) and a progression of a preexisting paresis cavernous angioma. J Neurosurg. 1991;75(5):709–714.
​ ​
in 3 patients. These results are comparable to previously   4. Dukatz T, Sarnthein J, Sitter H, et al. Quality of life after
reported results on DTI in BSCM (8.7%–9%).16,17,35 How- brainstem cavernoma surgery in 71 patients. Neurosurgery.
ever, in our patient sample we were able to include a higher 2011;69(3):689–695.
​ ​
  5. Hauck EF, Barnett SL, White JA, Samson D. Symptomatic
number of high-risk, deep-seated lesions without a pial brainstem cavernomas. Neurosurgery. 2009;64(1):61–71.
​ ​
surface (47%) than in previously reported studies.17 All of   6. Gross BABH, Batjer HH, Awad IA, Bendok BR. Brainstem cav-
our patients showed at least a partial regression of motor ernous malformations. Neurosurgery. 2009;64(5):E805–E818.
​ ​
deficits during follow-up, with a favorable overall outcome   7. Gross BA, Batjer HH, Awad IA, et al. Brainstem cavernous
in 50% (mRS score 0–2) and a stable or improved mRS malformations:​1390 surgical cases from the literature. World
score in 72% of the patients. Neurosurg. 2013;80(1-2):89–93.
​ ​
  8. Cao Z, Lv J, Wei X, Quan W. Appliance of preoperative dif-
fusion tensor imaging and fiber tractography in patients with
Perspective brainstem lesions. Neurol India. 2010;58(6):886–890.
​ ​
DTT is an emerging method that allows noninvasive   9. Berman J. Diffusion MR tractography as a tool for surgical
visualization of eloquent fibers and their localization on planning. Magn Reson Imaging Clin N Am. 2009;17(2):205–214.
​ ​
10. Sarubbo S, De Benedictis A, Merler S, et al. Towards a func-
the brainstem level. However, users must be aware of the tional atlas of human white matter. Hum Brain Mapp. 2015;​
limits of this method to prevent misinterpretation of find- 36(8):3117–3136.
​
ings. Because the DTI methodology enables the display of 11. Basser MJ, Mattiello J, LeBihan D. MR diffusion tensor
mainly long fibers, tractography results might be affected spectroscopy and imaging. Biophys J. 1994;66(1):259–267.
​ ​
by MRI artifacts (i.e., hemorrhage and hemosiderin depos- 12. Sundgren PC, Dong Q, Gómez-Hassan D, et al. Diffusion
its), and no information is provided on the localization of tensor imaging of the brain:​review of clinical applications.
cranial nuclei. To address this limitation, electrophysiologi- Neuroradiology. 2004;46(5):339–350.
​ ​
13. Basser PJ, Pajevic S, Pierpaoli C, et al. In vivo fiber tractography
cal data might provide valuable information about tract al- using DT-MRI data. Magn Reson Med. 2000;44(4):625–632.
​ ​
teration and conductivity. Furthermore, conventional DTI 14. Chen X, Weigel D, Ganslandt O, et al. Diffusion tensor-based
is to a certain degree operator dependent, and results are fiber tracking and intraoperative neuronavigation for the
prone to high interobserver variability. The latter might resection of a brainstem cavernous angioma. Surg Neurol.
be overcome by a standardized protocol with individual 2007;68(3):285–291.
​ ​
seed points and FAT determinations. Our proposed nTMS- 15. Faraji AH, Abhinav K, Jarbo K, et al. Longitudinal evalua-
based tractography protocol combines radiological with tion of corticospinal tract in patients with resected brainstem
electrophysiological data and allows for an objective visu- cavernous malformations using high-definition fiber trac-
tography and diffusion connectometry analysis:​preliminary
alization of relevant motor pathways on the brainstem level. experience. J Neurosurg. 2015;123(5):1133–1144.
​ ​
In the current series of patients, our data not only helped to 16. Flores BC, Whittemore AR, Samson DS, Barnett SL. The
guide surgical planning and intraoperative orientation, but utility of preoperative diffusion tensor imaging in the surgi-
also enabled us to define individual risk factors (distance to cal management of brainstem cavernous malformations. J
tracts, RMT, tract disruption) for postoperative morbidity. Neurosurg. 2015;122(3):653–662.
​ ​

720 J Neurosurg Volume 135 • September 2021

 
* | D*2 *34 *
 Zdunczyk et al.

17. Li D, Jiao YM, Wang L, et al. Surgical outcome of motor 32. Rosenstock T, Grittner U, Acker G, et al. Risk stratification in
deficits and neurological status in brainstem cavernous mal- motor area-related glioma surgery based on navigated trans-
formations based on preoperative diffusion tensor imaging:​ cranial magnetic stimulation data. J Neurosurg. 2017;​126(4):​
a prospective randomized clinical trial. J Neurosurg. 2018;​ 1227–1237.
130(1):​286–301. 33. Cantu C, Murillo-Bonilla L, Arauz A, et al. Predictive fac-
18. Ulrich NH, Kockro RA, Bellut D, et al. Brainstem cavernoma tors for intracerebral hemorrhage in patients with cavernous
surgery with the support of pre- and postoperative diffusion angiomas. Neurol Res. 2005;​27(3):​314–318.
tensor imaging:​initial experiences and clinical course of 23 34. Pandey P, Westbroek EM, Gooderham PA, Steinberg GK.
patients. Neurosurg Rev. 2014;​37(3):​481–492. Cavernous malformation of brainstem, thalamus, and basal
19. Januszewski J, Albert L, Black K, Dehdashti AR. The useful- ganglia:​a series of 176 patients. Neurosurgery. 2013;​72(4):​
ness of diffusion tensor imaging and tractography in surgery 573–589.
of brainstem cavernous malformations. World Neurosurg. 35. Garcia RM, Ivan ME, Lawton MT. Brainstem cavernous mal-
2016;​93:​377–388. formations:​surgical results in 104 patients and a proposed
20. Ruohonen J, Karhu J. Navigated transcranial magnetic stimu- grading system to predict neurological outcomes. Neurosur-
lation. Neurophysiol Clin. 2010;​40(1):​7–17. gery. 2015;​76(3):​265–278.
21. Picht T, Mularski S, Kuehn B, et al. Navigated transcranial 36. Walcott BP, Choudhri O, Lawton MT. Brainstem cavernous
magnetic stimulation for preoperative functional diagnostics in malformations:​Natural history versus surgical management.
brain tumor surgery. Neurosurgery. 2009;​65(6)(suppl):​93–99. J Clin Neurosci. 2016;​32:​164–165.
22. Lefaucheur JP, Picht T. The value of preoperative functional 37. Mori S, van Zijl PC. Fiber tracking:​principles and strategies—
cortical mapping using navigated TMS. Neurophysiol Clin. a technical review. NMR Biomed. 2002;​15(7-8):​468–480.
2016;​46(2):​125–133. 38. Leclercq D, Delmaire C, de Champfleur NM, et al. Diffusion
23. Picht T, Schmidt S, Brandt S, et al. Preoperative functional tractography:​methods, validation and applications in patients
mapping for rolandic brain tumor surgery:​comparison of with neurosurgical lesions. Neurosurg Clin N Am. 2011;​22(2):​
navigated transcranial magnetic stimulation to direct cortical 253–268, ix.
stimulation. Neurosurgery. 2011;​69(3):​581–588.
24. Picht T. Current and potential utility of transcranial magnetic
stimulation in the diagnostics before brain tumor surgery. Disclosures
CNS Oncol. 2014;​3(4):​299–310. T. Picht has served as a speaker for Nexstim Oy but is not a con-
25. Krieg SM, Shiban E, Buchmann N, et al. Utility of presurgi- tracted consultant.
cal navigated transcranial magnetic brain stimulation for the
resection of tumors in eloquent motor areas. J Neurosurg. Author Contributions
2012;​116(5):​994–1001.
26. Frey D, Strack V, Wiener E, et al. A new approach for corti- Conception and design: Vajkoczy, Zdunczyk. Acquisition of data:
cospinal tract reconstruction based on navigated transcranial Zdunczyk, Roth. Analysis and interpretation of data: Zdunczyk,
stimulation and standardized fractional anisotropy values. Roth. Drafting the article: Zdunczyk. Critically revising the
Neuroimage. 2012;​62(3):​1600–1609. article: Vajkoczy, Zdunczyk, Picht. Reviewed submitted version of
27. Jussen D, Zdunczyk A, Schmidt S, et al. Motor plasticity manuscript: Vajkoczy, Roth, Picht. Statistical analysis: Zdunczyk.
after extra-intracranial bypass surgery in occlusive cerebro- Study supervision: Zdunczyk.
vascular disease. Neurology. 2016;​87(1):​27–35.
28. Awiszus F, Feistner H, Urbach D, Bostock H. Characterisa- Supplemental Information
tion of paired-pulse transcranial magnetic stimulation condi- Previous Presentations
tions yielding intracortical inhibition or I-wave facilitation The results reported in this paper (resting motor threshold, frac-
using a threshold-hunting paradigm. Exp Brain Res. 1999;​ tional anisotropy threshold, distance to tracts) were presented
129(2):​317–324. orally at the annual scientific meeting of the German Society for
29. Nimsky C, Ganslandt O, Merhof D, et al. Intraoperative visu- Neurosurgery, Würzburg, Germany, May 15, 2019.
alization of the pyramidal tract by diffusion-tensor-imaging-
based fiber tracking. Neuroimage. 2006;​30(4):​1219–1229. Correspondence
30. Rosenstock T, Giampiccolo D, Schneider H, et al. Specific
Peter Vajkoczy: Charité–Universitätsmedizin Berlin, Germany.
DTI seeding and diffusivity-analysis improve the quality
[email protected].
and prognostic value of TMS-based deterministic DTI of the
pyramidal tract. Neuroimage Clin. 2017;​16:​276–285.
31. Kovanlikaya I, Firat Z, Kovanlikaya A, et al. Assessment of
the corticospinal tract alterations before and after resection
of brainstem lesions using diffusion tensor imaging (DTI)
and tractography at 3T. Eur J Radiol. 2011;​77(3):​383–391.

J Neurosurg Volume 135 • September 2021 721

 
* | D*2 *34 *
 CASE REPORT
J Neurosurg 135:722–726, 2021

Endoscopic endonasal transsphenoidal direct access

and Onyx embolization of a dural arteriovenous fistula
mimicking a carotid-cavernous fistula: case report
Patrick J. Karas, MD,1 Robert Y. North, MD, PhD,1 Visish M. Srinivasan, MD,1
Nathan R. Lindquist, MD,2 K. Kelly Gallagher, MD,2 Jan-Karl Burkhardt, MD,1
Daniel Yoshor, MD,1 and Peter Kan, MD1
1
Department of Neurosurgery and 2Bobby R. Alford Department of Otolaryngology, Baylor College of Medicine, Houston, Texas

The classic presentation of a carotid-cavernous fistula (CCF) is unilateral painful proptosis, chemosis, and vision loss.
Just as the goal of treatment for a dural arteriovenous fistula (dAVF) is obliteration of the entire fistulous connection and
the proximal draining vein, the modern treatment of CCF is endovascular occlusion of the cavernous sinus via a trans-
venous or transarterial route. Here, the authors present the case of a woman with a paracavernous dAVF mimicking the
clinical and radiographic presentation of a CCF. Without any endovascular route available to access the fistulous con-
nection and venous drainage, the authors devised a novel direct hybrid approach by performing an endoscopic endona-
sal transsphenoidal direct puncture and Onyx embolization of the fistula.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201737
KEYWORDS  dural arteriovenous fistula; carotid-cavernous malformation mimic; endoscopic endonasal
transsphenoidal; percutaneous embolization; vascular disorders

C arotid-cavernous fistulas (CCFs) are shunts that We devised a hybrid approach utilizing an endoscopic en-
allow blood to flow from the carotid artery (or donasal transsphenoidal approach, allowing us to directly
branches thereof) into the cavernous sinus.1 The re- puncture the venous pouch under stereotaxy and endo-
sulting arterialized cavernous sinus causes increased pres- scopic visualization, and completely embolize the fistula.
sure in the superior ophthalmic vein, leading to venous To our knowledge, this is the first successful endoscopic
congestion in the orbit. The classic clinical manifestation endonasal transsphenoidal direct puncture and emboliza-
is unilateral chemosis, proptosis, and vision loss. However, tion of a dAVF.
dural arteriovenous fistulas (dAVFs) not involving the cav-
ernous sinus can also manifest with similar clinical pre- Case Report
sentations if the venous drainage from the fistula causes
increased pressure within the superior ophthalmic vein History and Examination
and subsequent vascular congestion within the orbit.2 A 71-year-old right-handed woman was referred by
Historically, CCFs were treated with either sacrifice of an ophthalmologist to the emergency department with a
the offending carotid artery or surgical packing of the cav- 2-day history of acute-onset right eye pain, proptosis, and
ernous sinus.3 The modern treatment focuses on occlusion blurry vision. The patient’s medical history was notable
of the cavernous sinus through endovascular approaches, for warfarin use after cardiac valve replacement 7 years
usually through a transvenous route.4,5 Similarly, the treat- prior, aspirin use (81 mg), chronic kidney disease now
ment for dAVFs is obliteration of the fistulous connection with normal estimated glomerular filtration rate, and a
point and proximal venous drainage, either surgically or prior right-sided stroke with residual left upper-extremity
through an endovascular approach. spasticity. Her neurological examination was remarkable
Here, we present the case of a woman with a dAVF for right eye proptosis with conjunctival injection and an
whose clinical and radiographic presentation mimicked a afferent pupillary defect with decreased visual acuity. The
CCF. Initial endovascular treatment failed due to a lack remainder of her cranial nerve examination, including ex-
of venous access to the fistula and the inability to achieve traocular movements, was unremarkable. No bruit could
obliteration of the fistula from a transarterial treatment. be auscultated from the orbit or temporal region.

ABBREVIATIONS  CCF = carotid-cavernous fistula; dAVF = dural arteriovenous fistula.

SUBMITTED  May 10, 2020.  ACCEPTED  July 1, 2020.
INCLUDE WHEN CITING  Published online November 13, 2020; DOI: 10.3171/2020.7.JNS201737.

722 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *34 *
 Karas et al.

CT angiography of the orbit and brain (Fig. 1) revealed

dilated vascular channels in the region of the right cav-
ernous sinus, anterior and medial to the cavernous inter-
nal carotid artery. The superior ophthalmic vein was also
engorged and hyperdense. Cerebral angiography of the
right common carotid artery showed no abnormalities.
Left carotid injections (Fig. 2) showed venous filling dur-
ing the arterial phase in the region of the right cavernous
sinus with small indirect feeders from both internal and
external carotid origins. The fistula was low flow and had
no clear venous drainage through the superior or inferior
petrosal sinus. Rather, contrast was seen to slowly diffuse
into the proximal superior ophthalmic vein and proximal
basilar plexus before fully dissipating. The distal superior
ophthalmic vein appeared thrombosed.
Transvenous access via the right side was attempted;
however, the right internal jugular vein was chronically
occluded. The right cavernous sinus was then catheter-
ized through the left internal jugular vein and left inferior
petrosal sinus. Microcatheter injection of the cavernous
sinus revealed a symmetric cavernous sinus and no con-
nection to the venous fistula pouch (Fig. 2D and E). We
therefore diagnosed the patient with a paracavernous,
paraclinoid dural arteriovenous fistula, although a Bar- FIG. 1. Preoperative CT angiogram of the orbit (A) showing dilated hy-
row type D CCF with a loculated fistula pouch was also perdense superior ophthalmic vein (black arrow). We hypothesize that
a possibility. this vein was the primary venous drainage of the fistula and that sponta-
neous occlusion of the vein led to the patient’s acute onset of symptoms.
Treatment Axial (B), coronal (C), and sagittal (D) brain CT angiograms showing the
location of the fistula (white arrow) in the region of the right cavernous
Selective microcatheterization of the left internal max- sinus just anterior and medial to the cavernous segment of the carotid
illary artery was performed, gaining access just proximal artery. Copyright Patrick Karas. Published with permission.
to the arterial feeders to the fistula. We then performed
Onyx (Medtronic) embolization of the fistulous branches
with the aid of a dual-lumen Scepter C balloon (MicroVen-
tion) under inflation with the goal of pushing Onyx into with a 22-gauge spinal needle. Return of arterial blood
the fistulous pouch. However, we were unable to reach the as well as an intraoperative injection of contrast dye into
fistulous pouch with Onyx, resulting in incomplete treat- the venous pouch confirmed proper location of the spinal
ment of the arteriovenous fistula. Postembolization angi- needle within the fistulous pouch. We then proceeded to
ography revealed persistent filling of the fistula primarily inject Onyx under live fluoroscopy until the fistula was
from small branches arising off the internal carotid artery. completely obliterated. An intraoperative transradial an-
After failure of the initial endovascular treatment, we giogram of the left common carotid artery confirmed that
devised a combined open and endovascular approach, we had completely obliterated the fistula.
with the goal of completely embolizing the fistulous pouch
(Video 1). Postoperative Course
VIDEO 1. Patient presentation, diagnostic imaging findings, and By postoperative day 1, the patient noted significant
intraoperative video. Copyright Patrick Karas. Published with per- improvement in her right eye pain, visual acuity, and prop-
mission. Click here to view. tosis. By postoperative day 5, her proptosis and conjuncti-
Informed consent was obtained, and the patient agreed val injection was almost completely resolved (Fig. 4). The
to proceed with surgery. Surgery was performed in a hy- patient had an uneventful postoperative period. She was
restarted on warfarin and a therapeutic heparin bridge on
brid operating room equipped with monoplane fluorosco-
postoperative day 2 and discharged home on postoperative
py. The patient’s head was secured in a radiolucent skull day 7 once her warfarin reached a therapeutic level. Her
clamp, and StealthStation (Medtronic) frameless stereo- 3-month postoperative visit showed complete resolution of
taxy was registered using the head CT angiogram. We her symptoms with no recurrence.
performed an endoscopic endonasal transsphenoidal ap-
proach and removed the bone at the posterior sphenoid si-
nus overlying the venous pouch of the dAVF (Fig. 3). The Discussion
exact location of the fistula was located using frameless Noncavernous dAVFs that mimic the clinical presen-
stereotactic navigation based on the preoperative CT an- tation of CCFs are rare, occurring in only 2%–15% of
giogram. Precise puncture of the fistula through the dura patients with arteriovenous fistulas and clinical exami-
was important to avoid the adjacent cavernous segment nation findings suggestive of CCF.2,6 As in our case, dis-
of the internal carotid artery. We punctured the fistula tinguishing a paracavernous dAVF from a true CCF can

J Neurosurg Volume 135 • September 2021 723

 
* | D*2 *34 *
Karas et al.

be difficult, even with direct microcatheterization of the

cavernous sinus, and the diagnosis is presumptive despite
a normal and symmetrical cavernous sinus on the veno-
gram with no connection to the fistula. An isolated seg-
ment of the cavernous sinus that formed the venous pouch
was in the differential diagnosis. In CCF mimics, although
the fistulous connection is not at the cavernous sinus, the
pathophysiology underlying the clinical symptoms of che-
mosis, proptosis, and vision loss is the same: downstream
congestion, retrograde filling, and/or arterialized high-
flow blood in the superior ophthalmic vein causes venous
congestion in the orbit and the resulting classic clinical
phenotype.7
In general, direct high-flow CCFs caused by traumatic
or aneurysmal rupture classically present acutely, while
indirect lower-flow CCFs present more gradually as the
venous drainage pattern changes and pressure in the su-
perior ophthalmic vein increases. 5 Our patient presented
acutely despite having a dAVF analogous to a Barrow type
D indirect CCF. We hypothesize that the dAVF had been
present but asymptomatic for a long period of time. Acute
thrombosis of the superior ophthalmic vein (noted on
the preoperative CT scan of the orbit) likely triggered an
abrupt increase in intraorbital pressure from venous con-
gestion, causing the patient’s acute symptom onset.
The treatment of both CCF and dAVF focuses on iden-
tifying and obliterating the venous pouch of the fistula
and proximal draining veins.8,9 Unfortunately, we had no
transvenous access to the venous pouch to allow for embo-
lization of it and the proximal draining vein. Transarterial
access also failed as the small feeders arising off the distal
internal maxillary were too narrow and long to reach the
venous pouch with Onyx, and the feeders from the inter-
nal carotid artery were too small to safely embolize with-
out a high risk of reflux into the internal carotid artery.
Sacrificing the internal carotid artery is not an effective
option, with the additional risk of hemiplegia and aphasia.
We therefore considered other approaches.
Direct puncture of the superior ophthalmic vein can
be used to access the cavernous sinus and treat CCFs.10,11
Thrombosis of the distal superior ophthalmic vein pre-
vented attempting this approach in our case. Percutaneous
embolization via direct puncture of the cavernous sinus
through the superior orbital fissure has also been de-
scribed.12,13 We decided against this approach because the FIG. 2. A and B: Anteroposterior (A) and lateral (B) preoperative cere-
venous pouch of the fistula was small and did not commu- bral angiograms of the brain after left common carotid injection showing
nicate with the cavernous sinus. While the superior orbital the fistula in the region of the right cavernous sinus. Note the origin of
fissure provides access to the cavernous sinus, access to the right superior ophthalmic vein (black arrow) and the proximal basilar
adjacent structures is less guaranteed; we were concerned plexus (black arrowhead); there was no visualized distal contrast flow
that finding a small fistulous pouch adjacent to the cavern- through these structures. C: Magnified left common carotid injection
ous sinus would be prohibitively difficult. Craniotomy was showing multiple very small feeding arteries arising directly from both
the internal carotid artery and the internal maxillary artery (black arrow).
also considered but is significantly more invasive than our D: Transvenous angiogram of the cavernous sinus via the left inferior
ultimate approach. petrosal sinus, revealing no connection to the fistula and a symmetric
Given the proximity of the venous pouch to the poste- cavernous sinus, preventing endovascular access through a transve-
rior wall of the sphenoid sinus, we determined that a trans- nous approach. E: Superimposed arterial and venous runs illustrate the
sphenoidal approach would enable us to directly access juxtaposition of the arteriovenous fistula adjacent to the right cavernous
the venous pouch for Onyx embolization. Direct puncture sinus. Copyright Patrick Karas. Published with permission. Figure is
and embolization of the venous pouch has been success- available in color online only.
fully described for dAVFs of the skull base such as the
hypoglossal canal,14 but not via an endoscopic endonasal
transsphenoidal approach. Our ability to perform this ap-

724 J Neurosurg Volume 135 • September 2021

 
* | D*2 *34 *
 Karas et al.

FIG. 4. Preoperative photograph (upper) showing ptosis, chemosis,

and conjunctival injection. The patient’s ocular symptoms resolved
rapidly after surgery, with almost complete resolution by postoperative
day 7 (lower), and complete resolution at her 3-month follow-up visit.
Copyright Patrick Karas. Published with permission. Figure is available
in color online only.

be moved to perform fluoroscopy. Stereotactic navigation

FIG. 3. A: Intraoperative frameless stereotactic navigation was used was essential for localizing the venous pouch, which was
to localize the exact location of the arteriovenous fistula prior to bony adjacent to the cavernous sinus and carotid artery. Navi-
removal. B: After the overlying bone was removed, a spinal needle gation enabled us to remove only the bone directly over-
was used to puncture the fistula under direct endoscopic visualization lying the pouch to safely puncture the pouch rather than
(left), contrast was injected to confirm entrance into the fistula pouch, surrounding structures. We punctured the venous pouch
and Onyx was injected through the spinal needle to embolize the fistula with the spinal needle under direct endoscopic visualiza-
(right). C: An intraoperative postembolization angiogram confirmed that
the fistula was completely obliterated. D: The endoscopic and frame- tion while simultaneously confirming the location of the
less stereotactic monitors positioned above the patient’s head had to pouch with the navigation probe. After puncturing the fis-
be moved for the angiographic portion of the procedure in the hybrid tulous pouch, the needle was held in place by hand while
operating room. CP = carotid protuberance; OP = optic protuberance; the StealthStation terminal and endoscope terminals were
S = sella. Copyright Patrick Karas. Published with permission. Figure is removed to bring in the monoplane. At this point, prop-
available in color online only. er placement was confirmed by checking for continuous
backflow of arterialized blood from the needle and sub-
sequent injection of contrast into the venous pouch under
proach was dependent on a full understanding of the ve- fluoroscopy.
nous outflow of the fistula and the availability of a hybrid A second technical consideration was the proximity of
operating room. the fluoroscope’s cassette holder to the patient’s nose. The
Several technical challenges arose in the operating cassette was too close to the nose to be able to attach the
room. Because of the fixed operating room table, the pa- syringe directly onto the spinal needle. We therefore used
tient was positioned supine in a reverse Trendelenburg a short segment of an intravenous extension tube, which is
position rather than the traditional beach chair position typically included in a lumbar puncture set. However, the
commonly used for transsphenoidal approaches. The volume needed to prime the intravenous extension tube
StealthStation navigation terminal and endoscope moni- with dimethyl suboxide (DMSO) is much greater than the
tor terminals were placed at the head of the bed. How- volume generally needed during endovascular procedures,
ever, after puncturing the dAVF venous pouch, both the forcing us to prime the tube with two 1.0-ml volumes of
StealthStation and endoscopic monitor terminals had to DMSO prior to injecting Onyx. Similarly, more Onyx was

J Neurosurg Volume 135 • September 2021 725

 
* | D*2 *34 *
Karas et al.

needed to fill the dead space of the tubing. Standard in-   9. Ng PP, Higashida RT, Cullen S, et al. Endovascular strategies
travenous tubing is 4 mm (0.16 inches) in outer diameter. for carotid cavernous and intracerebral dural arteriovenous
The injection also needed to be performed expediently as fistulas. Neurosurg Focus. 2003;​15(4):​ECP1.
10. Miller NR, Monsein LH, Debrun GM, et al. Treatment of
DMSO was corrosive to the intravenous tubing. carotid-cavernous sinus fistulas using a superior ophthalmic
Ultimately, the hybrid endoscopic endonasal transsphe- vein approach. J Neurosurg. 1995;​83(5):​838–842.
noidal direct puncture and embolization of the dAVF was 11. Srinivasan VM, Sen AN, Kan P. Trans-superior ophthalmic
successful. The patient recovered rapidly from the proce- vein approach for treatment of carotid-cavernous fistula.
dure, and her ocular symptoms promptly resolved. Neurosurg Focus. 2019;​46(Suppl_2):​V4.
12. Teng MMH, Lirng JF, Chang T, et al. Embolization of ca-
rotid cavernous fistula by means of direct puncture through
Conclusions the superior orbital fissure. Radiology. 1995;​194(3):​705–711.
Paracavernous dAVFs can cause symptoms mimicking 13. Narayanan S, Murchison AP, Wojno TH, Dion JE. Percutane-
CCFs when the drainage pattern involves the cavernous ous trans-superior orbital fissure embolization of carotid-cav-
sinus or the superior ophthalmic vein. Treatment of these ernous fistulas:​technique and preliminary results. Ophthal
Plast Reconstr Surg. 2009;​25(4):​309–313.
lesions can be challenging if transvenous and transarterial 14. Diaz OM, Toledo MM, Roehm JOF, et al. Unique percutane-
routes are not favorable. We present a novel hybrid open ous direct puncture technique for occlusion of a hypoglossal
and endovascular approach via endoscopic endonasal canal dural arteriovenous fistula. J Neurointerv Surg. 2018;​
transsphenoidal direct puncture of the fistula to achieve 10(12):​1179–1182.
complete embolization of the venous pouch and proximal
venous drainage. Combining techniques and principles
from endonasal and endovascular surgery allowed us to Disclosures
safely achieve a successful outcome with minimal mor- The authors report no conflict of interest concerning the materi-
bidity to the patient. als or methods used in this study or the findings specified in this
paper.

References Author Contributions

  1. Barrow DL, Spector RH, Braun IF, et al. Classification and Conception and design: Kan, Karas, North, Gallagher, Burkhardt,
treatment of spontaneous carotid-cavernous sinus fistulas. J Yoshor. Acquisition of data: Kan, Karas, North, Lindquist, Galla-
Neurosurg. 1985;​62(2):​248–256. gher, Yoshor. Drafting the article: Kan, Karas. Critically revising
  2. Kobkitsuksakul C, Jiarakongmun P, Chanthanaphak E, the article: all authors. Reviewed submitted version of manu-
Pongpech S. Noncavernous arteriovenous shunts mimicking script: all authors. Approved the final version of the manuscript
carotid cavernous fistulae. Diagn Interv Radiol. 2016;​22(6):​ on behalf of all authors: Kan.
555–559.
  3. Dorrance GM, Loudenslager PE. Physiological consider- Supplemental Information
ations in the treatment of pulsating exophthalmos. Am J Videos
Ophthalmol. 1934;​17(12):​1099–1111.
  4. Chi CT, Nguyen D, Duc VT, et al. Direct traumatic carotid Video 1. https://vimeo.com/442459088.
cavernous fistula:​angiographic classification and treatment
strategies. Study of 172 cases. Interv Neuroradiol. 2014;​ Correspondence
20(4):​461–475. Peter Kan: Baylor College of Medicine, Houston, TX. ptkan@
  5. Ellis JA, Goldstein H, Connolly ES Jr, Meyers PM. Carotid- utmb.edu.
cavernous fistulas. Neurosurg Focus. 2012;​32(5):​E9.
  6. Hasuo K, Matsumoto S, Mihara F, et al. Dural non-cavernous
sinus arteriovenous fistulas symptomatically simulating
spontaneous carotid-cavernous fistulas:​an analysis of angio-
graphic findings. Radiat Med. 1997;​15(4):​203–208.
  7. Sedat J, Kominami S, Siriwimonmas S, et al. Extracavernous
arteriovenous fistulae. Report of five cases. Interv Neurora-
diol. 1999;​5(3):​235–243.
  8. Meyers PM, Halbach VV, Dowd CF, et al. Dural carotid cav-
ernous fistula:​definitive endovascular management and long-
term follow-up. Am J Ophthalmol. 2002;​134(1):​85–92.

726 J Neurosurg Volume 135 • September 2021

 
* | D*2 *34 *
 CLINICAL ARTICLE
J Neurosurg 135:727–732, 2021

A comparison of radial versus femoral artery access for

acute stroke interventions
Omaditya Khanna, MD, Lohit Velagapudi, BS, Somnath Das, BS, Ahmad Sweid, MD,
Nikolaos Mouchtouris, MD, Fadi Al Saiegh, MD, Michael B. Avery, MD, MSc, Nohra Chalouhi, MD,
Richard F. Schmidt, MD, Kalyan Sajja, MD, M. Reid Gooch, MD, Stavropoula Tjoumakaris, MD,
Robert H. Rosenwasser, MD, and Pascal M. Jabbour, MD
Department of Neurological Surgery, Thomas Jefferson University Hospital, Philadelphia, Pennsylvania

OBJECTIVE  In this study, the authors aimed to investigate procedural and clinical outcomes between radial and femo-
ral artery access in patients undergoing thrombectomy for acute stroke.
METHODS  The authors conducted a single-institution retrospective analysis of 104 patients who underwent mechanical
thrombectomy, 52 via transradial access and 52 via traditional transfemoral access. They analyzed various procedural
and clinical metrics between the two patient cohorts.
RESULTS  There was no difference between patient demographics or presenting symptoms of stroke severity between
patients treated via transradial or transfemoral access. The mean procedural time was similar between the two treatment
cohorts: 60.35 ± 36.81 minutes for the transradial group versus 65.50 ± 29.92 minutes for the transfemoral group (p =
0.451). The mean total fluoroscopy time for the procedure was similar between the two patient cohorts (20.31 ± 11.68 for
radial vs 18.49 ± 11.78 minutes for femoral, p = 0.898). The majority of patients underwent thrombolysis in cerebral in-
farction score 2b/3 revascularization, regardless of access site (92.3% for radial vs 94.2% for femoral, p = 0.696). There
was no significant difference in the incidence of access site or periprocedural complications between the transradial and
transfemoral cohorts.
CONCLUSIONS  Acute stroke intervention performed via transradial access is feasible and effective, with no significant
difference in procedural and clinical outcomes compared with traditional transfemoral access. Larger studies are re-
quired to further validate the efficacy and limitations of transradial access for neurointerventional procedures.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201174

H
KEYWORDS  radial artery catheterization; endovascular; stroke; angiogram; vascular disorders

istorically, endovascular procedures have been for performing a wide range of neuroendovascular proce-
performed via femoral artery catheterization, due dures with favorable patient satisfaction measures.4–7 In
to the fact that it is relatively easy to achieve ac- the field of interventional cardiology, several large, mul-
cess and the large vessel size allows for the use of wider- ticenter randomized controlled trials have demonstrated
bore catheters. Although femoral artery catheterization that radial artery catheterization is associated with fewer
is standard of care among neurointerventionalists, it has access site and procedural complications compared with
been shown to be associated with potential complica- femoral artery access,8,9 as well as improved patient sat-
tions such as pseudoaneurysm formation, retroperitoneal isfaction10 and lower overall costs.11 As such, practicing
hematoma, arteriovenous fistula, and artery occlusion.1–3 endovascular neurosurgeons at our institution have, of late,
Furthermore, femoral artery access is painful and un- begun to adopt transradial access for their interventional
comfortable for patients, who must remain supine for 4 procedures with increasing frequency, based on clinical
to 6 hours postoperatively before they are allowed to be factors more amenable to radial artery catheterization
mobilized. such as a difficult aortic arch configuration, tortuous neck
Radial artery access has started to gain more wide- vasculature, and severe obesity (Fig. 1).
spread use among neurointerventionalists, and it has been In this study, we present our institution’s case series of
shown to be a feasible alternative to femoral artery access 104 consecutive stroke interventions and report outcome

ABBREVIATIONS  ICA = internal carotid artery; NIHSS = National Institutes of Health Stroke Scale; TICI = thrombolysis in cerebral infarction; tPA = tissue plasminogen
activator.
SUBMITTED  April 8, 2020.  ACCEPTED  July 2, 2020.
INCLUDE WHEN CITING  Published online November 13, 2020; DOI: 10.3171/2020.7.JNS201174.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  727

* | D*2 *35 *
Khanna et al.

FIG. 1. Illustrative cases of aortic arch configurations more amenable to radial artery catheterization than to femoral access.
A: Angiogram obtained in a 38-year-old female with a basilar thrombus, illustrating that patients with vertebrobasilar occlusions
can be treated via radial artery access, as it offers a direct access to the vertebral artery (white arrow) via the subclavian artery
(red arrow). B: CT angiogram obtained in a 74-year-old patient who presented with a left M1 occlusion. Patients with a bovine
aortic arch configuration who require catheterization of the left ICA (red arrow) are excellent candidates for radial artery cath-
eterization, which affords a less tortuous course for access via the subclavian artery (white arrow) compared with femoral artery
catheterization. C: Radial artery catheterization is favorable for accessing the right ICA in patients with type 2 or 3 aortic arches,
requiring less tension to be built up in the guidewire. Figure is available in color online only.

metrics for the 52 cases that were performed transradi- with thrombectomy via femoral or radial access. Eligible
al versus 52 that were carried out using traditional femo- patients are taken for thrombectomy in an interventional
ral artery catheterization during the same time span. Of suite located adjacent to the CT scanner, which typically
note, our case series includes the very first procedures commences in short order (less than 15 minutes) after
performed transradially and illustrates the learning curve CTA is completed. There were no set inclusion or exclu-
associated with adopting a novel treatment paradigm. To sion criteria for determining whether to pursue femoral
our knowledge, this represents the largest case series that versus radial access, and this was left to the discretion of
compares radial versus femoral artery access for acute the attending neurosurgeon, although patients who were
stroke intervention. found to have type 3 aortic arches and tortuous neck vas-
culature or basilar occlusion, as seen on preprocedural
Methods CTA for stroke workup, were more likely to be selected
Study Design for transradial catheterization. We accessed patients’ elec-
tronic medical records and collected relevant clinical data
The study protocol was approved by our institution- pertaining to their clinical presentation, imaging findings
al review board, and the need for informed consent was on stroke workup, and various relevant procedural met-
waived. A retrospective review was conducted of a pro- rics.
spectively maintained single-institution patient database
of all acute stroke interventions spanning 1 year from
November 2018 (when our first transradial thrombectomy Radial and Femoral Artery Catheterization Techniques
was performed) until December 2019. A total of 104 con- All acute stroke interventions included in this study
secutive cases are included in the overall study cohort, were conducted using conscious sedation. For cases uti-
split evenly between 52 consecutive patients who under- lizing radial catheterization, the right wrist is prepped
went thrombectomy via transradial access and a series of and draped, and subcutaneous lidocaine is administered.
the first 52 consecutive patients who underwent thrombec- Radial artery catheterization is achieved using ultrasound
tomy via transfemoral access during the same time frame. guidance via double wall puncture and the Seldinger tech-
Patients who necessitated access site crossover to com- nique. A 7F slim sheath is introduced, followed by the in-
plete the thrombectomy were excluded from analysis (4 jection of a radial cocktail consisting of 2000 U of hepa-
patients required converting from radial to femoral access rin (same dose regardless of whether a patient received
and 2 patients from femoral to radial access). intravenous tissue plasminogen activator [tPA]), 5 mg of
All patients presenting with acute stroke undergo CTA nicardipine, and 200 µg of nitroglycerin to prevent arterial
at our institution. The CT angiogram is then reviewed spasm. A radial artery run is performed to evaluate the
by the neuroendovascular fellow and attending surgeon, arterial anatomy, allowing the navigation of a 0.038-inch
and a decision is made at that time whether to proceed Glidewire (Terumo Interventional Systems) into the bra-

728 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 Khanna et al.

TABLE 1. Patient demographics and clinical presentations of 104 analysis, Fisher’s exact test, independent-samples t-test, or
patients who underwent acute stroke intervention via radial or Mann-Whitney U-test, as appropriate; p < 0.05 was con-
femoral access sidered statistically significant. Statistical analysis was
Radial Access Femoral Access carried out with IBM SPSS (version 24.0, IBM Corp.).
(n = 52) (n = 52) p Value
Sex, M/F 23/29 19/33 0.424
Results
Mean age, yrs 72.9 ± 13.4 69.5 ± 16.9 0.200 A total of 104 consecutive cases were included in our
Mean NIHSS score 16.9 ± 7.70 13.12 ± 7.05 0.922
analysis, which included 52 patients who underwent trans-
femoral access and 52 patients who underwent transradial
Vessel occlusion access (Table 1). There was no significant difference in age
  Rt sided 48.1% 46.2% >0.999 or sex between the two cohorts. The mean National Insti-
  Lt sided 51.9% 53.8% >0.999 tutes of Health Stroke Scale (NIHSS) score for patients
 Basilar 13.5% 1.90% 0.027 undergoing thrombectomy via transradial access was 16.9
 MCA 76.9% 82.7% 0.464 ± 7.70 and for those who underwent intervention via femo-
 ACA 1.90% 3.80% >0.999 ral access it was 13.12 ± 7.05 (p = 0.922). There was no sig-
nificant difference between the radial and femoral groups
 ICA 5.80% 5.80% >0.999
for time from symptom onset to time of presentation and
IV tPA given 0.230 administration of intravenous tPA. There was no signifi-
 Yes 34.6% 46.2% cant difference in the laterality of intervention between
 No 65.4% 53.8% the two groups presenting with anterior circulation vessel
>6 hrs since onset 0.332 occlusions or the location of the thrombus itself. However,
 Yes 36.5% 46.0% patients with a basilar occlusion were more likely to be
 No 63.5% 54.0%
treated via transradial access (p = 0.027) (Fig. 2).
The mean procedural time (defined as the time be-
ACA = anterior cerebral artery; IV = intravenous; MCA = middle cerebral tween the first and last fluoroscopic image acquisition)
artery. was similar between the two treatment cohorts: 60.4 ±
Patients with posterior circulation occlusions were more likely to undergo 36.8 minutes for the transradial group versus 65.5 ± 29.9
thrombectomy via radial artery catheterization. Mean values are presented as
the mean ± SD. Boldface type indicates statistical significance.
minutes for the transfemoral group (p = 0.451). The need
for stent placement was not significant based on which ac-
cess site was used. The total volume of contrast dose ad-
ministered during the procedure was similar between the
chial artery. The sheath is then exchanged for a 6F Cook two patient cohorts (107.30 ± 50.6 ml for the radial group
Shuttle (Cook Medical) 90-cm-long sheath, or a 6F 90-cm vs 88.5 ± 48.8 ml for the femoral group, p = 0.650). Simi-
Ballast long sheath (Balt USA LLC) over the exchange lar revascularization outcomes were achieved between
length wire. A small incision is made to facilitate place- the two treatment cohorts: the majority of procedures
ment of the larger working sheath, as needed. A 5F Sim yielded thrombolysis in cerebral infarction (TICI) grade
select (Penumbra, Inc.) is placed inside the Shuttle and is 2b/3 revascularization in both the transradial (92.3%) and
used to navigate to the aortic arch and select the vessel of transfemoral (94.2%) groups (p = 0.696), and there was
interest. The Shuttle is advanced over the Sim catheter, no difference in the mean number of passes performed
parked in the distal common carotid artery or proximal to achieve these results. The vast majority of thrombecto-
vertebral artery. The Sim catheter is then removed, and mies in both treatment cohorts were performed via a stent
the distal guide catheter of choice is inserted inside the retriever, although transradial access was associated with
long sheath. Mechanical thrombectomy is then begun in an increased incidence of requiring a combination of tech-
earnest using the operator’s technique of choice. niques required for recanalization (51.9% vs 5.80%, p <
After the procedure is complete, the sheath is removed, 0.001) (Table 2).
and a radial artery compression device (TR Band, Terumo Thrombectomies performed via radial and femoral
Interventional Systems) is applied. The compression band catheterization had similar rates of access site compli-
is inflated with air and is incrementally deflated over the cations. Five patients (9.6%) were found to have large,
course of an hour. superficial hematomas at the access site in the femoral
Transfemoral procedures are performed using an 8F group compared with 1 patient in the radial group (1.92%)
sheath and conducted in accordance with the operator’s (p = 0.093). There was 1 patient (1.92%) who was found
preferences based on anatomy and thrombus location. to have a femoral artery pseudoaneurysm that required
Vessel compression after the procedure was achieved us- obliteration with thrombin injection. We did not observe
ing manual pressure or using Angio-Seal (Terumo Inter- any patients who experienced symptomatic radial artery
ventional Systems). occlusion, and no patients who underwent radial artery
access required any subsequent treatment for access site
Statistical Analysis complications. The incidence of hemorrhagic conversion
Data were analyzed using descriptive and bivariate sta- after thrombectomy was similarly low in both treatment
tistical methods. Depending on the variable type, normal- cohorts. Of note, all patients in both treatment cohorts
ity was tested using a Shapiro-Wilk analysis, chi-square who were found to have hemorrhagic conversion of stroke

J Neurosurg Volume 135 • September 2021 729

 
* | D*2 *35 *
Khanna et al.

FIG. 2. Illustrative case of a 77-year-old female presenting with progressive lethargy and intractable nausea/vomiting (NIHSS
score 21) who was found to have a basilar occlusion. A: CT angiogram showing a type 3 aortic arch and a tortuous innominate
and subclavian artery with a sharp-angled origin of the right vertebral artery (RT VA; red arrow). B: A right subclavian injection
shows a favorable geometry to catheterize the right vertebral artery. C and D: Angiograms obtained prethrombectomy (C) and
postthrombectomy (D) that was performed via a stent retriever, resulting in TICI grade 2b revascularization. Figure is available in
color online only.

were diagnosed on routine postprocedural imaging, and vast majority of practicing endovascular neurosurgeons
none required any subsequent treatment. Patients’ modi- still prefer femoral artery access. Our reluctance of using
fied Rankin Scale scores at the time of discharge from radial artery catheterization for cerebral angiography was
the hospital were similar between the radial and femoral primarily due to the perceived difficulty of safely navigat-
cohorts (3.08 ± 1.93 vs 2.96 ± 1.82, p = 0.831). ing catheters into the cerebral vasculature, which is made
even more challenging in patients with anomalous aortic
Discussion arch anatomy and tortuous vessels. In recent years, there
Currently, radial artery access is not routinely used for have been many advances in endovascular technology,14
neuroendovascular procedures. Despite early case reports with the development of more durable and pliable cath-
detailing the efficacy of the transradial approach,12,13 the eters. Furthermore, several small case series have been

TABLE 2. Summary of procedural and clinical outcomes and complications for patients
undergoing acute stroke intervention via radial or femoral access
Radial Access (n = 52) Femoral Access (n = 52) p Value
Thrombectomy technique
  Stent retrieval 34.6% 73.1% 0.001
 Aspiration 11.5% 21.2% 0.185
  Stent placement 1.9% 0% 0.315
  Combination 51.9% 5.80% <0.001
Mean procedure duration, mins 60.4 ± 36.8 65.5 ± 29.9 0.451
Mean contrast dose, ml 107.30 ± 50.6 88.5 ± 48.8 0.650
Mean no. of passes 1.96 ± 1.21 2.12 ± 1.30 0.591
TICI revascularization grade
 0 1.9% 1.3% 0.315
 1 1.9% 0% 0.315
 2a 3.8% 5.8% 0.647
 2b/3 92.3% 94.2% 0.696
Access complications
  Superficial hematoma, n (%) 1 (1.9) 5 (9.6) 0.093
  Pseudoaneurysm, n (%) 0 1 (1.92) 0.315
Hemorrhagic conversion, n (%) 2 (3.85%) 3 (5.77%) 0.647
Mean mRS score at discharge 3.08 ± 1.93 2.96 ± 1.82 0.831
mRS = modified Rankin Scale.
Mean values are presented as the mean ± SD. Boldface type indicates statistical significance.

730 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 Khanna et al.

published that detail the feasibility and utility of using clinical outcomes between the two groups.19 Their study
radial access to perform cerebral angiography and me- also shed light on aortic arch configurations and how ra-
chanical thrombectomy.7,15 This study—to our knowledge, dial artery access provided a more favorable trajectory in
the largest case series to date—highlights the safety and patients with type 2 or 3 aortic arches or in those who
efficacy of radial artery catheterization for acute stroke in- were found to have tortuous vasculature on preoperative
tervention, with results comparable to those of traditional CTA. Indeed, our experience with transradial mechanical
transfemoral access. thrombectomy for anterior circulation occlusions began
This study includes a patient cohort of 104 stroke in- with using this technique principally for patients with pos-
terventions that were performed via transradial or trans- terior circulation lesions or difficult aortic arch configura-
femoral access. There was no difference in the presenting tions, both of which present favorable geometry for cath-
severity of stroke burden between the two patient cohorts. eterizing desired vessels via radial artery access (Video 1).
All studied metrics for procedural outcomes, including VIDEO 1. Various illustrative cases of acute stroke interventions
procedure time, contrast dose, and TICI revascularization performed via radial artery catheterization. Copyright Omaditya
grades, were found to be comparable between the trans- Khanna. Published with permission. Click here to view.
radial and transfemoral groups. In the treatment of acute These aortic arch configurations can be evaluated on pre-
stroke, the time from symptom onset to revasculariza- procedural CTA, and we encourage practitioners inter-
tion is paramount in preventing permanent neurological ested in pursuing transradial thrombectomy to transition
deficits and achieving favorable clinical outcomes.16,17 Our their practice using these favorable arch configurations. At
case series, which comprises our institution’s initial expe- our institution, transradial access has now become the de-
rience with transradial thrombectomy, shows that there is fault choice for 1) basilar thrombectomy and 2) left-sided
relative equipoise between radial and femoral artery ac- internal carotid artery (ICA) thrombectomy with bovine
cess for duration of needle-to–cerebral reperfusion times. arch configurations. Furthermore, we have been success-
Although we initially began to pursue transradial catheter- ful in navigating tortuous anatomy via radial artery cath-
ization only for diagnostic angiography, once we became eterization using access catheters and guide systems that
familiar with this mode of access, we began incorporating were initially developed for transfemoral access. As more
it for our acute stroke interventions, and the results of this practitioners adopt the transradial route, we postulate that
study illustrate the swift, manageable learning curve as- there will be newer device development that will aid in
sociated with adopting radial catheterization into clinical achieving even improved outcomes.
practice, with similar clinical outcomes achieved.
The overall rate of access site complications was low Conclusions
for radial artery access: 1 patient was found to have a
superficial hematoma, which resolved with conservative In this study, we have shown that acute stroke inter-
treatment. No patient had persistent symptoms as a re- vention can be successfully performed via radial artery
sult of radial artery catheterization. In the transfemoral catheterization, with no difference in clinical outcomes or
group, 5 patients (9.62%) were found to have superficial periprocedural complications compared with traditional
hematomas (albeit none had retroperitoneal hematomas), transfemoral access. Further studies in a larger patient co-
and 1 patient was found to have femoral artery pseudo- hort will need to be performed across several treatment
aneurysms that required treatment via thrombin injection. centers to further elucidate equipoise between transradial
Although our case series is small, which, in part, limits and transfemoral access for mechanical thrombectomy;
achieving statistical significance, the dichotomy in access specifically, whether radial artery catheterization is asso-
site complications does serve to highlight the potential ciated with a decreased risk of access site complications,
for reducing access site complications when radial artery as has been convincingly shown in the field of interven-
catheterization is utilized over femoral access. Indeed, as tional cardiology. We hope that the findings of this study,
the cardiac literature has revealed across multiple, large- although limited in scope and case volume, will encour-
scale, multicenter trials, radial artery access is associated age other neurointerventionalists to pursue radial artery
with an overall significant decrease in access site compli- access in their practice, so that our field may advance to-
cations, which has served as the impetus for the field to ward performing a randomized clinical trial to determine
change its practice such that radial access is preferred over whether this is associated with lower morbidities and im-
femoral access when feasable.18 Larger case series, includ- proved clinical outcomes.
ing, potentially, randomized trials, are needed to assess
the incidence of access site complications for radial versus References
femoral catheterization in patients undergoing neuroendo-   1. Fifi JT, Meyers PM, Lavine SD, et al. Complications of mod-
vascular intervention. ern diagnostic cerebral angiography in an academic medical
Several studies have recently been published that high- center. J Vasc Interv Radiol. 2009;​20(4):​442–447.
light the feasibility and efficacy of using radial artery ac-   2. Fargen KM, Velat GJ, Lawson MF, et al. Occurrence of angi-
cess for performing a wide range of neurointerventional ographic femoral artery complications after vascular closure
procedures.4,6,7 Chen et al. presented a study of 51 patients, with Mynx and AngioSeal. J Neurointerv Surg. 2013;​5(2):​
161–164.
18 of whom underwent thrombectomy via radial artery   3. Sorenson TJ, Nicholson PJ, Hilditch CA, et al. A lesson from
access and the remaining 33 via traditional femoral ar- cardiology:​the argument for ultrasound-guided femoral
tery access and found that there was no significant differ- artery access in interventional neuroradiology. World Neuro-
ence in operative times, periprocedural complications, or surg. 2019;​126:​124–128.

J Neurosurg Volume 135 • September 2021 731

 
* | D*2 *35 *
Khanna et al.

  4. Chen SH, Snelling BM, Shah SS, et al. Transradial approach 17. Spiotta AM, Vargas J, Turner R, et al. The golden hour of
for flow diversion treatment of cerebral aneurysms:​a multi- stroke intervention:​effect of thrombectomy procedural time
center study. J Neurointerv Surg. 2019;​11(8):​796–800. in acute ischemic stroke on outcome. J Neurointerv Surg.
  5. Snelling BM, Sur S, Shah SS, et al. Transradial approach 2014;​6(7):​511–516.
for complex anterior and posterior circulation interventions:​ 18. Kolkailah AA, Alreshq RS, Muhammed AM, et al. Transra-
technical nuances and feasibility of using current devices. dial versus transfemoral approach for diagnostic coronary an-
Oper Neurosurg (Hagerstown). 2019;​17(3):​293–302. giography and percutaneous coronary intervention in people
  6. Khanna O, Sweid A, Mouchtouris N, et al. Radial artery with coronary artery disease. Cochrane Database Syst Rev.
catheterization for neuroendovascular procedures. Stroke. 2018;​4:​CD012318.
2019;​50(9):​2587–2590. 19. Chen SH, Snelling BM, Sur S, et al. Transradial versus trans-
  7. Snelling BM, Sur S, Shah SS, et al. Transradial cerebral angi- femoral access for anterior circulation mechanical throm-
ography:​techniques and outcomes. J Neurointerv Surg. 2018;​ bectomy:​comparison of technical and clinical outcomes. J
10(9):​874–881. Neurointerv Surg. 2019;​11(9):​874–878.
  8. Jolly SS, Yusuf S, Cairns J, et al. Radial versus femoral ac-
cess for coronary angiography and intervention in patients
with acute coronary syndromes (RIVAL):​a randomised, Disclosures
parallel group, multicentre trial. Lancet. 2011;​377(9775):​ The authors report no conflict of interest concerning the materi-
1409–1420. als or methods used in this study or the findings specified in this
  9. Andò G, Capodanno D. Radial versus femoral access in in- paper.
vasively managed patients with acute coronary syndrome:​a
systematic review and meta-analysis. Ann Intern Med. 2015;​ Author Contributions
163(12):​932–940.
10. Hess CN, Krucoff MW, Sheng S, et al. Comparison of qual- Conception and design: Jabbour, Khanna, Sweid, Mouchtouris.
ity-of-life measures after radial versus femoral artery access Acquisition of data: Khanna, Velagapudi, Das, Sweid. Analysis
for cardiac catheterization in women:​results of the Study and interpretation of data: Jabbour, Khanna, Velagapudi, Sweid,
of Access Site for Enhancement of Percutaneous Coronary Mouchtouris, Al Saiegh, Avery, Chalouhi, Schmidt, Sajja,
Intervention for Women quality-of-life substudy. Am Heart J. Gooch, Tjoumakaris, Rosenwasser. Drafting the article: Jabbour,
2015;​170(2):​371–379. Khanna. Critically revising the article: Jabbour, Khanna, Sweid,
11. Safley DM, Amin AP, House JA, et al. Comparison of costs Mouchtouris, Al Saiegh, Avery, Chalouhi, Schmidt, Sajja, Gooch,
between transradial and transfemoral percutaneous coronary Tjoumakaris, Rosenwasser. Reviewed submitted version of
intervention:​a cohort analysis from the Premier research manuscript: Khanna, Velagapudi, Das. Approved the final version
database. Am Heart J. 2013;​165(3):​303–309.e2 of the manuscript on behalf of all authors: Jabbour. Statistical
12. Matsumoto Y, Hongo K, Toriyama T, et al. Transradial ap- analysis: Velagapudi, Mouchtouris. Study supervision: Gooch,
proach for diagnostic selective cerebral angiography:​results Tjoumakaris, Rosenwasser.
of a consecutive series of 166 cases. AJNR Am J Neuroradiol.
2001;​22(4):​704–708. Supplemental Information
13. Levy EI, Boulos AS, Fessler RD, et al. Transradial cerebral Videos
angiography:​an alternative route. Neurosurgery. 2002;​51(2):​ Video 1. https://vimeo.com/443015912.
335–342.
14. Rosenwasser RH, Lang M, Tjoumakaris S, Jabbour P. Dis- Correspondence
ruptive innovation in neurovascular disease. Neurosurgery.
Pascal M. Jabbour: Thomas Jefferson University Hospital,
2017;​64(CN_suppl_1):​78–82.
Philadelphia, PA. [email protected].
15. McCarthy DJ, Chen SH, Brunet MC, et al. Distal radial
artery access in the anatomical snuffbox for neurointerven-
tions:​case report. World Neurosurg. 2019;​122:​355–359.
16. Alawieh A, Vargas J, Fargen KM, et al. Impact of procedure
time on outcomes of thrombectomy for stroke. J Am Coll
Cardiol. 2019;​73(8):​879–890.

732 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 CLINICAL ARTICLE
J Neurosurg 135:733–741, 2021

Outcomes of stereotactic radiosurgery for hemorrhagic

arteriovenous malformations with or without prior
resection or embolization
Mariko Kawashima, MD,1 Hirotaka Hasegawa, MD, PhD,1,2 Masahiro Shin, MD, PhD,1
Yuki Shinya, MD,1 Osamu Ishikawa, MD, PhD,1 Satoshi Koizumi, MD,1 Atsuto Katano, MD, PhD,3
Hirofumi Nakatomi, MD, PhD,1 and Nobuhito Saito, MD, PhD1

Departments of 1Neurosurgery and 3Radiology, University of Tokyo Hospital, Tokyo, Japan; and 2Department of Neurologic
Surgery, Mayo Clinic, Rochester, Minnesota

OBJECTIVE  The major concern about ruptured arteriovenous malformations (rAVMs) is recurrent hemorrhage, which
tends to preclude stereotactic radiosurgery (SRS) as a therapeutic modality for these brain malformations. In this study,
the authors aimed to clarify the role of SRS for rAVM as a stand-alone modality and an adjunct for a remnant nidus after
surgery or embolization.
METHODS  Data on 410 consecutive patients with rAVMs treated with SRS were analyzed. The patients were classified
into groups, according to prior interventions: SRS-alone, surgery and SRS (Surg-SRS), and embolization and SRS (Em-
bol-SRS) groups. The outcomes of the SRS-alone group were analyzed in comparison with those of the other two groups.
RESULTS  The obliteration rate was higher in the Surg-SRS group than in the SRS-alone group (5-year cumulative
rate 97% vs 79%, p < 0.001), whereas no significant difference was observed between the Embol-SRS and SRS-alone
groups. Prior resection (HR 1.78, 95% CI 1.30–2.43, p < 0.001), a maximum AVM diameter ≤ 20 mm (HR 1.81, 95% CI
1.43–2.30, p < 0.001), and a prescription dose ≥ 20 Gy (HR 2.04, 95% CI 1.28–3.27, p = 0.003) were associated with a
better obliteration rate, as demonstrated by multivariate Cox proportional hazards analyses. In the SRS-alone group, the
annual post-SRS hemorrhage rates were 1.5% within 5 years and 0.2% thereafter and the 10-year significant neurologi-
cal event–free rate was 95%; no intergroup difference was observed in either outcome. The exclusive performance of
SRS (SRS alone) was not a risk for post-SRS hemorrhage or for significant neurological events based on multivariate
analyses. These results were also confirmed with propensity score–matched analyses.
CONCLUSIONS  The treatment strategy for rAVMs should be tailored with due consideration of multiple factors associ-
ated with the patients. Stand-alone SRS is effective for hemorrhagic AVMs, and the risk of post-SRS hemorrhage was
low. SRS can also be favorably used for residual AVMs after initial interventions, especially after failed resection.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201502
KEYWORDS  Gamma Knife; long-term outcomes; ruptured arteriovenous malformation; stereotactic radiosurgery;

A
vascular disorders

dvances in imaging studies and improved access to widely considered reasonable, although recent prospective
these developments have led to an increase in the studies have demonstrated skepticism regarding the safety
incidental diagnosis of brain arteriovenous mal- of prophylaxis interventions for unruptured AVMs.10,11
formation (AVM); however, hemorrhagic stroke remains Stereotactic radiosurgery (SRS) effectively obliterates
the prominent reason for its diagnosis.1,2 In addition, hem- the nidus by inducing endothelial damage followed by the
orrhage due to rupture is associated with mortality and progressive thickening of intimal layers and the subsequent
morbidity rates of up to 29% and 34%, respectively.3–6 It thrombosis of nidal vessels over a couple of years.12 Al-
is well known that the risk of future bleeding increases though SRS is a reasonable minimally invasive therapeutic
in AVMs with a history of bleeding.7–9 Because of this, option for small to medium AVMs, some argue that a pe-
therapeutic interventions for ruptured AVMs (rAVMs) are riod of latency before obliteration of the nidus would pose

ABBREVIATIONS  AVM = arteriovenous malformation; DSA = digital subtraction angiography; mRBAS = modified radiosurgery-based AVM score; mRS = modified Rankin
Scale; rAVM = ruptured AVM; RIC = radiation-induced change; SMG = Spetzler-Martin grade; SNE = significant neurological event; SRS = stereotactic radiosurgery; VRAS
= Virginia Radiosurgery AVM Scale.
SUBMITTED  April 28, 2020.  ACCEPTED  July 13, 2020.
INCLUDE WHEN CITING  Published online December 4, 2020; DOI: 10.3171/2020.7.JNS201502.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  733

* | D*2 *35 *
Kawashima et al.

a certain risk of stroke, especially in rAVM cases,13,14 and cally confirmed intracranial hemorrhage that had occurred
thus resection or endovascular treatment is preferred. Ac- within or adjacent to the nidus and caused clinical symp-
cordingly, in cases of rAVMs, SRS has been mainly used toms. Radiation-induced changes (RICs) were defined
for posttreatment remnants or rAVMs in deep brain loca- as peripheral edema that was shown as a hyperintensity
tions for which surgical intervention is challenging. In other or hypoattenuation area on T2-weighted MRI or CT, re-
words, SRS alone for rAVM has been scarcely highlighted, spectively. Delayed adverse events such as cyst formation,
and actual outcomes have yet to be elucidated. Therefore, encapsulated hematoma, and tumorigenesis within an ir-
in this study, we aimed to assess the efficacy and safety of radiated field were all identified as such.17 A significant
SRS for rAVM as a stand-alone modality and as an adjunc- neurological event (SNE) was defined as any event causing
tive treatment following resection and embolization. a > 1-point decline in the modified Rankin Scale (mRS)
score and associated with SRS, further treatments for the
Methods AVM, or post-SRS hemorrhage. The radiographic results
were assessed not only by the attending neurosurgeons but
Patient Selection also by radiologists at our institution in a blinded manner,
Of 3064 procedures (in the SRS database) performed whereas the clinical assessment was performed only by
between 1990 and 2016 at our institution, 512 procedures the attending neurosurgeons.
in 475 patients with rAVMs were identified. AVMs with
radiographic findings suggesting previous hemorrhage Statistical Analysis
(hemosiderin deposition or cavity formation) without clin- First, baseline characteristics of the patients during
ical signs or symptoms were not considered true rAVMs. SRS were summarized and compared among the groups
After exclusions for planned volume-staged SRS (n = 28), using Fisher’s exact test and the Mann-Whitney U-test
a history of previous radiation (n = 7), and no radiological for categorical and continuous variables, respectively.
follow-up (n = 30), a total of 410 patients who had under- The outcomes of prior treatments had been prospectively
gone 411 initial radiosurgeries for rAVMs were included recorded, and changes in the mRS score were retrospec-
in our study. Notably, 1 patient had 2 separate rAVMs tively evaluated. The modified radiosurgery-based AVM
treated with SRS alone. score (mRBAS),18 Spetzler-Martin grade (SMG),19and
The patients were classified into three groups, depend- Virginia Radiosurgery AVM Scale (VRAS) score20 were
ing on their interventions before SRS: 1) SRS alone, 2) sur- calculated based on their definitions.
gery and SRS (Surg-SRS), and 3) embolization and SRS Next, the nidus obliteration, post-SRS hemorrhage, and
(Embol-SRS). Surgery was defined as any direct surgical SNE-free rates were calculated with the Kaplan-Meier
intervention to the nidus; hematoma evacuation without method and compared between groups using the log-
intervention to the nidus, ventricular drainage, and clip- rank test. The post-SRS hemorrhage rates were similarly
ping for a feeder aneurysm were excluded. Patients who evaluated using the person-year method, with the period
had undergone resection after embolization were allocated categorized in two phases—within 5 years from SRS (la-
to the Surg-SRS group. Embolization was defined as any tency phase) and thereafter. Factors potentially affecting
endovascular embolization to the nidus, feeders, or intra- the rates, as mentioned earlier, were evaluated using the
nidal aneurysm; however, embolization to a flow-related Cox proportional hazards model for bivariate and multi-
aneurysm was excluded. This study was approved by the variate analyses. Continuous variables were roughly di-
clinical research review board at the University of Tokyo, chotomized with their median values. The factors entered
and all participants provided written informed consent at into the multivariate model were selected using the for-
the time of treatment. ward stepwise selection method with a cutoff p value of
0.25. RICs were evaluated as crude rates and compared
Radiosurgical Techniques and Posttreatment Follow-Up using Fisher’s exact test and logistic regression analysis for
The Leksell Gamma Knife (Elekta AB) was used in all bivariate and multivariate analyses, respectively.
cases. Detailed procedures were described in previous arti- Potential differences between groups can lead to sig-
cles.14–16 Digital subtraction angiography (DSA) was solely nificant biases, and to address this issue, propensity score
used as the stereotactic imaging modality before Febru- matching was performed as a sensitivity analysis: 1) SRS-
ary 1991; thereafter, CT (March 1991–July 1996) and MRI alone group versus Surg-SRS group and 2) SRS-alone
(August 1996–2016) were used in combination to increase group versus Embol-SRS group. Propensity scores were
planning accuracy. Radiosurgical plans were created with generated using a binary logistic regression model with
commercially available planning software (KULA until the following variables: sex, patient age at SRS, maximum
1998 and Leksell GammaPlan thereafter; Elekta AB). A diameter and volume of AVM, deep drainage, eloquent
prescription dose of 20 Gy was selected for most cases, location, mRBAS, SMG, and VRAS score. Thereafter,
although the doses were lowered in cases of large AVMs one-to-one matching without any replacement was com-
(usually > 10 ml in volume) and eloquent AVMs based on pleted using the nearest neighbor match with a caliper of
the surgeons’ discretion. The patients were clinically and 0.05. The obliteration rates, post-SRS hemorrhage rates,
radiographically followed up with MRI every 6 months and SNE-free rates were evaluated with the Kaplan-Meier
before obliteration and every year thereafter. Nidus oblit- method and log-rank test.
eration was confirmed with DSA, although MRI was alter- A p value < 0.05 was considered statistically signifi-
natively used when DSA was contraindicated or rejected. cant. All analyses were performed using JMP Pro 14 soft-
Post-SRS hemorrhage was defined as any radiographi- ware (SAS Institute Inc.).

734 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 Kawashima et al.

TABLE 1. Baseline patient and AVM characteristics

Variable Overall SRS-Alone Group Surg-SRS Group Embol-SRS Group p Value*
No. of cases 411 306 60 45
Male sex, no. (%) 217 (53) 166 (54) 30 (50) 21 (47) 0.572/0.341
Median age at SRS in yrs (range) 28 (4–80) 30 (4–80) 24 (8–64) 28 (5–76) 0.113/0.681
Significant complications prior to SRS,† no. (%) 6 (10) 5 (11)
mRS score at SRS, no. (%)
 0–2 353 (86) 270 (88) 47 (78) 36 (80)
0.032/0.871
 3–5 58 (14) 36 (12) 13 (22) 9 (20)
Median max diameter in mm (range) 19 (3–68) 19 (3–68) 17 (7–48)‡ 25 (9–60)‡ 0.143/<0.001
Median nidus vol in ml (range) 1.4 (0.1–44.5) 1.4 (0.1–23.5) 1.0 (0.1–16.7)‡ 4.1 (0.2–44.5)‡ 0.052/<0.001
Median prescription dose in Gy (range) 20 (10–28) 20 (10–28) 20 (18–25.2) 20 (17–25.2) 0.014/0.399
Median mRBAS (range) 0.97 (0.18–5.47) 0.99 (0.18–3.28) 0.74 (0.25–2.13) 1.06 (0.37–5.47) <0.001/0.242
Eloquent location, no. (%) 277 (67) 213 (70) 33 (55) 32 (71) 0.036/0.864
Deep drainage, no. (%) 292 (71) 228 (75) 34 (57) 30 (67) 0.007/0.280
SMG, no. (%)
 I–II 171 (42) 121 (40) 33 (55) 17 (38)
0.032/0.871
  ≥III 240 (58) 185 (60) 27 (45) 28 (62)
VRAS score, no. (%)
 1–2 258 (63) 195 (64) 45 (75) 18 (40)
0.103/0.003
 3–4 153 (37) 111 (36) 15 (25) 27 (60)
Boldface type indicates statistical significance (p < 0.05).
* A p value for the SRS-alone versus Surg-SRS groups/SRS-alone versus Embol-SRS groups.
† Those causing a > 1-point decline in the mRS score were considered significant.
‡ Data at the time of SRS.

Results Nidus Obliteration

Baseline Characteristics Among the entire cohort, nidus obliteration was con-
Baseline characteristics are summarized in Table 1. The firmed in 296 (72%) cases at a median of 25 months after
SRS-alone, Surg-SRS, and Embol-SRS groups comprised SRS, 260 (88%) of whom underwent DSA. The cumula-
306 (74.5%), 60 (14.6%), and 45 (10.9%) cases, respective- tive obliteration rates were 61% and 81% at 3 and 5 years,
ly. The median follow-up periods were 111 months (range respectively (Fig. 1A). Thirty-two patients underwent sec-
1–351 months), 121 months (5–317 months), and 79 months ondary SRS at a median of 53 months after the initial treat-
(13–325 months) in the SRS-alone, Surg-SRS, and Embol- ment; of these patients, nidus obliteration was confirmed
SRS groups, respectively. The median intervals from hem- in 22, yielding a final crude obliteration rate of 77%.
orrhage to SRS were 3.7 months, 10.6 months (0.4 months The cumulative 5-year obliteration rates were 79%,
from hemorrhage to initial resection), and 8.7 months (2.0 97%, and 75% in the SRS-alone, Surg-SRS, and Embol-
months from onset to initial embolization) in the SRS- SRS groups, respectively (Fig. 1B). The rate was signifi-
alone, Surg-SRS, and Embol-SRS groups, respectively. cantly higher in the Surg-SRS group than in the SRS-alone
The median intervals from the completion of prior inter- (p < 0.001) and Embol-SRS (p = 0.003) groups; however,
ventions to SRS were 10.1 and 3.4 months in the Surg-SRS no difference was observed between the SRS-alone and
and Embol-SRS groups, respectively. Persistent complica- Embol-SRS groups (p = 0.544). The multivariate Cox pro-
tions of preceding treatments causing a > 1-point decline in portional hazards analyses demonstrated that prior resec-
the mRS score were observed in 6 (10%) and 5 (11%) cases tion (HR 1.78, 95% CI 1.30–2.43, p < 0.001), a maximum
in the Surg-SRS and Embol-SRS groups, respectively. AVM diameter ≤ 20 mm (HR 1.81, 95% CI 1.43–2.30, p
Compared to cases in the Surg-SRS group, cases in the < 0.001), and a prescription dose ≥ 20 Gy (HR 2.04, 95%
SRS-alone group were more likely to develop deep drain- CI 1.28–3.27, p = 0.003) were associated with a higher
age (57% vs 75%, p = 0.007) and be treated with slightly obliteration rate (Table 2).
lower marginal doses (p = 0.014). No difference in nidus
volume was observed between the SRS-alone and Surg- Post-SRS Hemorrhage
SRS groups (p = 0.052). Compared to cases in the Embol- Post-SRS hemorrhage was observed in 26 cases (6.3%)
SRS group, cases in the SRS-alone group were more likely either within 5 years after SRS (20 patients) or thereaf-
to have a smaller nidus and a smaller maximum diameter ter (6 patients), yielding an annual post-SRS hemorrhage
(median nidus volume 4.1 vs 1.4 ml, p < 0.001; median rate of 1.2% within 5 years after SRS and 0.2% thereaf-
maximum diameter 25 vs 19 mm, p < 0.001). ter. The annual post-SRS hemorrhage rates were 1.5% in

J Neurosurg Volume 135 • September 2021 735

 
* | D*2 *35 *
Kawashima et al.

FIG. 1. Kaplan-Meier curves showing cumulative obliteration rates after initial SRS for the entire cohort (A) and between groups
before propensity score matching (B). Figure is available in color online only.

the SRS-alone group (19 patients over 1270 patient-years), (2-point decline in the mRS score), 5 developed mild func-
0% in the Surg-SRS group (0 patients over 278 patient- tional disturbance (1-point decline in the mRS score), and
years), and 0.6% in the Embol-SRS group (1 patient over 6 developed transient or no functional disturbance. Post-
165 patient-years) within 5 years after SRS; however, the SRS hemorrhage occurred in 2 patients following nidus
rates were 0.2% in the SRS-alone group (4 patients over obliteration; one case was caused by a de novo nidus adja-
2261 patient-years), 0.2% in the Surg-SRS group (1 patient cent to the original one, which triggered a mild neurologi-
over 523 patient-years), and 0.3% in the Embol-SRS group cal decline.21 The other case occurred from an unknown
(1 patient over 349 patient-years) thereafter. etiology but at the same location as the original nidus, re-
The log-rank tests showed no significant difference in sulting in no functional decline.
the cumulative hemorrhage rates between groups (Fig. 2).
The multivariate Cox proportional hazards analysis dem-
onstrated that a prescription dose ≥ 20 Gy (HR 0.31, 95% Radiation-Induced Adverse Events and Neurological
CI 0.13–0.73, p = 0.007) was associated with a lower hem- Outcomes After SRS
orrhage rate (Table 3). In the entire cohort, symptomatic RICs were observed
Among the patients with post-SRS hemorrhage, 7 died, in 28 cases (6.8%), consisting of 22 patients (7.2%) in the
2 were severely disabled, 6 were moderately disabled SRS-alone group, 3 (5.0%) in the Surg-SRS group, and 3

TABLE 2. Factors associated with nidus obliteration after initial SRS

Bivariate Analysis Multivariate Analysis
Variable p Value HR (95% CI) p Value HR (95% CI)
Age <30 yrs 0.763 1.04 (0.82–1.30) — —
Male sex 0.158 0.85 (0.67–1.07) — —
Max AVM diameter ≤20 mm <0.001 1.91 (1.50–2.42) <0.001 1.81 (1.43–2.30)
Nidus vol ≤2 ml <0.001 1.85 (1.46–2.35) — —
Prescription dose ≥20 Gy <0.001 2.38 (1.49–3.79) 0.003 2.04 (1.28–3.27)
Deep drainage 0.170 0.84 (0.66–1.08) 0.272 0.87 (0.68–1.12)
Eloquent location 0.152 0.84 (0.66–1.07) — —
Prior resection <0.001 1.93 (1.42–2.63) <0.001 1.78 (1.30–2.43)
Prior embolization 0.291 0.81 (0.55–1.19) 0.783 0.95 (0.64–1.40)
SMG I–II <0.001 1.52 (1.21–1.92) — —
mRBAS <1.0 <0.001 0.63 (0.50–0.79) — —
VRAS score 1–2 <0.001 1.90 (1.49–2.42) — —
Boldface type indicates statistical significance (p < 0.05).

736 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 Kawashima et al.

patients (2.0%) in the SRS-alone group, none in the Surg-

SRS group, and 1 (2.2%) in the Embol-SRS group. In ad-
dition, no significant difference in disease-specific deaths
was observed among the groups.

Matched Cohort Analysis

Propensity score matching between the SRS-alone and
Surg-SRS groups yielded matched groups with 56 patients
in each group. There were no significant differences in
baseline characteristics between the two groups (Supple-
mentary Table 1). The log-rank test demonstrated that the
obliteration rate was significantly higher in the Surg-SRS
group than in the SRS-alone group (p = 0.034). The post-
SRS hemorrhage rates (p = 0.288) and SNE-free rates (p =
0.621) were not significantly different.
Propensity score matching between the SRS-alone and
FIG. 2. Kaplan-Meier curves showing the post-SRS hemorrhage rates. Embol-SRS groups yielded matched groups with 40 pa-
Figure is available in color online only.
tients in each group. There were no significant differences
in baseline variables between the matched cohorts (Sup-
plementary Table 2). There was no significant difference
(6.7%) in the Embol-SRS group. The multivariate logis- in the obliteration rates (p = 0.423), post-SRS hemorrhage
tic regression analysis revealed an association between a rates (p = 0.115), and SNE-free rates (p = 0.430) between
maximum AVM diameter ≤ 20 mm (HR 0.42, 95% CI the two groups.
0.19–0.95, p = 0.037) and a lower rate of symptomatic
RICs (Table 4).
Delayed adverse events were observed in 20 cases Discussion
(4.9%) at a median interval of 161 months after SRS, in- This study demonstrated that SRS alone provides a fa-
cluding chronic encapsulated hematoma in 19 patients vorable obliteration rate (79% at 5 years) with a reason-
and a malignant neoplasm in 1 patient. There was no sig- able safety profile. The post-SRS hemorrhage risk in the
nificant difference in the occurrence of delayed adverse SRS-alone group was low (1.5% within 5 years and 0.2%
events between the groups. thereafter), especially given that the reported annual hem-
In the entire cohort, SNEs were observed in 20 cas- orrhagic risk of untreated rAVMs ranges between 4.5%
es (4.9%), consisting of 17 (5.6%), 1 (1.7%), and 2 (4.4%) and 7.5%.7,22,23 Moreover, the decent SNE-free rate (95%
patients in the SRS-alone, Surg-SRS, and Embol-SRS at 10 years) further substantiates the safety of SRS alone.
groups, respectively. The 10-year SNE-free rates were Although numerous studies have reported AVM treatment
95% in the entire cohort, 95% in both the SRS-alone and outcomes, few recent studies have focused on rAVM,6,24–27
Embol-SRS groups, and 97% in the Surg-SRS group. No and very few of them have described the outcomes of SRS
significant differences were observed in the SNE-free alone for rAVM (Table 5). Hence, this study would serve
rates among the groups (Fig. 3). In the entire cohort, dis- as an important basis for selecting an optimal intervention
ease-specific deaths were observed in 7 patients (1.7%), 6 for rAVM.

TABLE 3. Factors associated with post-SRS hemorrhage

Bivariate Analysis Multivariate Analysis
Variable p Value HR (95% CI) p Value HR (95% CI)
Age <30 yrs 0.106 0.52 (0.24–1.15) — —
Male sex 0.432 1.37 (0.62–3.03) — —
Max diameter ≤20 mm 0.102 1.92 (0.88–4.17) 0.195 0.59 (0.27–1.31)
Nidus vol ≤2 ml 0.166 0.58 (0.27–1.25) — —
Prescription dose ≥20 Gy 0.001 0.25 (0.11–0.57) 0.007 0.31 (0.13–0.73)
Deep drainage 0.278 1.72 (0.65–4.55) — —
Eloquent location 0.193 1.91 (0.72–5.08) 0.337 1.62 (0.23–1.65)
Prior resection 0.139 0.22 (0.03–1.63) 0.230 0.29 (0.04–2.19)
Prior embolization 0.562 0.65 (0.15–2.76) 0.320 0.48 (0.11–2.06)
SMG I–II 0.149 0.53 (0.22–1.26) — —
mRBAS <1.0 0.040 0.43 (0.19–0.96) — —
VRAS score 1–2 0.083 0.50 (0.23–1.09) — —
Boldface type indicates statistical significance (p < 0.05).

J Neurosurg Volume 135 • September 2021 737

 
* | D*2 *35 *
Kawashima et al.

TABLE 4. Factors associated with symptomatic RIC

Bivariate Analysis Multivariate Analysis
Variable p Value HR (95% CI) p Value HR (95% CI)
Age <30 yrs 0.847 1.09 (0.50–2.35) — —
Male sex 1.000 0.99 (0.46–2.13) — —
Max AVM diameter ≤20 mm 0.017 0.38 (0.17–0.84) 0.037 0.42 (0.19–0.95)
Nidus vol ≤2 ml 0.110 0.50 (0.23–1.08) — —
Prescription dose ≥20 Gy 0.525 0.70 (0.23–2.12) — —
Deep drainage 0.086 2.57 (0.87–7.59) 0.107 2.45 (0.82–7.31)
Eloquent location 0.217 1.84 (0.73–4.65) — —
Prior resection 0.782 0.69 (0.20–2.35) 0.770 0.83 (0.23–2.92)
Prior embolization 1.000 1.03 (0.30–3.54) 0.703 1.28 (0.36–4.58)
SMG I–II 0.074 0.44 (0.18–1.07) — —
mRBAS <1.0 0.169 0.54 (0.25–1.19) — —
VRAS score 1–2 0.027 0.42 (0.19–0.91) — —
Boldface type indicates statistical significance (p < 0.05).

It should be clarified that the emergency management gical hemorrhage, and a 7%–15% chance of neurological
of acutely ill rAVM patients has to be considered sepa- decline.31–33 Surgical morbidities were observed in 10%
rately because patients in the current study do not directly of our cohort. Although SRS has been traditionally con-
reflect such patients. According to expert opinion, hemato- sidered the second-best method for treating rAVM and
ma evacuation and decompression are warranted to save a has been spared for deep-seated or eloquent niduses, it
life in cases of acute rupture depending on its severity.28 It seems acceptable to consider SRS as a reasonable alter-
is probably better both to resect small, superficial, nonelo- native stand-alone modality given the favorable outcomes
quent AVMs and to embolize in an acute or intermediate obtained in our study. In addition, the restorative use of
fashion any obviously accessible flow-related or intranidal SRS following surgery is equally feasible, as the multivari-
aneurysms.29,30 After the acute phase, definitive treatment ate analysis demonstrated an association between prior
for AVM should be performed, a situation similar to that surgical intervention and a better chance of obliteration
in the current study.26,30 compared to that with SRS alone. Although there were
The main advantage of surgical treatment is the im- some differences in baseline characteristics between the
mediate reduction of hemorrhage risk. Nevertheless, treatment groups, similar results were confirmed with the
a postoperative remnant is rare but possible, and a zero matched cohort analyses. The reason for the better oblit-
risk of postoperative hemorrhage is not always obtained. eration rate in the Surg-SRS group is unclear; however, the
Surgery-related complications can matter as well. Indeed, partial disconnection of blood inflow might have impeded
recent surgical series have reported an 87%–100% chance the intranidal flow, possibly facilitating radiation-induced
of complete obliteration, up to a 14% chance of postsur- intraluminal thrombosis and subsequent obliteration. On
the contrary, the more that procedures are performed, the
more likely treatment-related invasiveness and complica-
tions will become tangible; thus, a simpler treatment is
generally better, if feasible. SRS would be applicable as a
stand-alone or salvage treatment for a remnant nidus fol-
lowing a planned complete resection.
The interval between hemorrhage and SRS was rela-
tively longer (10.6 months) in this study than expected.
This is mainly because most of the patients were referred
to us after having undergone acute care and the subsequent
rehabilitation. Additionally, residual AVMs were discov-
ered on follow-up studies performed several months after
the detection of hemorrhage. Nevertheless, such findings
were unlikely to impact the outcomes of SRS.
While the post-SRS hemorrhage rate in the latency phase
was slightly better in the Embol-SRS group, no advanta-
geous or disadvantageous effect was observed in multivari-
ate analyses. In principle, embolization is an equally good
FIG. 3. Kaplan-Meier curves displaying the SNE-free rates. Figure is alternative; the success rate has been reported to be higher
available in color online only. than 90%, especially for SMG I–II AVMs.34,35 Regarding
738 J Neurosurg Volume 135 • September 2021
 
* | D*2 *35 *
 TABLE 5. Literature review of studies focusing on rAVM
Authors No. of Age AVM Initial Acute Functional FU Obliteration
& Year Cases (yrs) Grading Symptoms Management Mortality Outcome Scores (mos) Rate Notes
Shotar et 135 Mean, SMG: I–II, 67%; III, 20%; GCS: <5, 10%; <13, Surgery 44%, embol, 20.7% mRS: 0–2, 57%; NR NR Outcomes per treatment not
al., 201824 42 IV–V, 14% 41% 19% ≥3, 43% provided
Fukuda et 101 Mean, Mean max AVM diameter, NIHSS: 0, 26%; 1–9, Surgery 40%, embol 8% mRS: 0–2, 66%; NR NR Outcomes per treatment not
al., 20176 45 2.5 cm; deep venous 29%; ≥10, 45% 44%, surgery & ≥3, 34% provided
drainage, 43% embol 21%
Todnem et 16 Mean, SMG: III, 75%; IV, 25% mRS: 1–2, 33%; ≥3, Embol & SRS 100% 0% mRS: 0, 19%; 1–2, 45 57% Only SMG III or IV AVMs includ-
al., 201925 47 67% 63%; ≥3, 19% (crude rate) ed; no post-SRS hemorrhage
observed
Ding et 565 Median, SMG*: I–II, 44%; ≥III, NR SRS alone 62–79%,† 0.4% NR 57 64%/5 yrs Post-SRS hemorrhage: 2.0%/yr;
al., 201426 29 56%; median RBAS, surgery & SRS 17%, permanent radiation-induced
1.08 embol & SRS 21% morbidity: 5.1%
Tam et al., 33 Median, SMG: I–II, 61%; III–IV, GCS: 13–15, 67%; Urgent surgery 36%, 3% GOS: 5, 91%; 4, 89 73%/3 yrs Only pediatric patients included;
201927 12 30%; data missing, 9% 9–12, 9%; 3–8, 15%; interval surgery 21%, 3%; 3, 3%; 1, 3% (after SRS) outcomes per treatment
data missing, 9% SRS alone 33%, not provided; no post-SRS
SRS & others 12% hemorrhage observed
Present 411 Median, SMG*: I–II, 42%; ≥III, mRS*: 0, 36%; 1–2, SRS alone 75%, 2% SNE‡ rate, 5% 109 82%/5 yrs Post-SRS hemorrhage ≤5 yrs:

* | D*2 *35 *
study 28 58%; median RBAS, 49%; ≥3, 14% surgery & SRS 15%, 1.2%/yr; 10-year SNE-free
0.97 embol & SRS 11% rate: 95%
Embol = embolization; FU = follow-up; GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale; NIHSS = National Institutes of Health Stroke Scale; NR = data not reported; RBAS = radiosurgery-based AVM score.
* Parameters at the time of SRS.
 
† These data were not provided and were thus calculated from other provided values. Depending on the overlap between patients with prior interventions, SRS alone ranges between 62% and 79%.
‡ SNE was defined as a > 1-point decline in the mRS score that was associated with SRS, further treatments for the AVM, or post-SRS hemorrhage.

J Neurosurg Volume 135 • September 2021

739
Kawashima et al.
Kawashima et al.

the downsides, hemorrhagic complications are possible in evidenced by the provision of a 79% 5-year obliteration
up to 11% of cases, postprocedural ischemic lesions can oc- rate, 1.5% annual latency phase hemorrhage rate, and 95%
cur in up to 22% of cases, and overall morbidity and mortal- 10-year SNE-free rate. SRS can also be favorably used for
ity can be up to 14% and 3%, respectively.35–37 In the current residual AVMs after initial interventions; notably, excel-
study, several patients actually experienced morbidities as lent results were achieved in AVMs for which resection
well. Embolization with SRS, initially deemed a great com- had failed.
bination (because embolization could reduce the nidus size
down to the level at which SRS could be suitably applied), References
eventually became questionable because the obliteration
  1. Friedlander RM. Clinical practice. Arteriovenous malforma-
rate might be unfavorably decreased.38–40 However, this re- ​
tions of the brain. N Engl J Med. 2007;356(26):2704–2712.​
mains a matter of ongoing discussion, and this combination   2. Choi JH, Mohr JP. Brain arteriovenous malformations in
should not be abandoned yet.41 Indeed, in our study, obliter- ​
adults. Lancet Neurol. 2005;4(5):299–308. ​
ation rates in the Embol-SRS group were decent, although   3. Brown RD Jr, Wiebers DO, Forbes G, et al. The natural his-
the nidus size was larger than that in the SRS-alone group. tory of unruptured intracranial arteriovenous malformations.
Concurrently, it is worth shifting attention (regarding em- ​
J Neurosurg. 1988;68(3):352–357. ​
bolization) from size reduction to the targeted embolization   4. ApSimon HT, Reef H, Phadke RV, Popovic EA. A popula-
tion-based study of brain arteriovenous malformation:​long-
of possible bleeding points.42,43 The maximum size limit in ​
term treatment outcomes. Stroke. 2002;33(12):2794–2800.​
SRS for AVM has gradually increased in the last 2 decades.   5. Choi JH, Mast H, Sciacca RR, et al. Clinical outcome after
It has been indicated that single-session SRS with a regular first and recurrent hemorrhage in patients with untreated
dose is safe even for a large AVM (10–20 ml).44 For larger brain arteriovenous malformation. Stroke. 2006;37(5):1243– ​ ​
AVMs (> 20 ml), volume-staged SRS was adopted after the 1247.
mid-2000s.45 Therefore, pre-SRS embolization aiming for   6. Fukuda K, Majumdar M, Masoud H, et al. Multicenter as-
volume reduction may not be as important as before; the sessment of morbidity associated with cerebral arteriovenous
goal of pre-SRS embolization has been shifted to targeted malformation hemorrhages. J Neurointerv Surg. 2017;9(7): ​ ​
664–668.
embolization for possible bleeding points such as an intra-   7. Pollock BE, Flickinger JC, Lunsford LD, et al. Factors that
nidal aneurysm.29 predict the bleeding risk of cerebral arteriovenous malforma-
This study has several limitations. First, as a retrospec- ​
tions. Stroke. 1996;27(1):1–6. ​
tive study, it is inherently subject to bias resulting from the   8. Stapf C, Mast H, Sciacca RR, et al. Predictors of hemorrhage
selection of treatment modalities. To minimize the poten- in patients with untreated brain arteriovenous malformation.
tial bias, we performed not only multivariate analysis but ​ ​
Neurology. 2006;66(9):1350–1355.
also matched cohort analyses, which produced similar re-   9. Yen CP, Sheehan JP, Schwyzer L, Schlesinger D. Hemorrhage
risk of cerebral arteriovenous malformations before and dur-
sults. Hence, the obtained results are valid and reliable. Al- ing the latency period after GAMMA knife radiosurgery.
though a prospective study is desirable to evaluate the ef- ​ ​
Stroke. 2011;42(6):1691–1696.
ficacy of SRS for hemorrhagic AVMs, it would be difficult 10. Mohr JP, Parides MK, Stapf C, et al. Medical management
to conduct such a study given the scarcity of AVMs and with or without interventional therapy for unruptured brain
the need for emergency care in such cases. Therefore, a arteriovenous malformations (ARUBA):​a multicentre, non-
retrospective case series including the maximum possible ​
blinded, randomised trial. Lancet. 2014;383(9917):614–621. ​
number of patients is the best way forward. Second, since 11. Al-Shahi Salman R, White PM, Counsell CE, et al. Out-
high-grade AVMs requiring staged SRS were excluded, come after conservative management or intervention for
unruptured brain arteriovenous malformations. JAMA. 2014;​
this study did not provide any recommendations regarding ​
311(16):1661–1669.
their management. Such cases would likely require a mul- 12. Schneider BF, Eberhard DA, Steiner LE. Histopathology of
tidisciplinary treatment strategy involving the tolerance arteriovenous malformations after gamma knife radiosur-
of some mild adverse effects. Third, recent evolutions in ​
gery. J Neurosurg. 1997;87(3):352–357. ​
endovascular techniques were not considered because the 13. Karlsson B, Lax I, Söderman M. Risk for hemorrhage during
majority of patients in the Embol-SRS group were treated the 2-year latency period following gamma knife radiosur-
before the 2000s. Similarly, the efficacy of targeted em- gery for arteriovenous malformations. Int J Radiat Oncol
​ ​
Biol Phys. 2001;49(4):1045–1051.
bolization could not be referenced, as there were only a 14. Maruyama K, Kawahara N, Shin M, et al. The risk of hemor-
few such patients in the cohort. Further case accumula- rhage after radiosurgery for cerebral arteriovenous malfor-
tion and reanalysis would therefore be desirable. Despite ​
mations. N Engl J Med. 2005;352(2):146–153. ​
these limitations, this study is of scientific interest because 15. Hanakita S, Shin M, Koga T, et al. Risk reduction of cerebral
it demonstrated the long-term outcome of SRS for rAVM stroke after stereotactic radiosurgery for small unruptured
per treatment modality—not only as a sole treatment mo- brain arteriovenous malformations. Stroke. 2016;47(5):1247– ​ ​
dality, but also as a salvage treatment after failed resection 1252.
or embolization. 16. Shin M, Maruyama K, Kurita H, et al. Analysis of nidus
obliteration rates after gamma knife surgery for arteriovenous
malformations based on long-term follow-up data:​the Univer-
Conclusions ​
sity of Tokyo experience. J Neurosurg. 2004;101(1):18–24. ​
17. Hasegawa H, Hanakita S, Shin M, et al. A comprehensive
Although the definitive treatment for rAVM should be study of symptomatic late radiation-induced complications
determined after due consideration of a patient’s charac- after radiosurgery for brain arteriovenous malformation:​
teristics, AVM radiographic features, and in-facility cir- incidence, risk factors, and clinical outcomes. World Neuro-
cumstances, SRS alone would be a reasonable option, as surg. 2018;116:e556–e565.
​ ​

740 J Neurosurg Volume 135 • September 2021

 
* | D*2 *35 *
 Kawashima et al.

18. Pollock BE, Flickinger JC. Modification of the radiosurgery- 37. Saatci I, Geyik S, Yavuz K, Cekirge HS. Endovascular treat-
based arteriovenous malformation grading system. Neurosur- ment of brain arteriovenous malformations with prolonged
gery. 2008;​63(2):​239–243. intranidal Onyx injection technique:​long-term results in 350
19. Spetzler RF, Martin NA. A proposed grading system for arte- consecutive patients with completed endovascular treatment
riovenous malformations. J Neurosurg. 1986;​65(4):​476–483. course. J Neurosurg. 2011;​115(1):​78–88.
20. Starke RM, Yen CP, Ding D, Sheehan JP. A practical grading 38. Andrade-Souza YM, Ramani M, Scora D, et al. Embolization
scale for predicting outcome after radiosurgery for arteriove- before radiosurgery reduces the obliteration rate of arteriove-
nous malformations:​analysis of 1012 treated patients. J Neu- nous malformations. Neurosurgery. 2007;​60(3):​443–452.
rosurg. 2013;​119(4):​981–987. 39. Kano H, Kondziolka D, Flickinger JC, et al. Stereotactic
21. Kawashima M, Hasegawa H, Kurita H, et al. Ectopic recur- radiosurgery for arteriovenous malformations after emboliza-
rence of arteriovenous malformation after radiosurgery:​case tion:​a case-control study. J Neurosurg. 2012;​117(2):​265–275.
report and insight regarding pathogenesis. World Neurosurg. 40. Pollock BE, Flickinger JC, Lunsford LD, et al. Factors asso-
2020;​135:​63–67. ciated with successful arteriovenous malformation radiosur-
22. Gross BA, Du R. Natural history of cerebral arteriovenous gery. Neurosurgery. 1998;​42(6):​1239–1247.
malformations:​a meta-analysis. J Neurosurg. 2013;​118(2):​ 41. Hasegawa H, Yamamoto M, Shin M, Barfod BE. Gamma
437–443. knife radiosurgery for brain vascular malformations:​current
23. da Costa L, Wallace MC, Ter Brugge KG, et al. The natural evidence and future tasks. Ther Clin Risk Manag. 2019;​15:​
history and predictive features of hemorrhage from brain 1351–1367.
arteriovenous malformations. Stroke. 2009;​40(1):​100–105. 42. Alexander MD, Hippe DS, Cooke DL, et al. Targeted emboli-
24. Shotar E, Debarre M, Sourour NA, et al. Retrospective study zation of aneurysms associated with brain arteriovenous mal-
of long-term outcome after brain arteriovenous malformation formations at high risk for surgical resection:​a case-control
rupture:​the RAP score. J Neurosurg. 2018;​128(1):​78–85. study. Neurosurgery. 2018;​82(3):​343–349.
25. Todnem N, Ward A, Nahhas M, et al. A retrospective cohort 43. Sun Y, Jin H, Li Y, Tian Z. Target embolization of associated
analysis of hemorrhagic arteriovenous malformations treated aneurysms in ruptured arteriovenous malformations. World
with combined endovascular embolization and Gamma Knife Neurosurg. 2017;​101:​26–32.
stereotactic radiosurgery. World Neurosurg. 2019;​122:​e713– 44. Hasegawa H, Hanakita S, Shin M, et al. Re-evaluation of the
e722. size limitation in single-session stereotactic radiosurgery for
26. Ding D, Yen CP, Starke RM, et al. Radiosurgery for ruptured brain arteriovenous malformations:​detailed analyses on the
intracranial arteriovenous malformations. J Neurosurg. 2014;​ outcomes with focusing on radiosurgical doses. Neurosur-
121(2):​470–481. gery. 2020;​86(5):​685–696.
27. Tam KY, Lim K, Zhu CXL, et al. Long-term outcomes of 45. Hanakita S, Shin M, Koga T, et al. Outcomes of volume-
ruptured cerebral arteriovenous malformations in the paedi- staged radiosurgery for cerebral arteriovenous malforma-
atric population:​a retrospective review in a regional hospital tions larger than 20 cm3 with more than 3 years of follow-up.
in Hong Kong. J Clin Neurosci. 2019;​66:​66–70. World Neurosurg. 2016;​87:​242–249.
28. Kato Y, Dong VH, Chaddad F, et al. Expert consensus on the
management of brain arteriovenous malformations. Asian J
Neurosurg. 2019;​14(4):​1074–1081. Disclosures
29. Kano H, Kondziolka D, Flickinger JC, et al. Aneurysms in- The authors report no conflicts of interest concerning the materi-
crease the risk of rebleeding after stereotactic radiosurgery als or methods used in this study or the findings specified in this
for hemorrhagic arteriovenous malformations. Stroke. 2012;​ paper.
43(10):​2586–2591.
30. Ding D, Chen CJ, Starke RM, et al. Risk of brain arteriove- Author Contributions
nous malformation hemorrhage before and after stereotactic
radiosurgery. Stroke. 2019;​50(6):​1384–1391. Conception and design: Kawashima, Hasegawa. Acquisition of
31. Korja M, Bervini D, Assaad N, Morgan MK. Role of surgery data: Kawashima, Ishikawa, Koizumi, Katano. Analysis and inter-
in the management of brain arteriovenous malformations:​ pretation of data: Kawashima, Hasegawa. Drafting the article:
prospective cohort study. Stroke. 2014;​45(12):​3549–3555. Kawashima. Critically revising the article: Hasegawa, Shin,
32. Schramm J, Schaller K, Esche J, Boström A. Microsurgery Shinya. Reviewed submitted version of manuscript: Shin, Shinya,
for cerebral arteriovenous malformations:​subgroup outcomes Ishikawa, Koizumi, Katano. Study supervision: Shin, Nakatomi,
in a consecutive series of 288 cases. J Neurosurg. 2017;​ Saito.
126(4):​1056–1063.
33. Theofanis T, Chalouhi N, Dalyai R, et al. Microsurgery for Supplemental Information
cerebral arteriovenous malformations:​postoperative out- Online-Only Content
comes and predictors of complications in 264 cases. Neuro- Supplemental material is available with the online version of the
surg Focus. 2014;​37(3):​E10. article.
34. van Rooij WJ, Jacobs S, Sluzewski M, et al. Curative em- Supplementary Tables 1 and 2. https://thejns.org/doi/suppl/​
bolization of brain arteriovenous malformations with onyx:​ 10.3171/​​2020.7.JNS201502.
patient selection, embolization technique, and results. AJNR
Am J Neuroradiol. 2012;​33(7):​1299–1304. Correspondence
35. Katsaridis V, Papagiannaki C, Aimar E. Curative emboliza-
Mariko Kawashima: University of Tokyo Hospital, Tokyo, Japan.
tion of cerebral arteriovenous malformations (AVMs) with
[email protected].
Onyx in 101 patients. Neuroradiology. 2008;​50(7):​589–597.
36. Baharvahdat H, Blanc R, Termechi R, et al. Hemorrhagic
complications after endovascular treatment of cerebral ar-
teriovenous malformations. AJNR Am J Neuroradiol. 2014;​
35(5):​978–983.

J Neurosurg Volume 135 • September 2021 741

 
* | D*2 *35 *
 CLINICAL ARTICLE
J Neurosurg 135:742–750, 2021

Stereotactic radiosurgery with versus without prior Onyx

embolization for brain arteriovenous malformations
Ching-Jen Chen, MD,1 Dale Ding, MD,2 Cheng-Chia Lee, MD, PhD,3 Kathryn N. Kearns, BS,1
I. Jonathan Pomeraniec, MD, MBA,1 Christopher P. Cifarelli, MD, PhD,4
David E. Arsanious, MD,4 Roman Liscak, MD,5 Jaromir Hanuska, MD,5 Brian J. Williams, MD,2
Mehran B. Yusuf, MD,6 Shiao Y. Woo, MD,6 Natasha Ironside, MBChB,1
Rebecca M. Burke, MD, PhD,1 Ronald E. Warnick, MD,7 Daniel M. Trifiletti, MD,8
David Mathieu, MD,9 Monica Mureb, BS,10 Carolina Benjamin, MD,10 Douglas Kondziolka, MD,10
Caleb E. Feliciano, MD,11 Rafael Rodriguez-Mercado, MD,11 Kevin M. Cockroft, MD, MSc,12
Scott Simon, MD,12 Heath B. Mackley, MD,13 Samer G. Zammar, MD, MPH,12 Neel T. Patel, MD,12
Varun Padmanaban, MD,12 Nathan Beatson, BS,14 Anissa Saylany, BA,14 John Y. K. Lee, MD,14 and
Jason P. Sheehan, MD, PhD,1 on behalf of the International Radiosurgery Research Foundation
1
Department of Neurological Surgery, University of Virginia, Charlottesville, Virginia; Departments of 2Neurosurgery and
6
Radiation Oncology, University of Louisville, Kentucky; 3Department of Neurosurgery, Taipei Veterans General Hospital, Taipei,
Taiwan; 4Department of Neurosurgery, West Virginia University, Morgantown, West Virginia; 5Department of Neurosurgery, Na
Homolce Hospital, Prague, Czech Republic; 7Department of Neurosurgery, The Jewish Hospital, Cincinnati, Ohio; 8Department
of Radiation Oncology, The Mayo Clinic, Jacksonville, Florida; 9Department of Neurosurgery, University of Sherbrooke, Canada;
10
Department of Neurosurgery, New York University, New York, New York; 11Department of Neurosurgery, University of Puerto
Rico, San Juan, Puerto Rico; Departments of 12Neurosurgery and 13Radiation Oncology, Pennsylvania State University, Hershey,
Pennsylvania; and 14Department of Neurosurgery, University of Pennsylvania, Philadelphia, Pennsylvania

OBJECTIVE  Investigations of the combined effects of neoadjuvant Onyx embolization and stereotactic radiosurgery
(SRS) on brain arteriovenous malformations (AVMs) have not accounted for initial angioarchitectural features prior
to neuroendovascular intervention. The aim of this retrospective, multicenter matched cohort study is to compare the
outcomes of SRS with versus without upfront Onyx embolization for AVMs using de novo characteristics of the pre-
embolized nidus.
METHODS  The International Radiosurgery Research Foundation AVM databases from 1987 to 2018 were retrospec-
tively reviewed. Patients were categorized based on AVM treatment approach into Onyx embolization (OE) and SRS
(OE+SRS) or SRS alone (SRS-only) cohorts and then propensity score matched in a 1:1 ratio. The primary outcome was
AVM obliteration. Secondary outcomes were post-SRS hemorrhage, all-cause mortality, radiological and symptomatic
radiation-induced changes (RICs), and cyst formation. Comparisons were analyzed using crude rates and cumulative
probabilities adjusted for competing risk of death.
RESULTS  The matched OE+SRS and SRS-only cohorts each comprised 53 patients. Crude rates (37.7% vs 47.2%
for the OE+SRS vs SRS-only cohorts, respectively; OR 0.679, p = 0.327) and cumulative probabilities at 3, 4, 5, and 6
years (33.7%, 44.1%, 57.5%, and 65.7% for the OE+SRS cohort vs 34.8%, 45.5%, 59.0%, and 67.1% for the SRS-only
cohort, respectively; subhazard ratio 0.961, p = 0.896) of AVM obliteration were similar between the matched cohorts.
The secondary outcomes of the matched cohorts were also similar. Asymptomatic and symptomatic embolization-related
complication rates in the matched OE+SRS cohort were 18.9% and 9.4%, respectively.
CONCLUSIONS  Pre-SRS AVM embolization with Onyx does not appear to negatively influence outcomes after SRS.
These analyses, based on de novo nidal characteristics, thereby refute previous studies that found detrimental effects
of Onyx embolization on SRS-induced AVM obliteration. However, given the risks incurred by nidal embolization using

ABBREVIATIONS  AVM = arteriovenous malformation; CI = confidence interval; CIF = cumulative incidence function; DSA = digital subtraction angiography; OE = Onyx
embolization; OR = odds ratio; RIC = radiation-induced change; SHR = subhazard ratio; SRS = stereotactic radiosurgery.
SUBMITTED  May 10, 2020.  ACCEPTED  July 14, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS201731.

742 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *36 *
 Chen et al.

Onyx, this neoadjuvant intervention should be used judiciously in multimodal treatment strategies involving SRS for ap-
propriately selected large-volume or angioarchitecturally high-risk AVMs.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201731
KEYWORDS  arteriovenous malformation; stereotactic radiosurgery; embolization; stroke; endovascular; vascular
disorders

C stereotactic radiosurgery (SRS) de-

onventional Baseline Data and Variables
livered in a single session struggles to successfully Patient variables included age, sex, and prior AVM
treat large (diameter > 3 cm or volume > 12 ml) therapy (i.e., before embolization). AVM variables includ-
brain arteriovenous malformations (AVMs).1–3 Endovas- ed prior hemorrhage, volume, eloquent location, deep lo-
cular embolization has been used as a neoadjuvant inter- cation, and presence of deep venous drainage. AVM nidus
vention to render large nidi more amenable to SRS and volume prior to embolization was measured on catheter
target high-risk AVM-associated features, such as arterial digital subtraction angiography (DSA) using the ABC/2
aneurysms and high-flow intranidal arteriovenous fistu- method.13 Eloquent location was defined by Spetzler and
las.4–6 However, the role of upfront AVM embolization Martin as “sensorimotor, language, and visual cortices,
has come under scrutiny due to accumulating evidence of hypothalamus and thalamus, internal capsule, brainstem,
its potentially deleterious effect on obliteration rates after cerebellar peduncles, and deep cerebellar nuclei.”14 Deep
SRS.7 location was defined as thalamus, basal ganglia, and brain-
The etiology of lower post-SRS obliteration rates in stem.15 SRS data included margin dose.
embolized AVMs has been purported to stem from the
physical properties of ethylene vinyl alcohol copolymer Follow-Up and Outcomes
(Onyx, Medtronic Neurovascular), the most frequently The primary outcome was AVM obliteration on MRI
used embolysate in contemporary neurointerventional or DSA. Radiological follow-up was typically performed
procedures.8,9 However, previous studies that analyzed at 6-month intervals for the first 2 years and then annually
the effect of prior embolization on AVM SRS outcomes thereafter. Patients with evidence of AVM obliteration on
failed to account for the baseline nidal dimensions and follow-up noninvasive neuroimaging were recommended
angioarchitectural features before embolization.7,10–12 This to undergo confirmatory DSA. AVM obliteration was de-
represents a fundamental flaw that pervades the avail- fined as a lack of abnormal flow voids (based on MRI) or
able literature pertaining to the relationship between prior an absence of anomalous arteriovenous shunting (based
Onyx embolization and post-SRS outcomes. The aim of on DSA). In addition to radiological evidence of residual
this multicenter, retrospective matched cohort study is to arteriovenous shunting, the AVM was considered to be
compare the outcomes of SRS with versus without neoad- patent if any additional intervention targeting the nidus
juvant Onyx embolization using de novo characteristics of was performed after SRS. Time to obliteration was cen-
the pre-embolized nidus. sored at the time of obliteration, additional AVM interven-
tion, or loss to follow-up.
Methods Secondary outcomes were post-SRS hemorrhage, all-
This study follows the guidelines outlined in the cause mortality, radiological and symptomatic radiation-
Strengthening the Reporting of Observational Studies in induced changes (RICs), cyst formation, and asymptom-
Epidemiology (STROBE) statement, and it was approved atic and symptomatic embolization-related complications.
by the IRB of each participating center. Patient consent Post-SRS hemorrhage was defined as any AVM-related in-
was waived by each IRB. Institutions within the Interna- tracranial hemorrhage during the follow-up period, regard-
tional Radiosurgery Research Foundation were invited to less of associated neurological symptoms or lack thereof.
contribute AVM cases treated between 1987 and 2018. Radiological RIC was defined as the presence of perinidal
Each respective institution was independently responsible hyperintensities on T2-weighted or FLAIR MRI sequenc-
for verification and attestation of data accuracy. Individual es. Symptomatic RIC was defined as any new or worsening
patient data from each contributing center were de-identi- deterioration of neurological status in a patient with radio-
fied and pooled by an independent third party. logical RIC. Clinical and radiological follow-ups were ob-
The inclusion criteria were 1) treatment with single- tained concurrently, when feasible. When in-person follow-
session SRS (including Gamma Knife radiosurgery and up could not be obtained, clinical and neuroimaging data
linear accelerator–based radiosurgery) with or without from referring institutions or physicians were acquired and
upfront Onyx embolization; 2) available baseline data reviewed by the respective treatment center.
regarding patient, AVM, SRS, and embolization charac-
teristics; and 3) available outcome data after SRS. Exclu- Statistical Analysis
sion criteria were 1) AVMs embolized without employing All statistical analyses were performed using Stata (ver-
Onyx, and 2) treatment with dose- or volume-staged SRS. sion 15.1, StataCorp). Baseline characteristics and follow-
AVMs treated with combined endovascular embolization up durations were compared between the OE+SRS and
using Onyx as an embolic agent followed by SRS were SRS-only cohorts using Pearson’s chi-square or Fisher’s
categorized into the OE+SRS cohort, whereas AVMs exact tests for categorical variables, as appropriate, and
treated with SRS alone (i.e., without prior embolization) the Student t-test for continuous variables. To control for
were categorized into the SRS-only cohort. baseline differences, the OE+SRS and SRS-only cohorts

J Neurosurg Volume 135 • September 2021 743

 
* | D*2 *36 *
Chen et al.

TABLE 1. Comparison of baseline characteristics and follow-up durations between the unmatched OE+SRS versus
SRS-only cohorts
Characteristic OE+SRS Cohort, n = 68 SRS-Only Cohort, n = 1110 p Value
Mean age (SD), yrs 41.4 (17.6) 35.7 (16.6) 0.008
Females (%) 44/68 (64.7) 548/1110 (49.4) 0.014
Prior AVM surgery (%) 8/68 (11.8) 166/1104 (15.0) 0.461
Prior AVM fractionated radiation therapy (%) 3/68 (4.4) 4/959 (0.4) <0.001
Prior AVM hemorrhage (%) 39/68 (57.4) 571/1110 (51.4) 0.344
Deep AVM location (%) 7/68 (10.3) 276/1109 (24.9) 0.006
Eloquent AVM location (%) 35/62 (56.5) 610/1058 (57.7) 0.852
Deep venous drainage (%) 35/68 (51.5) 444/1109 (40.0) 0.062
Mean AVM volume (SD), ml 15.5 (16.4) 4.2 (7.9) <0.001
Mean SRS margin dose (SD), Gy 19.2 (2.7) 21.0 (3.6) <0.001
Mean mos radiological follow-up (SD) 41.7 (40.3) 49.3 (40.2) 0.140
Mean mos clinical follow-up (SD) 47.7 (40.6) 68.1 (50.3) 0.002
Boldface type indicates statistical significance.

were matched, without replacement in a 1:1 ratio, with a vs 0.4%, p < 0.001). Deep AVM location was more com-
caliper of 0.2 standard deviation of the logit of the pro- mon in the SRS-only cohort (24.9% vs 10.3%, p = 0.006).
pensity score using greedy matching. Propensity scores De novo AVM volume was larger in the OE+SRS cohort
were derived using a logistic regression model accounting (15.5 vs 4.2 ml, p < 0.001), whereas margin dose was higher
for AVM volumes. The PSMATCH2 package developed in the SRS-only cohort (mean 21.0 vs 19.2 Gy, p < 0.001).
for Stata was used for propensity score derivation and the The clinical follow-up duration was longer in the SRS-only
matching process.16 Standardized differences were used to cohort (mean 68.1 vs 47.4 months, p = 0.002).
assess the balance of baseline data, and differences < 0.20
between pretreatment characteristics of the matched co- Comparisons of Outcomes Between Unmatched OE+SRS
horts were considered an adequate balance. Versus SRS-Only Cohorts
The primary and secondary outcomes between the Table 2 compares the outcomes of the unmatched
OE+SRS versus SRS-only cohorts, before and after OE+SRS versus SRS-only cohorts. The AVM obliteration
matching, were compared using binary logistic regression rate was lower in the OE+SRS cohort (39.4% vs 64.0%,
analyses, and each of these comparisons were reported OR 0.365, 95% CI 0.219–0.607, p < 0.001). However, cu-
as odds ratios (ORs) with 95% confidence intervals (CIs).
mulative probabilities of obliteration at 3, 4, 5, and 6 years
Fisher’s exact test was performed for outcomes with zero
were similar between the two cohorts (35.0%, 48.6%,
frequencies. The cumulative rates of AVM obliteration,
post-SRS hemorrhage, radiological and symptomatic 54.4%, and 58.6% for the OE+SRS cohort vs 43.1%, 58.2%,
RICs, and cyst formation were compared between the two 64.2%, and 68.5% for the SRS-only cohort, respectively;
cohorts, before and after matching, using cumulative in- SHR 0.763, 95% CI 0.533–1.094, p = 0.141; Fig. 1A).
cidence functions (CIFs) and competing-risk regression The crude rates of each secondary outcome were simi-
models adjusting for the competing risk of death, and each lar between the unmatched cohorts. Cumulative prob-
of these comparisons was reported as a subhazard ratio abilities at 3, 4, 5, and 6 years of post-SRS hemorrhage
(SHR) with a 95% CI.17,18 CIFs and corresponding SHRs (6.5%, 8.7%, 11.0%, and 11.7% for the OE+SRS cohort vs
were not generated for outcomes with zero frequencies. 4.0%, 5.4%, 6.9%, and 7.3% for the SRS-only cohort, re-
Statistical significance was defined as p < 0.05, and all spectively; SHR 1.633, 95% CI 0.628–4.250, p = 0.314;
tests were two-tailed. Missing data were not imputed. Fig. 1B) and radiological RICs (24.5%, 25.0%, 26.1%, and
26.4% for the OE+SRS cohort vs 35.5%, 36.2%, 37.6%,
and 38.0% for the SRS-only cohort, respectively; SHR
Results 0.641, 95% CI 0.367–1.118, p = 0.117; Fig. 1C) were simi-
Overall Cohort Characteristics lar between the two cohorts. However, cumulative prob-
The overall study cohort comprised 1178 patients, in- abilities of symptomatic RICs at 3, 4, 5, and 6 years were
cluding 68 and 1110 patients in the OE+SRS and SRS-only significantly higher in the OE+SRS cohort (22.7%, 24.2%,
cohorts, respectively. Table 1 compares the baseline char- 25.6%, and 26.3% for the OE+SRS cohort vs 13.6%,
acteristics and follow-up durations between the unmatched 14.5%, 15.5%, and 15.9% for the SRS-only cohort, respec-
cohorts. Patients in the OE+SRS cohort were significantly tively; SHR 1.760, 95% CI 1.004–3.084, p = 0.048; Fig.
older (mean age 41.4 vs 35.7 years, p = 0.008) and more 1D). Cumulative probabilities of cyst formation at 3, 4, 5,
likely to be female (64.7% vs 49.4%, p = 0.014). The and 6 years were similar between the two cohorts (0.9%,
OE+SRS cohort had a greater proportion of patients who 0.9%, 2.9%, and 5.0% for the OE+SRS cohort vs 0.3%,
underwent prior AVM fractionated radiation therapy (4.4% 0.3%, 0.9%, and 1.5% for the SRS-only cohort, respec-

744 J Neurosurg Volume 135 • September 2021

 
* | D*2 *36 *
 Chen et al.

TABLE 2. Comparison of outcomes between the unmatched OE+SRS versus SRS-only cohorts
Outcome OE+SRS Cohort, n = 68 SRS-Only Cohort, n = 1110 OR (95% CI) p Value
Primary outcome
  AVM obliteration (%) 26/66 (39.4) 698/1090 (64.0) 0.365 (0.219–0.607) <0.001
Secondary outcomes (%)
  Post-SRS hemorrhage 5/61 (8.2) 65/1091 (6.0) 1.409 (0.546–3.639) 0.478
  All-cause mortality 1/66 (1.5) 31/1108 (2.8) 0.534 (0.072–3.977) 0.541
  Radiological RICs 20/63 (31.8) 373/1068 (34.9) 0.867 (0.502–1.495) 0.607
  Symptomatic RICs 13/63 (20.6) 141/1071 (13.2) 1.715 (0.908–3.237) 0.096
  Cyst formation 2/63 (3.2) 13/1079 (1.2) 2.689 (0.593–12.181) 0.200
  Asymptomatic embolization-related complications 16/68 (23.5) — — —
  Symptomatic embolization-related complications 5/68 (7.4) — — —
Boldface type indicates statistical significance.

tively; SHR 3.375, 95% CI 0.745–15.292, p = 0.115; Fig. of post-SRS hemorrhage at 3, 4, 5, and 6 years were simi-
1E). Asymptomatic and symptomatic embolization-relat- lar between the two cohorts (9.5%, 9.5%, 9.5%, and 9.5%
ed complication rates were 23.5% and 7.4%, respectively. for the OE+SRS cohort vs 9.9%, 9.9%, 9.9%, and 9.9%
for the SRS-only cohort, respectively; SHR 0.959, 95% CI
Matched Cohort Characteristics 0.265–3.478, p = 0.950; Fig. 2B). Cumulative probabili-
The matched OE+SRS and SRS-only cohorts each ties at 3 and 5 years of radiological (25.2% and 27.1% for
comprised 53 patients. Table 3 compares the baseline char- the OE+SRS cohort vs 38.8% and 41.4% for the SRS-only
acteristics and follow-up durations between the matched cohort; SHR 0.591, 95% CI 0.275–1.270, p = 0.178; Fig.
cohorts. After matching, patients in the OE+SRS cohort 2C) and symptomatic (25.6% and 28.8% for the OE+SRS
remained significantly older (mean age 43.1 vs 34.4 years, cohort vs 18.0% and 20.4% for the SRS-only cohort; SHR
p = 0.013) and more likely to have prior AVM hemorrhage 1.487, 95% CI 0.600–3.686, p = 0.392; Fig. 2D) RICs were
(60.4% vs 39.6%, p = 0.033). The clinical follow-up du- also similar between the two cohorts. Asymptomatic and
ration also remained significantly longer in the SRS-only symptomatic embolization-related complication rates
cohort (61.9 vs 43.2 months, p = 0.036). were 18.9% and 9.4%, respectively.
Table 4 summarizes the standardized differences in
baseline characteristics between the OE+SRS and SRS- Discussion
only cohorts before and after propensity score matching.
Imbalances in baseline characteristics after matching Volumetric reduction of large AVMs using emboliza-
were noted for age (0.404), prior AVM fractionated radia- tion to facilitate subsequent SRS of the smaller residual
tion therapy (0.209), prior AVM hemorrhage (0.350), and nidus is a conceptually appealing multimodal treatment,
deep location (−0.281). AVM volume and margin dose but this combined approach has been reported to yield a
were well balanced between the matched cohorts. Of the lower likelihood of obliteration compared to stand-alone
17 patients in the matched OE+SRS cohort with available SRS.6,7 As such, pre-SRS embolization has not become a
embolization results, volume reduction alone, flow reduc- mainstay therapy in AVM management. However, studies
tion alone, flow and volume reduction, flow reduction and supporting this notion performed their comparative analy-
aneurysm occlusion, and flow and volume reduction and ses with postembolization AVM characteristics, which in-
aneurysm occlusion were achieved in 11.8%, 5.9%, 47.1%, herently biased their findings against embolized AVMs.
5.9%, and 29.4%, respectively. That is, initially larger volumes and more complex angio-
architectures of AVMs selected for pre-SRS embolization
Comparisons of Outcomes Between Matched OE+SRS have not been accounted for by existing reports.
Versus SRS-Only Cohorts The predominant use of Onyx in contemporary AVM
Table 5 compares the outcomes of the matched OE+SRS embolization has raised concerns that it could affect AVM
versus SRS-only cohorts. Obliteration rates were similar be- SRS outcomes differently from other embolysates. The
tween the matched OE+SRS (37.7%) and SRS-only (47.2%) radiopaque component of Onyx, tantalum, may attenuate
cohorts (OR 0.679, 95% CI 0.313–1.472, p = 0.327). Cumu- the effective radiation dose via a shielding effect. Tanta-
lative probabilities of obliteration at 3, 4, 5, and 6 years were lum’s high atomic number of 73 causes significant image
also similar between the matched cohorts (33.7%, 44.1%, artifacts at typical kilovoltage CT energies, and therefore it
57.5%, and 65.7% for the OE+SRS cohort vs 34.8%, 45.5%, could distort radiosurgical dose distribution.9,19 Addition-
59.0%, and 67.1% for the SRS-only cohort, respectively; ally, extensive beam-hardening artifacts from tantalum on
SHR 0.961, 95% CI 0.530–1.743, p = 0.896; Fig. 2A). CT angiography and a similar hypointense appearance of
The crude rates of each secondary outcome were simi- Onyx to AVM flow voids on T2-weighted MRI can impair
lar between the matched cohorts. Cumulative probabilities the accuracy of SRS treatment planning by obscuring the

J Neurosurg Volume 135 • September 2021 745

 
* | D*2 *36 *
Chen et al.

FIG. 1. Comparisons of CIFs for AVM obliteration (p = 0.141) (A), post-SRS hemorrhage (p = 0.314) (B), radiological RICs (p =
0.117) (C), symptomatic RICs (p = 0.048) (D), and cyst development (p = 0.115) (E) between the unmatched OE+SRS versus SRS-
only cohorts.

borders of the residual nidus. Other plausible mechanisms In this multicenter, retrospective matched cohort study
for lower post-SRS obliteration rates of AVMs embolized we found no differences between the SRS outcomes of
with Onyx include embolization-induced angiogenesis, re- volumetrically comparable Onyx-embolized versus non-
canalization of embolized portions of a nidus not otherwise embolized AVMs. Our study is distinguished from the
targeted by SRS, and pseudoocclusion (i.e., angiographic flawed design of previous studies by its use of de novo
occlusion of a vessel that is functionally patent).20–26 In AVM characteristics (i.e., prior to embolization). Although
contrast, others have argued, based on simulation studies, obliteration was more likely in the unmatched SRS-only
that Onyx-mediated radiation dose alterations are negligi- cohort (p < 0.001), this difference was no longer significant
ble.19,27–30 As such, the risk-to-benefit profile of Onyx embo- in time-dependent analysis adjusting for competing risk of
lization preceding AVM SRS remains controversial. death. Both crude and time-dependent obliteration rates

746 J Neurosurg Volume 135 • September 2021

 
* | D*2 *36 *
 Chen et al.

TABLE 3. Comparison of baseline characteristics and follow-up durations between the matched OE+SRS versus SRS-
only cohorts
Characteristic OE+SRS Cohort, n = 53 SRS-Only Cohort, n = 53 p Value
Mean age (SD), yrs 43.1 (18.0) 34.4 (17.4) 0.013
Females (%) 33/53 (62.3) 27/53 (50.9) 0.240
Prior AVM surgery (%) 7/53 (13.2) 7/53 (13.2) 1.000
Prior AVM fractionated radiation therapy (%) 2/53 (3.8) 0/45 (0) 0.498
Prior AVM hemorrhage (%) 32/53 (60.4) 21/53 (39.6) 0.033
Deep AVM location (%) 6/53 (11.3) 12/53 (22.6) 0.121
Eloquent AVM location (%) 29/47 (61.7) 27/46 (58.7) 0.767
Deep venous drainage (%) 28/53 (52.8) 29/53 (54.7) 0.846
Mean AVM volume (SD), ml 10.3 (10.1) 10.8 (11.4) 0.794*
Mean SRS margin dose (SD), Gy 19.2 (2.7) 19.0 (3.4) 0.663
Mean mos radiological follow-up (SD) 37.2 (37.4) 44.4 (36.4) 0.319
Mean mos clinical follow-up (SD) 43.2 (37.5) 61.9 (50.5) 0.036
Boldface type indicates statistical significance.
* Matched covariates.

were similar between the matched OE+SRS and SRS-only OE+SRS cohort, both crude rates and cumulative prob-
cohorts. Therefore, pre-SRS AVM embolization with Onyx abilities of post-SRS hemorrhage were similar between
does not appear to reduce post-SRS obliteration rates. the two cohorts in unmatched and matched analyses. Al-
Despite our best attempts to balance the OE+SRS and though the cumulative probability of symptomatic RICs
SRS-only cohorts with propensity score matching, age and was higher in the unmatched OE+SRS cohort (p = 0.048),
prior AVM hemorrhage were significantly different be- cumulative probabilities of both radiological and symp-
tween the matched cohorts. Because initial nidus volume tomatic RICs were similar between the matched cohorts.
is generally accepted to be the most important AVM factor We were unable to elucidate whether the OE+SRS cohort’s
in SRS-induced obliteration, we prioritized this covariate higher proportion of prior AVM hemorrhage, which has
foremost in our matching process.31 Coincidentally, this been shown to protect against RICs, modified the rela-
approach also offered the optimal balance of the remain- tionship between Onyx embolization and RIC rates in
ing baseline covariates between the matched cohorts. In- the present analyses.32,33 Despite comparable RIC rates
clusion of other covariates in the derivation of propensity between the matched cohorts, detection of symptomatic
scores did not improve the balance of baseline characteris- SRS-related complications in embolized AVMs could be
tics relative to using AVM volume alone. Furthermore, the masked by embolization-related complications, which
effects of age and prior hemorrhage on obliteration rates were symptomatic and asymptomatic in 9% and 19% of
after AVM SRS may be inconsequential.32–35 the matched OE+SRS cohort, respectively.
Despite the higher proportion of ruptured AVMs in the Although the present study does not deem Onyx emboli-

TABLE 4. Standardized difference of baseline characteristics between OE+SRS versus SRS-only cohorts before and after propensity score
matching
Mean, Before Adjustment Mean, After Adjustment
OE+SRS SRS-Only Standardized Difference, OE+SRS SRS-Only Standardized Difference,
Characteristic Cohort Cohort Before Adjustment Cohort Cohort After Adjustment
Age 42.21 36.36 0.346 42.18 35.35 0.404
Sex 0.40 0.51 −0.223 0.41 0.46 −0.101
Prior AVM surgery 0.13 0.15 −0.052 0.14 0.19 −0.150
Prior AVM fractionated radiation therapy 0.02 0.00 0.136 0.02 0.00 0.209
Prior AVM hemorrhage 0.63 0.50 0.279 0.66 0.49 0.350
Deep AVM location 0.12 0.25 −0.344 0.14 0.24 −0.281
Eloquent AVM location 0.58 0.61 −0.073 0.59 0.59 −0.007
Deep venous drainage 0.50 0.48 0.041 0.52 0.59 −0.143
AVM volume 16.43 3.94 0.944 10.88 9.12 0.133
SRS margin dose 19.24 21.03 −0.572 19.35 19.35 0.000
Boldface type indicates standardized difference ≥ 0.20.

J Neurosurg Volume 135 • September 2021 747

 
* | D*2 *36 *
Chen et al.

TABLE 5. Comparison of outcomes between the matched OE+SRS versus SRS-only cohorts
Outcome OE+SRS Cohort, n = 53 SRS-Only Cohort, n = 53 OR (95% CI) p Value
Primary outcome
  AVM obliteration (%) 20/53 (37.7) 25/53 (47.2) 0.679 (0.313–1.472) 0.327
Secondary outcomes (%)
  Post-SRS hemorrhage 4/50 (8.0) 6/52 (11.5) 0.667 (0.176–2.519) 0.550
  All-cause mortality 1/52 (1.9) 0/52 (0) — 1.000*
  Radiological RICs 17/51 (33.3) 24/50 (48.0) 0.542 (0.242–1.211) 0.135
  Symptomatic RICs 11/51 (21.6) 10/50 (20.0) 1.100 (0.420–2.879) 0.846
 Cyst 2/51 (3.9) 0/51 (0) — 0.495*
  Asymptomatic embolization complications 10/53 (18.9) — — —
  Symptomatic embolization complications 5/53 (9.4) — — —
* Fisher’s exact test

zation detrimental to AVM SRS outcomes, we also failed to staged SRS approaches that can require longer implemen-
show a discrete benefit from the combination of both AVM tation periods.36 Embolization can also eliminate AVM-as-
therapies. Nevertheless, we believe that pre-SRS AVM em- sociated arterial aneurysms that, left untreated, can elevate
bolization with Onyx will retain a role in the treatment of the risk of post-SRS hemorrhage.5,37 Lastly, endovascular
carefully selected patients. Specifically, targeted emboli- occlusion of high-flow intranidal arteriovenous shunts,
zation of a large AVM with multiple compartments could which are relatively radioresistant, could improve the likeli-
simplify its morphology so as to facilitate treatment of the hood of SRS-induced obliteration.38 Although symptomatic
residual nidus with single-session SRS, thereby averting complications associated with pre-SRS AVM embolization

FIG. 2. Comparisons of CIFs for AVM obliteration (p = 0.896) (A), post-SRS hemorrhage (p = 0.950) (B), radiological RICs (p =
0.178) (C), and symptomatic RICs (p = 0.392) (D) between the matched OE+SRS versus SRS-only cohorts.

748 J Neurosurg Volume 135 • September 2021

 
* | D*2 *36 *
 Chen et al.

varied in symptomatology, severity, and duration, they oc- ciated with Onyx embolization of AVMs, use of this neo-
curred in nearly 10% of our matched OE+SRS cohort. In adjuvant intervention should be restricted to carefully se-
addition, asymptomatic complications occurred in 19%. lected large-volume or angioarchitecturally high-risk nidi
Taken together, the pre-SRS embolization-related compli- that are deemed appropriate candidates for multimodal
cation rate was relatively high. Therefore, the added risks treatment strategies employing SRS. Future studies de-
of pre-SRS embolization warrant careful consideration, rived from prospective, multicenter registries are necessary
and judicious patient selection is essential. Ultimately, the to clarify the role of combined embolization and SRS in
aggressiveness of pre-SRS embolization should be adjudi- AVM management, particularly in comparison to staged
cated by a multidisciplinary team comprising the operators SRS approaches.
performing embolization and those performing SRS.
We recognize that limitations of our study could impact References
its validity and generalizability. The accuracy and reliabil-   1. Kano H, Kondziolka D, Flickinger JC, et al. Stereotactic
ity of the analyses are dependent upon data reported by re- radiosurgery for arteriovenous malformations, Part 6:​Mul-
spective contributing centers and therefore may be suscep- tistaged volumetric management of large arteriovenous mal-
tible to reporting bias. Despite our best efforts to balance formations. J Neurosurg. 2012;​116(1):​54–65.
the pretreatment characteristics of the OE+SRS and SRS-   2. Ilyas A, Chen CJ, Ding D, et al. Volume-staged versus dose-
only cohorts through propensity score matching, several staged stereotactic radiosurgery outcomes for large brain
measured baseline covariates remained unbalanced and arteriovenous malformations:​a systematic review. J Neuro-
surg. 2018;​128(1):​154–164.
other unmeasured covariates may not be accounted for.   3. Patibandla MR, Ding D, Kano H, et al. Stereotactic radiosur-
The effects of these unbalanced covariates on the primary gery for Spetzler-Martin Grade IV and V arteriovenous mal-
and secondary outcomes are unquantifiable, but the resul- formations:​an international multicenter study. J Neurosurg.
tant potential selection bias should be acknowledged. We 2018;​129(2):​498–507.
could not ascertain the criteria or rationale for the utiliza-   4. Solomon RA, Connolly ES Jr. Arteriovenous malformations
tion of pre-SRS Onyx embolization in each case, which of the brain. N Engl J Med. 2017;​376(19):​1859–1866.
could reflect selection, treatment, and referral biases of the   5. Kano H, Kondziolka D, Flickinger JC, et al. Aneurysms in-
participating centers and treating physicians. However, we crease the risk of rebleeding after stereotactic radiosurgery
for hemorrhagic arteriovenous malformations. Stroke. 2012;​
were able to balance the most influential baseline charac- 43(10):​2586–2591.
teristics (i.e., nidus volume and margin dose) with respect   6. Ding D, Ilyas A, Sheehan JP. Contemporary management of
to AVM SRS outcomes.39 high-grade brain arteriovenous malformations. Neurosur-
Because pre-embolization MR images were not avail- gery. 2018;​65(CN_suppl_1):​24–33.
able for volumetric analysis in some cases, AVM volumes   7. Russell D, Peck T, Ding D, et al. Stereotactic radiosurgery
were calculated based on DSA using the ABC/2 method alone or combined with embolization for brain arteriovenous
for both cohorts to maintain consistency. This method has malformations:​a systematic review and meta-analysis. J
Neurosurg. 2018;​128(5):​1338–1348.
been generally accepted for intracerebral hematoma volu-   8. Lee CC, Chen CJ, Ball B, et al. Stereotactic radiosurgery
metry, but it has yet to be validated for AVMs, especially for arteriovenous malformations after Onyx embolization:​a
in those with large nidi and complex morphologies.40,41 Ad- case-control study. J Neurosurg. 2015;​123(1):​126–135.
ditionally, variations in unrecorded embolysate volume and   9. Andrade-Souza YM, Ramani M, Beachey DJ, et al. Liquid
degree of nidal devascularization may confound the results. embolisation material reduces the delivered radiation dose:​
Because AVM volume reduction by embolization was not a physical experiment. Acta Neurochir (Wien). 2008;​150(2):​
quantified in our data set, its effects on SRS outcomes could 161–164.
10. Kano H, Kondziolka D, Flickinger JC, et al. Stereotactic
not be analyzed. AVM obliteration was determined by MRI radiosurgery for arteriovenous malformations after emboliza-
alone in 8.3% of the cases, and the lack of angiographic tion:​a case-control study. J Neurosurg. 2012;​117(2):​265–275.
confirmation in a minority of the study cohort could have 11. Oermann EK, Ding D, Yen CP, et al. Effect of prior emboli-
overestimated the reported obliteration rates. However, zation on cerebral arteriovenous malformation radiosurgery
MRI has demonstrated reasonable accuracy for assessing outcomes:​a case-control study. Neurosurgery. 2015;​77(3):​
obliteration after AVM SRS compared to the gold stan- 406–417.
dard of DSA.42 Importantly, we were unable to detect small 12. Andrade-Souza YM, Ramani M, Scora D, et al. Embolization
before radiosurgery reduces the obliteration rate of arteriove-
differences in outcomes between the matched cohorts due nous malformations. Neurosurgery. 2007;​60(3):​443–452.
to their modest sample sizes. Lastly, our findings may not 13. Kothari RU, Brott T, Broderick JP, et al. The ABCs of mea-
be generalizable to small- and medium-volume AVMs or suring intracerebral hemorrhage volumes. Stroke. 1996;​27(8):​
those treated with staged SRS approaches.2,36 1304–1305.
14. Spetzler RF, Martin NA. A proposed grading system for arte-
riovenous malformations. J Neurosurg. 1986;​65(4):​476–483.
Conclusions 15. Wegner RE, Oysul K, Pollock BE, et al. A modified radio-
After accounting for de novo AVM characteristics, our surgery-based arteriovenous malformation grading scale and
findings refute previous studies that espoused an adverse its correlation with outcomes. Int J Radiat Oncol Biol Phys.
effect of Onyx embolization on SRS-induced nidal oblit- 2011;​79(4):​1147–1150.
16. Leuven E, Sianesi B. PSMATCH2:​Stata module to perform
eration. In a matched analysis of morphologically compa- full Mahalanobis and propensity score matching, common
rable nidi, AVMs embolized with Onyx appeared to have support graphing, and covariate imbalance testing. 2003. Sta-
SRS outcomes that were similar to those of nonembolized tistical Software Components. Accessed September 1, 2020.
lesions. However, given the neuroendovascular risks asso- http:​//ideas.repec.org/c/boc/bocode/s432001.html

J Neurosurg Volume 135 • September 2021 749

 
* | D*2 *36 *
Chen et al.

17. Fine JP, Gray RJ. A proportional hazards model for the 35. Chen CJ, Ding D, Kano H, et al. Stereotactic radiosurgery
subdistribution of a competing risk. J Am Stat Assoc. 1999;​ for pediatric versus adult brain arteriovenous malformations.
94(446):​496–509. Stroke. 2018;​49(8):​1939–1945.
18. Austin PC, Lee DS, Fine JP. Introduction to the analysis of 36. Seymour ZA, Chan JW, Sneed PK, et al. Dose response and
survival data in the presence of competing risks. Circulation. architecture in volume staged radiosurgery for large arterio-
2016;​133(6):​601–609. venous malformations:​a multi-institutional study. Radiother
19. Roberts DA, Balter JM, Chaudhary N, et al. Dosimetric Oncol. 2020;​144:​180–188.
measurements of Onyx embolization material for stereotactic 37. Ding D, Chen CJ, Starke RM, et al. Risk of brain arteriove-
radiosurgery. Med Phys. 2012;​39(11):​6672–6681. nous malformation hemorrhage before and after stereotactic
20. Valle RD, Zenteno M, Jaramillo J, et al. Definition of the key radiosurgery. Stroke. 2019;​50(6):​1384–1391.
target volume in radiosurgical management of arteriovenous 38. Taeshineetanakul P, Krings T, Geibprasert S, et al. Angio-
malformations:​a new dynamic concept based on angiograph- architecture determines obliteration rate after radiosurgery
ic circulation time. J Neurosurg. 2008;​109(suppl):​41–50. in brain arteriovenous malformations. Neurosurgery. 2012;​
21. Pollock BE, Kondziolka D, Lunsford LD, et al. Repeat stereo- 71(6):​1071–1079.
tactic radiosurgery of arteriovenous malformations:​factors 39. Flickinger JC, Pollock BE, Kondziolka D, Lunsford LD.
associated with incomplete obliteration. Neurosurgery. 1996;​ A dose-response analysis of arteriovenous malformation
38(2):​318–324. obliteration after radiosurgery. Int J Radiat Oncol Biol Phys.
22. Akakin A, Ozkan A, Akgun E, et al. Endovascular treatment 1996;​36(4):​873–879.
increases but gamma knife radiosurgery decreases angio- 40. Roark C, Vadlamudi V, Chaudhary N, et al. ABC/2 method
genic activity of arteriovenous malformations:​an in vivo does not accurately predict cerebral arteriovenous malforma-
experimental study using a rat cornea model. Neurosurgery. tion volume. Neurosurgery. 2018;​82(2):​220–225.
2010;​66(1):​121–130. 41. Webb AJ, Ullman NL, Morgan TC, et al. Accuracy of the
23. Buell TJ, Ding D, Starke RM, et al. Embolization-induced ABC/2 score for intracerebral hemorrhage:​systematic review
angiogenesis in cerebral arteriovenous malformations. J Clin and analysis of MISTIE, CLEAR-IVH, and CLEAR III.
Neurosci. 2014;​21(11):​1866–1871. Stroke. 2015;​46(9):​2470–2476.
24. Fournier D, Terbrugge K, Rodesch G, Lasjaunias P. Revascu- 42. Lee CC, Reardon MA, Ball BZ, et al. The predictive value
larization of brain arteriovenous malformations after emboli- of magnetic resonance imaging in evaluating intracranial
zation with bucrylate. Neuroradiology. 1990;​32(6):​497–501. arteriovenous malformation obliteration after stereotactic
25. Gobin YP, Laurent A, Merienne L, et al. Treatment of brain radiosurgery. J Neurosurg. 2015;​123(1):​136–144.
arteriovenous malformations by embolization and radiosur-
gery. J Neurosurg. 1996;​85(1):​19–28.
26. Rao VR, Mandalam KR, Gupta AK, et al. Dissolution of Disclosures
isobutyl 2-cyanoacrylate on long-term follow-up. AJNR Am J Dr. Liscak reports receiving consultant fees from Elekta AB.
Neuroradiol. 1989;​10(1):​135–141. Dr. Williams reports receiving consultant fees from Monteris
27. Bing F, Doucet R, Lacroix F, et al. Liquid embolization ma- Medical, Inc. Dr. Cockroft reports receiving consultant fees from
terial reduces the delivered radiation dose:​clinical myth or Medtronic and Minnetronix. Dr. Kondziolka reports receiving
reality? AJNR Am J Neuroradiol. 2012;​33(2):​320–322. research funding from Brainlab.
28. Mamalui-Hunter M, Jiang T, Rich KM, et al. Effect of liquid
embolic agents on Gamma Knife surgery dosimetry for arte- Author Contributions
riovenous malformations. Clinical article. J Neurosurg. 2011;​
115(2):​364–370. Conception and design: Chen. Acquisition of data: CC Lee,
29. Watanabe Y, Sandhu D, Warmington L, et al. Three-dimen- Kearns, Cifarelli, Arsanious, Liscak, Hanuska, Williams, Yusuf,
sional assessment of the effects of high-density embolization Woo, Warnick, Trifiletti, Mathieu, Mureb, Benjamin, Kondziolka,
material on the absorbed dose in the target for Gamma Knife Feliciano, Rodriguez-Mercado, Cockroft, Simon, Mackley, Zam-
radiosurgery of arteriovenous malformations. J Neurosurg. mar, Patel, Padmanaban, Beatson, Saylany, JYK Lee. Analysis
2016;​125(1)(suppl 1):​123–128. and interpretation of data: Sheehan, Chen, Ding, CC Lee. Draft-
30. Schlesinger DJ, Nordström H, Lundin A, et al. Dosimetric ing the article: Chen, Ding. Critically revising the article: all
effects of Onyx embolization on Gamma Knife arteriovenous authors. Reviewed submitted version of manuscript: all authors.
malformation dose distributions. J Neurosurg. 2016;​125(1) Approved the final version of the manuscript on behalf of all
(suppl 1):​114–122. authors: Sheehan. Statistical analysis: Chen. Administrative/tech-
31. Starke RM, Kano H, Ding D, et al. Stereotactic radiosurgery nical/material support: Sheehan. Study supervision: Sheehan.
for cerebral arteriovenous malformations:​evaluation of long-
term outcomes in a multicenter cohort. J Neurosurg. 2017;​ Correspondence
126(1):​36–44. Jason P. Sheehan: University of Virginia Health System,
32. Chen CJ, Lee CC, Ding D, et al. Stereotactic radiosurgery Charlottesville, VA. [email protected].
for unruptured versus ruptured pediatric brain arteriovenous
malformations. Stroke. 2019;​50(10):​2745–2751.
33. Ding D, Yen CP, Starke RM, et al. Effect of prior hemorrhage
on intracranial arteriovenous malformation radiosurgery out-
comes. Cerebrovasc Dis. 2015;​39(1):​53–62.
34. Chen CJ, Ding D, Kano H, et al. Effect of advanced age on
stereotactic radiosurgery outcomes for brain arteriovenous
malformations:​a multicenter matched cohort study. World
Neurosurg. 2018;​119:​e429–e440.

750 J Neurosurg Volume 135 • September 2021

 
* | D*2 *36 *
 CLINICAL ARTICLE
J Neurosurg 135:751–759, 2021

Intracranial EEG and laser interstitial thermal therapy

in MRI-negative insular and/or cingulate epilepsy: case
series
Elakkat D. Gireesh, MD,1 Kihyeong Lee, MD,1 Holly Skinner, DO,1 Joohee Seo, MD,1
Po-Ching Chen, PhD,1,4 Michael Westerveld, PhD,3 Richard D. Beegle, MD,5
Eduardo Castillo, PhD,4 and James Baumgartner, MD2
1
Epilepsy Center, Neuroscience Institute, AdventHealth; 2Department of Neurosurgery, Neuroscience Institute, AdventHealth;
3
Department of Neuropsychology, Neuroscience Institute, AdventHealth; 4MEG Center, Neuroscience Institute, AdventHealth;
and 5Department of Radiology, AdventHealth, Orlando, Florida

OBJECTIVE  The goal of this study was to assess the success rate and complications of stereo-electroencephalogra-
phy (sEEG) and laser interstitial thermal therapy (LITT) in the treatment of nonlesional refractory epilepsy in cingulate
and insular cortex.
METHODS  The authors retrospectively analyzed the treatment response in 9 successive patients who underwent insu-
lar or cingulate LITT for nonlesional refractory epilepsy at their center between 2011 and 2019. Localization of seizures
was based on inpatient video-EEG monitoring, neuropsychological testing, 3-T MRI, PET scan, magnetoencephalog-
raphy scan, and/or ictal SPECT scan. Eight patients underwent sEEG, and 1 patient had implantation of both sEEG
electrodes and subdural grids for localization of epileptogenic zones. LITT was performed in 5 insular cases (4 left and
1 right) and 3 cingulate cases (all left-sided). One patient also underwent both insular and cingulate LITT on the left side.
All of the patients who underwent insular LITT as well as 2 of the 3 who underwent cingulate LITT were right-hand domi-
nant. The patient who underwent insular plus cingulate LITT was also right-hand dominant.
RESULTS  Following LITT, 67% of the patients were seizure free (Engel class I) at follow-up (mean 1.35 years, range 0.6–
2.8 years). All patients responded favorably to treatment (Engel class I–III). Two patients developed small intracranial hem-
orrhages during the sEEG implantation that did not require surgical management. One patient developed a large intracra-
nial hemorrhage during an insular LITT procedure that did require surgical management. That patient experienced aphasia,
incoordination, and hemiparesis, which resolved with inpatient rehabilitation. No permanent neurological deficits were noted
in any of the patients at last follow-up. Neuropsychological status was stable in this cohort before and after LITT.
CONCLUSIONS  sEEG can be safely used to localize seizures originating from insular and cingulate cortex. LITT can
successfully treat seizures arising from these deep-seated structures. The insula and cingulum should be evaluated
more frequently for seizure onset zones.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201912
KEYWORDS  epilepsy surgery; insula; cingulate; laser ablation; stereo-EEG

A
pproximately 1% of the US population suffers from When insular or cingulate cortex is suspected as the
epilepsy,1 and 30%–35% of these patients have seizure onset zone, invasive monitoring of the structure
drug-resistant epilepsy (DRE).2 Patients with DRE would improve epilepsy management.6,7 Neither the insu-
can be treated with surgery designed to remove the seizure la nor the cingulate can be easily monitored using subdu-
onset zone or to disrupt the epileptic network. Currently, ral electrode arrays. Attempts to monitor these structures
long-term seizure control after epilepsy surgery ranges be- by using combinations of subdural and depth electrodes
tween 34% and 56% for extratemporal epilepsy.3–5 Failure have achieved inconsistent success.8 Recent advances in
to accurately identify the epileptogenic zone(s) and inad- stereo-electroencephalography (sEEG) monitoring have
equate or inaccurate surgical interventions may contribute helped to identify electrographic patterns from deep
to epilepsy surgery failures. structures, including insular and cingulate cortex.7,9–12

ABBREVIATIONS  DRE = drug-resistant epilepsy; FSIQ = full-scale IQ; iMRI = intraoperative MRI; LITT = laser interstitial thermal therapy; MEG = magnetoencephalogra-
phy; OR = operating room; RFTC = radiofrequency thermocoagulation; sEEG = stereo-electroencephalography; SMA = supplementary motor area.
SUBMITTED  May 21, 2020.  ACCEPTED  July 13, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS201912.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  751

* | D*2 *37 *
Gireesh et al.

Stereotactic robots with 3D navigational platforms allow equipped with a Siemens Somatom Open intraoperative
accurate and efficient placement of sEEG electrodes into CT. For sEEG procedures, electrode arrays were implant-
deep structures. ed as follows: skull bone fiducial, 10- to 14-mm titanium
The fact that insula lies deep within the sylvian fis- screw–based (DePuy Synthes) sterile registration was per-
sure and is adjacent to critical structures, including the formed using a Leksell frame (Elekta) and a ROSA robot
internal capsule, makes it difficult to approach surgi- (Zimmer Biomet). sEEG electrodes were placed stereotac-
cally. In the dominant hemisphere, the insula is covered tically through twist drill trephinations and anchor bolts
by eloquent frontal, temporal, and parietal opercula. The were inserted (electrodes and anchor bolts from PMT,
cingulate gyrus is challenging to access surgically be- Inc.) at planned entry sites along planned trajectories. For
cause of its location along the deep medial surface of the hybrid monitoring (sEEG plus subdural electrodes), sEEG
brain. Various minimally invasive approaches including electrodes were placed as described above; then the pa-
stereotactic radiofrequency thermocoagulation (RFTC)11 tient was taken out of the Leksell frame, and a craniotomy
and laser interstitial thermal therapy (LITT) have been was performed over the fronto-temporo-parietal region.
used as a surgical strategy for managing epilepsy arising Intraoperative electrocorticography was performed, after
from the insula. RFTC performed to treat lesional insular placing subdural grid and strip electrodes and addition-
epilepsy resulted in a seizure-free outcome in 53% of pa- al electrodes as needed. Eight patients underwent sEEG
tients, with a responder rate of 89% (described as Engel electrode implantation, and 1 patient had a hybrid elec-
class I–III). Of the patients treated with insular RFTC, trode array.
42% experienced perioperative neurological deficits. In Oblique trajectories were used to place insular sEEG
a recent study comparing LITT to open surgery for in- electrodes. The oblique trajectory allows multiple-con-
sular seizure foci, favorable outcomes were reported in tact6–8 coverage of insular gray matter with each electrode
71% (Engel class I and II) in the LITT group and 75% (Fig. 1). In cases in which intraoperative electrocorti-
of those who underwent open surgery.13 We report our cography suggested possible opercular involvement, ad-
epilepsy center’s experience in intracranial sEEG moni- ditional transverse sEEG electrodes were placed during
toring and subsequent LITT procedures in patients with surgery. The cingulate electrodes were placed transversely
MRI-negative epilepsy, localized to insular and/or cingu- in the initial cases. Because the area sampled with indi-
late regions. vidual sEEG contacts is small and our experience with
transversely placed cingulate electrode monitoring was
Methods inconclusive, we chose to monitor the cingulate gyrus
longitudinally in the latter cases (Fig. 1). With electrodes
Inclusion Criteria placed through the long axis of the cingulum, we could
Patients were selected using the following criteria: 1) more accurately assess seizure onset, spread, and the ex-
DRE (documented, continued seizures despite trial of tent of involvement.
at least two antiepileptic drugs); 2) clinical suspicion of After placing the electrodes, a postoperative CT scan
epilepsy originating from insular or cingulate regions; 3) was obtained to document electrode position and to deter-
negative MRI findings; 4) age 1–65 years; and 5) epilepsy mine the presence or absence of intracranial hemorrhage.
surgery between January 1, 2011, and January 30, 2019. Patients were admitted to the neuro-ICU throughout the
The institutional review board at AdventHealth Orlando period of intracranial EEG monitoring. After capturing
approved this retrospective chart review. the patient’s typical seizures, the case was discussed in a
multidisciplinary epilepsy conference. We used the same
Presurgical Workup skull bone fiducials for both sEEG and LITT fiber navi-
Presurgical evaluation included a scalp video-EEG gation, and therefore doing both procedures in the same
evaluation; MRI of brain including diffusion tensor imag- hospital visit eliminated the need for reimaging.
ing sequences; attempted ictal SPECT, PET, and magneto-
encephalography (MEG); and neuropsychological evalu- LITT Surgery
ation to localize the epileptogenic zones. MEG included Patients were taken to an intraoperative MRI (iMRI)
sensorimotor and language mapping. All cases were dis- OR suite with a 3-T Siemens Vario iMR imager. The sEEG
cussed at our multidisciplinary epilepsy surgery confer- electrodes and anchor bolts were removed using a sterile
ence, and treatments were planned based on conference technique. The patients were then placed in an iMRI head
consensus. Wada testing and functional MRI were done if holder for 3-point fixation, and a ROSA robot was attached
deemed necessary based on conference decision. to the Mayfield head holder by using a Monteris connector.
The skull bone fiducials were then registered to the ROSA
Electrode Placement and Intracranial EEG Monitoring navigation system by using a sterile registration probe.
For sEEG electrode placement, planning was per- An entry point and target point were then chosen for each
formed with ROSA planning software, using contrast- LITT trajectory by using the ROSA navigation software.
enhanced T1-weighted MR images to identify and avoid Using the ROSA drill guide attachment and the Monteris
vascular structures. Eight patients had intracranial moni- drill guide and drill bit, a twist drill trephination (4.5 mm)
toring with sEEG electrodes. The remaining patient was was created along the desired trajectory at the entry point.
evaluated using a hybrid of grid and sEEG electrodes, A Monteris minibolt was then advanced along the trajecto-
with depth electrodes placed into the insular regions. ry through the trephination with a Monteris minibolt hand
Electrodes were placed in an operating room (OR) driver. The accuracies of the trajectory and target were

752 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Gireesh et al.

FIG. 1. Sagittal T2-weighted FLAIR MR images demonstrating insular and cingulate electrode position and the sEEG traces from
the corresponding electrodes. A and B: Example of typical anterior (left cingulate) and posterior (left posterior cingulate) cingulate
electrode placement. C and D: Example of typical insular anterior (left anterior oblique insular) and posterior (left posterior oblique
insular) insular electrode placement. E: Seizures noted to be evolving near electrodes LAOI 2–4 and LPOI 3–5. LAOI = left
anterior oblique insular; LC = left cingulate; LPC = left posterior cingulate; LPF = left posterior frontal; LPOI = left posterior oblique
insular. Figure is available in color online only.

confirmed using Stryker neuronavigation registered to the robot was used to direct the laser fiber to achieve maximal
skull bone fiducials. If multiple trajectories were deemed ablation of epileptogenic tissue. This process was repeated
necessary, multiple minibolts were placed before ablation until all minibolt-based trajectories had been ablated. An
was initiated. Once all planned minibolts were in place, intraoperative T1-weighted MRI sequence with half-dose
the ROSA robot was detached from the iMRI head holder contrast was then obtained to estimate the extent of abla-
and removed from the OR. A Monteris LITT robot was tion achieved, and additional ablations were planned and
attached to the minibolt, and a Monteris 3.3-mm Side Fire performed if deemed necessary.
navigable LITT fiber was advanced through the robot and For cingulate ablations, the surgical approach for laser
minibolt to target depth. The iMRI imager was then moved fiber placement was identical to that for the insular fiber
into the OR and, under continuous MRI thermometry, la- placement. Because the cingulate gyrus is long and cy-
ser ablation was performed sequentially from the deep- lindrical, a 3.3-mm Monteris diffuse tip laser fiber was
est point along the trajectory to the most superficial point used to perform the ablation. Two cingulate ablations
along the trajectory by using the Monteris software and were completed through a single longitudinal trajectory.
workstation (Supplementary Fig. S1). The Monteris LITT One cingulate LITT procedure was accomplished via 3

J Neurosurg Volume 135 • September 2021 753

 
* | D*2 *37 *
Gireesh et al.

TABLE 1. Clinical profiles in 9 patients with insular and/or cingulate epilepsy

Age at
Case Seizure Seizure Risk Neuro Prior
No. Age Handedness Sex Onset Frequency Medications Factors Exam Surgery
1 7 yrs Rt F 6 yrs 10/day OXC, VPA, CLP NR Cognitive deficits None
2 46 yrs Rt M 27 yrs 1/night CBZ, ZNS, GBP Trauma Cognitive deficits None
3 19 yrs Rt M 3 wks 5–7/wk PHT, CBZ, OXCDevelopmen- Cognitive deficits Lt temporal lobectomy
tal delay
4 22 yrs Rt F 16 yrs 2–3/wk LTG, TPM, LEV, PMP NR Normal None
5 20 yrs Rt F 4 yrs 1–3/wk TPM, CBZ Low Apgar Normal None
score at birth
6 9 yrs Rt F 6 yrs 5–10/day LEV, LTG, VPA Family history Normal None
of epilepsy
7 18 yrs Rt M 12 yrs 1–3/mo LTG, OXC, PHT, LEV, CLB NR Rt-sided Lt frontoparietal topectomy
weakness
8 18 yrs Lt F 4 mos >5/wk PB, TPM, LTG, VPA, Autism Cognitive deficits Corpus callosotomy, lt pari-
ZNS, OXC etal topectomy, lt frontal MST
9 6 yrs Rt F 5 yrs 10–20/night TPM, OXC, CLP NR Cognitive deficits None
CBM = clobazam; CBZ = carbamazepine; CLP = clonazepam; GBP = gabapentin; LEV = leviteracetam; LTG = lamotrigine; MST = multiple subpial transections; neuro
exam = neurological examination; NR = none reported; OXC = oxcarbazepine; PB = phenobarbital; PHT = phenytoin; PMP = perampanel; TPM = topiramate; VPA =
valproic acid; ZNS = zonisamide.

orthogonal trajectories, because the desired longitudinal cluded semiology, lack of clear lateralization of the scalp
approach was not possible due to limiting paramedian EEG findings, and epileptiform discharges noted from in-
vascular anatomy. sular regions in the MEG sequence. The ictal onset zones
noted in the scalp EEG included frontal, temporal, mid-
Results line, central, or posterior temporal/parietal regions. One
of the insular cases had beta wave activity noted in the
Population Studied midline. The MEG findings indicated an insular epilepto-
Age at seizure onset for these patients ranged from 3 genic zone in 3 of the 6 insular cases.
weeks to 27 years (average 8.5 years). The average age Indications for implanting cingulate electrodes includ-
at the time of surgery was 18.3 years (range 6–46 years). ed seizure semiology, including early motor symptoms
There were 6 females and 3 males in this cohort of pa- and unclear lateralization of the ictal scalp EEG. No clear
tients. The frequency of the seizures varied from multiple MEG findings could be established in the patients with
seizures in a day to 1–3 seizures per month. The risk fac- cingulate onset epilepsy.
tors for epilepsy included trauma, developmental delay,
autism, and family history of seizures. In the patient with
sEEG Monitoring
a family history of seizures, epilepsy was reported in the
maternal grandmother, aunt, uncle, and two cousins. No We selected only the patients undergoing insular LITT
genetic abnormalities were identified based on whole- during the period of study. The total number of patients
exome sequencing. All patients had been treated with at undergoing sEEG in our center during this period was 76.
least three antiseizure medications and met criteria for Also, 172 patients underwent grid implantation during the
DRE. Three patients underwent prior unsuccessful epi- same period. Four of the 5 insular cases had sEEG moni-
lepsy surgeries. The detailed clinical profiles of this group toring done bilaterally. The case without bilateral insular
of patients are listed in Table 1. monitoring was evaluated with subdural grids and depth
electrodes. Two of the 3 cingulate cases had bilateral
Presurgical Workup sEEG monitoring. The patient who underwent both insu-
The semiology of our insular epilepsy cases included lar and cingulate LITT had bilateral insular and left-sided
activity arrest, staring, lip smacking, contralateral facial cingulate sEEG monitoring. Representative sEEG record-
and arm twitching, head drop, contralateral dystonic pos- ings with corresponding electrode positions are shown in
turing, tingling sensation in throat, and tingling sensation Fig. 1. The sEEG findings are noted in Table 2. The de-
in the contralateral arm. One of the patients had predomi- tails of sEEG electrodes and subdural grid coverage can
nantly nocturnal seizures. Cingular seizures were com- be found in Table 3.
monly associated with contralateral dystonic movements
(noted in 2 of the 3 cases). One patient with cingulate epi- LITT Procedure
lepsy had hypermotor seizures. For LITT, trajectories of the sEEG electrodes with ac-
Indications for implanting electrodes in the insula in- tive contacts were used. We used the Monteris minibolt

754 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Gireesh et al.

TABLE 2. Presurgical workup and sEEG findings in 9 patients with insular and/or cingulate epilepsy
Case Semiology Interictal
No. Semiology Localization EEG Ictal EEG PET SPECT MEG sEEG Seizure Onset
1 Nocturnal; head version to lt Rt frontal Beta waves Beta waves, Lt Lt temporal, Lt frontal, Lt anterior insular
over rt frontal midline insular lt insular lt insular
2 During sleep, dystonic Lt frontal, lt pa- NL NL NL NL Lt insular, Lt insular, lt cingulate
posturing of rt arm/leg rietal, lt insular lt perisylvian
3 Staring, activity arrest, lip smack- Lt temporal, lt Lt central, Lt frontal, Bilat Rt temporal Lt temporal, Lt insular, lt temporal
ing, rt facial & arm twitching insular lt temporal lt temporal lt insular
4 Hypermotor (rt > lt dystonic Lt temporal, lt Lt frontal Lt central, NL Lt > rt Lt frontal Rt frontal, rt insular
posturing) frontal lt parietal
5 Activity arrest, partial awareness, Lt Bilat frontal NL NL Lt NL Lt posterior insular
staring, rt-sided dystonic sharp wave
posturing
6 Tingling sensation in rt forearm, Lt Lt posterior Lt posterior NL Lt Lt Lt anterior insular
stiffening of rt arm
7 Rt head deviation, rt-sided Lt frontal Lt central, Lt central NL Lt > rt Lt frontal Lt cingulate
dystonia lt temporal
8 Arousal, frightened, rt arm & Lt frontal NL Lt NL NL Not Lt midcingulate spreading
leg stiffening, head to rt available to lt posterior insular
9 Hypermotor, flailing, drooling Frontal Bilat Lt NL NL Rt insular Lt midcingulate
NL = nonlocalizing.

and robot for ablation, with a directional laser fiber, typi- Complications
cally starting at the lowest power setting. When planned sEEG-Related Complications
trajectories did not accomplish adequate ablation (3 cas- sEEG-related bleeding was noted in 2 of the 9 cases.
es), we used additional trajectories to expand the extent of
ablation. Five patients had insular LITT (right side in 1, These included 1 case (the patient in case 7) of 3-mm
left side in 4). Three patients underwent cingulate cortex asymptomatic subdural hematoma and 1 case (the patient
LITT only, all on the left side. One of the patients had in case 2) of intraparenchymal bleed (29 mm in longest
LITT on both left cingulate and insular regions. A typical dimension). The intraparenchymal bleed was not related
example of the MRI changes seen immediately following to the insular sEEG electrodes and was located subcorti-
insular LITT and subsequent MRI findings 8 months later cally in the left parietal lobe, resulting in mild right hemi-
is shown in Fig. 2. The average volume of LITT in insular paresis. No lasting neurocognitive deficits were noted in
cases was 13.2 cm3 (range 5.3–29.2 cm3), and it was 15.3 association with these hemorrhages. One of the patients
cm3 (range 2.4–39.4 cm3) in cingulate cases (Table 3). with cingulate foci developed a CSF leak, which was de-

TABLE 3. sEEG coverage, LITT procedures, and surgical outcomes in 9 patients with insular and/or cingulate epilepsy
Surgical
Case Total No. of Insular Implantation Cingulate Implantation LITT No. of LITT LITT Outcomes Follow-Up
No. Electrode Contacts Side No. of Contacts Side No. of Contacts Location Trajectories Vol (cm3) (Engel class) (mos)
1 190 Bilat 24 Bilat 18 Lt insular 3 11.9 IA 22.67
2 190 Bilat 24 Lt 26 Lt insular, 2 29.2, 39.4 IB 15.7
lt cingulate
3 116 subdural, 16 Lt 6 NA NA Lt insular 3 11.5 IIA 16.07
depth (grid electrodes)
4 215 Bilat 18 Bilat 8 Rt insular 2 7.9 IA 34.27
5 250 Bilat 32 Bilat 48 Lt insular 3 5.3 IA 12.33
6 212 Bilat 28 Lt 16 Lt insular 2 * IA 10.2
7 98 NA NA Lt 24 Lt cingulate 1 11.6 IIIA 15.3
8 168 Bilat 20 Bilat 15 Lt cingulate 3 8 IIB 13.8
9 202 Bilat 24 Bilat 12 Lt cingulate 1 2.4 IB 19.5
NA = not applicable.
* Not available due to hemorrhage.

J Neurosurg Volume 135 • September 2021 755

 
* | D*2 *37 *
Gireesh et al.

the 6-month follow-up, she had very mild slowing in fine

finger movements of her right hand. At the 1-year follow-
up, her fine finger movements were symmetrical, and she
walked and ran without difficulty. She regained baseline
levels of expressive and receptive language skill.

Seizure Outcomes
Postoperative patient follow-up ranged from 10 months
to 34 months (Table 3). Six of the 9 patients (67%) achieved
complete seizure freedom (Engel class I), and all patients
had a favorable seizure response (Engel class I–III). Four of
the 5 patients with insular epilepsy, and 1 of 3 patients with
cingulate epilepsy were seizure free (Engel class I). The
patient who underwent both insular and cingulate LITT
also achieved seizure freedom. Two patients (cases 4 and
5) had partial seizure control (Engel class III) after initial
LITT treatment. They later underwent additional surgical
procedures (the patient in case 4 had left anterior cingulate
topectomy and corpus callosotomy, and the patient in case
5 had a second LITT; Fig. 3) before achieving seizure free-
dom. No significant statistical difference was noted in the
different parameters between the patients who had better
versus relatively worse outcome (Supplementary Table S1).

Neuropsychological Outcomes
All patients underwent comprehensive neuropsycholog-
FIG. 2. Case 1. Chronological MRI changes after insular LITT. A and B: ical evaluation (Supplementary Table S2). The full-scale
iMRI sequences obtained immediately after LITT. C and D: MR images IQ (FSIQ) score prior to surgery ranged from 64 to 108
obtained 8 months after LITT. Panels A and C are axial T2 images. Pan- (available only for 7 patients) and was 58–100 (5 patients)
els B and D are axial gradient echo images. after surgery. No significant differences were noted when
the pre-FSIQ and post-FSIQ scores were compared in
the available pairs (n = 5, p = 0.4). There were no definite
tected after implantation and was repaired without inci- trends in the ABAS (Adaptive Behavior Assessment Sys-
dent. tem) scores before and after surgery.
LITT-Associated Complications
A supplementary motor area (SMA) syndrome was Discussion
noted in the patient (case 2) who underwent both insu- Key Results
lar and cingulate LITT, leading to transient hemiparesis, Following intracranial monitoring and LITT of the in-
hemineglect, and word-finding difficulty. These symp- sula and cingulum, 6 of 9 patients in our cohort became
toms resolved with therapy. Two patients who underwent seizure free (Engel class I), 2 of 9 reached an Engel class
insular LITT (cases 4 and 5) experienced a mild transient II outcome, and 1 of 9 achieved an Engel class III out-
SMA-like syndrome, which included expressive aphasia. come with MRI-negative epilepsy. Previous studies have
One of the patients who underwent insular LITT (case 6) reported the technique and safety of the sEEG electrode
experienced a large intracerebral hemorrhage during the implantation for insular epilepsy.14 The patients in this
LITT that was detected by abrupt unexpected changes on group suffered from long-standing intractable epilepsy.
thermometry. Due to these changes, LITT was stopped Three of our 9 patients had undergone unsuccessful epi-
and MRI performed, revealing a large intracerebral hem- lepsy surgery (Table 1) previously. In those patients, the
orrhage. Immediately after the completion of the LITT initial workup and intracranial subdural electrode arrays
procedure, the patient underwent an emergency large placed during the initial phase 2 surgical treatment did not
decompressive craniectomy with placement of a dural include insular or cingulate electrodes. All 3 patients ex-
expansion graft and an intracranial pressure monitor. At perienced improved seizure control after sEEG and LITT
that time, the brain was soft and pulsatile. Angiography targeting these structures. This experience is consistent
was postponed out of concern that the intracerebral hem- with previous reports that clinical manifestations and the
orrhage might distort the vasculature. Postoperatively, the preliminary workups of this patient population can be
patient was extubated and found to have a mixed aphasia misleading.15
and right-sided weakness. She remained awake and able Previous surgical series report variable degrees of sei-
to follow some commands with an intracranial pressure zure freedom, ranging from 56%16 to 69%,17 following
below 15 mm Hg postoperatively. After inpatient rehabili- insular resection. Most of the patients in these series had
tation, she underwent angiography (negative findings) and lesions in or near insular cortex. In a pediatric study com-
replacement of her bone flap (Supplementary Fig. S2). At paring open resection versus LITT for insular epilepsy, le-
756 J Neurosurg Volume 135 • September 2021
 
* | D*2 *37 *
 Gireesh et al.

FIG. 3. Case 5. LITT of the left insula. A–C: Sagittal views showing the initial insular electrode implantation and the results of the
first LITT. D–F: Sagittal views showing the second insular electrode implantation and the results of the second LITT. Panels A and
D illustrate left anterior insular electrode locations. Panels B and E illustrate left posterior insular electrode locations. Panels C and
F are T2-weighted FLAIR images obtained after the second LITT. The sEEG recordings from the posterior insular electrode (yel-
low, panel E) demonstrated epileptogenic activity. Figure is available in color online only.

sions were identified on MR images in 60% of patients, ablation, the patient’s seizure semiology changed, with
and Engel class I outcome was reported in 43% of subjects resolution of hypermotor activity, but whole-body stiffen-
with at least a 1-year follow-up. In that cohort, 3 of the 6 ing continued. She subsequently underwent reevaluation
nonlesional cases achieved Engel class I outcome. In an- with bilateral subdural grids and sEEG electrodes. Be-
other recent study in which RFTC-based surgery was used cause monitoring provided poor seizure lateralization, she
to treat both lesional and nonlesional insular epilepsy, 53% underwent a corpus callosotomy with her electrode array
of patients achieved seizure freedom.11 This study reported in place. Following callosotomy, her seizures arose from
Engel class I outcome in 6 of their 15 MRI-negative cases. the left anterior cingulate. An open cingulate topectomy
In most of our cases, LITT was performed in the domi- was performed, resulting in a transient expressive aphasia
nant hemisphere. No permanent deficits were noted after that persisted for 3 months. During her recovery she was
the procedures. Our 3 cingulate cases underwent LITT of able to text but could not speak. She ultimately recovered
the left cingulate cortex. Our combined insular and cingu- expressive language skills and achieved complete seizure
late case also underwent left insular and cingulate LITT. freedom (Engel class IA).
To the best of our knowledge, there have been only case The other patient (case 5), who achieved partial seizure
reports on nonlesional cingulate LITT.18,19 In our series, control (Engel class III) after initial LITT in the left in-
the 3 patients (cases 7–9) with only left cingulate seizure sular cortex, underwent additional sEEG monitoring that
onset had favorable seizure outcome (Engel class I–III) demonstrated epileptogenic activity in insular tissue adja-
following cingulate LITT. The patient (case 2) with both cent to the initial ablation. That patient achieved complete
left cingulate and insular seizure onsets who underwent seizure freedom (Engel class IA) after an additional LITT
both left insular and cingulate LITT achieved Engel class procedure ablating the epileptogenic insular tissue adja-
IB outcome. cent to prior ablations (Fig. 3). This experience supports
One of our patients who underwent insular LITT (case the report by Mullatti et al. that the extent of complex ab-
4) experienced only a partial improvement in seizure out- lations should be supported by neurophysiological data.11
come following insular LITT because of insufficient initial One patient (case 7) experienced reduced but persistent
sEEG sampling. This patient underwent sEEG monitoring seizure activity following cingulate LITT. Review of the
early in our sEEG experience, before we began sampling postoperative MRI suggested that the initial LITT did not
from the cingulate gyrus. Her seizure semiology consisted completely ablate the targeted cingulate tissue. Additional
of whole-body stiffening followed by hypermotor activ- LITT was offered, but the patient declined additional sur-
ity including bicycling behavior and body rotation. With a gery.
limited sEEG array, her seizure onset was found to origi- The rate of any perioperative intracranial hemorrhage
nate from the right insula. Following right-sided insular associated with sEEG in our series was 22.2% (2 of 9),
J Neurosurg Volume 135 • September 2021 757
 
* | D*2 *37 *
Gireesh et al.

and the rate of symptomatic hemorrhage was 11.1% (1 semiologies in our series included contralateral dystonic
of 9). The bleeding associated with the sEEG implanta- posturing as well as hypermotor activity. These patterns of
tion in our cohort is similar to that reported in a recent seizure activity may be caused by the spread of the seizure
analysis.20 Also, a large intraparenchymal hemorrhage oc- activity originating from the insula and cingulum.
curred during LITT when multiple trajectories were used Because insular seizure activity is known to spread rap-
for an insular ablation (case 6, Supplementary Fig. S2). idly between hemispheres, we implanted sEEG electrodes
This hemorrhage required immediate cessation of LITT bilaterally in 5 of the 6 patients who underwent insular
and emergency craniectomy with a large duraplasty. That LITT.24 Previous series reported that open resection of the
patient recovered to preoperative neurological status with insula is associated with a risk of transient neurological
rehabilitation over a 12-month postoperative period and comorbidities as high as 75%.22 In our LITT cohort, tran-
Engel class IA seizure outcome. It may be reasonable to sient neurological deficits occurred in 4 of our 9 patients
consider emergency decompressive craniectomy with du- (44%). Even though we performed the majority (5 of 6) of
raplasty when LITT-associated hemorrhages occur. Rates our ablations in the dominant hemisphere, there was no
of perioperative hemorrhage following sEEG and LITT significant change in neuropsychological test scores fol-
in nonlesional insular and cingulate epilepsy are not well lowing LITT.
described. In a large series of 549 patients in whom sEEG
electrodes were implanted, 105 (19.1%) experienced hem- Limitations
orrhage, with 16.9% asymptomatic and 2.2% being symp- Ours is a small retrospective case series with insuffi-
tomatic.20 Reported rates of hemorrhage following LITT cient case volume to allow statistical analysis. The cohort
in the hippocampus and for hypothalamic hamartomas is heterogeneous both demographically and clinically.
fall in the range of 7%–8%.21 It is possible that rates of One-third of the patients underwent prior unsuccessful
hemorrhage from sEEG and/or LITT will be higher than epilepsy surgery. All of our cases were nonlesional, and no
in other structures, such as the hippocampus, due to the insular or cingulate tissue specimens were obtained. We
highly vasculature nature of the insula. The cause of hem- are therefore unable to determine the pathological cause
orrhage in this case is unclear, and the cerebral angiogram of seizures in our series. A prior case series reported that
was unrevealing. Large blood vessels are thought to have 5 of 6 patients with insular epilepsy had focal cortical dys-
enough blood flow to effectively transfer heat and prevent plasia.25
rupture, and small vessels are likely to get ablated. How- In one of our cases, LITT was repeated in insular tissue
ever, there may be middle-sized vessels of unknown diam- immediately adjacent to the prior ablation cavity after the
eter that do not possess the heat sink protective properties patient experienced recurrent seizure activity. In cases of
of larger vessels, thereby leading to rupture and bleeding.11 DRE without a clear lesion and with limited sEEG moni-
Case series involving patients with insular or cingulate toring volumes, it can be challenging to determine how
epilepsy surgery often describe transient neurological defi- much tissue to ablate with LITT. It also suggests that in
cits including paresis, dysphasia, alteration of taste, smell, some cases, a limited LITT of the insular region may not
hearing, pain, and thermal perceptions. These transient be adequate to achieve seizure freedom.
symptoms can occur in up to 75% of cases.22 In our series,
4 patients (44.4%) developed neurological deficits follow-
ing LITT and in 3 of 4 (33.3% of all patients) these defi-
Conclusions
cits were transient. Three patients who underwent insular The nonlesional insular and cingulate DRE cases can
LITT experienced transient expressive aphasia postproce- be effectively localized with sEEG and treated with LITT.
dure. The patient in case 5 experienced a severe mixed The complication rates associated with LITT in these lo-
expressive aphasia worse than receptive aphasia following cations are comparable to those following open insular
her initial LITT procedure (Fig. 3). At her second LITT, resection. The rates of hemorrhage may be higher than
she had a 48-hour period of marked word-finding diffi- reported in open insular resection and LITT in other loca-
culty during which she self-corrected with frustration. The tions in the brain, which may have to be considered while
patient in case 2 in our series underwent both insular and making therapeutic decisions. An insular or cingulate sei-
cingulate LITT based on the sEEG findings indicating sei- zure onset zone in the dominant hemisphere is not a con-
zure onset from both locations. In this patient, LITT on traindication to LITT. We conclude that sEEG followed by
both insular and cingulate cortex was done on the same LITT is a reasonable treatment option for DRE originat-
day and he developed a transient SMA syndrome. ing from these difficult-to-access regions of the cerebral
cortex, particularly for patients with nonlesional epilepsy.
Interpretation
The increased availability of sEEG has improved our References
understanding of cingulate and insular epilepsy, which has   1. Centers for Disease Control and Prevention (CDC). Epilepsy
led to increased surgical treatment for DRE arising from in adults and access to care—United States, 2010. MMWR
insular and cingulate regions. This experience has revealed Morb Mortal Wkly Rep. 2012;61(45):909–913.
​ ​
the complexity of clinical semiology for both insular and   2. Del Felice A, Beghi E, Boero G, et al. Early versus late
remission in a cohort of patients with newly diagnosed epi-
cingulate seizures.23 The semiology for patients with insu- lepsy. Epilepsia. 2010;51(1):37–42.
​ ​
lar epilepsy in our cohort included activity arrest, staring,   3. Englot DJ, Rolston JD, Wang DD, et al. Seizure outcomes
dystonic posturing, head drop, hypermotor activity, tin- after temporal lobectomy in pediatric patients. J Neurosurg
gling sensation in the throat, and lip smacking. Cingulate Pediatr. 2013;12(2):134–141.
​ ​

758 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Gireesh et al.

  4. Ansari SF, Tubbs RS, Terry CL, Cohen-Gadol AA. Surgery 20. McGovern RA, Ruggieri P, Bulacio J, et al. Risk analysis of
for extratemporal nonlesional epilepsy in adults:​an outcome hemorrhage in stereo-electroencephalography procedures.
meta-analysis. Acta Neurochir (Wien). 2010;​152(8):​1299– Epilepsia. 2019;​60(3):​571–580.
1305. 21. Shukla ND, Ho AL, Pendharkar AV, et al. Laser interstitial
  5. Jobst BC, Cascino GD. Resective epilepsy surgery for thermal therapy for the treatment of epilepsy:​evidence to
drug-resistant focal epilepsy:​a review. JAMA. 2015;​313(3):​ date. Neuropsychiatr Dis Treat. 2017;​13:​2469–2475.
285–293. 22. Bouthillier A, Nguyen DK. Epilepsy surgeries requiring an
  6. Desai A, Bekelis K, Darcey TM, Roberts DW. Surgical tech- operculoinsular cortectomy:​operative technique and results.
niques for investigating the role of the insula in epilepsy:​a Neurosurgery. 2017;​81(4):​602–612.
review. Neurosurg Focus. 2012;​32(3):​E6. 23. Nguyen DK, Nguyen DB, Malak R, et al. Revisiting the role
  7. Afif A, Becq G, Mertens P. Definition of a stereotactic of the insula in refractory partial epilepsy. Epilepsia. 2009;​
3-dimensional magnetic resonance imaging template of the 50(3):​510–520.
human insula. Neurosurgery. 2013;​72(1)(Suppl Operative):​ 24. Afif A, Minotti L, Kahane P, Hoffmann D. Anatomofunc-
35–46. tional organization of the insular cortex:​a study using intra-
  8. Surbeck W, Bouthillier A, Weil AG, et al. The combination cerebral electrical stimulation in epileptic patients. Epilepsia.
of subdural and depth electrodes for intracranial EEG inves- 2010;​51(11):​2305–2315.
tigation of suspected insular (perisylvian) epilepsy. Epilepsia. 25. Gras-Combe G, Minotti L, Hoffmann D, et al. Surgery for
2011;​52(3):​458–466. nontumoral insular epilepsy explored by stereoelectroen-
  9. Guénot M, Isnard J, Ryvlin P, et al. SEEG-guided RF ther- cephalography. Neurosurgery. 2016;​79(4):​578–588.
mocoagulation of epileptic foci:​feasibility, safety, and pre-
liminary results. Epilepsia. 2004;​45(11):​1368–1374.
10. Cardinale F, Cossu M, Castana L, et al. Stereoelectroen- Disclosures
cephalography:​surgical methodology, safety, and stereotactic Dr. Baumgartner has served as a paid consultant for PMT Corp.,
application accuracy in 500 procedures. Neurosurgery. 2013;​ iMRIs, Monteris, and Zimmer-Biomet.
72(3):​353–366.
11. Mullatti N, Landre E, Mellerio C, et al. Stereotactic thermo- Author Contributions
coagulation for insular epilepsy:​lessons from successes and
failures. Epilepsia. 2019;​60(8):​1565–1579. Conception and design: Gireesh, Baumgartner. Acquisition
12. Alkawadri R, So NK, Van Ness PC, Alexopoulos AV. Cingu- of data: Gireesh, Westerveld, Beegle, Baumgartner. Analysis
late epilepsy:​report of 3 electroclinical subtypes with surgi- and interpretation of data: Gireesh, Chen. Drafting the article:
cal outcomes. JAMA Neurol. 2013;​70(8):​995–1002. Gireesh, Chen. Critically revising the article: Lee, Skinner, Seo,
13. Hale AT, Sen S, Haider AS, et al. Open resection versus laser Chen, Westerveld, Beegle, Castillo, Baumgartner. Reviewed sub-
interstitial thermal therapy for the treatment of pediatric in- mitted version of manuscript: Gireesh. Approved the final version
sular epilepsy. Neurosurgery. 2019;​85(4):​E730–E736. of the manuscript on behalf of all authors: Gireesh. Statistical
14. Alomar S, Mullin JP, Smithason S, Gonzalez-Martinez J. analysis: Gireesh. Study supervision: Gireesh, Baumgartner.
Indications, technique, and safety profile of insular stereo-
electroencephalography electrode implantation in medically Supplemental Information
intractable epilepsy. J Neurosurg. 2018;​128(4):​1147–1157. Online-Only Content
15. Obaid S, Zerouali Y, Nguyen DK. Insular epilepsy:​semiol- Supplemental material is available with the online version of the
ogy and noninvasive investigations. J Clin Neurophysiol. article.
2017;​34(4):​315–323. Supplementary Tables and Figures. https://thejns.org/doi/
16. Freri E, Matricardi S, Gozzo F, et al. Perisylvian, including suppl/​10.3171/​2020.7.JNS201912.
insular, childhood epilepsy:​presurgical workup and surgical
outcome. Epilepsia. 2017;​58(8):​1360–1369. Previous Presentations
17. Weil AG, Le NM, Jayakar P, et al. Medically resistant pedi-
atric insular-opercular/perisylvian epilepsy. Part 2:​outcome Presented as “Intracranial EEG and LASER Ablation in Non-
following resective surgery. J Neurosurg Pediatr. 2016;​18(5):​ Lesional Insular and Cingulate Epilepsy—Case Series” at the
523–535. American Epilepsy Society conference in Baltimore, on Decem-
18. Upadhyayula PS, Rennert RC, Hoshide R, et al. Laser abla- ber 7, 2019.
tion of a nonlesional cingulate gyrus epileptogenic zone
using robotic-assisted stereotactic EEG localization:​a case Correspondence
report. Stereotact Funct Neurosurg. 2019;​97(1):​10–17. Elakkat D. Gireesh: Neuroscience Institute, AdventHealth, Orlan-
19. Marashly A, Loman MM, Lew SM. Stereotactic laser abla- do, FL. [email protected].
tion for nonlesional cingulate epilepsy:​case report. J Neuro-
surg Pediatr. 2018;​22(5):​481–488.

J Neurosurg Volume 135 • September 2021 759

 
* | D*2 *37 *
 CLINICAL ARTICLE
J Neurosurg 135:760–769, 2021

Resective temporal lobe surgery in refractory temporal

lobe epilepsy: prognostic factors of postoperative seizure
outcome
Valeri Borger, MD,1 Motaz Hamed, MD,1 Julia Taube, MSc,2 Gülsah Aydin, MD,1 Inja Ilic, MD,1
Matthias Schneider, MD,1 Patrick Schuss, MD, PhD,1 Erdem Güresir, MD, PhD,1
Albert Becker, MD, PhD,3 Christoph Helmstaedter, PhD,2 Christian E. Elger, MD, PhD,2 and
Hartmut Vatter, MD, PhD1
Departments of 1Neurosurgery and 2Epileptology, and 3Institute of Neuropathology, University Hospital Bonn, Germany

OBJECTIVE  Temporal lobe epilepsy (TLE) is one of the most common forms of epilepsy. In approximately 30% of
patients, seizures are refractory to drug treatment. Despite the achievements of modern presurgical evaluation in recent
years, the presurgical prediction of seizure outcome remains difficult. The aim of this study was to evaluate the seizure
outcome in patients with drug-refractory TLE who underwent resective temporal lobe surgery (rTLS) and to determine
features associated with unfavorable postsurgical seizure outcome.
METHODS  Patients with medically refractory TLE who underwent rTLS between 2012 and 2017 were reviewed from the
prospectively collected epilepsy surgery database. A retrospective analysis of clinical, radiological, neuropsychological,
histopathological, and perioperative findings of 161 patients was performed. The patients were divided into two groups
according to seizure outcome (group I, International League Against Epilepsy [ILAE] class 1; group II, ILAE class ≥ 2).
For identification of independent risk factors for unfavorable postoperative seizure outcome (ILAE class ≥ 2), a multivari-
ate logistic regression analysis was performed.
RESULTS  Seizure freedom (ILAE class 1) was achieved in 121 patients (75.2%). The neuropsychological evaluation
demonstrated that losses in cognitive performance were more pronounced in verbal memory after resections in the left
temporal lobe and in nonverbal memory after right-sided resections, whereas attention improved after surgery. Overall,
postoperative visual field deficits (VFDs) were common and occurred in 51% of patients. There was no statistically sig-
nificant difference in the incidence of VFD in patients with selective surgical procedures compared to the patients with
nonselective procedures. The lack of MRI lesions and placement of depth electrodes were preoperatively identified as
predictors for unfavorable seizure outcome.
CONCLUSIONS  rTLS is an effective treatment method in patients with refractory TLE. However, patients with a lack of
MRI lesions and placement of depth electrodes prior to rTLS are at higher risk for an unfavorable postsurgical seizure
outcome.
https://thejns.org/doi/abs/10.3171/2020.7.JNS20284
KEYWORDS  epilepsy surgery; hippocampal gliosis; temporal lobe epilepsy; seizure outcome

T
emporal lobe epilepsy (TLE) is one of the most ral lobe surgery (rTLS) has been a reasonable option for
common forms of epilepsy, first described by treatment in these patients.2 In a meta-analysis including
Hughlings-Jackson. In approximately 30% of pa- 32 studies with 2250 patients, Engel et al. reported that
tients, seizures are refractory to drug treatment.1 Since the after rTLS, seizure freedom was achieved in 65% of pa-
first randomized controlled trial by Wiebe et al. showed tients with TLE.3 In a recently published review, Englot
significantly improved outcomes with epilepsy surgery and Chang reported that the existing data favoring sur-
over drug treatment in refractory TLE, resective tempo- gery for appropriately selected candidates with refractory

ABBREVIATIONS  AHE = amygdalohippocampectomy; ATL = anterior temporal lobectomy; CA = cornu ammonis; CI = confidence interval; DCS-R = Diagnostikum für Zere-
bralschäden–Revised; EEG = electroencephalography; FCD = focal cortical dysplasia; HG = hippocampal gliosis; HS = hippocampal sclerosis; ILAE = International League
Against Epilepsy; OR = odds ratio; rTLS = resective temporal lobe surgery; sAHE = selective AHE; TLE = temporal lobe epilepsy; VFD = visual field deficit; VLMT = verbal
learning and memory test.
SUBMITTED  January 29, 2020.  ACCEPTED  July 31, 2020.
INCLUDE WHEN CITING  Published online January 8, 2021; DOI: 10.3171/2020.7.JNS20284.

760 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *37 *
 Borger et al.

TLE are convincing and suggest that a cure is possible in were pathologies within the temporal lobe without the in-
some patients with this disorder.4 Despite this fact and all volvement of the hippocampus, such as gliosis, gangliogli-
achievements of modern presurgical evaluation in recent oma, cavernoma, and focal cortical dysplasia (FCD). The
years, the presurgical prediction of seizure outcome re- diagnosis of HG was histologically confirmed as reactive
mains difficult. The aim of this study was to evaluate sei- astrogliosis without neuronal loss within the resected hip-
zure outcome in patients with drug-refractory TLE who pocampus. Neoplastic lesions were classified according to
underwent rTLS at our center and to determine features the WHO classification.9 The FCD was classified accord-
associated with unfavorable postsurgical seizure outcome. ing to the new ILAE classification.10 The last category was
no specific pathological changes.
Methods
Surgical Outcome Analysis
Population and Presurgical Evaluation
After rTLS, the outcome was assessed during follow-
Patients with TLE who underwent rTLS between 2012 up visits at 6 and 12 months. Patients with a follow-up pe-
and 2017 were reviewed from the prospectively conducted riod less than 12 months were not included in this study.
epilepsy surgery database at our center. The formation of At the 12-month visit, all patients underwent a thorough
this database was approved by the local ethics commit- clinical examination, evaluation of seizure outcome,
tee. Given the retrospective nature of the study, written video-EEG recording, 3-T MRI, and neuropsychologi-
informed consent was not required. cal reassessment. The postoperative seizure outcome was
During the studied period, rTLS was performed in a
assessed according to the ILAE classification.11 The pa-
total of 184 patients. There were 23 cases lost to follow-up;
tients were divided into two groups according to seizure
the most common reason was treatment of patients from
outcome: group I (ILAE class 1) and group II (ILAE class
abroad who moved back to their countries after surgery.
≥ 2).
These patients did not complete the follow-up visits at our
Surgically associated complications were assessed
center. Because we included only patients with completed
during the postoperative course of treatment. The clini-
follow-up at 12 months after rTLS, complete data sets for
cally relevant events requiring surgical revision, such as
161 consecutive patients were available. All patients suf-
bleeding complications and surgical site infections, were
fered from medically refractory TLE and had undergone
adequate treatment with at least two first-line antiepileptic analyzed. Furthermore, relevant newly occurring neuro-
drugs. A retrospective analysis of clinical, radiological, logical deficits such as motor deficits, aphasia, and cranial
histopathological, and perioperative findings was per- nerve palsy were assessed and analyzed.
formed. All patients underwent MRI within 2–3 days post-
All patients were preoperatively assessed in the depart- operatively to detect the extent of resection of desired
ment of epileptology in a similar fashion and were consid- structures. Incomplete resection was determined in cases
ered to be suitable for surgery.5,6 The evaluation included in which a substantial remnant of target tissue was not
detailed history of seizures, medical history, high resolu- reached by resection as confirmed by postoperative MRI,
tion 3-T MRI, neuropsychological assessment, and video- whether it was from imprecise delimitation or surgical
electroencephalography (EEG) monitoring using continu- and functional limitation. The target structures were
ous recordings. In patients with absent or several lesions different according to the surgical procedure. In candi-
on MRI, PET and SPECT were performed to identify a dates who underwent selective amygdalohippocampec-
seizure focus. In cases with inconclusive findings, invasive tomy (sAHE) or anterior temporal lobectomy (ATL) with
EEG monitoring was performed using stereotactically im- amygdalohippocampectomy (ATL with AHE), the resec-
planted depth electrodes.7 tion was stated as incomplete if mesial temporal struc-
tures such as the hippocampus, amygdala, or superior
part of the uncus were insufficiently removed. In patients
Surgical Procedures who underwent temporal lesionectomy or ATL without
All surgical procedures were performed while patients AHE, the resection was incomplete if any parts of the ep-
were under general anesthesia using intraoperative neu- ileptogenic lesion were not addressed by the surgery. Af-
ronavigation and intraoperative neurophysiological moni- ter completion of the presurgical evaluation, the extent of
toring with motor evoked and somatosensory evoked po- resection and the desired structures that were intended to
tentials. The goal of surgery was to remove temporal and be removed were defined by the responsible epileptologist
temporomesial structures, including the lesion depicted on and reviewed by the interdisciplinary epilepsy surgery
MRI or anatomical area with presumed seizure focus. conference. In cases where needed, resection masks were
generated and included in the intraoperative neuronaviga-
Histopathological Examination tion. Following surgery, the resection was matched with
The resected tissue was obtained from all patients in the presurgical resection mask to confirm the proper ex-
this study. Standardized neuropathological analysis was tent of resection. The postoperative MRI was performed
performed in all preserved specimens by local neuropa- in each patient within 2–3 days to confirm the extent of
thologists. The histopathological findings were differ- resection of addressed structures or lesions and rule out
entiated into three categories. First were hippocampal surgical complications such as bleeding, infarction, and
pathologies such as hippocampal sclerosis (HS) or hip- damage to brain tissue along the surgical approach. The
pocampal gliosis (HG) according to the International postsurgical MR images were analyzed by experienced
League Against Epilepsy (ILAE) classification.8 Next neuroradiologists.

J Neurosurg Volume 135 • September 2021 761

* | D*2 *37 *
Borger et al.

Neuropsychological Assessment for TLE were included in this analysis. There were 85
The neuropsychological evaluation focused on tests of males (52.8%). The surgery was performed in 81 patients
verbal and nonverbal memory representing temporal lobe (50.3%) on the left side and in 80 (49.7%) on the right.
functions. In addition, attention and executive functions, The mean age at epilepsy onset was 17.32 ± 13.09 years,
visuospatial abilities, and language and motor functions and the mean age at surgery was 36.1 ± 14.96 years. The
were considered. Verbal memory was measured via the mean duration of epilepsy was 19.1 ± 13.77 years. Ac-
verbal learning and memory test (VLMT). For visual cording to seizure outcome, 121 patients were assigned to
learning, the Diagnostikum für Zerebralschäden–Revised group I (ILAE class 1) and 40 to group II (ILAE class
(DCS-R) was applied. Parallel versions of the VLMT ≥ 2). Regarding basic clinical characteristics such as age
and DCS-R were used to minimize practice effects at the at seizure onset, age at surgery, duration of epilepsy, and
follow-up. Attention was assessed by the EpiTrack screen- side of surgery, there were no statistically significant dif-
ing tool and the d2 Aufmerksamkeitsbelastungstest. Lan- ferences (Table 1). During the presurgical evaluation, the
guage assessment comprised the BNT and the Token Test. invasive evaluation using depth electrodes was performed
Visuospatial abilities were evaluated via Leistungspruef- in significantly more patients in group II compared to
system subtest 7 and Wechsler Adult Intelligence Scale group I (15 [37.5%] vs 21 [17.4%], p = 0.015). The analysis
block design. The tests and their references are described of preoperative MRI revealed the evidence of HS as the
in previous articles.12 Pre- and postoperative test results most common radiological pathology in both groups (79
from each cognitive domain were summarized and classi- [65.3%] in group I vs 20 [50.0%] in group II, nonsignifi-
fied into a 5-point scale, ranging from severely impaired cant difference). The distribution of other lesions is shown
to above average (severely impaired = 0, at least two test in Table 1. A negative MR image without any lesions was
scores > 2 standard deviations below the mean of the nor- found significantly more often in group II compared with
mative sample; impaired = 1, at least two test scores > 1 group I (8 [20%] vs 6 [5%], p = 0.007).
standard deviation below the mean; borderline = 2, one
test score below the mean; unimpaired = 3, no test score Surgical Procedures
> 1 standard deviation below the mean; above average = Overall, the leading surgical procedure performed was
4, at least two test scores > 1 standard deviation above transsylvian sAHE in 91 of 161 patients. The ATL with
the mean). The distance between two subsequent catego- AHE was performed in 30 of 161 patients and without
ries approximately corresponds to one standard deviation AHE in 15 of 161 patients. In 20 of 161 patients, a tailored
from the mean standardized score across all test scores of lesionectomy without AHE was performed followed by
the respective domain. lesionectomy with AHE in 5 of 161 patients. As shown
in Table 2, the analysis revealed no differences related to
Ophthalmological Examination surgical procedure between the two outcome groups.
Visual fields were examined in each patient pre- and
postoperatively using kinetic Goldmann perimetry. A new Histopathological Examination
postoperatively diagnosed visual field deficit (VFD) was The overview of the histopathological findings is shown
classified as a superior quadrantanopia or homonymous in Table 1. There were significantly more patients with HS
hemianopia. in the group with favorable seizure outcome (77 [63.6%] in
group I vs 17 [42.5%] in group II, p = 0.026). Furthermore,
Statistical Analysis the prevalence of HG was significantly higher in group II
Statistical data analysis was performed using the SPSS compared with group I (11 [27.5%] vs 10 [8.3%], p = 0.05).
software package (IBM SPSS Statistics for Windows, ver- In regard to other histopathological findings, the groups
sion 25.0., IBM Corp.). Associations between parametric did not differ significantly (Table 1).
variables were analyzed using an unpaired, two-tailed
Student t-test. For analysis of associations between non- Surgical Outcome Analysis
parametric variables, the Mann-Whitney U-test was used. Of the 161 patients, at 6- and 12-month follow-up visits
Associations of categorical variables were compared using after rTLS a favorable seizure outcome with seizure free-
the chi-square or Fisher exact test. Results with p values < dom (ILAE class 1) was achieved in 121 patients (75.2%).
0.05 were considered statistically significant. For identifi- The proportion of patients with ILAE class 2–6 was not
cation of independent risk factors for unfavorable postop- significantly different for each ILAE class at the 6- and
erative seizure outcome (ILAE class ≥ 2), a multivariate 12-month follow-ups, respectively (Table 3). The analysis
logistic regression analysis was performed including the of surgical complications revealed an overall complica-
variables with significant p values in univariate analysis. tion rate of 11.8%. The overall rate of revision surgery was
The results of the analysis were presented by logistic re- 8.1% (Table 2). Surgical site infections were the most fre-
gression as odds ratio (OR) with a 95% confidence interval quent complication (in 9 [5.6%] of 161 patients), followed
(CI). by bleeding complications (6 [3.7%] of 161 patients). The
comparison of the two outcome groups revealed no sig-
nificant differences, either for overall complication rate or
Results rate of surgical revision, or for bleeding complications and
Population and Presurgical Evaluation infections in each group (Table 2). Transient motor neuro-
In total, data from 161 patients who underwent rTLS logical deficits such as paresis and hemiparesis occurred

762 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Borger et al.

TABLE 1. Patient demographics and characteristics according to ILAE seizure outcome class
Characteristic Overall Group I (ILAE class 1) Group II (ILAE class 2–6) p Value
No. of patients 161 121 40
Sex, n (%)
 Male 85 (52.8) 62 (51.2) 23 (57.5) NS
 Female 76 (47.2) 59 (48.8) 17 (42.5) NS
Mean age at epilepsy onset ± SD, yrs 17.32 ± 13.09 17.9 ± 13.59 15.56 ± 11.4 NS
Mean duration of epilepsy ± SD, yrs 19.1 ± 13.77 19.06 ± 13.8 19.4 ± 13.71 NS
Mean age at surgery ± SD, yrs 36.1 ± 14.96 36.66 ± 15.53 34.58 ± 13.12 NS
Site of surgery, n (%) NS
 Lt 81 (50.3) 56 (46.3) 25 (62.5) NS
 Rt 80 (49.7) 65 (53.7) 15 (37.5) NS
Invasive presurgical evaluation w/ depth electrodes, n (%) 36 (22.4) 21 (17.4) 15 (37.5) 0.015
Preop MRI findings, n (%)
  Unilateral HS 99 (61.5) 79 (65.3) 20 (50.0) NS
  Hippocampal lesions other than HS 27 (16.8) 20 (16.5) 7 (17.5) NS
  Temporal lesion w/o hippocampal involvement 21 (13.0) 16 (13.2) 5 (12.5) NS
  No lesion 14 (8.7) 6 (5.0) 8 (20.0) 0.007
Histology of hippocampus, n (%)
 HS 94 (58.4) 77 (63.6) 17 (42.5) 0.026
 HG 21 (13.0) 10 (8.3) 11 (27.5) 0.005
Histology of TL tissue w/o hippocampus, n (%)
  Temporal gliosis 11 (6.8) 6 (5.0) 5 (12.5) NS
 Ganglioglioma 9 (5.6) 7 (5.8) 2 (5.0) NS
 Cavernoma 6 (3.7) 5 (4.1) 1 (2.5) NS
  FCD type I 2 (1.2) 2 (1.7) 0 (0.0) NS
 Other 6 (3.7) 5 (4.1) 1 (2.5) NS
  No specific histopathological changes 12 (7.4) 9 (7.4) 3 (7.5) NS
NS = nonsignificant; TL = temporal lobe.

in 8 (4.9%) of 161 patients and were not significantly dif- improved after surgery (F [1, 99] = 12.561, p < 0.01, η2 =
ferent between groups I and II (6 [4.9%] vs 2 [5%], non- 0.113). Language and visuospatial abilities did not show
significant). The postoperative MRI showed that desired significant changes.
extent of resection was significantly often not achieved in Consistent with the group-level analysis, individual-
group II compared to group I (4 [10%] in group II vs 2 level analysis indicated that verbal memory decline was
[1.7%] in group I, p = 0.034). more frequent after left rTLS (63%) than after right rTLS
(38%). Visual memory was worse for 48% of the patients
Neuropsychological Outcome after right-sided and for 25% after left-sided resections
Before surgery, visual memory was impaired in 66% of (χ2 [6] = 11.373, p < 0.05). Deteriorations of visuospatial
patients, followed by verbal memory, language, and atten- abilities and language were noted in 14%–19% of cases.
tion in approximately 50% each. Visuospatial functions Attention improved after surgery (39% vs 16%). Figure 3
were affected in 39% of cases. Preoperatively, there were displays the number of patients with significant individual
no significant differences in performance between left and changes, corrected for floor effects.
right TLE (p = 0.29–0.80; Fig. 1).
Group-level analysis, by means of repeated-measures Postoperative Visual Field Impairment
ANOVA, revealed an interaction effect of visual memory
and surgical side (F [1,99] = 4.752, p = 0.032, η2 = 0.046). Overall, postoperative VFDs were common and oc-
Patient performance was worse after right-sided resec- curred in 82 (51%) of 161 patients. The most frequent
tions (Fig. 2, left). A significant main effect of surgery (F VFD was superior quadrantanopia (40%). Homonymous
[1,101] = 10.831, p < 0.01, η2 = 0.097) and a significant hemianopia occurred in 11% of the patients. There was no
main effect of surgical side (F [1,101] = 4.379, p < 0.05, η2 statistically significant difference in the overall incidence
= 0.042) were found for verbal memory. There was a trend rate of VFD in patients with selective versus nonselective
for an interaction of side and surgery (F [1,101] = 3.127, procedures, either for superior quadrantanopia or for hom-
p = 0.08, η2 = 0.03; Fig. 2, right). Attention significantly onymous hemianopia (Table 4).

J Neurosurg Volume 135 • September 2021 763

* | D*2 *37 *
Borger et al.

TABLE 2. Distribution of surgical modality and complications a favorable seizure outcome (ILAE class 1) was achieved
during the perioperative course of treatment according to in 75% of patients 1 year after surgery, which is consis-
seizure outcome tent with published data.16–20 Schmeiser et al. reported on
Group I, Group II, a series of 458 patients with TLE who were treated with
Variable Overall n = 121 n = 40 different surgical approaches. They found no differences
in short- and long-term seizure outcomes in regard to sur-
Surgery modality gical approach.16 Other studies addressing this aspect have
 sAHE 91 (56.5) 67 (55.4) 24 (60.0) shown comparable results regarding the seizure outcome
  ATL w/ AHE 30 (18.6) 21 (17.3) 9 (22.5) between standard temporal lobectomy and sAHE.21,22 The
  ATL w/o AHE 15 (9.3) 11 (9.1) 4 (10.0) systematic review and meta-analysis by Josephson et al.
shows that ATL is slightly more effective than sAHE re-
  LE w/ AHE 5 (3.1) 4 (3.3) 1 (2.5)
garding seizure outcome.18 Some authors have reported
  LE only 20 (12.5) 18 (14.9) 2 (5.0) that sAHE may carry the risk of seizure recurrence in
Overall surgical complications 19 (11.8) 12 (9.9) 7 (17.5) patients with an unrecognized lateral temporal epilepto-
  Bleeding complication 6 (3.7) 4 (3.3) 2 (5.0) genic zone.23 The analysis in this current series revealed
  Surgical site infection 9 (5.6) 5 (4.1) 4 (10.0) that surgical modality did not have an impact on seizure
Overall revision surgery 13 (8.1) 8 (6.6) 5 (12.5) outcome. The overall complication rate in our series was
11.8%, and a revision surgery was required in 8.1% of all
LE = lesionectomy. patients. However, the reported complication rates are
Data are given as number (%). All statistical comparisons between the two difficult to compare due to different surgical approach-
groups for each variable were nonsignificant.
es, different underlying pathologies, and heterogeneous
study populations. Surgical site infections were the most
common complications (5.6%) in our series, followed by
Multivariate Logistic Regression Analysis bleeding complications (3.7%). The reported rate of infec-
We performed a stepwise multivariate logistic regres- tions ranges between 1.5% and 8.5%.24,25 The occurrence
sion analysis using the variables “invasive preoperative rate of new postoperative motor deficits (hemiparesis) as
evaluation,” “evidence of a lesion in preoperative MRI,” reported by Erba et al.25 was 4.3%, and 1.2% in the series
“histopathological evidence of HS,” “histopathological by Schmeiser et al.16 In our series, hemiparesis occurred
evidence of HG,” and “extent of resection on postopera- in 4.9% of patients and was completely resolved in all pa-
tive MRI” to find independent predictors for unfavorable tients during the observation period. With respect to newly
seizure outcome (ILAE class ≥ 2). The analysis showed occurring neurological deficits, the comparison of patients
that the histopathological evidence of HG (OR 4.99, 95% with favorable and unfavorable outcomes in our series re-
CI 1.9–13.1, p = 0.001) and incomplete resection (OR 9.08, vealed no impact on seizure outcome. VFDs are a com-
95% CI 1.6–52.5, p = 0.014) were independent and sig- mon side effect after TLS. Due to inconsistent and differ-
nificant predictors for unfavorable seizure outcome after ent definitions, the reported rate of VFDs has a very wide
rTLS in TLE (Table 5). range (between 1.5% and 69%).24 In their study, Schmeiser
et al. reported on a large cohort of patients suffering from
TLE (overall rate of 73%).26 In patients who underwent
Discussion ATL the overall rate was 83%, and in patients who under-
Resective epilepsy surgery is an established treatment went transsylvian sAHE the rate was 74%. In the current
option in patients with focal refractory epilepsy, particu- series, the overall rate of postoperative VFD is consistent
larly those with TLE.2 However, although it is effective, with the reported literature. However, our results did not
it has been demonstrated that seizure freedom rates de- reveal any differences between selective and nonselective
crease over time after surgery.13 There are studies report- surgical procedures. Due to the fact that at our institution
ing that surgical treatment for TLE fails to provide a sei- the sAHE was performed exclusively via a transsylvian
zure-free outcome in 20%–30% of these patients.14,15 The approach, there are some limitations with regard to com-
reasons behind failure of surgical treatment are multiple parability of the data with other studies.
and comparison with existing data is difficult because of As mesiotemporal and neocortical structures play an
methodological issues. In this study, we tried to identify important role in memory function, postoperative memo-
factors associated with unfavorable seizure outcome in pa- ry impairment is a major sequela after rTLS. In the current
tients with TLE who underwent rTLS. In the present study, series, left TLE patients were generally more impaired

TABLE 3. Seizure outcome according to ILAE classification at 6- and 12-month follow-up visits

Follow-Up ILAE Classification

(mos) 1 2 3 4 5 6 Total
6 121 (75.2) 9 (5.6) 11 (6.8) 10 (6.2) 8 (5.0) 2 (1.2) 161
12 121 (75.2) 10 (6.2) 8 (5.0) 10 (6.2) 10 (6.2) 2 (1.2) 161
Data are given as number (%).

764 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Borger et al.

FIG. 1. Comparative histogram demonstrates the results of the preoperative cognitive performance. The results from each cogni-
tive domain are summarized and classified into a 5-point scale ranging from severely impaired to above average. The values
represent cumulative percentage of performance categories in each tested cognitive domain according to the side of the TLE.
Impaired = cumulative percentages of impaired and severely impaired performance categories; unimpaired = cumulative percent-
ages of unimpaired and above-average performance categories; borderline = percentages of borderline performance categories.
Visual memory was impaired in 66% of patients, followed by verbal memory, language, and attention in approximately 50% each,
respectively. Visuospatial functions were affected in 39% of cases. Preoperatively there were no significant differences in perfor-
mance between left and right TLE (p > 0.29–0.80).

than right TLE patients, and verbal learning and memory In a meta-analysis on a total of 3511 patients reported
deteriorated similarly in both groups. Language and vi- by Tonini et al., the authors found that intracranial moni-
suospatial abilities did not show significant changes. Our toring was a predictor for unfavorable seizure outcome.29
findings are consistent with comparable previously pub- In accordance with these results, in the current series we
lished studies in which deterioration of verbal memory found significantly more patients in the group with unfa-
has been observed after left-sided resections and visual vorable seizure outcome who underwent invasive presur-
memory deterioration has been observed after right-sided gical evaluation with depth electrodes. Interestingly, in
resections.16,27 Selective attention significantly improved multivariate logistic regression analysis in our series, this
after surgery, which could be due to the relatively high variable failed to be an independent predictor for an un-
number of seizure-free patients in our study cohort.28 favorable seizure outcome. According to published data,

FIG. 2. Performance in visual and verbal memory before and after surgery according to the side of the resection. Group-level
analysis revealed an interaction effect of visual memory and surgical side (F [1,99] = 4.752, p = 0.032, η2 = 0.046). Patient perfor-
mance was worse after right-sided resections than after left-sided resections (left). In addition, a significant main effect of surgery
(F [1,101] = 10.831, p < 0.01, η2 = 0.097) and a significant main effect of surgical side (F [1,101] = 4.379, p < 0.05, η2 = 0.042) was
found for verbal learning and memory. There was a trend for an interaction of side and surgery (F [1,101] = 3.127, p = 0.08, η2 =
0.03) (right).

J Neurosurg Volume 135 • September 2021 765

* | D*2 *37 *
Borger et al.

FIG. 3. Postoperative changes in performance categories according to side of resection. The histogram displays the number of
patients with significant individual changes, corrected for floor effects. To account for floor effects, patients with the lowest pos-
sible baseline score without postoperative change were filtered. We identified 13 patients with floor effects in verbal memory, 15
patients in visual memory, 3 patients in attention, 2 patients in visuospatial abilities, and 1 in the language domain. This revealed
a higher rate of postoperative decline in verbal memory after left-sided resections (χ2 [2] = 9.160, p = 0.01). The other findings
remained the same as in the whole sample. Considering ceiling effects, the results were not significantly different from the results
obtained from the whole sample. The bars for verbal and visual memory exclude patients with floor effects. The asterisks repre-
sent significant postoperative changes.

about 20%–30% of patients with TLE have normal MRI Menezes et al.34 and Roberts et al.35 showing that seizure
without epileptogenic lesions.30,31 The reported rates of sei- outcome in patients with normal MRI is comparable to
zure-free outcome following rTLS in these patients varied that in patients with abnormal MRI. Ivanovic et al. showed
widely, between 20% and 80%.32,33 In our studied popula- similar results in their analysis.33
tion, the overall rate of patients with MRI-negative TLE Regarding the histopathological findings, there are
was 8.7%. There were significantly more patients (20%) several studies suggesting that HS and its distinct pattern
with negative MRI in the group with unfavorable seizure may predict surgical outcome in patients with TLE.36–38
outcome compared to the 5% in the group with favorable According to the ILAE Task Force, neuronal loss may
seizure outcome. In contrast to the data published by To- affect all of the areas of the cornu ammonis (CA; HS
nini et al., in the multivariate logistic regression analysis ILAE type 1), predominantly CA1 (HS ILAE type 2), or
in the present series, negative MRI also failed to be an predominantly CA4 (HS ILAE type 3).8 Another pattern
independent prognostic factor for unfavorable seizure out- described in surgical specimens is astrogliosis without
come. These findings suggest that a normal MRI and the neuronal loss, and it is called “no hippocampal sclerosis,
need for invasive presurgical evaluation are not always as- gliosis only.” It is unclear whether HG precedes neuronal
sociated with worse postoperative seizure outcome. This loss leading to HS or whether it is a distinct disease entity.
suggestion can be supported by data reported by Sotero de The data evaluating the impact of HG on seizure outcome

TABLE 4. Occurrence of VFDs according to surgical procedures

Surgical Procedures
Variable Overall Selective Nonselective
No. of procedures 161 91 70
No VFD 79 (49) 39 (43) 40 (57)
VFD w/ superior quadrantanopia 65 (40) 43 (47) 22 (32)
VFD w/ homonymous hemianopia 17 (11) 9 (10) 8 (11)
Total 161 (100) 91 (100) 70 (100)
Data are given as number (%). All comparisons of VFDs between surgical groups were nonsignificant.

766 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Borger et al.

TABLE 5. Multivariate logistic regression analysis of factors related to unfavorable seizure outcome (ILAE class 2–6)
Seizure Outcome Multivariate Analysis
Factor Analyzed Group I, n = 121 Group II, n = 40 OR 95% CI p Value
Invasive presurgical evaluation w/ depth electrodes 21 (17.4%) 15 (37.5%) 2.01 0.8–5.1 0.144
Preop MRI w/o lesion 6 (5%) 8 (20%) 1.44 0.3–6.5 0.633
Histopathological evidence of HS 77 (63.6%) 17 (42.5%) 1.60 0.7–3.8 0.295
Histopathological evidence of HG 10 (8.3%) 11 (27.5%) 4.99 1.9–13.1 0.001
Incomplete resection 2 (1.7%) 4 (10%) 9.08 1.6–52.5 0.014

in patients with TLE following resective surgery is scarce. series shows significantly more patients with incomplete
The majority of the literature is focused on evaluation of resection on postoperative MRI in the group with unfavor-
the impact of HS on postsurgical seizure outcome. How- able seizure outcome. The data in the current study also
ever, the identification of HG as an independent predic- revealed the evidence of HG and an incomplete resection
tor for lack of seizure freedom in our series is an aspect of the epileptogenic lesion as independent predictors of un-
that is underrepresented in the literature. Since ILAE de- favorable postoperative seizure outcome. Furthermore, the
veloped a consensus classification of HS, several reports analysis shows that even though there were nonsignificant
have been published to rule out the impact of subtypes differences, there was a much higher proportion of gliosis
of HS on postoperative seizure outcome. In their recently in the specimen obtained from temporal lobe tissue among
published series on 307 cases with TLE and HS, Gales patients without effects on the hippocampus and with un-
et al. found no clear correlation between HS subtype and favorable seizure outcome compared to the seizure-free
epilepsy surgery outcome.39 Similar results were found by group. This could be caused by the fact that neocortical
Deleo et al.40 and Savitr Sastri et al.,41 who showed no sig- temporal lesions without clearly circumscribed pathology
nificant difference in short-term seizure outcome between are less resectable. Thus, such lesions might possibly have
patients with different HS subtypes. In their recently pub- worse results for reasons related to the lesion itself, not to
lished series, Hattingen et al. found that patients with hip- the surgery.
pocampal “gliosis only” according to the ILAE classifi-
cation have distinct histopathological and MRI patterns Study Limitations
compared with HS.42 In the current series, we did not The present study has several limitations. One of the
distinguish between patient HS subtypes and seizure out- strengths of the present series is a relatively large study
come. However, the analysis of histopathological features population, which was treated at a high-volume center in
in our series revealed HS as the most frequent pathology. a standardized fashion. Our study suffers from the risk of
Furthermore, there were significantly more patients with bias inherent to retrospective cohort analysis. Even though
HS in the group with favorable seizure outcome. In con- data analysis was retrospective, data acquisition was pro-
trast, HG was found significantly more often in patients spective. However, the implementation of standardized
with unfavorable outcome. Yet, only HG was identified to neurosurgical approaches and strict variable definitions
be an independent predictor for unfavorable outcome in might mitigate some of the shortcomings of a retrospec-
the multivariate logistic regression analysis. These find- tive study design.
ings may support the suggestion made by Hattingen et
al., who identified HG as a distinct entity in patients with
TLE. In our opinion, this finding is important given that Conclusions
several reports have recently been published describing Our analysis shows that rTLS is an effective treatment
features with the potential to distinguish between HG and method in patients with refractory TLE. However, patients
HS on preoperative MRI using novel methods of neuro- with a lack of lesions on MRI and placement of depth
imaging.43 Further progress in neuroimaging may allow electrodes prior to rTLS are at higher risk for an unfavor-
us to detect the underlying pathology within the hippo- able postsurgical seizure outcome. Therefore, these facts
campus more precisely on the preoperative scan. Thus, should be carefully taken into account, and each of these
the fact that HG independently predicts seizure outcome patients needs an individual approach during the selection
is novel in relationship to prior publications. process for surgery.
The insufficient resection of epileptogenic structures
is an obvious reason for continued seizures after epilepsy
surgery.29,44 The resection may prove difficult with struc- References
tures involving eloquent brain area or those not easy to   1. Engel J Jr. Surgical treatment for epilepsy:​too little, too late?
access surgically. There are several series reporting that JAMA. 2008;​300(21):​2548–2550.
  2. Wiebe S, Blume WT, Girvin JP, Eliasziw M. A randomized,
further resection of residual epileptogenic structures can controlled trial of surgery for temporal-lobe epilepsy. N Engl
result in a seizure-free outcome.12,45 This fact supports the J Med. 2001;​345(5):​311–318.
suggestion that a subgroup of patients fail rTLS for TLE   3. Engel J Jr, Wiebe S, French J, et al. Practice parameter:​tem-
because of incomplete resection of mesial temporal struc- poral lobe and localized neocortical resections for epilepsy:​
tures. In accordance with these results, the analysis in our report of the Quality Standards Subcommittee of the Ameri-

J Neurosurg Volume 135 • September 2021 767

* | D*2 *37 *
Borger et al.

can Academy of Neurology, in association with the American potomy for the treatment of medial temporal lobe epilepsy.
Epilepsy Society and the American Association of Neuro- Epilepsia. 1999;40(10):1408–1416.
​ ​
logical Surgeons. Neurology. 2003;​60(4):​538–547. 24. Georgiadis I, Kapsalaki EZ, Fountas KN. Temporal lobe
  4. Englot DJ, Chang EF. Rates and predictors of seizure free- resective surgery for medically intractable epilepsy:​a review
dom in resective epilepsy surgery:​an update. Neurosurg Rev. of complications and side effects. Epilepsy Res Treat. 2013;​
2014;​37(3):​389–405. 2013:752195.
​
  5. Bien CG, Raabe AL, Schramm J, et al. Trends in presurgical 25. Erba G, Winston KR, Adler JR, et al. Temporal lobectomy
evaluation and surgical treatment of epilepsy at one centre for complex partial seizures that began in childhood. Surg
from 1988–2009. J Neurol Neurosurg Psychiatry. 2013;​84(1):​ Neurol. 1992;38(6):424–432.
​ ​
54–61. 26. Schmeiser B, Daniel M, Kogias E, et al. Visual field defects
  6. Kral T, Clusmann H, Urbach J, et al. Preoperative evaluation following different resective procedures for mesiotemporal
for epilepsy surgery (Bonn Algorithm). Zentralbl Neurochir. lobe epilepsy. Epilepsy Behav. 2017;76:39–45.
​ ​
2002;​63(3):​106–110. 27. Helmstaedter C, Richter S, Röske S, et al. Differential effects
  7. Wellmer J, von der Groeben F, Klarmann U, et al. Risks and of temporal pole resection with amygdalohippocampectomy
benefits of invasive epilepsy surgery workup with implanted versus selective amygdalohippocampectomy on material-spe-
subdural and depth electrodes. Epilepsia. 2012;​53(8):​1322–1332. cific memory in patients with mesial temporal lobe epilepsy.
  8. Blümcke I, Thom M, Aronica E, et al. International consen- Epilepsia. 2008;49(1):88–97.
​ ​
sus classification of hippocampal sclerosis in temporal lobe 28. Helmstaedter C, Elger CE, Witt JA. The effect of quantitative
epilepsy:​a Task Force report from the ILAE Commission on and qualitative antiepileptic drug changes on cognitive recov-
Diagnostic Methods. Epilepsia. 2013;​54(7):​1315–1329. ery after epilepsy surgery. Seizure. 2016;36:63–69.
​ ​
  9. Louis DN, Perry A, Reifenberger G, et al. The 2016 World 29. Tonini C, Beghi E, Berg AT, et al. Predictors of epilepsy
Health Organization Classification of Tumors of the Central surgery outcome:​a meta-analysis. Epilepsy Res. 2004;62(1):
​ ​
Nervous System:​a summary. Acta Neuropathol. 2016;​131(6):​ 75–87.
803–820. 30. Carne RP, O’Brien TJ, Kilpatrick CJ, et al. MRI-negative
10. Blümcke I, Thom M, Aronica E, et al. The clinicopathologic PET-positive temporal lobe epilepsy:​a distinct surgically
spectrum of focal cortical dysplasias:​a consensus classifica- remediable syndrome. Brain. 2004;127(Pt​ 10):2276–2285.
​
tion proposed by an ad hoc Task Force of the ILAE Diagnos- 31. Téllez-Zenteno JF, Hernández Ronquillo L, Moien-Afshari F,
tic Methods Commission. Epilepsia. 2011;​52(1):​158–174. Wiebe S. Surgical outcomes in lesional and non-lesional epi-
11. Wieser HG, Blume WT, Fish D, et al. ILAE Commission lepsy:​a systematic review and meta-analysis. Epilepsy Res.
Report. Proposal for a new classification of outcome with 2010;89(2-3):310–318.
​ ​
respect to epileptic seizures following epilepsy surgery. Epi- 32. Bien CG, Szinay M, Wagner J, et al. Characteristics and
lepsia. 2001;​42(2):​282–286. surgical outcomes of patients with refractory magnetic reso-
12. Grote A, Witt JA, Surges R, et al. A second chance--reoper- nance imaging-negative epilepsies. Arch Neurol. 2009;66(12):
​ ​
ation in patients with failed surgery for intractable epilepsy:​ 1491–1499.
long-term outcome, neuropsychology and complications. J 33. Ivanovic J, Larsson PG, Østby Y, et al. Seizure outcomes of
Neurol Neurosurg Psychiatry. 2016;​87(4):​379–385. temporal lobe epilepsy surgery in patients with normal MRI
13. Sadek AR, Gray WP. Chopping and changing:​long-term re- and without specific histopathology. Acta Neurochir (Wien).
sults of epilepsy surgery. Lancet. 2011;​378(9800):​1360–1362. 2017;159(5):757–766.
​ ​
14. McIntosh AM, Kalnins RM, Mitchell LA, et al. Temporal lo- 34. Sotero de Menezes MA, Connolly M, Bolanos A, et al. Tem-
bectomy:​long-term seizure outcome, late recurrence and risks poral lobectomy in early childhood:​the need for long-term
for seizure recurrence. Brain. 2004;​127(Pt 9):​2018–2030. follow-up. J Child Neurol. 2001;16(8):585–590.
​ ​
15. Jeha LE, Najm IM, Bingaman WE, et al. Predictors of out- 35. Roberts DW, Jobst BC, Siegel AM, et al. Investigation of
come after temporal lobectomy for the treatment of intrac- extra-temporal epilepsy. Stereotact Funct Neurosurg. 2001;​
table epilepsy. Neurology. 2006;​66(12):​1938–1940. 77(1-4):216–218.
​
16. Schmeiser B, Wagner K, Schulze-Bonhage A, et al. Surgical 36. de Lanerolle NC, Kim JH, Williamson A, et al. A retrospec-
treatment of mesiotemporal lobe epilepsy:​which approach is tive analysis of hippocampal pathology in human temporal
favorable? Neurosurgery. 2017;​81(6):​992–1004. lobe epilepsy:​evidence for distinctive patient subcategories.
17. Hu WH, Zhang C, Zhang K, et al. Selective amygdalohippo- Epilepsia. 2003;44(5):677–687.
​ ​
campectomy versus anterior temporal lobectomy in the man- 37. Blümcke I, Pauli E, Clusmann H, et al. A new clinico-path-
agement of mesial temporal lobe epilepsy:​a meta-analysis of ological classification system for mesial temporal sclerosis.
comparative studies. J Neurosurg. 2013;​119(5):​1089–1097. Acta Neuropathol. 2007;113(3):235–244.
​ ​
18. Josephson CB, Dykeman J, Fiest KM, et al. Systematic re- 38. Thom M, Liagkouras I, Elliot KJ, et al. Reliability of patterns
view and meta-analysis of standard vs selective temporal of hippocampal sclerosis as predictors of postsurgical out-
lobe epilepsy surgery. Neurology. 2013;​80(18):​1669–1676. come. Epilepsia. 2010;51(9):1801–1808.
​ ​
19. Malikova H, Kramska L, Vojtech Z, et al. Different surgical 39. Gales JM, Jehi L, Nowacki A, Prayson RA. The role of his-
approaches for mesial temporal epilepsy:​resection extent, topathologic subtype in the setting of hippocampal sclerosis-
seizure, and neuropsychological outcomes. Stereotact Funct associated mesial temporal lobe epilepsy. Hum Pathol. 2017;​
Neurosurg. 2014;​92(6):​372–380. 63:79–88.
​
20. Wendling AS, Hirsch E, Wisniewski I, et al. Selective amyg- 40. Deleo F, Garbelli R, Milesi G, et al. Short- and long-term
dalohippocampectomy versus standard temporal lobectomy surgical outcomes of temporal lobe epilepsy associated with
in patients with mesial temporal lobe epilepsy and unilateral hippocampal sclerosis:​relationships with neuropathology.
hippocampal sclerosis. Epilepsy Res. 2013;​104(1-2):​94–104. ​ ​
Epilepsia. 2016;57(2):306–315.
21. Schramm J. Temporal lobe epilepsy surgery and the quest for 41. Savitr Sastri BV, Arivazhagan A, Sinha S, et al. Clinico-
optimal extent of resection:​a review. Epilepsia. 2008;​49(8):​ pathological factors influencing surgical outcome in drug
1296–1307. resistant epilepsy secondary to mesial temporal sclerosis. J
22. Yaşargil MG, Krayenbühl N, Roth P, et al. The selective Neurol Sci. 2014;340(1-2):183–190.
​ ​
amygdalohippocampectomy for intractable temporal limbic 42. Hattingen E, Enkirch SJ, Jurcoane A, et al. Hippocampal
seizures. J Neurosurg. 2010;​112(1):​168–185. “gliosis only” on MR imaging represents a distinct entity in
23. Parrent AG, Blume WT. Stereotactic amygdalohippocam- ​ ​
epilepsy patients. Neuroradiology. 2018;60(2):161–168.

768 J Neurosurg Volume 135 • September 2021

 
* | D*2 *37 *
 Borger et al.

43. Bernhardt BC, Fadaie F, Liu M, et al. Temporal lobe epilep- Güresir, Becker, Helmstaedter, Elger, Vatter. Reviewed submitted
sy:​hippocampal pathology modulates connectome topology version of manuscript: Hamed, Taube, Aydin, Ilic, Schneider,
and controllability. Neurology. 2019;​92(19):​e2209–e2220. Schuss, Güresir, Becker, Helmstaedter, Elger, Vatter. Approved
44. Paolicchi JM, Jayakar P, Dean P, et al. Predictors of outcome the final version of the manuscript on behalf of all authors:
in pediatric epilepsy surgery. Neurology. 2000;​54(3):​642– Borger. Statistical analysis: Borger, Taube. Administrative/
647. technical/material support: Helmstaedter, Elger, Vatter.
45. Harroud A, Bouthillier A, Weil AG, Nguyen DK. Temporal
lobe epilepsy surgery failures:​a review. Epilepsy Res Treat. Supplemental Information
2012;​2012:​201651. Previous Presentations
Parts of this work were presented orally at the 69th Annual Meet-
Disclosures ing of the German Society of Neurosurgery in Münster, Germany,
June 3–6, 2018, and at the 13th European Congress on Epileptol-
Dr. Helmstaedter reports receiving honoraria from UCB, Eisai, ogy in Vienna, Austria, August 26–30, 2018.
Precisis, and GW; receiving royalties from EpiTrack and Neu-
roCog FX; and being a consultant to Eisai, UCB, Precisis, and Correspondence
GW. Ms. Taube received support of non–study-related clinical or
research effort from the University Hospital Bonn Department of Valeri Borger: University Hospital Bonn, Germany.
Epileptology (European Reference Network EpiCARE Grant). valeri.borger@​ukbonn.de.

Author Contributions
Conception and design: Borger, Vatter. Acquisition of data:
Borger, Hamed, Taube, Aydin, Ilic, Schneider, Becker. Analysis
and interpretation of data: Borger, Taube, Helmstaedter, Vatter.
Drafting the article: Borger. Critically revising the article: Schuss,

J Neurosurg Volume 135 • September 2021 769

* | D*2 *37 *
 CLINICAL ARTICLE
J Neurosurg 135:770–782, 2021

MRI-guided stereotactic laser corpus callosotomy for

epilepsy: distinct methods and outcomes
Christopher W. Rich, MD,1 Rebecca E. Fasano, MD,2 Faical Isbaine, PhD,3 Amit M. Saindane, MD,4
Deqiang Qiu, PhD,4 Daniel J. Curry, MD,5 Robert E. Gross, MD, PhD,3 and Jon T. Willie, MD, PhD3,6
1
Emory University School of Medicine; Departments of 2Neurology, 3Neurosurgery, and 4Radiology, Emory University School
of Medicine, Atlanta, Georgia; 5Department of Neurosurgery, Texas Children’s Hospital, Houston, Texas; and 6Department of
Neurological Surgery, Washington University School of Medicine, St. Louis, Missouri

OBJECTIVE  Several small series have described stereotactic MRI-guided laser interstitial thermal therapy for partial
callosotomy of astatic and generalized tonic-clonic (GTC) seizures, especially in association with Lennox-Gastaut syn-
drome. Larger case series and comparison of distinct stereotactic methods for stereotactic laser corpus callosotomy
(SLCC), however, are currently lacking. The objective of this study was to report seizure outcomes in a series of adult
patients with epilepsy following anterior, posterior, and complete SLCC procedures and to compare the results achieved
with a frameless stereotactic surgical robot versus direct MRI guidance frames.
METHODS  The authors retrospectively reviewed sequential adult epilepsy surgery patients who underwent SLCC pro-
cedures at a single institution. They describe workflows, stereotactic errors, percentage disconnection, hospitalization
durations, adverse events, and seizure outcomes after performing anterior, posterior, and complete SLCC procedures
using a frameless stereotactic surgical robot versus direct MRI guidance platforms.
RESULTS  Thirteen patients underwent 15 SLCC procedures. The median age at surgery was 29 years (range 20–49
years), the median duration of epilepsy was 21 years (range 9–48 years), and median postablation follow-up was 20
months (range 4–44 months). Ten patients underwent anterior SLCC with a median 73% (range 33%–80%) midsagittal
length of callosum acutely ablated. Following anterior SLCC, 6 of 10 patients achieved meaningful (> 50%) reduction of
target seizures. Four patients underwent posterior (completion) SLCC following prior anterior callosotomy, and 1 patient
underwent complete SLCC as a single procedure; 3 of these 5 patients experienced meaningful reduction of target
seizures. Overall, 8 of 10 patients in whom astatic seizures were targeted and treated by anterior and/or posterior SLCC
experienced meaningful improvement. SLCC procedures with direct MRI guidance (n = 7) versus a frameless surgical
robot (n = 8) yielded median radial accuracies of 1.1 mm (range 0.2–2.0 mm) versus 2.4 mm (range 0.6–6.1 mm; p =
0.0011). The most serious adverse event was a clinically significant intraparenchymal hemorrhage in a patient who un-
derwent the robotic technique.
CONCLUSIONS  This is the largest reported series of SLCC for epilepsy to date. SLCC provides seizure outcomes
comparable to open surgery outcomes reported in the literature. Direct MRI guidance is more accurate, which has the
potential to reduce the risks of SLCC. Methodological advancements and larger studies are needed.
https://thejns.org/doi/abs/10.3171/2020.7.JNS20498
KEYWORDS  stereotactic laser ablation; laser interstitial thermal therapy; MRI; surgical robotics; corpus callosotomy;
Lennox-Gastaut syndrome; generalized epilepsy; astatic seizures; atonic seizures; drop attacks

C orpus callosotomy, which inhibits interhemispher- served for patients with continuation of the targeted sei-
ic spread of seizures, has been practiced in select- zures.4 Callosotomy is an effective treatment of astatic
ed patients with debilitating motor seizures since (including tonic, atonic, and otherwise unspecified “drop
the 1940s.1,2 Anterior callosotomy including the isthmus3 attack”) seizures, which meta-analyses suggest to be su-
is widely considered the first stage of callosotomy, with perior to vagus nerve stimulation (VNS).5 Open anterior
subsequent complete division (including the splenium) re- two-thirds callosotomy and complete callosotomy are ex-

ABBREVIATIONS  DBS = deep brain stimulation; GTC = generalized tonic-clonic; LGS = Lennox-Gastaut syndrome; LOS = length of stay; MRg-LITT = MRI-guided laser
interstitial thermal therapy; ROSA = Robotic Stereotactic Assistant; SGE = symptomatic generalized epilepsy; SLCC = stereotactic laser corpus callosotomy; T2IR = T2
inversion recovery; VNS = vagus nerve stimulation; VP = ventriculoperitoneal.
SUBMITTED  February 18, 2020.  ACCEPTED  July 15, 2020.
INCLUDE WHEN CITING  Published online January 22, 2021; DOI: 10.3171/2020.7.JNS20498.

770 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

* | D*2 *38 *
 Rich et al.

pected to provide greater than 50% reduction in drop at- stereotactic surgical robot (ROSA), we began using it for
tacks in 45% and 90% of patients, respectively, based on cases with multiple trajectories (anterior callosotomies),
systematic review of outcomes in pediatric patients.6–10 expecting to possibly reduce the procedure time. Choice
Open corpus callosotomy is usually performed via of approach was determined by this consideration rather
frontal craniotomy and microscopic interhemispheric than patient selection. In contrast, we performed all pos-
dissection.11 Reported surgical complications include in- terior completion callosotomies using the direct MRI-
tracranial hemorrhage, retraction injury, venous infarcts, guided technique, as there was no anticipated benefit of
bone flap osteomyelitis, subgaleal fluid collection, hydro- the robotic technique for single-trajectory cases. ROSA
cephalus, and sagittal sinus tearing,6,12–14 and hospitaliza- procedures involve imaging coregistration by methods in-
tions and recovery periods can be lengthy. Open posterior cluding either scalp surface reflectance or bone fiducials.
completion callosotomy can yield additional seizure ben- In other ROSA procedures (stereoelectroencephalogra-
efit but carries added risks of vascular injuries and neu- phy), we found that bone fiducials did not benefit accuracy,
rological deficits. In addition, persistent symptomatic dis- so all our ROSA SLCC procedures utilized scalp surface
connection syndrome is more likely in patients following coregistration.
total corpus callosotomy (15.4%) compared with anterior
two-thirds corpus callosotomy (6.9%);6 thus, less invasive MRg-LITT by Direct MRI-Guided Technique
approaches are desirable. Patients underwent endotracheal general anesthesia in
Stereotactic MRI-guided laser interstitial thermal ther- an interventional MRI suite (1.5-T Espree, Siemens, Inc.)
apy (MRg-LITT) is an emerging minimally invasive al- and were positioned supine with shoulder bump and later-
ternative to open epilepsy surgery, especially for focal and al head turn in pins and flanked by flexible radiofrequency
deep lesions.15–20 Stereotactic laser corpus callosotomy receive-only head coils (Fig. 1). After sterile scalp prepa-
(SLCC) based on MRg-LITT was successfully performed ration and draping, adhesive fiducial grids were placed
first by Curry (personal communication, November 2013) over the approximate entry sites at nondominant para-
and further described in a handful of subsequent case re- sagittal frontal and parietal locations. Volumetric imag-
ports and small series.21–28 MRg-LITT procedures can be ing, consisting of T2 inversion recovery (T2IR)–weighted
performed by various stereotactic techniques, but direct and gadolinium-enhanced (postcontrast) 3D T1-weighted
comparison of approaches is lacking. MRI, was obtained and used to plan trajectories with a
In this paper, we report on 13 adult patients who un- stereotactic workstation (ClearPoint Neuro, Inc.).
derwent 15 SLCC procedures, the largest sequential se- For anterior callosotomies, we planned 3–4 trajecto-
ries at a single surgical epilepsy center. These procedures ries targeting the genu, anterior body, and posterior body
included anterior (“two-thirds”) SLCC, posterior comple- with isthmus (Fig. 1). Entries were 2–3 cm off the midline,
tion SLCC after prior open or laser anterior callosotomy, passed through cingulate structures, pierced the callosum
and a complete SLCC performed as a single procedure. within 1 cm from midline, and terminated at or just be-
In addition, we directly compare the stereotactic accura- yond the midline callosum. We planned a margin of 4–6
cies of two approaches: direct MRI guidance (ClearPoint mm in diameter to avoid visible blood vessels. For poste-
Neuro, Inc.) versus stereotactic bolts placed by stereotac- rior completion callosotomies, only the remaining spleni-
tic surgical robot (Robotic Stereotactic Assistant [ROSA], um was targeted, avoiding the fornices and deep cerebral
Zimmer Biomet, Inc.). veins (Fig. 2). Based on an assumed ablation diameter es-
timate of approximately 18 mm, trajectories were planned
Methods to confluently ablate the anteroposterior parasagittal ex-
Study Population tent of the targeted portion of the corpus callosum.
Fiducial grids were used to mark final entry sites on
This retrospective study was approved by the Emory the scalp, and overlying ClearPoint “Scalpmount” base
University IRB. A search of a spreadsheet database of all frames were affixed percutaneously to the skull with lag
adult epilepsy surgery patients treated by two surgeons screws and stabilizing offset pins. Subsequent imaging
(J.T.W. and R.E.G.) was used to identify those who had was iteratively processed by the ClearPoint workstation to
undergone SLCC by either surgeon. Medical records were provide prescriptions to align each platform to the desired
reviewed for age, sex, epilepsy duration, typical seizure trajectory using x, y, pitch, and roll adjustments. For each
types and frequencies, EEG results, treatment history, trajectory, we performed a stereotactic 3.2-mm twist-hole
MRI brain results, seizure types targeted for therapy (e.g., craniostomy with a hand drill via a stab incision, realigned
astatic vs generalized tonic-clonic [GTC] seizures), dura- the frame to the desired trajectory as needed, and punc-
tions of surgery and hospitalization, adverse events, and tured the dura. A ceramic stylet was placed to target depth
seizure outcomes. and its location was verified with imaging. The ceramic
stylet was replaced with the stiffened Visualase cannula,
Surgical Methods and then the optical fiber with a 10-mm diffusing tip was
All MRg-LITT utilized the Visualase Thermal Ther- placed within the cannula and irrigating tubing was at-
apy System (Medtronic, Inc.).18,29 The two distinct work- tached. MR-based thermal imaging was obtained in two
flows utilized for stereotactic techniques in this study are planes along the length of the cannula. Using real-time
illustrated in the Supplemental Figure. The earliest cases thermal imaging and the Visualase workstation, a series
were performed by direct MRI-guided technique (Clear- of confluent ablations 1 cm in length were created through
Point Neuro, Inc.). When we subsequently obtained a the corpus callosum along each trajectory. We strove to

J Neurosurg Volume 135 • September 2021 771

 
*38 *
Rich et al.

FIG. 1. Alternative surgical approaches for anterior two-thirds SLCC. A: Intraoperative photograph of patient 5 undergoing drilling
and placement of bolts with a stereotactic surgical robot (ROSA). B: Patient 5 positioned and draped in a 1.5-T interventional MRI
machine with a stereotactic bolt holding the Visualase laser interstitial thermal therapy system. C: Patient 6 positioned supine with
head turned laterally and pinned to expose nondominant hemisphere for direct MRI guidance. The head is flanked by flexible MRI
head coils. D: Patient’s head prepped and draped with MRI-compatible fiducial grids over intended entry sites. E: Disposable
MRI guidance platforms are mounted overlying desired entry sites with percutaneous screws. A frame is aligned with imag-
ing, a percutaneous twist-drill craniotomy is performed, and the Visualase laser cannula is placed stereotactically to the target.
F–H: Visualase workstation screenshots of parasagittal T1 anatomical views merged with real-time thermal damage maps (orange
pixels) along the genu (F), body (G), and isthmus (H) of the corpus callosum in patient 7. I–K: Sagittal (I), coronal (J), and axial (K)
T1-weighted contrast-enhanced MR images of patient 6 showing the extent of ablation relative to coregistered diffusion imaging
models of fiber pathways (colored fibers emphasized with yellow arrows) seeded from the region of the central sulcus (primary mo-
tor pathways).

772 J Neurosurg Volume 135 • September 2021

*38 *
 Rich et al.

FIG. 2. Posterior completion SLCC of the splenium. An MRI guidance platform (ClearPoint Neuro, Inc.) was used to ablate the re-
maining splenium to complete the callosotomy in a patient with recurrent atonic seizures after an initially successful open anterior
corpus callosotomy. A and B: Parasagittal (A) and paracoronal (B) T1-weighted MR images of laser device placement in the sple-
nium. Note the proximity of the fornix (labeled), the ablation of which should be avoided. Dashed white lines indicate the locations
of the subsequent axial images. C and D: Axial T1-weighted MR images that demonstrate full anterior-posterior ablation (dotted
orange outlines) of the splenium and proximity to the fornix. E: Coronal inversion recovery MR image demonstrates ablation
disconnecting the full superior-inferior extent of the splenium (arrows). F: Pre- and postablation resting-state functional connectiv-
ity MRI results under similar levels of inhaled anesthetic show reduced hemispheric connectivity immediately following posterior
completion callosotomy. The heat map uses the color red to represent the greatest level of connectivity between two locations.

J Neurosurg Volume 135 • September 2021 773

 
*38 *
Rich et al.

avoid thermal injury of cortices outside the callosum (i.e., We separately report these results for 1) anterior SLCC, 2)
cingulum) using low-temperature thermal limits. More posterior SLCC (completion following prior anterior cal-
than one trajectory could typically be performed by sim- losotomy by either open or laser approach), and 3) com-
ply realigning the frame to a new trajectory and drilling a plete SLCC.
new craniostomy. Because a sterile field was maintained, For statistical analysis, two nonparametric tests were
a single laser apparatus was typically reused for multiple used for assessment of significance. Comparison of all
trajectories. FLAIR, diffusion, T2IR, and postcontrast T1- continuous data was completed using the Mann-Whitney
weighted MRI were used to verify confluent ablations. U-test, where sample sizes were more than 5 and p val-
ues were generated following conversion to a z-score. This
MRg-LITT by Frameless Stereotactic Robot-Assisted included comparison of ClearPoint and ROSA methods
Technique with respect to trajectories required, percentage of length
Under endotracheal general anesthesia in the operat- of callosum ablated, and lengths of stay (LOSs). The Fish-
ing room, patients were positioned supine with the head er exact test was used for categorical variables (seizure
in a Radionics CRW (Integra, Inc.) universal head holder outcomes, adverse events, and persistent sequelae) when
frame base, which was then attached using a Mayfield sample sizes were larger than 2.
adapter (Integra, Inc.) to the stereotactic surgical robot A literature review of MRg-LITT of the corpus cal-
(ROSA; Fig. 1). Using prior volumetric CT and gadolini- losum for epilepsy was performed by a PubMed search
um-enhanced MRI, trajectories were planned as described combining the search terms “corpus callosotomy” and
in the previous section, but with ROSA planning software. “laser interstitial thermal therapy.” Seven identified stud-
After reflectance-based coregistration of the skin surface ies were reviewed for preoperative diagnosis, epilepsy
to historical CT and MR imaging, and sterile preparation duration, prior VNS therapy, duration of hospitalization,
and draping, a 3.2-mm twist-hole craniostomy was per- complications, and outcomes as defined by the authors of
formed through a stab incision for each trajectory under each individual study.
stereotactic control of the robotic arm. Durotomies were
completed with a stylet, and titanium Visualase bolts were Results
screwed into each craniostomy over an aligning rod us- Patient Characteristics
ing robotic stereotactic control. The bolts and surround- From 2014 to 2018, 13 adult patients underwent 15
ing sterile field were sealed under nonpermeable adhesive SLCC procedures. Specifically, 10 patients underwent an-
drapes (Tegaderm and Ioban), and the patient was trans- terior SLCC, 1 (case 8) underwent complete SLCC, 2 (cas-
ported to the interventional MRI suite where a sterile field es 2 and 11) underwent posterior SLCC following prior
was recovered by removing the drape to expose a bolt that anterior SLCC, and 2 (cases 1 and 4) underwent posterior
was resterilized with betadine. The stiffened Visualase SLCC following prior remote open anterior corpus cal-
cannula was inserted to depth through a bolt, the laser fi- losotomy. Nine patients carried a diagnosis of symptom-
ber and coolant cannula were assembled, and MRI was atic generalized epilepsy (SGE; including Lennox-Gastaut
performed to verify placement. After each ablation trajec- syndrome [LGS]), and 4 patients were diagnosed as having
tory, with the sterile field maintained, the apparatus was poorly localized focal epilepsy with rapidly generalizing
withdrawn, inspected, and then used through the next bolt. features (Table 1). Astatic seizures were the most common
Thus, for most cases, a single laser apparatus was used to indication, although 3 patients (cases 2, 5, and 11) under-
complete all ablation trajectories. went SLCC to palliate GTC seizures. The median age at
the time of surgery was 29 years (range 20–49 years). The
Data Analysis and Literature Review median duration of epilepsy at time of the procedure was
Radial error, a standard measure of stereotactic accu- 21 years (range 9–48 years).
racy, is the difference (perpendicular distance) between Many patients underwent multiple surgical procedures
the center of the intended trajectory at the target to the (Tables 1 and 2). The majority (all but cases 5, 7, 8, and
center of the actual device placement on subsequent im- 10) underwent prior VNS. Four patients underwent pos-
aging.30 We retrospectively measured radial errors at the terior SLCC after a prior anterior callosotomy. One such
target using T1-weighted MRI from 17 of 25 recoverable patient with LGS (case 1) who had undergone remote open
ROSA trajectories and 14 of 15 recoverable ClearPoint tra- anterior callosotomy also underwent left stereotactic laser
jectories that had remained saved in the respective work- amygdalohippocampotomy (MRg-LITT) for mesial tem-
stations. To assess the extent of callosum ablated in each poral sclerosis, and then 1 year later underwent posterior
procedure, we measured the internal perimeter length of (completion) SLCC for recurrent astatic seizures. One
the callosum from the inferior aspect of the genu to the in- patient (case 5) underwent prior open anterior temporal
ferior aspect of the splenium, and calculated the analogous lobectomy before undergoing posterior SLCC to palliate
length of ablation as a percentage of the total. persistent secondary GTC seizures. After separate ante-
Although callosotomy may benefit multiple seizure rior and posterior SLCC procedures, case 11 subsequently
types,5,27,31 we report outcomes with respect to the tar- underwent bilateral centromedian thalamic nucleus deep
geted astatic and GTC seizures as resolved (no further brain stimulation (DBS). Case 13 underwent craniotomy,
documented seizures of the targeted subtype), meaning- cranioplasty, and ventriculoperitoneal (VP) shunt place-
fully improved (> 50% reduction but not completely re- ment to manage a hemorrhagic complication of SLCC
solved), or not meaningfully improved (< 50% reduction). (described below).

774 J Neurosurg Volume 135 • September 2021

*38 *
 TABLE 1. Preoperative patient characteristics with outcomes, complications, and additional surgical procedures for all 13 patients
Case No.
Variable 1 2.1 2.2 3 4 5 6 7 8 9 10 11.1 11.2 12 13
Epilepsy SGE/ SGE/ SGE/ SGE/LGS SGE/ Focal + SGE/LGS Focal + T-C Focal + T-C SGE/ SGE/ Focal + Focal SGE/LGS SGE/LGS
diagnosis LGS LGS LGS LGS T-C LGS LGS T-C + T-C
Target sz Astatic T-C T-C Astatic Astatic T-C Astatic Astatic Astatic Astatic Astatic T-C T-C Astatic Astatic
type
Age at sur- 20 22 22 44 29 32 47 49 39 22 23 22 22 32 43
gery (yrs)
Epilepsy du- 10 21 21 43 26 19 40 48 37 19 9 20 20 11 38
ration (yrs)
VNS at time Yes Yes Yes Yes Yes No Yes No No Yes No Yes Yes Yes Yes
of surgery
Surgery Post Ant Post Ant SLCC Post Ant Ant SLCC Ant SLCC Comp SLCC Ant Ant Ant Post Ant SLCC Attempted/
SLCC SLCC SLCC SLCC SLCC SLCC SLCC SLCC SLCC aborted ant SLCC
(hemorrhagic dis-
ruption of isthmus,
rt cingulum, pre-
SMA, & SMA)
Technique/ CP/1 CP/3 CP/2 CP/4 CP/1 ROSA/4 CP/3 ROSA/3 ROSA/5 ROSA/3 ROSA/3 ROSA/3 CP/1 ROSA/3 ROSA/1
trajectories
Trajectory 0.3 0.2, 0.2, 1.8, 1.9, 1.1, 0.4 6.1, 5.0, 2, 2, 1.8 0.6, 3.0, 2.3 2.3, 1.8, 2.4, 2.5, UTR 0.8, 0.8, 0.7 0.8, UTR 3.5
radial errors 1.1, 0.9 1.8 2.9, 3.2, 3.0 UTR 2.4
(mm) UTR UTR
FU (mos) 44 4 44 43 20 33 41 24 20 24 20 5 8 6 15

*38 *
Length of 170 513 456 597 273 546 595 381 531 571 441 531 159 702 64
surgery
(mins)
Length of 170 257 228 149 273 137 198 127 106 190 147 177 159 234 64
surgery
(mins)/trajec-
 
tory (no.)
Total hospital 2 1 4 14 9 2 3 8 16 9 21 9 2 3 76
LOS (days)
Complica- None None None SMA syndrome None None None Mild lt- Mild lt hemipare- None Tran- Transient None Transient lt Intraparenchymal
tions (ablation to sided neglect sis secondary to sient lt diffuse leg weakness; hemorrhage w/
cingulate on secondary to ablation of post hemipa- weak- transient intraventricular
dominant side); some nondom- nondominant resis (5 ness incontinence; extension
reintubation inant cingulum cingulate, respi- days) transient

J Neurosurg Volume 135 • September 2021

ablation ratory failure dysarthria
CONTINUED ON PAGE 776 »

775
Rich et al.
 776
Rich et al.

» CONTINUED FROM PAGE 775

TABLE 1. Preoperative patient characteristics with outcomes, complications, and additional surgical procedures for all 13 patients
Case No.
Variable 1 2.1 2.2 3 4 5 6 7 8 9 10 11.1 11.2 12 13
Persistent None None None Dysarthria, Agraph- None None Subjective Dysarthria None None None None None Exacerbation of
sequelae mild rt ia worsened baseline lt hemi-
hemiparesis, executive paresis, abolition
incontinence function & gait of self-injurious

J Neurosurg Volume 135 • September 2021

dysfunction behavior (finger/
nail-biting, hair-
pulling)
Other SLAH, Open Open VP shunt & DBS* Craniotomy
procedures open ant CC ATL rt temporal evacuation of
ant CC craniectomy IPH, cranioplas-
secondary ty, VP shunt*
to herpes
encephalitis
1-year Res NA NA Res Unch Unch Res Impr Impr Impr Unch NA NA NA Res
outcome:
astatic sz
Last Res NA NA Res Impr Unch Unch Impr Impr Impr Unch NA NA Res Res
available
outcome:
astatic sz
1-year Res Unch Unch Unch NA Unch Res Res Impr NA Res NA NA NA Res
outcome:
GTC sz
Last Impr Unch Unch Impr NA Unch Unch Res Impr NA Res Unch Unch Unch Res
available
outcome:

*38 *
GTC sz
Ant = anterior; ATL = anterior temporal lobectomy; CC = corpus callosotomy; Comp = complete; CP = ClearPoint; FU = follow-up; Impr = improved; IPH = intraparenchymal hemorrhage; NA = not applicable; Post = poste-
rior; Res = resolved; SLAH = stereotactic laser amygdalohippocampotomy; SMA = supplementary motor area; sz = seizure; T-C = tonic-clonic; Unch = unchanged; UTR = unable to recover.
* Additional procedure followed SLCC.
 Rich et al.

TABLE 2. Cases grouped by procedure type

Procedure Type
Variable All SLCC Anterior SLCC* Posterior SLCC (completion) Complete SLCC 3/3
No. of cases 15 10 4 1
Median length of procedure (range), mins 513 (64–702) 539 (64–702) 222 (159–456) 531
No. of median trajectories (range) 3 (1–5) 3 (1–4) 1 (1–2) 5
Median follow-up (range), mos 20 (4–44) 22 (4–43) 32 (8–44) 20
Median total hospital LOS (range), days 8 (1–76) 9 (1–76) 3 (3–9) 16
Cases w/ adverse events 7 6 0 1
Cases w/ persistent sequelae 5 3 1 1
Estimated collosum length ablated 78% (33–100%) 73% (33–80%) 100% (98–100%) 100%
Outcome: targeted sz (no.)†
 Resolved 3 3 1 0
 Improved 5 3 1 1
Outcome: all szs (no.)
 Resolved 3 3 0 0
 Improved 2 1 1 1
* Intention to treat.
† Targeted seizure type was either astatic or GTC (detailed in Table 1).

SLCC Procedure Characteristics rological complaints at last follow-up. Specifically, patient

By intention-to-treat analysis, 10 patients underwent 3 was transiently reintubated for respiratory failure but had
anterior SLCC procedures with a median of 3 trajectories persistent mild right hemiparesis, dysarthria, and incon-
(range 1–4 trajectories) completed (Table 1) and a median tinence. Patient 4 reported persistent agraphia following
estimated 73% (range 33%–80%) of the callosal length ab- posterior completion SLCC. Patient 7 reported worsening
lated. One of these cases, an outlier (case 13), had only one executive function and gait ataxia. Patient 8, who under-
trajectory, and the procedure was aborted due to hemor- went complete SLCC in a single surgical session, was tran-
rhage (detailed below), achieving only an estimated 33% of siently reintubated for respiratory failure and experienced
the callosum disconnected (Table 2). The overall median transient hemiparesis followed by persistent dysarthria.
duration of completed anterior SLCC procedures was 546 Patient 13 experienced a series of complications as a
minutes (range 381–702 minutes), excluding the aborted result of attempted anterior SLCC. Following placement
case (case 13; Table 2). Anterior SLCC procedures were of a laser device using ROSA, MRI revealed a 3.5-mm
associated with a median acute hospital LOS of 9 days device deviation from the intended trajectory and a right
(range 1–76 days). frontal hemorrhage in the dorsal cingulate region with
When three complete callosotomies were performed in intraventricular extension (Fig. 3). The procedure was
either one or two stages, a total of 4–5 trajectories were aborted without any ablation, and the patient underwent
used. For posterior SLCC procedures, a median of 1 tra-
jectory (range 1–2 trajectories) was used to complete the
callosotomy in all cases. The median duration of poste-
rior completion SLCC procedures was 222 minutes (range
159–456 minutes). Posterior completion SLCC after a pri-
or anterior procedure was associated with a median LOS
of 3 days (range 3–9 days). The only complete SLCC per-
formed in a single surgical session (case 8) had a duration
of 531 minutes, and this procedure was associated with a
16-day LOS.

Adverse Events
Adverse events were found to be neurological or car-
diorespiratory in nature and were categorized as transient FIG. 3. Hemorrhagic complication of attempted SLCC. A and B: Axial
or persistent. We observed no cranial infections or wound (A) and sagittal (B) MR images from patient 13 in which initial placement
of the laser cannula via a stereotactic robotic approach was associated
complications. Neurological symptoms such as weakness, with deflection and intraparenchymal hemorrhage with intraventricular
neglect, abulia, or dysarthria were common but typically extension. Following recovery, imaging revealed injury of the isthmus of
transient (Table 1) and may be attributed to acute discon- the corpus callosum (B) as well as dorsal cingulum and medial frontal
nection.23,32 Seven of 15 SLCC procedures were associated lobe white matter. The patient experienced exacerbation of baseline left
with adverse events. Five patients exhibited persistent neu- hemiparesis in addition to complete seizure control.

J Neurosurg Volume 135 • September 2021 777

 
*38 *
Rich et al.

TABLE 3. Comparison of SLCC techniques

Length of Total No. of No. of Median
SLCC No. of Procedure Length of Procedure/ Hospital LOS Adverse Persistent No. of Radial Error
Method Procedures (range), mins Trajectory (range), mins (range), days Events* Sequelae Trajectories (range), mm*
CP
 Ant 3 595 (513–597) 198 (149–257) 3 (1–14) 1 2 10 1.8 (0.2–2)
 Post 4 222 (159–456) 222 (159–456) 3 (2–9) 0 1 5 0.4 (0.2–0.9)
 Total 7 456 (159–597) 198 (149–273) 2.5 (1–14) 1 3 15 (14 retrieved) 1.1 (0.2–2.0)
ROSA
 Ant 6 539 (381–702) 162 (127–234) 8.5 (3–21) 4 1 19 2.4 (0.6–6.1)
 Comp 1 531 106.2 16 1 1 5 2.4 (1.8–3.2)
 Aborted 1 64 64 76 1 1 1 3.5
 Total 8 531 (381–702) 147 (106–234) 9 (2–76) 6 3 25 (17 retrieved) 2.4 (0.6–6.1)
Statistical analysis was completed using two nonparametric tests for assessment of significance. All continuous data were analyzed using the Mann-Whitney U-test,
where sample sizes were > 5 and p values were generated following conversion to a z-score. The Fisher exact test was used for categorical variables (seizure out-
comes, adverse events, and persistent sequelae) where sample sizes were > 2. No significance was found in variables outside of those annotated.
* Significant difference favoring ClearPoint found using p < 0.05.

decompressive craniotomy to evacuate the hemorrhage. each stereotactic approach (7 by ClearPoint and 8 by
During the same prolonged hospital admission, she under- ROSA) precluded meaningful statistical comparisons
went tracheostomy, cranioplasty, and VP shunt placement. of procedure duration, hospital LOS, complications, and
The patient suffered persistent exacerbation of baseline seizure outcomes (Table 3). However, the large number
left hemiparesis, persistent injury of the right frontal lobe of trajectories performed in these cases provided the op-
white matter, and division of the isthmus of the corpus cal- portunity for meaningful comparison of stereotactic ac-
losum, which could explain subsequent resolution of clini- curacy with each method. When compared on the basis of
cal seizure activity. available trajectory data, 14 ClearPoint trajectories had a
Notably, despite the expectation that posterior calloso- median radial error of 1.1 mm (range 0.2–1.9 mm), while
tomy might increase the incidence of disconnection syn- 17 ROSA trajectories had a larger median radial error of
drome, only 1 of 4 posterior SLCC procedures (completion 2.4 mm, with a considerably wider range of errors (range
with ClearPoint) yielded a new neurological complaint. 0.7–6.1 mm; p = 0.0011). Greater errors with ROSA sug-
gest that it is a less accurate approach for performing cal-
Seizure Outcomes losotomy compared with ClearPoint (Table 3, Fig. 4). No-
Outcomes are categorized by targeted seizure type tably, the most severe complication, an intraparenchymal
and procedure (Table 2). The median duration of follow- hemorrhage in case 13, was associated with an inaccurate
up was 20 months (range 4–44 months) for SLCC. Of 13 ROSA trajectory.
patients who underwent SLCC procedures, 10 suffered
preoperatively from primarily astatic seizures, and all but
2 had GTCs as well. At the last available follow-up, 6 of
10 patients who underwent anterior SLCC experienced an
improvement of more than 50% with respect to targeted
seizures, and 4 of these patients exhibited complete resolu-
tion. Of the 7 patients with astatic seizures who underwent
anterior SLCC, 5 of 7 were improved, with 3 of these ex-
periencing complete seizure resolution. Notably, all 7 had
improvement in at least one seizure type, with none show-
ing exacerbation.
Of 5 patients who underwent posterior SLCC for per-
sistent seizures or upfront complete SLCC, 3 experienced
more than a 50% meaningful benefit of targeted seizures, 1
of whom experienced resolution. All 3 patients with astatic
seizures who underwent completion were improved, with
1 resolved. Improvement or resolution was achieved in 8
of 10 patients with primarily astatic seizures treated by
anterior and/or posterior SLCC. FIG. 4. Comparison analysis of trajectory radial errors obtained using
direct MRI guidance platforms (ClearPoint) versus a stereotactic surgi-
cal robot (ROSA) for SLCC. The Mann-Whitney U-test was performed
ClearPoint Versus ROSA and found a significant reduction in radial error when using ClearPoint
In general, the small numbers of cases performed by compared with ROSA (p = 0.0011).

778 J Neurosurg Volume 135 • September 2021

*38 *
 TABLE 4. Literature review with reported outcomes and complications for SLCC
Median Median
Median Age Epilepsy Median Length of Median
Authors No. of at Surgery Duration VNS Follow-Up Surgery LOS Complications/Adverse Reported
& Year Cases (range), yrs (range), yrs Diagnosis Therapy (mos) (mins) (days) Events Outcomes
Ho et al., Post (splenium) 30 29 NR No 4 NR NR None Resolution of drop attacks,* >50% reduction
201622 SLCC (n = 1) in all szs
Ball et al., Ant SLCC 21 NR Atonic sz not Yes 5 NR NR Transient gait difficulty Resolution of drop attacks, decreased overall
201821 (n = 1) otherwise (n = 1) sz frequency
specified
Karsy et Ant SLCC 17 17 LGS No 12 NR NR None Significant reduction in sz frequency &
al., 201824 (n = 1) severity
Lehner et Ant SLCC 28.2 (21–44) 19.2 (11–28) 3 focal 3 yes, 24 (n = 4), NR NR Inaccurate fiber placement 100% reduction of szs (n = 3), 83% (n = 1),
al., 201925 (n = 5) epilepsy, 1 2 no 1 (n = 1) requiring repeat pro- 0% (n = 1), 80% resolution of drop attacks
LGS, 1 NR cedure/transient SMA (4/5)
syndrome (n = 1), 8-mm
hemorrhage (n = 1)
Palma et Ant SLCC 12 (1–23) NR 2 focal 1 yes, 39 (33–48) NR NR None Complete resolution of drop attacks in all 3
al., 201826 (n = 1), post epilepsy, 2 no pts w/ reduction in overall sz frequency
SLCC (n = 2) 1 NR
Tao et al., Ant SLCC 32.5 (25–40) 26.5 (23–30) 2 LGS 2 yes 12.5 (7–18) NR NR Transient hypersomnia Freedom of disabling szs & drop attacks (n =
201927 (n = 2) (n = 1) 1), 90% reduction of disabling szs (n = 1)
Huang Post (splenium) 24 (8–40) 19.5 (8–34) 2 focal 2 yes, 17 (12–34) NR 2 (1–8) Disconnection syndrome Adult pts: EC I (n = 2, targeted sz = atonic),
et al., SLCC (n = 4), epilepsy, 4 no (n = 1), asymptomatic EC II (n = 1, targeted sz = GTC); pediatric
201923† post (splenium, 3 LGS, 1 extension of ablation pts: EC III (n = 1, targeted sz = atonic), EC
rostrum, genu) LGS & focal into post thalamus IV (n = 2, 1 GTC & 1 atonic)

*38 *
SLCC (n = 2) epilepsy (n = 1)
Tao et al., Ant SLCC 33 (11–52) 26 (10–49) 10 LGS 9 yes, 19 (6–40) NR NR Asymptomatic intracere- Ant SLCC:
202028 (n = 10), post 1 no bral hemorrhage 100% sz freedom (2/10), 80–100% (4/10),
SLCC (n = 3), 13 (n = 1), significant 50–80% (2/10); 100% reduction in drop
cases in 10 pts aggression (n = 1) attacks (5/10), 80–100% (3/10)
Post SLCC:
 
100% sz freedom (1/3), 80–100% (2/3); 100%
reduction in drop attacks (1/2)
Present 15 (13 pts) 29 (20–49) 21 (9–48) 10 LGS, 3 9 yes, 20 (4–44) Ant Ant Adverse events 46.7% 1-yr outcome (atonic szs): resolved 4/10,
study focal 4 no SLCC SLCC (7/15), permanent impr 3/10, unch 3/10; outcome (atonic
epilepsy 546, post 9, post sequelae 33% (5/15; szs): resolved 3/11, impr 5/11, unch 3/11;
SLCC SLCC details in Table 1)† 1-yr outcome (GTC szs): resolved 5/10,
222 3 impr 1/10, unch 4/10; outcome (GTC szs):
resolved 5/13, impr 1/13, unch 7/13

J Neurosurg Volume 135 • September 2021

EC = Engel class; NR = not reported.
* Authors reported as atonic seizures.
† All patients received prior partial corpus callosotomy.

779
Rich et al.
Rich et al.

Discussion SLCC may contribute to increased morbidity relative to

This is the largest reported series of SLCC for epilepsy other procedures with fewer trajectories. Despite adverse
to date. We report the outcomes of SLCC and comparative events, anterior SLCC was associated with an acute hos-
accuracies of two distinct stereotactic approaches and their pitalization LOS shorter than what we generally observe
associated workflows. The limitations of this study include with open callosotomy in adults (R.E.G., unpublished ob-
a single-center retrospective case-series design, small sam- servations). While SLCC can be associated with accept-
ple size, and reliance upon patient reporting of seizures. able morbidity, there is room for improvement.
In 1940, Van Wagenen and Herren2 observed a de- Compared with ROSA, the ClearPoint single-site work-
crease in seizures after damage to the callosum, including flow provided superior stereotactic accuracy. The robot-
patients with partial or complete callosal division. Ante- ic approach may be associated with an accumulation of
rior two-thirds callosotomy may be most effective when stereotactic errors from coregistration procedures (using
transection includes the callosal isthmus,3,7 which may re- scalp surface laser reflectance in our cases), deflections of
late to the variable but relatively posterior location of hu- drilling at nonorthogonal angles, and stereotactic bolt in-
man decussating motor fibers compared with nonhuman accuracy.43 Small superficial deflections can greatly ampli-
primates.33–35 Neuropsychological studies of patients with fy over long trajectories. Indeed, the serious hemorrhage
callosotomies in the 1970s determined that sectioning of we observed was associated with a 3.5-mm device deflec-
the anterior callosum through the isthmus but sparing the tion following the robotic technique. Workflows involv-
splenium resulted in fewer deficits than with complete cal- ing device placement in the operating room followed by
losal sectioning.36–38 Thus, anterior callosotomy sparing subsequent CT or MR imaging may only allow adequate
the splenium is generally the preferred initial approach, assessment of accuracy after a device is placed into the
with posterior callosotomy being reserved for patients brain. If trajectories or ablations prove problematic once
with persistent seizures or for patients with relatively poor the patient has been transferred to the MRI suite, the phy-
baseline functional status. sician must consider returning to the operating room or
Extensive experience with open callosotomy indicates aborting aspects of the procedure.
that astatic seizures are the most improved: worthwhile In contrast, use of direct MRI-guided stereotaxy and
(> 50%) reduction in astatic seizures has been reported a single-site workflow provides several advantages (Sup-
in 45% and 90% of patients following anterior and total plemental Figure). Direct MRI guidance with ClearPoint
corpus callosotomy, respectively.6–10 Likewise, recent case allows real-time recognition and mitigation of potential
reports and series using MRg-LITT for callosotomy sug- inaccuracy at any point before, during, and after placing
gest 70%–100% reductions in astatic seizures (Table 4).21– a device into the brain. Relatedly, ClearPoint mitigates in-
28
By comparison, we observed incremental seizure ben- accuracies from nonorthogonal drilling, because the laser
efits from anterior and posterior SLCC procedures in the device is guided and secured by a platform overlying a
majority of patients, and 80% of patients with primarily craniostomy and the trajectory does not directly depend
astatic seizures treated by anterior and/or posterior SLCC upon the accuracy of the craniostomy or bolt direction. Fi-
experienced meaningful benefit or resolution (Tables 2 nally, ClearPoint mitigates interdependence of anticipated
and 4). We also observed meaningful reductions in focal trajectories. At any stage in the ablation procedure, sub-
and GTC seizures in individual patients, consistent with sequent trajectory plans can take into account the results
some open callosotomy reports.39–42 of prior actions. In one case (patient 6), the entire anterior
SLCC procedures were associated with higher rates of SLCC was performed from a series of trajectories through
adverse events, particularly transient respiratory failure, one craniostomy site because each subsequent trajectory
than have generally been reported for open callosotomy in could be planned after directly observing what had been
children. Notably, we have observed similar rates of tran- accomplished by prior actions. The predominant disad-
sient respiratory failure in our prior limited institutional vantages of the MRI-based procedure are the prolonged
experience of open anterior corpus callosotomy in adults use of a diagnostic scanner and limited availability of in-
(R.E.G., unpublished observations). It is unclear whether terventional/intraoperative MRI suites that are located in a
these findings reflect differences in reporting or whether limited number of surgical epilepsy centers.
callosotomy can be more morbid in adults than in children.
During SLCC, transgression or indirect collateral injury of Conclusions
dorsal cingulum and/or motor fibers may impact networks MRI-guided SLCC is a potentially safe and effective
associated with arousal, motivation, and/or motor control. alternative to traditional open corpus callosotomy. In the
Increased attention to protecting such pathways via alter- future, automated trajectory planning algorithms may
native trajectories might decrease the morbidity of the la- provide safer, faster, and more consistent trajectory place-
ser approach. It seems prudent to plan trajectories through ment.44,45 Use of advanced imaging such as tractography
the nondominant hemisphere to minimize risks to domi- could facilitate patient-specific disconnection procedures.
nant motor and/or language networks. We also observed Partial callosotomies, particularly of the isthmus, might
one severe hemorrhagic complication during an aborted benefit patients with astatic seizures while further reduc-
anterior SLCC. This is the most clinically significant hem- ing the risks of adverse events and cognitive deficits.
orrhage we have observed in a cumulative experience of
more than 300 stereotactic MRg-LITT cases (J.T.W. and
R.E.G., unpublished observations). The relatively large References
number of stereotactic trajectories undertaken in anterior   1. Mathews MS, Linskey ME, Binder DK. William P. van Wa-

780 J Neurosurg Volume 135 • September 2021

* | D*2 *38 *
 Rich et al.

genen and the first corpus callosotomies for epilepsy. J Neu- 22. Ho AL, Miller KJ, Cartmell S, et al. Stereotactic laser abla-
rosurg. 2008;​108(3):​608–613. tion of the splenium for intractable epilepsy. Epilepsy Behav
  2. Van Wagenen W, Herren RY. Surgical division of commis- Case Rep. 2016;​5:​23–26.
sural pathways in the corpus callosum:​relation to spread of 23. Huang Y, Yecies D, Bruckert L, et al. Stereotactic laser abla-
an epileptic attack. Arch NeurPsych. 1940;​44(4):​740–759. tion for completion corpus callosotomy. J Neurosurg Pediatr.
  3. Huck FR, Radvany J, Avila JO, et al. Anterior callosotomy in 2019;​24(4):​433–441.
epileptics with multiform seizures and bilateral synchronous 24. Karsy M, Patel DM, Halvorson K, et al. Anterior two-thirds
spike and wave EEG pattern. Acta Neurochir Suppl (Wien). corpus callosotomy via stereotactic laser ablation. Neurosurg
1980;​30:​127–135. Focus. 2018;​44(VideoSuppl2):​V2.
  4. Spencer SS, Spencer DD, Sass K, et al. Anterior, total, and 25. Lehner KR, Yeagle EM, Argyelan M, et al. Validation of
two-stage corpus callosum section:​differential and incre- corpus callosotomy after laser interstitial thermal therapy:​a
mental seizure responses. Epilepsia. 1993;​34(3):​561–567. multimodal approach. J Neurosurg. 2019;​131(4):​1095–1105.
  5. Lancman G, Virk M, Shao H, et al. Vagus nerve stimulation 26. Palma AE, Wicks RT, Popli G, Couture DE. Corpus calloso-
vs. corpus callosotomy in the treatment of Lennox-Gastaut tomy via laser interstitial thermal therapy:​a case series. J
syndrome:​a meta-analysis. Seizure. 2013;​22(1):​3 –8. Neurosurg Pediatr. 2018;​23(3):​303–307.
  6. Graham D, Tisdall MM, Gill D. Corpus callosotomy out- 27. Tao JX, Issa NP, Wu S, et al. Interstitial stereotactic laser
comes in pediatric patients:​a systematic review. Epilepsia. anterior corpus callosotomy:​a report of 2 cases with opera-
2016;​57(7):​1053–1068. tive technique and effectiveness. Neurosurgery. 2019;​85(3):​
  7. Pinard JM, Delalande O, Chiron C, et al. Callosotomy for epi- E569–E574.
lepsy after West syndrome. Epilepsia. 1999;​40(12):​1727–1734. 28. Tao JX, Satzer D, Issa NP, et al. Stereotactic laser anterior
  8. Rahimi SY, Park YD, Witcher MR, et al. Corpus callosotomy corpus callosotomy for Lennox-Gastaut syndrome. Epilepsia.
for treatment of pediatric epilepsy in the modern era. Pediatr 2020;​61(6):​1190–1200.
Neurosurg. 2007;​43(3):​202–208. 29. Jethwa PR, Barrese JC, Gowda A, et al. Magnetic resonance
  9. Shimizu H. Our experience with pediatric epilepsy surgery thermometry-guided laser-induced thermal therapy for intra-
focusing on corpus callosotomy and hemispherotomy. Epi- cranial neoplasms:​initial experience. Neurosurgery. 2012;​
lepsia. 2005;​46(suppl 1):​30–31. 71(1)(Suppl Operative):​133–145.
10. Turanli G, Yalnizoğlu D, Genç-Açikgöz D, et al. Outcome 30. Burchiel KJ, McCartney S, Lee A, Raslan AM. Accuracy of
and long term follow-up after corpus callosotomy in child- deep brain stimulation electrode placement using intraopera-
hood onset intractable epilepsy. Childs Nerv Syst. 2006;​ tive computed tomography without microelectrode recording.
22(10):​1322–1327. J Neurosurg. 2013;​119(2):​301–306.
11. Wilson DH, Reeves A, Gazzaniga M. Division of the corpus 31. Malmgren K, Rydenhag B, Hallböök T. Reappraisal of cor-
callosum for uncontrollable epilepsy. Neurology. 1978;​28(7):​ pus callosotomy. Curr Opin Neurol. 2015;​28(2):​175–181.
649–653. 32. Bower RS, Wirrell E, Nwojo M, et al. Seizure outcomes after
12. Jalilian L, Limbrick DD, Steger-May K, et al. Complete corpus callosotomy for drop attacks. Neurosurgery. 2013;​
versus anterior two-thirds corpus callosotomy in children:​ 73(6):​993–1000.
analysis of outcome. J Neurosurg Pediatr. 2010;​6(3):​257–266. 33. Cipolloni PB, Pandya DN. Topography and trajectories of
13. Kasasbeh AS, Smyth MD, Steger-May K, et al. Outcomes commissural fibers of the superior temporal region in the
after anterior or complete corpus callosotomy in children. rhesus monkey. Exp Brain Res. 1985;​57(2):​381–389.
Neurosurgery. 2014;​74(1):​17–28. 34. Hofer S, Frahm J. Topography of the human corpus callosum
14. Stigsdotter-Broman L, Olsson I, Flink R, et al. Long-term fol- revisited—comprehensive fiber tractography using diffusion
low-up after callosotomy—a prospective, population based, tensor magnetic resonance imaging. Neuroimage. 2006;​32(3):​
observational study. Epilepsia. 2014;​55(2):​316–321. 989–994.
15. Ellis JA, Mejia Munne JC, Wang SH, et al. Staged laser inter- 35. Zarei M, Johansen-Berg H, Smith S, et al. Functional anat-
stitial thermal therapy and topectomy for complete oblitera- omy of interhemispheric cortical connections in the human
tion of complex focal cortical dysplasias. J Clin Neurosci. brain. J Anat. 2006;​209(3):​311–320.
2016;​31:​224–228. 36. Gazzaniga MS, Risse GL, Springer SP, et al. Psychologic and
16. McCracken DJ, Willie JT, Fernald BA, et al. Magnetic reso- neurologic consequences of partial and complete cerebral
nance thermometry-guided stereotactic laser ablation of cav- commissurotomy. Neurology. 1975;​25(1):​10–15.
ernous malformations in drug-resistant epilepsy:​imaging and 37. Gordon HW, Bogen JE, Sperry RW. Absence of deconnexion
clinical results. Oper Neurosurg (Hagerstown). 2016;​12(1):​ syndrome in two patients with partial section of the neocom-
39–48. missures. Brain. 1971;​94(2):​327–336.
17. Thompson SA, Kalamangalam GP, Tandon N. Intracranial 38. Risse GL, Gates J, Lund G, et al. Interhemispheric transfer
evaluation and laser ablation for epilepsy with periventricular in patients with incomplete section of the corpus callosum.
nodular heterotopia. Seizure. 2016;​41:​211–216. Anatomic verification with magnetic resonance imaging.
18. Willie JT, Laxpati NG, Drane DL, et al. Real-time magnetic Arch Neurol. 1989;​46(4):​437–443.
resonance-guided stereotactic laser amygdalohippocampot- 39. Oguni H, Olivier A, Andermann F, Comair J. Anterior cal-
omy for mesial temporal lobe epilepsy. Neurosurgery. 2014;​ losotomy in the treatment of medically intractable epilepsies:​
74(6):​569–585. a study of 43 patients with a mean follow-up of 39 months.
19. Willie JT, Malcolm JG, Stern MA, et al. Safety and effective- Ann Neurol. 1991;​30(3):​357–364.
ness of stereotactic laser ablation for epileptogenic cerebral 40. Rathore C, Abraham M, Rao RM, et al. Outcome after
cavernous malformations. Epilepsia. 2019;​60(2):​220–232. corpus callosotomy in children with injurious drop attacks
20. Boerwinkle VL, Foldes ST, Torrisi SJ, et al. Subcentimeter and severe mental retardation. Brain Dev. 2007;​29(9):​577–
epilepsy surgery targets by resting state functional magnetic 585.
resonance imaging can improve outcomes in hypothalamic 41. Reutens DC, Bye AM, Hopkins IJ, et al. Corpus callosotomy
hamartoma. Epilepsia. 2018;​59(12):​2284–2295. for intractable epilepsy:​seizure outcome and prognostic fac-
21. Ball T, Sharma M, White AC, Neimat JS. Anterior corpus tors. Epilepsia. 1993;​34(5):​904–909.
callosotomy using laser interstitial thermal therapy for re- 42. Sunaga S, Shimizu H, Sugano H. Long-term follow-up of
fractory epilepsy. Stereotact Funct Neurosurg. 2018;​96(6):​ seizure outcomes after corpus callosotomy. Seizure. 2009;​
406–411. 18(2):​124–128.

J Neurosurg Volume 135 • September 2021 781

 
*38 *
Rich et al.

43. Brandmeir NJ, Savaliya S, Rohatgi P, Sather M. The compar- Gross. Analysis and interpretation of data: Willie, Rich, Fasano,
ative accuracy of the ROSA stereotactic robot across a wide Isbaine, Saindane, Qiu, Gross. Drafting the article: Willie, Rich,
range of clinical applications and registration techniques. J Fasano. Critically revising the article: Willie, Rich, Fasano,
Robot Surg. 2018;​12(1):​157–163. Saindane, Qiu, Curry, Gross. Reviewed submitted version of man-
44. Vakharia VN, Sparks R, Li K, et al. Automated trajectory uscript: all authors. Approved the final version of the manuscript
planning for laser interstitial thermal therapy in mesial tem- on behalf of all authors: Willie. Statistical analysis: Rich, Isbaine,
poral lobe epilepsy. Epilepsia. 2018;​59(4):​814–824. Qiu. Administrative/technical/material support: Willie, Rich,
45. Vakharia VN, Sparks RE, Vos SB, et al. Computer-assisted Gross. Study supervision: Willie, Rich, Gross.
planning for minimally invasive anterior two-thirds laser
corpus callosotomy:​a feasibility study with probabilistic Supplemental Information
tractography validation. Neuroimage Clin. 2020;​25:​102174. Online-Only Content
Supplemental material is available with the online version of the
Disclosures article.
Supplemental Figure. https://thejns.org/doi/suppl/​10.3171/​
Dr. Curry reports being a consultant to Medtronic, Inc. Dr. Gross 2020.7.JNS20498.
reports being a consultant to Medtronic, and receiving support of
non–study-related clinical or research effort from Medtronic. Dr. Correspondence
Willie reports being a consultant to Medtronic, ClearPoint Neuro,
NeuroPace, AIMM Therapeutics, and NICO Medical, and receiv- Jon T. Willie: Washington University School of Medicine and
ing support of non–study-related clinical or research effort from Barnes-Jewish Hospital, St. Louis, MO. [email protected].
Medtronic, NeuroPace, and NICO Medical.

Author Contributions
Conception and design: Willie, Rich, Fasano, Curry, Gross.
Acquisition of data: Willie, Rich, Fasano, Isbaine, Saindane, Qiu,

782 J Neurosurg Volume 135 • September 2021

*38 *
 CLINICAL ARTICLE
J Neurosurg 135:783–791, 2021

Outcomes following surgical management of vagus

nerve stimulator–related infection: a retrospective
multi-institutional study
Hirotaka Hasegawa, MD, PhD,1,2 Jamie J. Van Gompel, MD,1 W. Richard Marsh, MD,1
Robert E. Wharen Jr., MD,3 Richard S. Zimmerman, MD,4 David B. Burkholder, MD,5
Brian N. Lundstrom, MD, PhD,5 Jeffrey W. Britton, MD,5 and Fredric B. Meyer, MD1

Departments of 1Neurologic Surgery and 5Neurology, Mayo Clinic, Rochester, Minnesota; 2Department of Neurosurgery,
The University of Tokyo Hospital, Tokyo, Japan; 3Department of Neurologic Surgery, Mayo Clinic, Jacksonville, Florida; and
4
Department of Neurologic Surgery, Mayo Clinic, Phoenix, Arizona

OBJECTIVE  Surgical site infection (SSI) is a rare but significant complication after vagus nerve stimulator (VNS) place-
ment. Treatment options range from antibiotic therapy alone to hardware removal. The optimal therapeutic strategy re-
mains open to debate. Therefore, the authors conducted this retrospective multicenter analysis to provide insight into the
optimal management of VNS-related SSI (VNS-SSI).
METHODS  Under institutional review board approval and utilizing an institutional database with 641 patients who had
undergone 808 VNS-related placement surgeries and 31 patients who had undergone VNS-related hardware removal
surgeries, the authors retrospectively analyzed VNS-SSI.
RESULTS  Sixteen cases of VNS-SSI were identified; 12 of them had undergone the original VNS placement procedure
at the authors’ institutions. Thus, the incidence of VNS-SSI was calculated as 1.5%. The mean (± standard deviation)
time from the most recent VNS-related surgeries to infection was 42 (± 27) days. Methicillin-sensitive staphylococcus
was the usual causative bacteria (58%). Initial treatments included antibiotics with or without nonsurgical procedures (n =
6), nonremoval open surgeries for irrigation (n = 3), generator removal (n = 3), and total or near-total removal of hardware
(n = 4). Although 2 patients were successfully treated with antibiotics alone or combined with generator removal, removal
of both the generator and leads was eventually required in 14 patients. Mild swallowing difficulties and hoarseness oc-
curred in 2 patients with eventual resolution.
CONCLUSIONS  Removal of the VNS including electrode leads combined with antibiotic administration is the definitive
treatment but has a risk of causing dysphagia. If the surgeon finds dense scarring around the vagus nerve, the prudent
approach is to snip the electrode close to the nerve as opposed to attempting to unwind the lead completely.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201385
KEYWORDS  epilepsy; hardware infection; surgical complication; vagus nerve stimulator

T
he incidence of epilepsy from all etiologies is ap- procedures for medically intractable epilepsy, having been
proximately 1% worldwide.1 Approximately a third performed in more than 100,000 patients as of 2018. This
of these cases are medically refractory and may number will continue to grow given that vagus nerve
require surgical intervention.1,2 Among the variety of stimulation is now available for treatment-resistant depres-
epilepsy surgeries, vagus nerve stimulation is an effective sion.1,13–16
treatment.3–5 Overall, it has been reported that 35%–50% Although placement of a VNS is considered minimally
of patients have > 50% seizure reduction within a couple invasive, the possible complications are variable, including
of months to a year from vagus nerve stimulator (VNS) infection, vocal cord damage, laryngopharyngeal dysfunc-
implantation, with a minority becoming seizure free.3,6–12 tion, postoperative hematoma, carotid artery injury, asep-
Since its approval in 1997, VNS placement has become tic reaction, cable discomfort, generator displacement, and
available worldwide and is now one of the most common hardware-related complications such as lead fracture, mal-

ABBREVIATIONS  GR = generator removal; MSS = methicillin-sensitive staphylococcus; NTR = near-total removal; SSI = surgical site infection; TR = total removal; VNS =
vagus nerve stimulator; VNS-SSI = VNS-related SSI.
SUBMITTED  April 20, 2020.  ACCEPTED  July 14, 2020.
INCLUDE WHEN CITING  Published online December 18, 2020; DOI: 10.3171/2020.7.JNS201385.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  783

*38 *
Hasegawa et al.

function, or disconnection.1,9,17 In addition, VNS-related present or if any parts of foreign bodies remained in place.
surgical site infection (VNS-SSI) has been reported at a A combination of broad-spectrum antibiotics was initially
frequency of 2.6%–8%3,9,​17–20 and often prompts hardware used and then narrowed depending on the causative mi-
removal in light of the long-established principle of surgi- croorganisms and their sensitivities.
cal treatment for foreign body infection. Indeed, in many
previous reports, hardware removal for VNS-SSI was Data Analyses
dogmatically performed and was usually very effective in Data on patients, therapeutic interventions, surgical
terms of infection control.3,21–25 However, while antibiotic morbidities of hardware removal, infection outcomes,
therapy may not always be effective, surgical removal of the and postinfection epilepsy management were collected
cervical electrodes carries some risk, specifically potential and studied. Surgical records in all the patients were thor-
injury to the vagus nerve. Thus, a few studies have posed a oughly reviewed. Therapeutic interventions for VNS-SSI
question regarding the necessity of hardware removal;26–29 were categorized as follows, according to their invasive-
however, the data and case numbers are quite limited. The ness: 1) antibiotics with or without percutaneous proce-
purpose of this retrospective multicenter analysis, includ- dures (aspiration, etc.), 2) nonremoval open procedures
ing three affiliate tertiary care centers, was to provide in- (debridement and open irrigation), 3) GR, and 4) NTR or
sight into the optimum management of VNS-SSI. TR. Interventions 1–3 were considered as lead-preserva-
tion measures.
Methods
Patient Selection Results
The study was approved by the Mayo Clinic Institu- Baseline Characteristics
tional Review Board; informed consent was not required A total of 16 patients were found to have VNS-SSI and
on the basis of the minimal risk associated with the ret- thus were included in the study. Detailed baseline charac-
rospective study design. From a database including cases teristics of the patients are summarized in Table 1. Briefly,
from three of our institutions from 1997 to 2019, we iden- there were 9 males and 7 females. The mean (± standard
tified 641 patients with 808 VNS-related implantation sur- deviation) age at VNS placement was 28 (± 11) years and
geries (596 VNS placements/replacements and 212 gen- the mean BMI was 26 (± 6) kg/m2. Only 1 patient had
erator replacements) and 31 patients with 36 VNS-related diabetes mellitus, and 4 patients (25%) were obese as
hardware removal surgeries (Fig. 1). Among these cases, defined by a BMI > 30 kg/m2. No patient had additional
we identified patients who had developed VNS-SSI and brain hardware including shunts or any other clear infec-
included them in this study. The diagnosis of VNS-SSI tion risk factors. The diagnoses leading to intractable epi-
was made if a patient had undergone VNS-related proce- lepsy included Lennox-Gastaut syndrome (n = 2), cortical
dures and developed one or more signs of SSI, such as dysplasia (n = 1), Rasmussen’s encephalitis (n = 1), dysem-
purulence, positive culture, and superficial signs including bryoplastic neuroepithelial tumor (n = 1), mesial temporal
wound dehiscence, swelling, drainage, and redness. The sclerosis (n = 1), microcephaly (n = 1), and other refrac-
presence of redness alone was not necessarily considered tory seizure without a specified underlining condition (n
to be infection, as this could be caused by a reaction to = 9). All the initial VNS placements were done on the left
stitches/staples or dressings. side. The VNS implantation procedures were complicated
The data included patients who had undergone multiple in 1 patient (case 8) because of an intraoperative pneu-
failed attempts to control VNS-SSI at outside hospitals mothorax in the contralateral side; this patient had severe
and were then referred to one of the study sites. In general, chronic global static encephalopathy secondary to micro-
if there was uncertainty about an infection, observation cephaly. The right tension pneumothorax was discovered
with oral antibiotics was first recommended. Upon confir- immediately after the surgery, as his oxygen level dropped
mation of VNS-SSI, hardware removal was generally rec- significantly. Decompression was performed in the opera-
ommended, though the actual type of surgery and extent tive suite with chest tube placement. Since the affected
of removal were at the discretion of attending physicians, side was contralateral, the pneumothorax was thought to
based on a patient’s clinical course and condition. Genera- be unrelated to the VNS insertion procedure. Another pa-
tor removal (GR) alone was performed in several cases in tient (case 3), without any sign of fever, tenderness, drain-
which the infection was limited to the local chest incision age, redness, or any other sign suggestive of infection 2
and did not track up to the cervical region. When there weeks after surgery, developed a postoperative subacute
was evidence of a cervical infection, attempts were made subcutaneous hematoma in the chest, which eventually
to unwrap the electrode lead from the vagus nerve (total required needle aspiration 3 weeks following surgery.
removal [TR]). In cases with extensive scarring, small This case demonstrated a hematoma without purulence;
parts of the spiral electrodes were left in situ to minimize nevertheless, Staphylococcus aureus grew from the cul-
vagus nerve injury (near-total removal [NTR]). Periop- ture at that time, which was thought to be contamination.
erative intravenous antibiotics were held until specimens By then, the patient had been without any signs of infec-
were collected. Postoperative antibiotics were individual- tion for more than 1 month. However, 2 months following
ized through discussion with the infectious disease team surgery, the patient was found to have wound dehiscence
and thus varied from person to person. Generally, intra- and thus underwent hardware removal. Both patients had
venous antibiotics were preferred if fever with or without demonstrated self-mutilating behavior and thus tended to
any other signs or symptoms suggesting bacteremia was scratch at their incisions.

784 J Neurosurg Volume 135 • September 2021

 
*38 *
 Hasegawa et al.

FIG. 1. A flowchart of case accumulation in the present study.

Establishment of VNS-SSI required TR. Overall, TR was performed in 12 patients;

Among the 16 cases of VNS-SSI, 11 occurred after VNS-SSI was eventually cleared in all of them, including
initial VNS placement and 5 occurred after generator re- 1 patient who had developed a large pectoral abscess and
placement, with 2 of the former and 2 of the latter having needed revision debridement and irrigation because of a
undergone the original surgery leading to infection at an recurrent chest pocket infection.
outside hospital (Fig. 1). Therefore, the incidence of infec- The success rate of each treatment was as follows: 17%
tion for surgeries performed at our institutions was 1.5% (1/6) for antibiotics with or without percutaneous proce-
(12/808). Among the institutional cases, the incidence of dures, 0% (0/4) for debridement and irrigation, 20% (1/5)
VNS-SSI did not differ between whole system placement for GR, 67% (2/3) for NTR, and 100% (12/12) for TR. Sat-
and generator replacement (1.5% [9/596] and 1.4% [3/212], isfactory infection control was finally achieved in all the
respectively, p = 0.99, Fisher’s exact test). The mean time patients. The postoperative complications included mild
to VNS-SSI was 42 (± 27) days from the most recent VNS- hoarseness and difficulty/discomfort with swallowing in 2
related surgery and 681 (± 1360) days from initial place- patients who had undergone TR (2/12 [17%]), which was
ment. Wound dehiscence, redness, swelling, and drainage fully cleared in both patients within a year. Intraopera-
were the typical symptoms of VNS-SSI; occasionally, tive laceration of the jugular vein occurred in 1 patient in
wound breakdown with exposure of hardware was the whom TR had been attempted, which was successfully
presenting symptom (Table 1). The data on causative bac- controlled without any neurological or nonneurological
teria were available in 12 patients; methicillin-sensitive sequelae. This patient ended up going no further than
staphylococcus (MSS) was identified in 7 patients (58%), NTR. Otherwise, no perioperative complication or death
methicillin-resistant S. aureus in 1 (8%), and unspecified was observed.
gram-positive coccus in 1 (8%). In 1 patient with MSS,
Enterobacter and Pseudomonas were also identified. Cul- Epilepsy Management After VNS Infection
ture was negative in 2 patients (17%). The intensity and frequency of seizures were unchanged
after clinical confirmation of VNS-SSI as long as the VNSs
Actual Management for VNS-SSI remained on, though the observation period was too short
The therapeutic interventions against VNS-SSI are in 4 patients, as their VNSs were removed either partially
summarized in Table 2. The mean follow-up period from or totally right after the VNS-SSI was evident. Also, 1 pa-
the procedure/surgery for VNS-SSI to the last clinical visit tient developed VNS-SSI before the system was activated.
was 80 (± 52) months. Five neurosurgeons were involved One patient without hardware removal remained on
in the treatment of VNS-SSI; lead-preservation measures vagus nerve stimulation; in the other patients, seizure con-
were initially taken in 12 patients, including antibiotics trol went back to their baseline level once the VNS was
with or without percutaneous procedures in 6, debridement removed. After the removal, 13 patients were managed
and irrigation in 3, and GR in 3, whereas direct TR or NTR medically, whereas repeat VNS placement was performed
was initially performed in 4 patients. The lead-preserva- in 2 patients in whom stimulation had been quite effec-
tion measures were successful in 1 patient exclusively after tive, following an extensive discussion with the infectious
20 days of oral antibiotics in whom the chest incision had disease team. The times from surgical removal to the re-
revealed a 2-cm superficial opening with some erythema peat placement were 25 and 185 days, respectively. With
without any fluid collection (the deeper layer seemed to be support from head and neck surgeons, the placement was
uncontaminated) and in 1 patient after GR with a subse- successfully done on the left side (infected side) without
quent 4 weeks of intravenous antibiotic therapy; all of the complication in all the patients. An additional 7- to 10-day
other 14 patients required TR or NTR. Overall, NTR was course of oral cephalexin was given on top of the standard
performed in 3 patients, and VNS-SSI was cleared in 2 perioperative intravenous antibiotics. During the follow-
of them; the third patient (case 12) had recurrent infection up period of 124 and 167 months thereafter, those patients
despite a subsequent 8 weeks of antibiotic therapy and thus were free from subsequent infection.

J Neurosurg Volume 135 • September 2021 785

 
*38 *
Hasegawa et al.

TABLE 1. Characteristics of patients with VNS-related infections

Age Days From Most Infection
Case (yrs),* BMI PSx for Complication at Infection Origin of Recent VNS-Related After Initial
No. Sex DM Smoking (kg/m2)* Epilepsy Initial Implantation Symptoms Infection Surgery to Infection Implantation
1 50, F Yes Yes 38 Temporal — Wound dehiscence, Chest 27 Yes
lobectomy redness, drainage
2 26, F No Yes 20 Subdural — Wound dehiscence, Neck 51 Yes
electrodes redness, drainage
3 40, M No No 26 None Postop subcutaneous Wound dehiscence, Neck 79 Yes
hematoma 3 wks later redness, drainage
4 17, M No No 26 Temporal — Redness, wound Chest 18 No
lobectomy dehiscence, exposed
generator
5 31, F No No 35 None —† Wound dehiscence Chest 92 Yes
6 23, F No No 19 ND — Redness, swelling, Chest 17 Yes
drainage
7 13, F No No 20 Subdural — Drainage from both cra- Neck 34 Yes
electrodes nial & neck incisions
8 37, M No No 18 ND Pneumothorax Wound dehiscence, Neck 76 Yes
exposed tie-down
9 35, M No No 29 SEEG — Wound dehiscence, Chest 10 No
redness, swelling,
drainage
10 22, M No Yes 20 Temporal — Redness, swelling Neck 80 Yes
lobectomy
11 15, F No No 33 None — Wound dehiscence, Chest 14 Yes
redness, drainage
12 43, M No Yes 31 RNS, MST — Swelling Chest 31 Yes
13 18, M No No 23 None ND† ND ND 33 No
14 33, F No No ND None ND† Pain, discomfort, drainage Chest 23 Yes
15 30, M No No 25 None ND† Redness, swelling Chest 34 No
16 15, M No No 21 None — Wound dehiscence, Chest 36 No
exposed generator
DM = diabetes mellitus; MST = multiple subpial transection; ND = no data; PSx = previous surgery; RNS = responsive neurostimulation; SEEG = stereoelectroencephalography.
* At the time of VNS placement.
† Done at an outside hospital.

Discussion hiscence, redness, swelling, fluid collection, and drainage.

To date, several reports have described VNS-related Infection can arise from both the chest and neck incision;
infection; most of them are case reports or small case se- our data showed a mild inclination toward a chest origin
ries or briefly describe VNS-SSI as part of comprehen- (67% vs 33%), and in fact, some studies have shown a
sive studies including 10 or fewer patients (Table 3).17,20,​ similar trend.20,26,27 However, this finding may be simply
26–32
In the present study, we analyzed in detail 16 patients attributable to the greater number of interventions per-
with VNS-SSI according to our large-volume database, formed at the site of chest incision given the need for gen-
which included 808 VNS hardware placement surgeries, erator replacement. In our cohort, 1 patient (case 3) devel-
describing the clinical characteristics of VNS-SSI and oped a postoperative hematoma and another patient (case
focusing on the outcomes of both nonsurgical and surgi- 8) developed pneumothorax. The hematoma required a
cal interventions as well as their complications. Overall, needle aspiration; thus, there is a possibility that contami-
the incidence of VNS-SSI in the present study was 1.5%, nation could have occurred as a result of that intervention.
which is slightly less than previous reports but within vari- Regarding the contralateral pneumothorax, there was no
ance.3,17–19,​26,​27,​31,32 If VNS-SSI was appropriately managed, convincing evidence indicating an association with infec-
it did not result in severe neurological sequelae or death tion. Notably, both patients had exhibited self-mutilating
in our cohort. VNS-SSI mostly occurred within 3 months behavior and tended to scratch their incision, which may
from the initial placement or generator replacement; spe- have been associated with the infection.
cial attention should be provided if a patient develops any Given our findings, we summarized an optimal treat-
signs of infection within this time period, including de- ment algorithm for VNS-SSI (Fig. 2). Regarding the ini-

786 J Neurosurg Volume 135 • September 2021

 
*38 *
 TABLE 2. Therapeutic courses and outcomes of hardware removal
FU Mos Days From
Case Causative Microscope/ Infection From VNS Replacement Removal to
No. Microorganism Loupes* Therapeutic Interventions† Control Surgical Complications Removal VNS, Side Replacement VNS
1 ND Microscope Abx (PO Augmentin × 10 days)→TR & Abx (PO cefalexin × 1 wk) Good None 130 Yes, lt 185
2 MSS Microscope D&I & Abx (PO, ND, 5 wks)→TR & Abx (PO cefalexin × 2 wks) Good None 77 Med tx
3 MSS Microscope TR & Abx (PO cefalexin × 2 wks) Good None 183 Med tx
4 GPC Loupes GR & reimplantation & Abx (IV, short course)→TR & Abx (PO Good None 58 Med tx
cefalexin × 2 wks)
5 MSS Loupes Abx (PO levofloxacin→clindamycin→cefdinir × 1 mo)→D&I & Good None 4 Med tx
Abx (PO, ND)→GR & Abx (PO clindamycin × 10 days)→NTR
& Abx (IV ceftriaxone × 2 wks & PO TMP-SMX × 1 yr)
6 MRSA Loupes GR & Abx (IV vancomycin & cefepime × 4 days)→TR & Abx (IV Good None 60 Med tx
vancomycin × 4 wks)
7 MSS, Loupes Abx (PO Augmentin × 2 wks)→NTR & Abx (IV meropenem × 8 Good None 74 Med tx
Enterobacter, wks)
Pseudomonas
8 MSS Microscope D&I & Abx (PO Bactrim × 4 yrs)→TR & Abx (PO TMP-SMX × 1 mo) Good None 61 Med tx
9 MSS Microscope GR & Abx (PO TMP-SMX × 3 wks)→TR & Abx (PO TMP-SMX Good Mild hoarseness, mild 119 Med tx
× 2 wks) swallowing difficulty
10 MSS Microscope Asp & Abx (PO, cefalexin × 2 wks→Bactrim × 4 Good None 168 Yes, lt 25
days→levofloxacin & Augmentin × 2 wks→cefalexin &
TMP-SMX × 1 mo)→TR & Abx (PO cefalexin × 10 days)
11 ND Abx (PO clindamycin × 10 days→PO cefalexin × 10 days) Good No surgery performed 86‡ Continue on

*38 *
VNS
12 MSS Loupes NTR & Abx (IV nafcillin × 3 wks→PO cefalexin × 5 wks)→TR & Good Mild hoarseness persisted > 120 Med tx
Abx (IV nafcillin × 2 wks→PO cefalexin × 2 wks) 1 yr, discomfort w/ swallowing
13 ND Loupes Multiple D&I & Abx (ND)→TR & Abx (PO cephalexin × 2 wks) Good None (vocal code 21 Med tx
compromised before surgery)
14 Negative culture Loupes TR & Abx (PO ciprofloxacin × 10 days) Reinfected, None 3 Med tx
 
treated w/
D&I & Abx
15 ND Loupes Abx (PO augmentin × 2 wks)→GR & Abx (IV, vancomycin × 2 Good None 44 Med tx
wks→daptomycin × 2 wks)
16 Negative culture Microscope TR & Abx (PO cephalexin × 10 days) Good None 69 Med tx
Abx = antibiotics; Asp = aspiration; D&I = debridement and irrigation; FU = follow-up; GPC = gram-positive coccus; IV = intravenous; Med tx = medical treatment; MRSA = methicillin-resistant Staphylococcus aureus; PO =
per os; TMP-SMX = trimethoprim/sulfamethoxazole.
* At the time of VNS removal.

J Neurosurg Volume 135 • September 2021

† Perioperative short-course antibiotics were not included.
‡ Months from confirmation of infection (surgical removal not needed).

787
Hasegawa et al.
Hasegawa et al.

TABLE 3. Literature review of studies describing VNS infection

Time From Attempts for Lead-Preservation No.*

Recent [no. requiring TR/NTR]
VNS-Related Abx w/ or w/o GR (directly Directly Final FU After
Authors No. of Incidence of Procedures Nonremoval or after Proceeded Extent of Surgical Interventions
& Year Cases Infection to Infection Procedures failed Abx) by TR/NTR Removal Morbidities for Infection
Révész et al., 10 0.8% (1990–1999), 99 days† 10 [8] 5 [2] 3 None 2, None NA
201617 3.3% (2000–2014) GR 3, TR 5
Wozniak et al., 6 2.9% 1–8 wks 4 [1] 2 GR 3, TR 3 NA 10 mos, 5.5
201126 yrs, 7.5 yrs
after GR
Elliott et al., 9 NA NA NA NA NA None 2, None NA
201130 TR 7
Air et al., 200927 10 5.2% 29.8 days 7 [5] 3 None 2, None NA
TR/NTR 8
Liechty et al., 1 NA NA 1‡ [0] GR 1 None NA
200629
Patel & 3 NA 9–37 days 2 [2] 1 [1] 1 TR 3 NA NA
Edwards, 200420
Smyth et al., 6 7.1% <4 wks 3 [0] 3 None 3, NA NA
200332 TR 3
Ortler et al., 1 NA 5 wks 1 [1] TR 1 None NA
200128
Murphy et al., 1 4.2% 3 mos 1 [1] TR 1 NA 38 days
199831
Present study 16 1.5% 42 days 9 [8] 5 [4] 4 None 1, Mild swallowing 80 mos
GR 1, TR/ problem w/ hoarse-
NTR 14 ness, 2/12 TR
NA = not assessed.
* There may be overlap between “Abx with or without nonremoval procedures” and “GR.”
† Time from implantation to surgery due to infection, wound revision, or explantation.
‡ Battery replacement with concomitant use of sump antibiotic irrigation system.

tial treatment, some studies have argued that VNS-SSI the extent of removal, GR alone appears to have a low suc-
can be managed solely with antibiotics.17,19,​27,​30,32 Indeed, cess rate and should be cautiously considered perhaps with
in our cohort, infection was cleared with antibiotics alone an extended trial of antibiotics. Some authors have sug-
in only 1 patient, suggesting that empirical antibiotic gested lead-preserving procedures in conjunction with an
therapy may prevent propagation of infection if it stays intensive antibiotic regimen and repeat local irrigation.26–29
superficial and local. Also, despite a low ultimate rate However, these recommendations are based on case re-
of success with antibiotic treatment alone, attempting ports or small case series and thus require further research
antibiotic treatment before escalating the intervention to to more firmly establish their efficacy and safety.
hardware removal did not appear to be associated with Complete hardware removal including the electrode is
an increase in complications. Therefore, antibiotic mono- the definitive treatment, as confirmed in this report. On
therapy would be a reasonable prehospital option when the other hand, the incidence of transient vocal cord/swal-
infection seems superficial and localized at one incision lowing problems was 17% after TR and 0% after NTR.
site. In such cases, however, longer observation should In fact, removal/replacement attributable to noninfectious
be mandated, as infection can recur even > 1 year after reasons has been reported to be as high as 33%.9,33–37 For-
the initiation of therapy.27 Because most VNS-SSIs are tunately, these problems were cleared within a year in
caused by pan-sensitive MSS,26,27 there are multiple op- our patients. Given our experience, we recommend the
tions for an appropriate antibiotic choice. Since we have surgeon attempt complete VNS removal, but if scarring
not aggressively attempted nonsurgical management, in- around the vagus nerve is prohibitive, it is prudent and
tensive antibiotic therapies may show a better success rate safer to snip the electrode close to the nerve, irrigate copi-
to save hardware than our results. This is worth evaluation ously with antibiotic-containing saline, and then resort to
in future studies. antibiotics as advised by the infectious disease team. This
If infection extends to VNS hardware, removal of that approach has worked well and limits the risk of signifi-
hardware is recommended, as infection, in fact, persisted cant nerve injury. We should also reiterate the importance
in all cases that had nonremoval management. Regarding of pursuing the standard management of foreign body in-
788 J Neurosurg Volume 135 • September 2021
 
*38 *
 Hasegawa et al.

FIG. 2. A proposed treatment algorithm for VNS-SSI is shown. As long as the infection is limited to the superficial layer and does
not reach the hardware, it would be plausible to manage it medically with oral (PO) antibiotics (Abx). Physicians may perform
needle aspiration to determine whether purulence is present. If the patient develops any signs of deep infection, such as significant
continuous drainage, purulence, fever, exposed hardware, or any other signs suggestive of bacteremia, hardware removal should
be considered. Generator removal (GR) could be considered if only the chest incision is involved; however, it should be explained
to the patient that the ultimate probability of cure is not high and that the patient will need a prolonged course of intravenous (IV)
Abx followed by oral therapy. Total removal (TR) is recommended in cases of deep neck infection; however, near-total removal
(NTR) with snipping of the leads as close to the vagus nerve as possible can be considered if there is concern about the possibility
of extensive scarring around the nerve. Similar to GR, NTR would require a postoperative intensive Abx regimen. Given our expe-
rience, debridement and irrigation may not be recommended because of the extremely high rate of failure.

fection on top of hardware removal, including sufficient be done in a patient after NTR (in whom the vagus nerve
debridement, copious irrigation, minimizing dead space leads would remain in place). However, close follow-up
by placing a drain and applying appropriate compression, and cardiac monitoring would be desirable for possible
and the perioperative use of appropriate antibiotics with cardiac side effects, and preoperative assessment of vocal
sufficient bacterial coverage. In cases of recurrent infec- cord function is important to avoid the risk of bilateral
tion after TR, meticulous re-debridement and irrigation vocal cord palsy. In theory, the battery could be placed in
would be recommended. the right side while the leads are placed in the left side if
The feasibility of repeat VNS placement in the set- right-sided battery placement is deemed inappropriate. In
ting of a preceding infection is yet to be determined. Our summary, the timing and side of placement should be de-
small number of cases suggests that repeat placement termined after due consideration of the above-mentioned
can be safely done once infection is completely cleared. points.
Therefore, it can be considered for those in whom VNS is
This study has several limitations. First, although it re-
known to be highly effective, especially if their epilepsy
control worsens following VNS removal. Though we did ports the largest number of VNS-SSI cases with over 800
not observe vocal cord palsy, its risk in VNS revision is VNS-related surgery cases examined, VNS-SSI was un-
reported to be slightly higher than that in a first implanta- common, occurring in only 16 patients. Second, there may
tion.9 Regarding the timing, repeat VNS placement was have been patients who were treated at an outside hospital,
successfully performed at around 1 month from hardware which could have resulted in an underestimation of the in-
removal in 1 patient in the present cohort, suggesting that cidence of VNS-SSI; however, that number is likely very
early placement may be possible if TR is done and fol- small, as the infection rate in our study was similar to pre-
lowed by appropriate postoperative antibiotic therapy; viously published rates. Third, the most important man-
however, this number is too small to draw any conclu- agement of VNS-SSI is primary prevention. Therefore,
sions. Right-sided placement is a possible alternative,38–40 determining those at greatest risk would be important and
and indeed, it may be necessary if reimplantation needs to should be the subject of future research.
J Neurosurg Volume 135 • September 2021 789
 
*38 *
Hasegawa et al.

Conclusions 18. Ramsay RE, Uthman BM, Augustinsson LE, et al. Vagus
nerve stimulation for treatment of partial seizures:​2. Safety,
VNS-SSI is a rare but significant complication of VNS side effects, and tolerability. Epilepsia. 1994;35(3):627–636.​ ​
implantation. Our data demonstrate that in cases other than 19. Ben-Menachem E. Vagus nerve stimulation, side effects, and
a localized superficial site infection, it is best to surgically long-term safety. J Clin Neurophysiol. 2001;18(5):415–418. ​ ​
remove as much of the VNS apparatus as safety permits. If 20. Patel NC, Edwards MS. Vagal nerve stimulator pocket infec-
there is dense scarring at the vagus nerve electrode wrap, ​
tions. Pediatr Infect Dis J. 2004;23(7):681–683. ​
then it is prudent to leave the spiral electrodes in situ and 21. Patwardhan RV, Stong B, Bebin EM, et al. Efficacy of vagal
nerve stimulation in children with medically refractory epi-
treat aggressively with antibiotics. lepsy. Neurosurgery. 2000;47(6):1353–1358.
​ ​
22. Yu C, Ramgopal S, Libenson M, et al. Outcomes of vagal
References nerve stimulation in a pediatric population:​a single center
experience. Seizure. 2014;23(2):105–111.
​ ​
  1. Wheless JW, Gienapp AJ, Ryvlin P. Vagus nerve stimulation
23. Kabir SM, Rajaraman C, Rittey C, et al. Vagus nerve stimu-
(VNS) therapy update. Epilepsy Behav. 2018;​88S:​2–10.
lation in children with intractable epilepsy:​indications,
  2. Spencer S, Huh L. Outcomes of epilepsy surgery in adults
complications and outcome. Childs Nerv Syst. 2009;25(9): ​ ​
and children. Lancet Neurol. 2008;​7(6):​525–537.
  3. DeGiorgio CM, Schachter SC, Handforth A, et al. 1097–1100.
Prospective long-term study of vagus nerve stimulation for 24. Galbarriatu L, Pomposo I, Aurrecoechea J, et al. Vagus nerve
the treatment of refractory seizures. Epilepsia. 2000;​41(9):​ stimulation therapy for treatment-resistant epilepsy:​a 15-year
1195–1200. experience at a single institution. Clin Neurol Neurosurg.
  4. Usami K, Kawai K, Sonoo M, Saito N. Scalp-recorded evoked 2015;137:89–93.
​ ​
potentials as a marker for afferent nerve impulse in clinical 25. Fernandez L, Gedela S, Tamber M, Sogawa Y. Vagus nerve
vagus nerve stimulation. Brain Stimul. 2013;​6(4):​615–623. stimulation in children less than 3 years with medically in-
  5. Amar AP, Heck CN, Levy ML, et al. An institutional experi- tractable epilepsy. Epilepsy Res. 2015;112:37–42. ​ ​
ence with cervical vagus nerve trunk stimulation for medi- 26. Wozniak SE, Thompson EM, Selden NR. Vagal nerve stimu-
cally refractory epilepsy:​rationale, technique, and outcome. lator infection:​a lead-salvage protocol. J Neurosurg Pediatr.
Neurosurgery. 1998;​43(6):​1265–1280. ​ ​
2011;7(6):671–675.
  6. Englot DJ, Rolston JD, Wright CW, et al. Rates and predic- 27. Air EL, Ghomri YM, Tyagi R, et al. Management of vagal
tors of seizure freedom with vagus nerve stimulation for nerve stimulator infections:​do they need to be removed? J
intractable epilepsy. Neurosurgery. 2016;​79(3):​345–353. Neurosurg Pediatr. 2009;3(1):73–78.
​ ​
  7. Elliott RE, Morsi A, Tanweer O, et al. Efficacy of vagus 28. Ortler M, Luef G, Kofler A, et al. Deep wound infection after
nerve stimulation over time:​review of 65 consecutive pa- vagus nerve stimulator implantation:​treatment without re-
tients with treatment-resistant epilepsy treated with VNS > moval of the device. Epilepsia. 2001;42(1):133–135.
​ ​
10 years. Epilepsy Behav. 2011;​20(3):​478–483. 29. Liechty PG, Tubbs RS, Blount JP. The use of a sump antibi-
  8. Connor DE Jr, Nixon M, Nanda A, Guthikonda B. Vagal otic irrigation system to save infected hardware in a patient
nerve stimulation for the treatment of medically refractory with a vagal nerve stimulator:​technical note. Surg Neurol.
epilepsy:​a review of the current literature. Neurosurg Focus. 2006;65(1):48–50.
​ ​
2012;​32(3):​E12. 30. Elliott RE, Morsi A, Kalhorn SP, et al. Vagus nerve stimu-
  9. Giordano F, Zicca A, Barba C, et al. Vagus nerve stimulation:​ lation in 436 consecutive patients with treatment-resistant
surgical technique of implantation and revision and related epilepsy:​long-term outcomes and predictors of response.
morbidity. Epilepsia. 2017;​58(suppl 1):​85–90. ​
Epilepsy Behav. 2011;20(1):57–63. ​
10. Kawai K, Tanaka T, Baba H, et al. Outcome of vagus nerve 31. Murphy JV, Hornig GW, Schallert GS, Tilton CL. Adverse
stimulation for drug-resistant epilepsy:​the first three years of events in children receiving intermittent left vagal nerve
a prospective Japanese registry. Epileptic Disord. 2017;​19(3):​ stimulation. Pediatr Neurol. 1998;19(1):42–44.
​ ​
327–338. 32. Smyth MD, Tubbs RS, Bebin EM, et al. Complications of
11. García-Pallero MA, García-Navarrete E, Torres CV, et al. chronic vagus nerve stimulation for epilepsy in children. J
Effectiveness of vagal nerve stimulation in medication- ​ ​
Neurosurg. 2003;99(3):500–503.
resistant epilepsy. Comparison between patients with and 33. Rychlicki F, Zamponi N, Cesaroni E, et al. Complications of
without medication changes. Acta Neurochir (Wien). 2017;​ vagal nerve stimulation for epilepsy in children. Neurosurg
159(1):​131–136. Rev. 2006;29(2):103–107.
​ ​
12. Kawai K, Shimizu H, Maehara T, Murakami H. Outcome of 34. Aalbers MW, Rijkers K, Klinkenberg S, et al. Vagus nerve
long-term vagus nerve stimulation for intractable epilepsy. stimulation lead removal or replacement:​surgical technique,
Neurol Med Chir (Tokyo). 2002;​42(11):​481–490. institutional experience, and literature overview. Acta Neuro-
13. Conway CR, Sheline YI, Chibnall JT, et al. Cerebral blood ​ ​
chir (Wien). 2015;157(11):1917–1924.
flow changes during vagus nerve stimulation for depression. 35. Ortler M, Unterhofer C, Dobesberger J, et al. Complete re-
Psychiatry Res. 2006;​146(2):​179–184. moval of vagus nerve stimulator generator and electrodes. J
14. Conway CR, Sheline YI, Chibnall JT, et al. Brain blood-flow ​
Neurosurg Pediatr. 2010;5(2):191–194. ​
change with acute vagus nerve stimulation in treatment-refrac- 36. Vassilyadi M, Strawsburg RH. Delayed onset of vocal cord
tory major depressive disorder. Brain Stimul. 2012;​5(2):​163–171. paralysis after explantation of a vagus nerve stimulator in a
15. Zobel A, Joe A, Freymann N, et al. Changes in regional ce- ​
child. Childs Nerv Syst. 2003;19(4):261–263. ​
rebral blood flow by therapeutic vagus nerve stimulation in 37. Couch JD, Gilman AM, Doyle WK. Long-term expectations
depression:​an exploratory approach. Psychiatry Res. 2005;​ of vagus nerve stimulation:​a look at battery replacement and
139(3):​165–179. revision surgery. Neurosurgery. 2016;78(1):42–46.​ ​
16. Conway CR, Xiong W. The mechanism of action of vagus 38. McGregor A, Wheless J, Baumgartner J, Bettis D. Right-
nerve stimulation in treatment-resistant depression:​current con- sided vagus nerve stimulation as a treatment for refractory
ceptualizations. Psychiatr Clin North Am. 2018;​41(3):​395–407. epilepsy in humans. Epilepsia. 2005;46(1):91–96.​ ​
17. Révész D, Rydenhag B, Ben-Menachem E. Complications 39. Navas M, Navarrete EG, Pascual JM, et al. Treatment of
and safety of vagus nerve stimulation:​25 years of experience refractory epilepsy in adult patients with right-sided vagus
at a single center. J Neurosurg Pediatr. 2016;​18(1):​97–104. nerve stimulation. Epilepsy Res. 2010;90(1-2):1–7. ​ ​

790 J Neurosurg Volume 135 • September 2021

 
*38 *
 Hasegawa et al.

40. Spuck S, Nowak G, Renneberg A, et al. Right-sided vagus Gompel, Marsh, Wharen, Zimmerman, Burkholder, Lundstrom,
nerve stimulation in humans:​an effective therapy? Epilepsy Britton. Approved the final version of the manuscript on behalf of
Res. 2008;​82(2-3):​232–234. all authors: Meyer. Statistical analysis: Hasegawa. Study supervi-
sion: Meyer, Van Gompel.

Disclosures Correspondence
B.N.L. is a named inventor for intellectual property licensed Fredric B. Meyer: Mayo Clinic, Rochester, MN. meyer.fredric@
to Cadence Neuroscience Inc., which is co-owned by the Mayo mayo.edu.
Clinic. B.N.L. has waived contractual rights to royalties. B.N.L. is
an investigator for the Medtronic Deep Brain Stimulation Therapy
for Epilepsy Post-Approval Study (EPAS). B.N.L. is an investiga-
tor for Mayo Clinic Medtronic NIH Public Private Partnership
(UH3-NS95495). 

Author Contributions
Conception and design: Hasegawa. Acquisition of data: Hasegawa.
Analysis and interpretation of data: Hasegawa. Drafting the arti-
cle: Hasegawa. Critically revising the article: Meyer, Van Gompel,
Marsh, Wharen. Reviewed submitted version of manuscript: Van

J Neurosurg Volume 135 • September 2021 791

 
*38 *
 CLINICAL ARTICLE
J Neurosurg 135:792–798, 2021

MR-guided focused ultrasound pallidotomy for

Parkinson’s disease: safety and feasibility
Howard M. Eisenberg, MD,1 Vibhor Krishna, MD, SM,2 W. Jeffrey Elias, MD,3
G. Rees Cosgrove, MD,4 Dheeraj Gandhi, MD,5 Charlene E. Aldrich, RN, MSN,1 and
Paul S. Fishman, MD, PhD6

Departments of 1Neurosurgery, 5Diagnostic Radiology and Nuclear Medicine, and 6Neurology, University of Maryland School
of Medicine, Baltimore, Maryland; 2Department of Neurosurgery, Ohio State University Medical Center, Columbus, Ohio;
3
Department of Neurosurgery, University of Virginia Health Sciences Center, Charlottesville, Virginia; and 4Department of
Neurosurgery, Brigham and Women’s Hospital, Boston, Massachusetts

OBJECTIVE  Stereotactic radiofrequency pallidotomy has demonstrated improvement in motor fluctuations in patients
with Parkinson’s disease (PD), particularly levodopa (L-dopa)–induced dyskinesias. The authors aimed to determine
whether or not unilateral pallidotomy with MR-guided focused ultrasound (MRgFUS) could safely improve Unified Dyski-
nesia Rating Scale (UDysRS; the primary outcome measure) scores over baseline scores in patients with PD.
METHODS  Twenty patients with PD and L-dopa responsiveness, asymmetrical motor signs, and motor fluctuations,
including dyskinesias, participated in a 1-year multicenter open-label trial of unilateral MRgFUS ablation of the globus
pallidus internus.
RESULTS  The sonication procedure was successfully completed in all 20 enrolled patients. MRgFUS-related adverse
neurological events were generally mild and transient, including visual field deficit (n = 1), dysarthria (n = 4, 2 mild and
2 moderate), cognitive disturbance (n = 1), fine motor deficit (n = 2), and facial weakness (n = 1). Although 3 adverse
events (AEs) were rated as severe (transient sonication-related pain in 2, nausea/vomiting in 1), no AE fulfilled US
FDA criteria for a Serious Adverse Effect. Total UDysRS, the primary outcome measure, improved 59% after treatment
(baseline mean score 36.1, 95% CI 4.88; at 3 months 14.2, 95% CI 5.72, p < 0.0001), which was sustained throughout
the study (at 12 months 20.5, 95% CI 7.39, 43% improvement, p < 0.0001). The severity of motor signs on the treated
side (Movement Disorder Society version of the United Parkinson’s Disease Rating Scale [MDS-UPDRS] part III) in the
“off” medication state also significantly improved (baseline mean score 20.0, 95% CI 2.4; at 3 months 10.6, 95% CI 1.86,
44.5% improvement, p < 0.0001; at 12 months 10.4, 95% CI 2.11, 45.2% improvement, p > 0.0001). The vast majority of
patients showed a clinically meaningful level of improvement on the impairment component of the UDysRS or the motor
component of the UPDRS, while 1 patient showed clinically meaningful worsening on the UPDRS at month 3.
CONCLUSIONS  This study supports the feasibility and preliminary efficacy of MRgFUS pallidotomy in the treatment of
patients with PD and motor fluctuations, including dyskinesias. These preliminary data support continued investigation,
and a placebo-controlled, blinded trial is in progress.
Clinical trial registration no.: NCT02263885 (clinicaltrials.gov)
https://thejns.org/doi/abs/10.3171/2020.6.JNS192773
KEYWORDS  MR-guided focused ultrasound; MRgFUS; pallidotomy; Parkinson’s disease; dyskinesia; functional

S
neurosurgery
tereotactic surgery targeting the globus pallidus in- (L-dopa–induced dyskinesias).4,5 Cardinal motor signs of
ternus (GPi) has been shown to benefit motor symp- PD such as tremor, bradykinesia, and rigidity are also im-
toms of Parkinson’s disease (PD).1–3 Radiofrequency proved contralateral to a pallidotomy, specifically in the
pallidotomy results in a particularly substantial decrease “off” L-dopa state in patients with a fluctuating response
in involuntary movements known as dyskinesias, induced to L-dopa.6–8
by long-term treatment with carbidopa/levodopa (L-dopa) Deep brain stimulation (DBS) of the GPi was intro-

ABBREVIATIONS  AE = adverse event; DBS = deep brain stimulation; FGATIR = fast gray matter acquisition T1 inversion recovery; GPi = globus pallidus internus; L-dopa
= levodopa; MDS-UPDRS = Movement Disorder Society version of the UPDRS; MRgFUS = MR-guided focused ultrasound; PD = Parkinson’s disease; SAE = Serious
Adverse Effect; UDysRS = Unified Dyskinesia Rating Scale; UPDRS = United Parkinson’s Disease Rating Scale.
SUBMITTED  October 11, 2019.  ACCEPTED  June 29, 2020.
INCLUDE WHEN CITING  Published online November 27, 2020; DOI: 10.3171/2020.6.JNS192773.

792 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*39 *
 Eisenberg et al.

duced in part to reduce the neurological complications or greater on the Hoehn and Yahr scale; poorly controlled
of radiofrequency pallidotomy, such as weakness, imbal- depression (Beck Depression Inventory score > 14); and
ance, cognitive abnormalities, and visual field defects.9–14 significant medical or psychiatric comorbidities. Patients
Beneficial effects of GPi DBS that are similar to those with vascular or structural lesions on MRI, a skull density
of pallidotomy are well documented using this strategy, ratio (i.e., the ratio of density of cortical bone to density of
which employs high-frequency stimulation rather than a cancellous bone) of 0.40 or less, claustrophobia, or inabil-
destructive brain lesion to ameliorate motor symptoms ity to lie flat were also excluded.
of PD.15–18 DBS is still an open surgical procedure, and
similar to radiofrequency, pallidotomy shares the risk of Evaluation and Testing
intracranial bleeding and infection.19,20 Another limitation Preprocedure evaluations included physical and neuro-
of DBS is that some appropriate patients find this form of logical examinations, comprehensive neuropsychological
open brain surgery, with the permanent implantation of its testing, formal visual field testingS-UPDRS in both “on”
associated hardware, unacceptable.21 and “off” states, and the Unified Dyskinesia Rating Scale
MR-guided focused ultrasound (MRgFUS) is an in- (UDysRS).
cisionless method that can create a highly targeted brain Imaging evaluation obtained prior to the procedure in-
ablation. MRgFUS of the ventral intermediate nucleus of cluded a noncontrast head CT scan as well as MRI of the
the thalamus was initially FDA approved for the treat- brain. CT images reconstructed using a C-filter in bone
ment of medically refractory essential tremor, and very windows were utilized for skull density measurements,
recently for tremor-predominant PD as well. MRgFUS identifying and marking calcifications, and procedural
has also been utilized in the treatment of other motor as- planning. Multiplanar MRI was performed for procedural
pects of PD, including dyskinesias.22–25 An initial patient planning. The following sequences were used for target-
report of successful unilateral lesioning of the GPi was ing and preplanning purposes: fast gray matter acquisi-
followed by a small open-label study of PD patients with tion T1 inversion recovery (FGATIR) in 15 patients, 3D
dose fluctuation including dyskinesia, which showed simi- T1-weighted gradient echo in 3 patients, and T2-weighted
lar results to radiofrequency pallidotomy.26,27 On the basis fast spin echo in 2 patients. The axial images were refor-
of previous experience with radiofrequency pallidotomy matted or obtained in the commissural (anterior commis-
and MRgFUS for PD and related movement disorders, sure–posterior commissure) plane.
we initiated a trial of MRgFUS unilateral pallidotomy for
patients with asymmetrical motor signs and functionally
MRgFUS Pallidotomy Procedure
interfering motor fluctuations, including bothersome dys-
kinesias. MRgFUS pallidotomy was performed using the In-
Sightec ExAblate 4000 Transcranial System interfaced
with a 3T GE Medical System MRI machine. The patient’s
Methods head was shaved on the day of the procedure. After ad-
Study Population and Criteria ministration of local anesthetic, an Integra CRW stereo-
Patients with a diagnosis of PD in the study had the tactic frame was placed on the patient’s head. A silicone
following inclusion criteria: age between 40 and 80 years, membrane bag was placed around the head and connected
and L-dopa–responsive PD with at least a 30% difference to a circulating degassed and cooled water bath to create
in motor score severity on part III of the Movement Dis- an air-free interface between the transducers and the head.
order Society version of the United Parkinson’s Disease The patient was placed supine and the head was affixed to
Rating Scale (MDS-UPDRS) between “on” and “off” the focused ultrasound device throughout the procedure.
medication states. Patients had a minimum severity in the Intraoperative MRI of the brain was performed, and the
“off” state with a motor score of at least 30 on part III intraoperative images were coregistered with the preopera-
of the MDS-UPDRS. This study was registered with the tive planning study. Targeting was performed using a com-
ClinicalTrials.gov database (http://clinicaltrials.gov), and bination of standard stereotactic coordinates for the GPi
its registration no. is NCT02263885. (20 mm lateral of midline, 3–4 mm anterior of midpoint,
PD motor symptoms were either unilateral or markedly and 3 mm interior of the intercommissural line) and direct
asymmetrical, with predominant disability from one side imaging of the GPi (Fig. 1 left). In 12 patients, direct im-
of the body to the extent that the investigator felt there was aging of the GPi using FGATIR MRI sequences as well
a likelihood of improvement with unilateral treatment. All as tractography of the corticospinal tract was utilized to
patients had functionally interfering motor fluctuations adjust the ablation location. Initially, low-energy sonica-
with dyskinesia on optimal and stable medical therapy tions were delivered to the target GPi, contralateral to the
with a score of at least 3 in response to question 4.2 of the side of the desired treatment effect. As has been previously
MDS-UPDRS. described, focused ultrasound lesioning was performed by
Exclusion criteria included suspicion of another related initially heating an approximately 2-mm-diameter volume
neurodegenerative disease (such as progressive supranu- of tissue with short low-energy sonications generating a
clear palsy, multisystem atrophy, or dementia with Lewy 40°C–45°C target temperature.23 The energy delivered
bodies); impaired cognition with a Montreal Cognitive was gradually increased, progressing incrementally to
Assessment score of less than 21; impairment of speech or higher ablative temperatures, with each therapeutic sonica-
swallowing (score of 3 or 4 on question 5 of part II of the tion followed by patient evaluation for both improvement
MDS-UPDRS); impairment of balance with a score of 3 of motor symptoms of PD (rigidity, tremor, bradykinesia,

J Neurosurg Volume 135 • September 2021 793

 
*39 *
Eisenberg et al.

TABLE 1. Demographics and baseline characteristics

Variable Value
Sex (n = 20)
 Men 13
 Women 7
Race (n = 20)
 White 18
 Black 0
 Hispanic 2
 Other 0
Age, yrs (n = 20)
 Mean 56.4
 SD 11.3
FIG. 1. Axial FGATIR sequence at day 1 (left) and 6 months (right)  Max 73.9
after treatment. The lesion (large arrow in both panels) is demonstrated  Min 34.5
as hyperintensity confined to the ventral lateral aspect of the GPi, with Yrs since diagnosis (n = 6)
a smaller size of the T2 hyperintense lesions and resolution of the per-  Mean 9.9
ilesional edema noted later on the follow-up image (right). The smaller  SD 6.4
arrow in the left panel indicates the internal lamina between the external
 Max 19.7
and internal segments of the globus pallidus.
 Min 2.7
Daily L-dopa equivalents (n = 19), mg
 Mean 1039.4
and dystonia, because patients were treated in the “off”
 SD 601.3
medication state) and potential off-target adverse effects
 Max 2500
such as dysarthria, weakness, or vision changes. Transient
 Min 200
improvement without off-target adverse effects with sub-
lethal sonications was followed by sonications with in-
creased energy to reach target temperatures greater than
56°C. The size and location of the ablative zone were also
continuously monitored using MR thermometry. body structure or function, excluding trivial impairment
Multiple lesions (typically 2–3), usually contiguous, or damage.
were created guided by the boundaries of the GPi on
MRI and the clinical response of the patient. Patients Results
were awake during the procedure but were medicated as Patient Demographics
needed for procedure-related pain and nausea. Following
the treatment session, the stereotactic frame was removed. Of 34 patients initially screened, 20 were enrolled and
The patients were observed overnight and discharged treated in the study. These patients included 13 men and 7
from the hospital the following day. A postprocedure MRI women, of whom 18 were White and 2 were Hispanic, with
of the brain was acquired prior to discharge to evaluate a mean age of 56.4 years (range 35–74 years). The mean
the location and size of the lesion within the pallidum. Not time from diagnosis was 9.9 years, and the mean daily L-
all patients attained the target temperature of 56°C, but dopa equivalent of therapy was 1039 ± 601 mg (Table 1).
all treated patients had a lesion detectable by MRI im- All 20 of the treated patients completed the 3-month visit.
mediately after the procedure and at least one sonication One patient withdrew following the 3-month visit for per-
greater than 51° (Fig. 1). sonal reasons. Nineteen patients were expected to attend
the 6-month visit; 1 was unable to attend but continued
Adverse Events participation. One patient withdrew following the 6-month
visit to have an alternative treatment. For the 12-month
The severity of adverse events (AEs) was predefined visit, 18 patients were expected to attend but 1 was un-
as mild (minor inconvenience, not affecting daily routine able to attend. All patient and visit data are included in the
activities), moderate (bothersome, interferes with routine analysis and figures.
daily activities), or severe (incapacitating, cannot per-
form activities of daily living). The severity rating (mild,
MRgFUS Treatment Characteristics
moderate, or severe) reflects the maximal severity of the
occurrence at any time. Serious Adverse Effects (SAEs) Patients received a mean of 15 ± 3 sonications during
fulfilled the FDA definition of serious, which is an injury treatment with a mean power of 605.6 ± 164.9 W. The
or illness that 1) is life-threatening; 2) results in perma- mean maximum sonication power was 1045.1 ± 233.3 W.
nent impairment of a body function or permanent damage
to a body structure; or 3) necessitates medical or surgi- Clinical Outcomes: Efficacy
cal intervention to preclude permanent impairment of a The primary efficacy outcome measure, the total
body function or permanent damage to a body structure. UDysRS score at 3 months, significantly improved from a
Permanent means irreversible impairment or damage to a baseline score of 36.1 ± 11.12 (95% CI 4.88) to 14.2 ± 13.06

794 J Neurosurg Volume 135 • September 2021

 
*39 *
 Eisenberg et al.

FIG. 2. Clinical outcome measures following unilateral MRgFUS pallidotomy. Mean UDysRS total scores (A), part III scores (B),
part III (motor) scores for the treated extremities (C), and part IV (motor complications) scores (D) at baseline and at 1, 3, 6, and 12
months after treatment. B = baseline. Figure is available in color online only.

(95% CI 5.72), an improvement of 59% (p < 0.0001 using a ment mean value of 11.1 ± 5.61 (95% CI 2.46) improved to
paired t-test). This improvement continued throughout the 6.4 ± 4.15 (95% CI 1.82, p = 0.007) by 3 months and was
duration of the study with a mean total UDysRS score at maintained at this value (6.4 ± 4.31, 95% CI 1.99) at 12
12 months of 20.5 ± 15.99 (95% CI 7.39), a 43% improve- months, representing a 42% improvement (Fig. 2D). No dif-
ment (p < 0.0001; Fig. 2A). ference in results could be discerned between centers; how-
Similar improvement was seen in part III of the ever, the study was not powered to detect such differences.
UDysRS, which is the examiner rating of dyskinesia in- In these patients with asymmetrical motor signs, even
tensity, with a reduction from a mean of 10.2 ± 6.17 (95% the unilateral treatment provided resulted in significant
CI 2.0) to 4.5 ± 4.87 (95% CI 2.14) at 3 months, an im- improvement in motor aspects of experiences of daily liv-
provement of 56.4%. This improvement also persisted, ing as assessed by part II of the MDS-UPDRS. Treatment
with a mean value of 4.4 ± 5.17 at 12 months (95% CI was followed by a significant reduction in mean score
2.39), for a reduction of 68.6% (Fig. 2B). from 14.0 ± 7.53 (95% CI 3.30) at baseline to 7.4 ± 5.04
The efficacy of the treatment of motor signs of PD was (95% CI 2.21, p = 0007) at 3 months, an improvement of
assessed through the motor examination section (part III) 47%. This group change was not sustained, worsening to
of the MDS-UPDRS for those portions related to the treat- 12.1 ± 6.5 (95% CI 2.99) by 12 months after baseline.
ed side (9 items for arm and leg) in the “off” medication
state. There was a significant reduction of scores from a
baseline of 20.0 ± 5.62 (95% CI 2.46) to 10.6 ± 4.24 (95% Individual Patient Therapeutic Benefit or Worsening:
CI 1.86) at 3 months after treatment, representing a 44.5% Responder Analyses
improvement (p < 0.0001). This significant reduction was To obtain a measure of individual patient treatment suc-
sustained at 12 months after treatment, with a mean score cess or failure, responder analysis was performed. Although
of 10.4 ± 4.57 (95% CI 2.11), representing a 45.2% reduc- most patients tolerate a dyskinetic state better than the bra-
tion from the baseline value (p < 0.0001; Fig. 2C). dykinesia state, there is individual patient variation as to
Motor complications of PD were assessed with part IV the relative importance of improving cardinal Parkinsonian
of the MDS-UPDRS capturing data regarding duration and signs over the dyskinetic state. Achieving either of these out-
functional impact of both dyskinesias and “off” periods. comes may be important depending on the patient’s treat-
After treatment there was a significant and persistent im- ment objectives. For this responder analysis, a responder is a
provement in scores on this measure. The baseline pretreat- patient with a minimal clinically important improvement on

J Neurosurg Volume 135 • September 2021 795

 
*39 *
Eisenberg et al.

either the UDysRS part III impairment score (2.32 points, ments in both “on” and “off” medication ratings. As the
while worsening scores are 2.76) or the MDS-UPDRS part study’s primary clinical endpoint, total UDysRS assess-
III motor examination (3.25 points, while worsening scores ments were improved by 59% at 3 months following the
are 4.63), as defined in previous studies.30–32 procedure. Similarly, the motor signs of PD, as assessed
The majority of patients were responders at all post- from part III of the UPDRS contralateral to the lesion,
procedure evaluations, based on the UDysRS part III im- improved by 44% at 3 months and were maintained at 1
pairment scores, suggesting a meaningful clinical benefit year. Safety is suggested by the mild and transient nature
in reduction in dyskinesias (70% at 3 months, 68% at 6 of the AEs reported.
months, 76% at 12 months). The majority of patients were Stereotactic lesioning of the GPi had been shown in
also responders at all postprocedure evaluations based on multiple studies to result in significant and long-lasting
the MDS-UPDRS part III motor examination scores, sug- improvement of motor signs and symptoms of PD, includ-
gesting clinically meaningful relief of cardinal signs of ing dose fluctuations and L-dopa–induced dyskinesia.4–8
PD, including tremor, bradykinesia, and rigidity (80% at Although DBS is the current surgical standard of care for
3 months, 68% at 6 months, 76% at 12 months). The vast PD, preferred over radiofrequency pallidotomy, it has risks
majority of patients had clinically important levels of im- associated with all open surgical procedures of the brain,
provement at all postprocedure evaluations on at least one particularly bleeding and infection.19,20
of these two outcome measures (90% at 3 months, 95% at Previous studies involving patients with essential
6 months, 94% at 12 months). One patient had a clinically tremor and tremor-predominant PD have established that
important level of worsening at only the 3-month evalua- MRgFUS thalamotomy is an emerging alternative to DBS
tion, based on the MDS-UPDRS part III. for patients with a medically refractory tremor.22–25 This
In this study, screening scores were also used as baseline study was designed to investigate if MRgFUS pallidotomy
scores, so that follow-up scores for all outcome measures could provide benefits like radiofrequency pallidotomy
were compared to this single time point. In this setting it and pallidal DBS, while reproducing the high level of
is possible that the initial single score may represent an safety and tolerance of MRgFUS of the thalamus seen in
atypically severe patient state and that improvement seen in patients with essential tremor.33
follow-up scores may be influenced by the regression to the Our experience is also consistent with the previous lim-
mean phenomenon. We feel that this phenomenon did not ited reports of MRgFUS pallidotomy in PD. A 55-year-
substantively influence the significant improvement seen in old woman with PD was initially reported by Na et al. to
this study because improvement was noted in multiple out- show a 62% reduction in baseline UDysRS score 3 months
come measures over multiple follow-up time points. A trend after a unilateral MRgFUS-mediated pallidotomy and a
of diminishing improvement over time in the UDysRS 61.9% reduction in motor “off” scores with no AEs.29 In a
score would be expected with disease progression. recently published study by Jung et al.,28 10 patients with
PD underwent unilateral MRgFUS pallidotomy. Of the 8
Safety and Tolerability patients who successfully completed the procedure (in 2
A total of 61 AEs were reported by 17 patients (3 pa- patients thermal ablation could not be achieved, likely due
tients reported no AEs). Twenty-five of these AEs (41%) to poor transcranial acoustic transmission), there was an
were reported to be unrelated to the procedure, although improvement in total UDysRS by 52.7% and a 30.2% re-
4 AEs (all mild) were considered to be related to place- duction in UPDRS “off” scores at 6 months.28 This is very
ment of the stereotactic frame (headache, facial edema). similar to the percentage reductions in the total UDysRS
Ten AEs were considered to be PD related. and MDS-UPDRS part III in the “off” state in our study.
Of the 36 procedure-related AEs, the most common This magnitude of improvement in motor “off” state and
were nausea/vomiting (n = 3), headache (n = 3), and sonica- dyskinesia is also comparable to that seen in studies of
tion-related head pain (n = 7). Seventeen of the AEs were radiofrequency open stereotactic pallidotomy.4–8 This
transient, which included the only severe AEs (2 with tran- amount of reduction in dyskinesias and motor “off” scores
sient sonication-related head pain, 1 with transient nausea persisted through the 12-month duration of the study and
and vomiting, 5% of total). Neurological AEs related to the is also clinically important, along with a meaningful re-
procedure included visual field deficit (1 mild, transient), duction in functional interference by motor symptoms in
dysarthria (4 total, 2 mild and 2 moderate), cognitive dis- the majority of treated patients.30–32
turbance (1 mild), fine motor deficit (2 mild), facial weak- The safety profile of our study of MRgFUS pallidot-
ness (1 mild), and balance difficulties (1 moderate). omy is also consistent with previous work investigating
Twenty AEs persisted and continue to be followed, in- its application for tremor. There were no SAEs (as de-
cluding fine motor difficulties (1 mild), dysarthria (1 mild, fined by the US FDA) in our 20 patients who underwent
2 moderate), and balance difficulties (1 mild), none of the procedure. Of the 61 AEs reported, only 3 (5%) were
which were rated as severe. No AEs fulfilled FDA criteria rated as severe, and all of these were transient (headache,
of an SAE. pain, nausea). Because the intent of MRgFUS is to create
a permanent brain ablation, possible neurological deficits
due to damage to brain tissue adjacent to the target are
Discussion a clear concern. Neurological deficits accounted for only
This preliminary study of unilateral MRgFUS palli- 10 of the 61 reported AEs, 7 of which were considered
dotomy for the motor fluctuations and L-dopa–induced mild and none were rated as severe. Five of these deficits
dyskinesia associated with PD demonstrated improve- persisted for at least the duration of the study. With regard

796 J Neurosurg Volume 135 • September 2021

 
*39 *
 Eisenberg et al.

to neurological deficits that had been previously associ-   9. Biousse V, Newman NJ, Carroll C, et al. Visual fields in pa-
ated with radiofrequency pallidotomy, only 1 patient had tients with posterior GPi pallidotomy. Neurology. 1998;​50(1):​
a mild visual field abnormality by confrontation that was 258–265.
10. Higuchi Y, Iacono RP. Surgical complications in patients
not detectable by later formal visual field analysis. Anoth- with Parkinson’s disease after posteroventral pallidotomy.
er patient noted memory difficulties and showed decline Neurosurgery. 2003;​52(3):​558–571.
on one subtest of a neuropsychological battery, but im- 11. Olzak M, Laskowska I, Jelonek J, et al. Psychomotor and
provement on another subtest. This experience is consis- executive functioning after unilateral posteroventral pallidot-
tent with an analysis of the safety of MRgFUS-mediated omy in patients with Parkinson’s disease. J Neurol Sci. 2006;​
thalamotomy for essential tremor, where the vast majority 248(1-2):​97–103.
of thalamotomy-related neurological AEs were mild and 12. Palur RS, Berk C, Schulzer M, Honey CR. A metaanalysis
rarely severe.33 comparing the results of pallidotomy performed using micro-
electrode recording or macroelectrode stimulation. J Neuro-
As expected for this incisionless procedure, no intra- surg. 2002;​96(6):​1058–1062.
cranial bleeding or infection occurred. The results of this 13. Samuel M, Caputo E, Brooks DJ, et al. A study of medial
initial study compare very favorably to those of previous pallidotomy for Parkinson’s disease:​clinical outcome, MRI
studies of radiofrequency pallidotomy for PD, in which up location and complications. Brain. 1998;​121(pt 1):​59–75.
to several percent of patients showed reportable neurologi- 14. Schmand B, de Bie RM, Koning-Haanstra M, et al. Unilateral
cal deficits that were serious and persistent. Although the pallidotomy in PD:​a controlled study of cognitive and behav-
total number of patients who have undergone MRgFUS ioral effects. The Netherlands Pallidotomy Study (NEPAS)
pallidotomy, including our study, is relatively small, the group. Neurology. 2000;​54(5):​1058–1064.
15. Blomstedt P, Hariz GM, Hariz MI. Pallidotomy versus pal-
current experience suggests an acceptable safety profile. lidal stimulation. Parkinsonism Relat Disord. 2006;​12(5):​
This early experience is also encouraging because applica- 296–301.
tion of advancing imaging technology with improvements 16. Jiménez F, Velasco F, Carrillo-Ruiz JD, et al. Comparative
in targeting has the potential to improve the safety and evaluation of the effects of unilateral lesion versus electrical
efficacy of the procedure.34 stimulation of the globus pallidus internus in advanced Par-
kinson’s disease. Stereotact Funct Neurosurg. 2006;​84(2-3):​
64–71.
Conclusions 17. Krack P, Martinez-Fernandez R, Del Alamo M, Obeso JA.
The results of this initial study with regard to both effi- Current applications and limitations of surgical treatments
cacy and safety support our ongoing approach of expanded for movement disorders. Mov Disord. 2017;​32(1):​36–52.
clinical investigation in the form of an adequately powered, 18. Merello M, Nouzeilles MI, Kuzis G, et al. Unilateral radio-
frequency lesion versus electrostimulation of posteroventral
blinded, controlled study of MRgFUS pallidotomy in the pallidum:​a prospective randomized comparison. Mov Dis-
treatment of patients with significantly asymmetrical PD ord. 1999;​14(1):​50–56.
and a fluctuating motor response to medications, includ- 19. Patel DM, Walker HC, Brooks R, et al. Adverse events as-
ing dyskinesia. Successful development of this technology sociated with deep brain stimulation for movement disorders:​
will provide patients with PD with a new, less-invasive analysis of 510 consecutive cases. Neurosurgery. 2015;​11(2)
treatment option. (suppl 2):​190–199.
20. Sillay KA, Larson PS, Starr PA. Deep brain stimulator hard-
ware-related infections:​incidence and management in a large
References series. Neurosurgery. 2008;​62(2):​360–367.
  1. Alkhani A, Lozano AM. Pallidotomy for Parkinson disease:​ 21. Kim MR, Yun JY, Jeon B, et al. Patients’ reluctance to un-
a review of contemporary literature. J Neurosurg. 2001;​94(1):​ dergo deep brain stimulation for Parkinson’s disease. Parkin-
43–49. sonism Relat Disord. 2016;​23(9):​91–94.
  2. Cif L, Hariz M. Seventy years with the globus pallidus:​pal- 22. Chang JW, Park CK, Lipsman N, et al. A prospective trial of
lidal surgery for movement disorders between 1947 and 2017. magnetic resonance-guided focused ultrasound thalamotomy
Mov Disord. 2017;​32(7):​972–982. for essential tremor:​results at the 2-year follow-up. Ann Neu-
  3. Gross RE. What happened to posteroventral pallidotomy for rol. 2018;​83(1):​107–114.
Parkinson’s disease and dystonia? Neurotherapeutics. 2008;​ 23. Elias WJ, Lipsman N, Ondo WG, et al. A randomized trial of
5(2):​281–293. focused ultrasound thalamotomy for essential tremor. N Engl
  4. de Bie RM, de Haan RJ, Nijssen PC, et al. Unilateral palli- J Med. 2016;​375(8):​730–739.
dotomy in Parkinson’s disease:​a randomised, single-blind, 24. Bond AE, Shah BB, Huss DS, et al. Safety and efficacy
multicentre trial. Lancet. 1999;​354(9191):​1665–1669. of focused ultrasound thalamotomy for patients with
  5. Vitek JL, Bakay RA, Freeman A, et al. Randomized trial of medication-refractory, tremor-dominant Parkinson disease:​
pallidotomy versus medical therapy for Parkinson’s disease. a randomized clinical trial. JAMA Neurol. 2017;​74(12):​1412–
Ann Neurol. 2003;​53(5):​558–569. 1418.
  6. Eskandar EN, Shinobu LA, Penney JB Jr, et al. Stereotactic 25. Zaaroor M, Sinai A, Goldsher D, et al. Magnetic resonance-
pallidotomy performed without using microelectrode guid- guided focused ultrasound thalamotomy for tremor:​a report
ance in patients with Parkinson’s disease:​surgical technique of 30 Parkinson’s disease and essential tremor cases. J Neu-
and 2-year results. J Neurosurg. 2000;​92(3):​375–383. rosurg. 2018;​128(1):​202–210.
  7. Kondziolka D, Bonaroti E, Baser S, et al. Outcomes after 26. Magara A, Bühler R, Moser D, et al. First experience with
stereotactically guided pallidotomy for advanced Parkinson’s MR-guided focused ultrasound in the treatment of Parkin-
disease. J Neurosurg. 1999;​90(2):​197–202. son’s disease. J Ther Ultrasound. 2014;​2(1):​11.
  8. Shannon KM, Penn RD, Kroin JS, et al. Stereotactic palli- 27. Martínez-Fernández R, Rodríguez-Rojas R, Del Álamo M,
dotomy for the treatment of Parkinson’s disease. Efficacy and et al. Focused ultrasound subthalamotomy in patients with
adverse effects at 6 months in 26 patients. Neurology. 1998;​ asymmetric Parkinson’s disease:​a pilot study. Lancet Neurol.
50(2):​434–438. 2018;​17(1):​54–63.

J Neurosurg Volume 135 • September 2021 797

 
*39 *
Eisenberg et al.

28. Jung NY, Park CK, Kim M, et al. The efficacy and limits of Author Contributions
magnetic resonance-guided focused ultrasound pallidotomy Conception and design: Fishman, Eisenberg, Elias. Acquisition
for Parkinson’s disease:​a phase I clinical trial. J Neurosurg. of data: all authors. Analysis and interpretation of data: Fishman,
2019;​130(6):​1853–1861. Eisenberg, Krishna, Elias, Cosgrove, Gandhi. Drafting the article:
29. Na YC, Chang WS, Jung HH, et al. Unilateral magnetic reso- Fishman, Eisenberg. Critically revising the article: Fishman,
nance-guided focused ultrasound pallidotomy for Parkinson Eisenberg, Krishna, Elias, Cosgrove, Gandhi. Reviewed submitted
disease. Neurology. 2015;​85(6):​549–551. version of manuscript: Fishman, Eisenberg, Krishna, Elias, Cos-
30. Horváth K, Aschermann Z, Ács P, et al. Minimal clinically grove. Approved the final version of the manuscript on behalf of
important difference on the Motor Examination part of MDS- all authors: Fishman. Administrative/technical/material support:
UPDRS. Parkinsonism Relat Disord. 2015;​21(12):​1421–1426. Elias, Cosgrove, Gandhi, Aldrich. Study supervision: Eisenberg,
31. Horváth K, Aschermann Z, Kovács M, et al. Minimal clini- Krishna, Elias, Cosgrove, Aldrich.
cally important differences for the experiences of daily liv-
ing parts of Movement Disorder Society-sponsored Unified Correspondence
Parkinson’s Disease Rating Scale. Mov Disord. 2017;​32(5):​
789–793. Paul S. Fishman: University of Maryland School of Medicine,
32. Mestre TA, Beaulieu-Boire I, Aquino CC, et al. What is a Baltimore. [email protected].
clinically important change in the Unified Dyskinesia Rating
Scale in Parkinson’s disease? Parkinsonism Relat Disord.
2015;​21(11):​1349–1354.
33. Fishman PS, Elias WJ, Ghanouni P, et al. Neurological ad-
verse event profile of magnetic resonance imaging-guided
focused ultrasound thalamotomy for essential tremor. Mov
Disord. 2018;​33(5):​843–847.
34. Krishna V, Sammartino F, Agrawal P, et al. Prospective
tractography-based targeting for improved safety of focused
ultrasound thalamotomy. Neurosurgery. 2019;​84(1):​160–168.

Disclosures
Funding for the study presented was provided by InSightec Inc.,
the Focused Ultrasound Foundation, and the Michael J. Fox Foun-
dation for Parkinson’s Research. Dr. Elias reports being a consul-
tant for Second Sight.

798 J Neurosurg Volume 135 • September 2021

 
*39 *
 CLINICAL ARTICLE
J Neurosurg 135:799–805, 2021

Unilateral pallidothalamic tractotomy for akinetic-rigid

Parkinson’s disease: a prospective open-label study
Shiro Horisawa, MD,1 Atsushi Fukui, MD, PhD,1 Hayato Yamahata, MD,1 Yukiko Tanaka, MD,1
Atsushi Kuwano, MD,1 Oji Momosaki, MD,1 Mutsumi Iijima, MD, PhD,2 Magi Nanke, MD,3
Takakazu Kawamata, MD, PhD,1 and Takaomi Taira, MD, PhD1

Departments of 1Neurosurgery and 2Neurology, and 3School of Medicine, Tokyo Women’s Medical University, Tokyo, Japan

OBJECTIVE  Neurosurgical ablation is an effective treatment for medically refractory motor symptoms of Parkinson’s
disease (PD). A limited number of studies have reported the effect of ablation of the pallidothalamic tract for PD. In this
study, the authors evaluated the safety and efficacy of unilateral pallidothalamic tractotomy for akinetic-rigid (AR)–PD.
METHODS  Fourteen AR-PD patients, who were enrolled in this prospective open-label study, underwent unilateral palli-
dothalamic tractotomy. The Movement Disorder Society–Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) Part
III and Part IV (dyskinesia and dystonia) scores and levodopa equivalent daily dose (LEDD) were evaluated at baseline
and at 3 and 12 months postoperatively.
RESULTS  Of the 14 patients enrolled in the study, 4 were lost to follow-up and 10 were analyzed. The total MDS-
UPDRS Part III score significantly improved from 45 ± 4.6 at baseline to 32.9 ± 4.8 at 12 months postoperatively (p =
0.005). Contralateral side rigidity and bradykinesia significantly improved from 4.4 ± 0.5 and 10.4 ± 1.5 at baseline to 1.7
± 0.4 (p = 0.005) and 5.2 ± 1.4 (p = 0.011) at 12 months, respectively. While posture significantly improved with a 20%
reduction in scores (p = 0.038), no significant improvement was found in gait (p = 0.066). Dyskinesia and dystonia were
improved with a 79.2% (p = 0.0012) and 91.7% (p = 0.041) reduction in scores, respectively. No significant change was
found in the LEDD. Hypophonia was noted in 2 patients, eyelid apraxia was noted in 1 patient, and a reduced response
to levodopa, which resulted in an increase in the daily dose of levodopa, was noted in 3 patients. No serious permanent
neurological deficits were observed.
CONCLUSIONS  Unilateral pallidothalamic tractotomy improved contralateral side rigidity and bradykinesia, dyskinesia,
and dystonia in patients with AR-PD.
Clinical trial registration no.: UMIN000031138 (umin.ac.jp)
https://thejns.org/doi/abs/10.3171/2020.7.JNS201547
KEYWORDS  Parkinson’s disease; pallidothalamic tract; campotomy; stereotactic neurosurgery; functional
neurosurgery

N
eurosurgical ablation and deep brain stimulation The pallidothalamic tract consists of two distinct fiber
(DBS) are the current standard treatments for the bundles including the ansa lenticularis and the lenticu-
cardinal motor symptoms in Parkinson’s disease lar fasciculus, which connect the internal segment of the
(PD).1–5 Although the mainstay surgical treatment is DBS, globus pallidus (GPi) to thalamic subnuclei such as the
ablative procedures are still an important option for pa- ventral anterior nucleus and the ventrolateral nucleus.9,10
tients who cannot access DBS because of economic or Both fiber bundles originate from different parts of the
geographic limitations. Additionally, increased attention GPi, course across the internal capsule, and merge at the
has been paid to recent advancements in the field of mini- field of Forel where they form one fiber, which is called the
mally invasive ablative procedures, including MRI-guided “thalamic fasciculus.”11 Ablation of the field of Forel, also
focused ultrasound.6–8 called “campotomy,” was first performed for the treatment

ABBREVIATIONS  AC-PC = anterior commissure–posterior commissure; AR = akinetic-rigid; DBS = deep brain stimulation; GPi = internal segment of the globus pallidus;
LEDD = levodopa equivalent daily dose; MDS-UPDRS = Movement Disorder Society–Unified Parkinson’s Disease Rating Scale; MX = mixed; PD = Parkinson’s disease;
PTT = pallidothalamic tractotomy; STN = subthalamic nucleus; TD = tremor dominant.
SUBMITTED  April 29, 2020.  ACCEPTED  July 21, 2020.
INCLUDE WHEN CITING  Published online January 15, 2021; DOI: 10.3171/2020.7.JNS201547.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  799

*39 *
Horisawa et al.

of PD in the 1960s by Spiegel et al. and was later revived was performed on the side most affected by parkinsonian
by Jeanmonod et al., who renamed the procedure “palli- motor symptoms. With the patient under local anesthesia,
dothalamic tractotomy” (PTT).12,13 Jeanmonod’s group a Leksell stereotactic frame (Elekta) was fixed to their
detailed the effect of PTT with the use of radiofrequency head. Axial (1-mm slice) T1-weighted and axial (1-mm
and focused ultrasound ablation on the cardinal motor slice) and coronal (2-mm slice) T2-weighted MR images
symptoms of PD.6,12,14 However, only two research groups were used to plan the PTT surgery. The lateral part of the
have published reports investigating PTT/campotomy for pallidothalamic tract is long and corresponds to the tha-
PD.6,12,15 Additionally, these studies mainly evaluated PD lamic fasciculus on Morel’s atlas. According to Morel’s
with tremor. Akinetic-rigid (AR)–PD without tremor is the atlas, the width of the pallidothalamic tract is 6 mm at
subtype of PD with the poorest prognosis, and the efficacy the anterior commissure–posterior commissure (AC-PC)
of PTT/campotomy for AR-PD has not been evaluated. To plane. Therefore, we set two stereotactic ablative targets to
that end, in the present study, we report the prospective ef- cover the pallidothalamic tract (Fig. 1). The medial target
ficacy and safety of unilateral PTT for AR-PD. was set at 8–10 mm lateral and 0.5–2.5 mm inferior to the
midpoint of the AC-PC and at the midpoint of the AC-PC.
The lateral target was set at 10–13 mm lateral, 1.0–2.0 mm
Methods inferior, and 1 mm posterior to the midpoint of the AC-
Study Design PC. A monopolar radiofrequency electrode (1.0-mm-di-
This was a prospective, open-label noncontrolled study ameter tip with an uninsulated length of 2.0–4.0 mm) and
performed at Tokyo Women’s Medical University Hospi- a Leksell Neuro Generator (Elekta) were used to perform
tal. The protocol was reviewed and approved by the Eth- macrostimulation (130 Hz, 100-µsec pulse width, up to 5
ics Committees of Tokyo Women’s Medical University mA) and ablation. Before ablation, macrostimulation was
Hospital. All patients provided written informed consent performed to confirm adverse effects. A single ablation at
before inclusion in this study. The study is registered in the 70°C for 40 seconds was made at each target.
University Hospital Medical Information Network Clini-
cal Trials Registry (umin.ac.jp), and its clinical trial regis- Statistical Analysis
try number is UMIN000031138. A Shapiro-Wilk normality test was performed to con-
firm the normality of the data. As the data were nonnor-
Patients mally distributed, Friedman’s tests were used to determine
Patients with medically refractory PD who declined to the change in MDS-UPDRS scores and the LEDD across
undergo DBS surgery were enrolled. The diagnosis of PD the three study visits: at baseline and at 3 and 12 months
was confirmed by a neurologist or neurosurgeon special- postoperatively. Wilcoxon’s signed-rank tests were also
izing in movement disorders (T.T., M.I., S.H.). The exclu- used to compare the evaluated variables between baseline
sion criteria were as follows: comorbid neurodegenerative and 12 months. Statistical analysis was performed using
disorders, history of intracranial surgery, cerebral stroke, the JMP statistical package, version 13.0.0 (SAS Institute
traumatic brain injury, and antiplatelet or anticoagulant Inc.). All statistical tests were two-tailed, and a p value <
medications. 0.05 was determined to be statistically significant. Values
are expressed as the mean ± standard deviation, unless in-
Evaluations dicated otherwise.
The primary endpoint was the change in Movement
Disorder Society–Unified Parkinson’s Disease Rating Results
Scale (MDS-UPDRS) scores after unilateral PTT. The Patients
MDS-UPDRS Part III (off-medication state) and Part IV, From December 2016 to April 2019, 14 patients with
specifically dyskinesia items 4.1–4.2 and dystonia item PD were enrolled in this study. Of these, 4 were lost to
4.6, were assessed. Evaluations were performed at base- follow-up; therefore, 10 patients (5 male, 5 female) were
line and at 3 and 12 months postoperatively. To interpret included in the final analysis. All the patients had AR-
surgical outcomes clinically, subscores of the MDS- PD. The mean age at disease onset was 52.1 ± 7.1 years.
UPDRS were also evaluated independently as follows: The mean disease duration was 12.6 ± 5.0 years. The
contralateral side (surgical side) and ipsilateral side rigid- mean preoperative Hoehn and Yahr stage was 3.2 ± 0.6.
ity (item 3.3) and bradykinesia (items 3.4–3.8), gait (items Detailed information regarding patient characteristics is
3.10–3.11), and posture (items 3.12–3.13). The levodopa summarized in Table 1.
equivalent daily dose (LEDD) was also evaluated. Anti-
parkinsonian drugs were not changed before the surgery. Clinical Assessments
After surgery, an adjustment of antiparkinsonian drugs A summary of the clinical assessments is given in Ta-
was allowed based on the patient’s neurological condition. ble 2 and Fig. 2. The total MDS-UPDRS Part III score
Postoperative MRI was performed on the day of surgery was significantly improved from 45.0 ± 4.6 at baseline to
and 3 months postoperatively. 32.9 ± 4.8 at 12 months postoperatively (p = 0.005; 26.9%
reduction). Contralateral side rigidity and bradykinesia
Surgical Procedure were significantly improved from 4.4 ± 0.5 and 10.4 ± 1.5
Leksell SurgiPlan (Elekta) was used to plan the ste- at baseline to 1.7 ± 0.4 (p = 0.005) and 5.2 ± 1.4 (p = 0.011)
reotactic targets and electrode trajectory. Unilateral PTT at 12 months, respectively (Video 1).

800 J Neurosurg Volume 135 • September 2021

 
*39 *
 Horisawa et al.

FIG. 1. Anatomical location of the pallidothalamic tract shown by postoperative MRI overlaying Morel’s atlas. A: Two lesions (ar-
rows) on the thalamic fasciculus (ft), which correspond to the pallidothalamic tract at the AC-PC plane. B: The same lesions at 0.9
mm below the AC-PC plane. The lesions were set just lateral to the mammillothalamic tract (mtt; arrowhead). al = ansa lenticularis;
fl = lenticular fasciculus.

VIDEO 1. Pre- and postoperative bradykinesia. Copyright Shiro with a 79.2% (p = 0.0012; Video 3) and a 91.7% (p = 0.041)
Horisawa. Published with permission. Click here to view. reduction in scores at 12 months.
No significant change was found in ipsilateral side ri- VIDEO 3. Pre- and postoperative dyskinesia. Copyright Shiro
gidity and bradykinesia. Both gait and posture were sig- Horisawa. Published with permission. Click here to view.
nificantly improved as compared among the three time The LEDD was 575 ± 79.7 mg at baseline and 575 ± 92.2
points (Friedman test), but only posture was improved in mg at 12 months, which was not significantly different.
a comparison between baseline and 12 months postopera- Postoperative MRI showed coagulated lesions located
tively (Video 2). above the subthalamic nucleus (STN) in all patients, as
VIDEO 2. Pre- and postoperative gait in the medication-off period. shown in Fig. 3.
Copyright Shiro Horisawa. Published with permission. Click here to
view.
Adverse Events
Dyskinesia and dystonia were significantly improved All adverse events are listed in Table 3. Six patients
developed somnolence during the surgery, which sponta-
neously improved the following morning. A reduced re-
TABLE 1. Summary of characteristics of 10 patients with AR-PD sponse to the oral intake of levodopa, including a shorter
Variable Value effect duration and a longer time to become effective,
developed within 1 week after the surgery in 3 patients.
No. of patients The extended off duration resulted in an increase of their
 Men 5 daily dose and dose frequency of levodopa. Two patients
 Women 5 developed hypophonia, which was mild and did not inter-
Age at disease onset in yrs 52.1 ± 7.1 fere with their daily communication. A chronic subdural
Disease duration in yrs 12.6 ± 5.0 hematoma developed in 1 patient at 2 months postsurgery.
The hematoma caused mild hemiparesis, and hematoma
Preop Hoehn & Yahr stage 3.2 ± 0.6
irrigation surgery was performed, which resulted in com-
Side of surgery, n plete recovery of neurological deficits.
 Rt 6
 Lt 4 Patients Lost to Follow-Up
Values are expressed as the mean ± standard deviation, unless indicated Four patients were lost to follow-up; 2 lived in remote
otherwise. locations, and the reason for the loss of the other 2 pa-

J Neurosurg Volume 135 • September 2021 801

 
*39 *
Horisawa et al.

TABLE 2. Clinical outcomes of 10 patients who underwent PTT for AR-PD

MDS-UPDRS Baseline 3 Mos p Value* 12 Mos p Value† p Value‡
Part III
  Total score 45 ± 4.6 31.5 ± 4.6 0.005 32.9 ± 4.8 0.005 <0.001
  Contralat side
   Rigidity (item 3.3) 4.4 ± 0.5 1.7 ± 0.4 0.005 1.7 ± 0.4 0.005 <0.001
   Bradykinesia (items 3.4–3.8) 10.4 ± 1.5 5.4 ± 1.3 0.008 5.2 ± 1.4 0.011 <0.001
  Ipsilat side
   Rigidity (item 3.3) 3.4 ± 0.6 3 ± 0.6 0.194 3 ± 0.6 0.194 0.368
   Bradykinesia (items 3.4–3.8) 7.6 ± 1.3 7.3 ± 1.3 0.180 7.3 ± 1.3 0.180 0.135
  Gait (items 3.10–3.11) 4.6 ± 0.5 3 ± 0.6 0.016 3.7 ± 0.6 0.066 0.004
  Posture (items 3.12–3.13) 4 ± 0.4 2.9 ± 0.5 0.020 3.2 ± 0.5 0.038 0.006
Part IV
  Dyskinesia (items 4.1–4.2) 4.8 ± 0.4 0.8 ± 0.3 0.011 1 ± 0.3 0.012 <0.001
  Dystonia (item 4.6) 2.4 ± 0.4 0.2 ± 0.2 0.041 0.2 ± 0.2 0.041 0.007
LEDD (mg) 575 ± 79.7 577.8 ± 106 0.893 575 ± 92.2 0.99 0.651
Hoehn & Yahr stage 3.2 ± 0.2 3 ± 0.3
Data are presented as mean ± standard deviation, unless indicated otherwise. Significance was accepted at p < 0.05.
Boldface type indicates statistical significance. Bonferroni correction was used for multiple comparisons.
* Comparison of baseline and 3-month postoperative scores, Wilcoxon signed-rank test.
† Comparison of baseline and 12-month postoperative scores, Wilcoxon signed-rank test.
‡ Comparison of baseline, 3-month postoperative, and 12-month postoperative scores, Friedman test.

FIG. 2. Pre- and postoperative subscores of MDS-UPDRS Part III. Contralateral side rigidity (A) and bradykinesia (B), posture
(C), dyskinesia (D), and dystonia (D) were significantly improved 12 months after surgery. Gait (C) also improved 12 months after
surgery without statistical significance for the comparison of mean baseline and 12-month postoperative scores (Wilcoxon signed-
rank test). Error bars represent the standard error of the mean.

802 J Neurosurg Volume 135 • September 2021

 
*39 *
 Horisawa et al.

TABLE 3. Adverse events

No. of Patients
Event Temporary Persistent
Somnolence 6
Hypophonia 2
Drooling 1
Eyelid apraxia 1
Fatigue 1
Decreased response to levodopa 3
Chronic subdural hemorrhage 1

tions and quality of life, in a well-designed prospective

study. In their report, a 43.9% improvement was observed
FIG. 3. Follow-up T1- and T2-weighted MR images at 0 (A and B) and 3 in the clinician-scored motor evaluation, a 65.7% improve-
(C and D) months after PTT showing the lesion above the subthalamic ment in tremor, an 87.8% improvement in rigidity, a 68%
nucleus (white arrows). improvement in bradykinesia, an 82.4% improvement in
dyskinesia, and a 31.9% improvement in gait in 12 cases
at the 2-year follow-up. Additionally, the patients’ quality
tients was unknown. The mean follow-up period was 2 of life was also improved by 37.8%. The previous studies
months (range 1–3 months). The pre- and postoperative reported a greater degree of improvement in the clinician-
total MDS-UPDRS Part III scores at each last available scored motor evaluation (MDS-UPDRS Part III score)
evaluation were 35 ± 3.4 and 26.5 ± 3.3, respectively than we did in the current study.6,12,15
(24.3% improvement). No significant postoperative com- A possible reason for this difference in motor outcomes
plications were observed in these patients. may be patient selection. PD has three subtypes: tremor
dominant (TD), AR, and mixed (MX).16 Several studies
have demonstrated morphological and physiological dif-
Discussion ferences between AR-PD and TD-PD.17–20 Rapid disease
In this study, we sought to evaluate prospectively the progression, poorer response to levodopa, and the likeli-
efficacy and safety of unilateral PTT for AR-PD. One year hood of dementia are greater in AR-PD than in TD-PD
after surgery, patients showed improvements in the clini- and MX-PD.16,21 Loss of dopaminergic neurons in the cau-
cian-scored motor evaluation, contralateral side rigidity dal GPi in AR-PD is more pronounced and widespread
and bradykinesia (50% improvement), dyskinesia (79.2% than in TD-PD and MX-PD.22 AR-PD or postural instabil-
improvement), and dystonia (91.7% improvement). While ity and gait difficulty subtypes have been reported to be
posture significantly improved with a 20% reduction in less likely to respond to GPi-DBS or STN-DBS compared
scores (p = 0.038), no significant improvement was found to TD-PD.23,24 In Godinho and colleagues’ study,15 8 pa-
in gait (p = 0.066). The LEDD was not changed. In terms tients (66.7%) had TD-PD and 4 had AR-PD. In Aufenberg
of the procedure’s safety, 1 patient had a chronic subdural and associates’ study,12 no patient had pure AR-PD; all the
hematoma 2 months after the surgery. No other serious evaluated patients had significant resting tremor. In Gal-
adverse events were reported. lay and colleagues’ study, 5 patients (10%) had AR-PD, 9
Two recent studies using radiofrequency PTT/campot- (18%) had TD-PD, and 37 (72%) had MX-PD.6 Conversely,
omy for PD have reported superior results in treating the all patients in the present study had AR-PD without sig-
cardinal motor symptoms of PD including rigidity, tremor, nificant resting tremor. This suggests that the response to
bradykinesia, gait, and posture.12,15 Aufenberg et al. re- surgery may be different among AR-, TD-, and MX-PD
ported the effects of radiofrequency PTT in 22 unilateral patients.
and 19 bilateral cases, observing an 87% improvement in A significant reduction in the LEDD is an important and
tremor, an 82.4% improvement in rigidity, 58% improve- valuable issue for patients and is the reason for the prefer-
ment in bradykinesia, 88% improvement in dyskinesia, ence for STN-DBS over GPi-DBS. Our study showed no
and 33% improvement in axial symptoms at 22 months’ significant reduction in LEDD after unilateral PTT, which
follow-up.12 Additionally, they reported a 68% reduction is concordant with findings in Godinho and colleagues’
in the LEDD. Gallay et al.6 reported unilateral/bilateral study,15 which reported a 9.3% reduction in LEDD. In our
PTT using focused ultrasound for 51 patients with PD. study, 3 patients had a reduction in LEDD at 3 months,
The results were similar to those of Aufenberg et al.,12 and 2 of the 3 returned to their preoperative LEDD by 1
showing an 84% improvement in tremor, 70% in rigid- year after surgery. Another 3 patients exhibited a reduced
ity, 73% in distal limb bradykinesia, and 55% reduction response to levodopa, including a shorter levodopa effect
of levodopa intake. Similarly, Godinho et al.15 reported a duration and a longer time to become effective, which re-
detailed analysis of the effects of unilateral campotomy on sulted in an increased LEDD to maintain their preopera-
motor disturbances, as well as neuropsychological func- tive daily living activities. Two of these patients required

J Neurosurg Volume 135 • September 2021 803

 
*39 *
Horisawa et al.

their dose frequency to be increased from 3 to 4 times per ment of outcomes and introduced a bias in the results. In
day. The preoperative and 3-month postoperative LEDDs addition, the limited sample size reduced the power of the
in the 3 patients were 900, 750, and 450 mg; and 1200, study to detect small to moderate treatment effects. Tract-
900, and 500 mg, respectively. The LEDD was unchanged ography can improve stereotactic targeting; however, we
at the 1-year evaluation. The etiology of this reduced re- were unable to visualize the course of the pallidothalamic
sponse to levodopa after surgery is not known. However, tract using tractography. Finally, the 1-year follow-up is
a similar phenomenon was reported by Intemann et al.,25 relatively short for PD, which is a slowly progressive dis-
suggesting that a gradual reduction in responsiveness to ease.
levodopa may be a side effect of pallidotomy. In particular,
bilateral interventions may be associated with a significant Conclusions
prolonged reduction in LEDD. Only Jeanmonod’s groups
(both the Aufenberg et al.12 and the Gallay et al.6 studies) This study revealed improved contralateral side rigid-
have performed bilateral PTT by radiofrequency and fo- ity and bradykinesia, and dyskinesia following unilateral
cused ultrasound ablation. However, they did not report on PTT for patients with AR-PD. No serious adverse effects
the differences in clinical results between unilateral and were associated with the procedure. In the future, a larger
bilateral PTT. Aufenberg et al.12 reported a 68% decrease sample of patients together with a longer follow-up and
in LEDD at an average of 22 months’ postoperative evalu- blinded assessments is needed to elaborate more precisely
ation after radiofrequency PTT, including bilateral proce- the safety and efficacy of PTT for AR-PD.
dures. Gallay et al.6 reported a 55% reduction in LEDD
at the 1-year postoperative evaluation after focused ultra- References
sound PTT, including bilateral procedures. These results   1. Follett KA, Weaver FM, Stern M, et al. Pallidal versus sub-
are markedly different from those of both the Godinho et thalamic deep-brain stimulation for Parkinson’s disease. N
al. study15 and our present study, both of which included Engl J Med. 2010;362(22):2077–2091.
​ ​
only unilateral procedures.   2. Deuschl G, Schade-Brittinger C, Krack P, et al. A random-
Among the various complications noted in the present ized trial of deep-brain stimulation for Parkinson’s disease. N
study, we consider the decreased response to levodopa to Engl J Med. 2006;355(9):896–908.​ ​
  3. Obeso JA, Olanow CW, Rodriguez-Oroz MC, et al. Deep-
be the most serious adverse effect, and this has not been brain stimulation of the subthalamic nucleus or the pars in-
reported by previous studies using PTT. Somnolence was terna of the globus pallidus in Parkinson’s disease. N Engl J
noted in 60% of patients, consistent with previous re- Med. 2001;345(13):956–963.
​ ​
ports.6,15 Apraxia of lid opening was noted in 1 patient.   4. Lozano AM, Lang AE, Galvez-Jimenez N, et al. Effect of
While the present study supports the safety of unilateral GPi pallidotomy on motor function in Parkinson’s disease.
PTT/campotomy, the safety of bilateral PTT/campotomy Lancet. 1995;346(8987):1383–1387.
​ ​
remains unknown. In our previous study of unilateral PTT   5. Laitinen LV, Bergenheim AT, Hariz MI. Leksell’s postero-
ventral pallidotomy in the treatment of Parkinson’s disease. J
combined with contralateral pallidotomy for dystonia, we Neurosurg. 1992;76(1):53–61.
​ ​
reported that more than half (54.5%) of the patients expe-   6. Gallay MN, Moser D, Rossi F, et al. MRgFUS pallidotha-
rienced significant prolonged hypophonia.26 In the present lamic tractotomy for chronic therapy-resistant Parkinson’s
study, only 2 patients (20%) reported significant hypopho- disease in 51 consecutive patients:​single center experience.
nia, which persisted at the 1-year evaluation. However, the Front Surg. 2020;6:76. ​ ​
degree of hypophonia was very mild and did not interfere   7. Jung NY, Park CK, Kim M, et al. The efficacy and limits of
with their daily communication. Patients with PD are magnetic resonance-guided focused ultrasound pallidotomy
likely to have speech and swallowing problems includ- for Parkinson’s disease:​a Phase I clinical trial. J Neurosurg.
2019;130(6):1853–1861.
​ ​
ing hypophonia, which suggests that they are vulnerable   8. Martínez-Fernández R, Rodríguez-Rojas R, Del Álamo M,
to surgical complications following bilateral procedures. et al. Focused ultrasound subthalamotomy in patients with
Similarly, Gallay et al.6 reported a significant worse rate of asymmetric Parkinson’s disease:​a pilot study. Lancet Neurol.
hypophonia after bilateral PTT using focused ultrasound. 2018;17(1):54–63.
​ ​
Therefore, we would not perform bilateral PTT for PD.   9. Chung BS, Park JS. Whole course of pallidothalamic tracts
The effects of unilateral PTT are expected to decrease identified on the sectioned images and surface models. Clin
gradually with time. For those patients who require addi- Anat. 2020;33(1):66–76.
​ ​
tional surgery to provide further improvement, DBS may 10. Gallay MN, Jeanmonod D, Liu J, Morel A. Human pallido-
thalamic and cerebellothalamic tracts:​anatomical basis for
be better to ensure both safety and efficacy. We did not functional stereotactic neurosurgery. Brain Struct Funct.
evaluate psychiatric and neuropsychological conditions af- 2008;212(6):443–463.
​ ​
ter surgery. While Godinho et al. have previously report- 11. Neudorfer C, Maarouf M. Neuroanatomical background and
ed that these conditions did not significantly deteriorate, functional considerations for stereotactic interventions in the
Gallay et al. reported a transient anxio-depressive state in H fields of Forel. Brain Struct Funct. 2018;223(1):17–30.
​ ​
1 patient (2%).6,15 Further, Aufenberg et al. reported that 12. Aufenberg C, Sarnthein J, Morel A, et al. A revival of Spie-
anxio-depressive and reduced activity states were ob- gel’s campotomy:​long term results of the stereotactic palli-
dothalamic tractotomy against the parkinsonian thalamocor-
served in 6 (14.6%) and 4 (9.8%) patients, respectively.12 A tical dysrhythmia. Thalamus Relat Syst. 2005;3(2):121–132.
​ ​
larger number of patients is needed to elucidate the effects 13. Spiegel EA, Wycis HT, Szekely EG, et al. Campotomy in
of PTT on patients’ psychiatric and psychological states. various extrapyramidal disorders. J Neurosurg. 1963;20: ​ ​
This study has some limitations. The open-label trial 871–884.
design may have influenced reporting and the measure- 14. Magara A, Bühler R, Moser D, et al. First experience with

804 J Neurosurg Volume 135 • September 2021

 
*39 *
 Horisawa et al.

MR-guided focused ultrasound in the treatment of Parkin- 25. Intemann PM, Masterman D, Subramanian I, et al. Staged
son’s disease. J Ther Ultrasound. 2014;​2:​11. bilateral pallidotomy for treatment of Parkinson disease. J
15. Godinho F, Magnin M, Filho PT, et al. Stereotactic lesion in Neurosurg. 2001;​94(3):​437–444.
the Forel’s field H:​a 2-years prospective open-label study on 26. Horisawa S, Fukui A, Tanaka Y, et al. Pallidothalamic trac-
motor and nonmotor symptoms, neuropsychological func- totomy (Forel’s field H1-tomy) for dystonia:​preliminary re-
tions, and quality of life in Parkinson disease. Neurosurgery. sults. World Neurosurg. 2019;​129:​e851–e856.
2019;​85(4):​E650–E659.
16. Rajput AH, Voll A, Rajput ML, et al. Course in Parkinson
disease subtypes:​a 39-year clinicopathologic study. Neurol- Disclosures
ogy. 2009;​73(3):​206–212. The authors report no conflict of interest concerning the materi-
17. Zhang J, Wei L, Hu X, et al. Akinetic-rigid and tremor- als or methods used in this study or the findings specified in this
dominant Parkinson’s disease patients show different patterns paper.
of intrinsic brain activity. Parkinsonism Relat Disord. 2015;​
21(1):​23–30. Author Contributions
18. Schillaci O, Chiaravalloti A, Pierantozzi M, et al. Different
patterns of nigrostriatal degeneration in tremor type versus Conception and design: Horisawa, Iijima, Kawamata, Taira.
the akinetic-rigid and mixed types of Parkinson’s disease Acquisition of data: Horisawa, Fukui, Yamahata, Tanaka,
at the early stages:​molecular imaging with 123I-FP-CIT Kuwano, Momosaki, Iijima, Nanke, Taira. Analysis and
SPECT. Int J Mol Med. 2011;​28(5):​881–886. interpretation of data: Horisawa, Taira. Drafting the article:
19. Kaasinen V, Kinos M, Joutsa J, et al. Differences in striatal Horisawa. Critically revising the article: Horisawa. Reviewed
dopamine transporter density between tremor dominant and submitted version of manuscript: Horisawa. Approved the final
non-tremor Parkinson’s disease. Eur J Nucl Med Mol Imag- version of the manuscript on behalf of all authors: Horisawa.
ing. 2014;​41(10):​1931–1937. Statistical analysis: Horisawa, Fukui. Administrative/technical/
20. Benninger DH, Thees S, Kollias SS, et al. Morphological dif- material support: Iijima, Taira. Study supervision: Iijima,
ferences in Parkinson’s disease with and without rest tremor. Kawamata, Taira.
J Neurol. 2009;​256(2):​256–263.
21. Rajput AH, Rajput ML, Ferguson LW, Rajput A. Baseline Supplemental Information
motor findings and Parkinson disease prognostic subtypes. Videos
Neurology. 2017;​89(2):​138–143. Video 1. https://vimeo.com/453250738.
22. Rajput AH, Sitte HH, Rajput A, et al. Globus pallidus dopa- Video 2. https://vimeo.com/453250752.
mine and Parkinson motor subtypes:​clinical and brain bio- Video 3. https://vimeo.com/453250758.
chemical correlation. Neurology. 2008;​70(16 Pt 2):​1403–1410.
23. Tsuboi T, Lemos Melo Lobo Jofili Lopes J, Patel B, et al. Par- Correspondence
kinson’s disease motor subtypes and bilateral GPi deep brain
Shiro Horisawa: Neurological Institute, Tokyo Women’s Medical
stimulation:​one-year outcomes. Parkinsonism Relat Disord.
University, Tokyo, Japan. [email protected].
2020;​75:​7–13.
24. Katz M, Luciano MS, Carlson K, et al. Differential effects of
deep brain stimulation target on motor subtypes in Parkin-
son’s disease. Ann Neurol. 2015;​77(4):​710–719.

J Neurosurg Volume 135 • September 2021 805

 
*39 *
 TECHNICAL NOTE
J Neurosurg 135:806–814, 2021

DBS targeting for essential tremor using intersectional

dentato-rubro-thalamic tractography and direct proton
density visualization of the VIM: technical note on 2 cases
*Melanie A. Morrison, PhD,2 Anthony T. Lee, MD, PhD,1 Alastair J. Martin, PhD,2
Cameron Dietiker, MD,3 Ethan G. Brown, MD,3 and Doris D. Wang, MD, PhD1
1
Department of Neurological Surgery, Weill Institute for Neurosciences, University of California, San Francisco; 2Department of
Radiology & Biomedical Imaging, University of California, San Francisco; and 3Department of Neurology, Movement Disorders
and Neuromodulation Center, Weill Institute for Neurosciences, University of California, San Francisco, California

OBJECTIVE  Direct visualization of the ventral intermediate nucleus (VIM) of the thalamus on standard MRI sequences
remains elusive. Therefore, deep brain stimulation (DBS) surgery for essential tremor (ET) indirectly targets the VIM
using atlas-derived consensus coordinates and requires awake intraoperative testing to confirm clinical benefits. The
objective of this study was to evaluate the utility of proton density (PD)–weighted MRI and tractography of the intersect-
ing dentato-rubro-thalamic tract (DRTT) for direct “intersectional” targeting of the VIM in ET.
METHODS  DBS targets were selected by identifying the VIM on PD-weighted images relative to the DRTT in 2 patients
with ET. Tremor reduction was confirmed with intraoperative clinical testing. Intended target coordinates based on the
direct intersectional targeting technique were compared with consensus coordinates obtained with indirect targeting.
Pre- and postoperative tremor scores were assessed using the Fahn-Tolosa-Marin tremor rating scale (TRS).
RESULTS  Planned DBS coordinates based on direct versus indirect targeting of the VIM differed in both the antero-
posterior (range 0 to 2.3) and lateral (range −0.7 to 1) directions. For 1 patient, indirect targeting—without PD-weighted
visualization of the VIM and DRTT—would have likely resulted in suboptimal electrode placement within the VIM. At the
3-month follow-up, both patients demonstrated significant improvement in tremor symptoms subjectively and according
to the TRS (case 1: 68%, case 2: 72%).
CONCLUSIONS  Direct intersectional targeting of the VIM using PD-weighted imaging and DRTT tractography is a
feasible method for DBS placement in patients with ET. These advanced targeting techniques can supplement awake
intraoperative testing or be used independently in asleep cases to improve surgical efficiency and confidence.
https://thejns.org/doi/abs/10.3171/2020.8.JNS201378
KEYWORDS  essential tremor; deep brain stimulation; surgical targeting; ventral intermediate nucleus;
dentato-rubro-thalamic tractography; proton density magnetic resonance imaging; electrophysiological mapping;
functional neurosurgery; surgical technique

E
ssential tremor (ET) is the most common move- respond to medical treatment, and tremor reduction aver-
ment disorder, with a global lifetime prevalence of ages 50%.5 Surgical approaches used for medically refrac-
0.9% and a prevalence of 4.6% in people 65 years tory ET include either ablative procedures using radiofre-
or older.1 Classic ET typically manifests as postural or ki- quency, radiosurgery, or MRI-guided focused ultrasound
netic tremor most commonly in the hands and forearms, (MgFUS) to lesion the ventral intermediate nucleus (VIM)
with or without head tremor, and less commonly in the of the thalamus or a neuromodulation procedure through
lower limbs and voice.2,3 Primidone and propranolol are deep brain stimulation (DBS). Of these options, DBS for
the two pharmacological agents with the best level of ET is considered the current gold standard with tremor
evidence for treating ET;4 however, only 50% of patients reductions of 60%–90% and low complication rates.4–8

ABBREVIATIONS  AC-PC = anterior commissure–posterior commissure; AP = anteroposterior; DBS = deep brain stimulation; DRTT = dentato-rubro-thalamic tract; DTI =
diffusion tensor imaging; ET = essential tremor; MgFUS = MRI-guided focused ultrasound; PD = proton density; TRS = Fahn-Tolosa-Marin tremor rating scale; VIM = ventral
intermediate nucleus.
SUBMITTED  April 20, 2020.  ACCEPTED  August 3, 2020.
INCLUDE WHEN CITING  Published online January 15, 2021; DOI: 10.3171/2020.8.JNS201378.
* M.A.M. and A.T.L. contributed equally to this work.

806 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*40 *
 Morrison et al.

However, habituation and long-term adverse effects such sequences were derived from optimized T2-weighted se-
as dysarthria and gait ataxia occur in a subset of patients quences using parameters that have previously been shown
after several years of chronic stimulation.9 Whether stim- to be effective for visualizing the VIM.28,41 To confirm elec-
ulation of specific thalamic nuclei or direct targeting of trode placement, standard T1- and T2-weighted sequences
fiber tracts can improve long-term outcomes is unclear.10 were reacquired on the 1.5-T Philips scanner the day after
The classic DBS target for ET is the VIM, a region ap- surgery. Imaging parameters are listed in Table 1.
proximately 4 × 4 × 6 mm in size and not directly visible Subjects also underwent CT imaging prior to surgery
on traditional MRI sequences.11–13 The VIM receives input on a GE Discovery CT750 HD scanner. Whole-brain
principally from the contralateral dentate nucleus through axial images were reconstructed using the BonePlus filter
the dentato-rubro-thalamic tract (DRTT) and sends ef- and reformatted along the anterior commissure–posterior
ferent connections to the motor and premotor cortices.14 commissure (AC-PC) plane with a voxel resolution of 0.5
Precise targeting of the VIM is essential for effective DBS × 0.5 × 0.6 mm (spacing between slices = 0.6 mm) and a
outcomes, as deviations greater than 2 mm from the opti- matrix of 512 × 512.
mal lead placement can halve the likelihood of achieving
good tremor control.15,16 Optimal placement is achieved DRTT Reconstruction
via a combination of indirect targeting using atlas-derived MR fusion of all pairs of sequences was performed
consensus coordinates superimposed onto a patient’s MRI using the image fusion module of the Brainlab Elements
scan and intraoperative recording and testing. However, software (version 4.0). Next, automated segmentation of
intraoperative testing depends on patient participation in deep brain structures was performed using the brain map-
and tolerance of the awake procedure, and indirect tar- ping module of the software. For both cases, the automati-
geting is subject to significant anatomical and functional cally segmented contour of each nucleus of interest—den-
variability among patients.17–21 More recent approaches tate nucleus and red nucleus, and VIM—was reviewed and
attempt to identify the VIM by demarcating its location manually adjusted using T2-weighted and PD-weighted
based on prospective tractography of the medial lemnis- image sequences, respectively. As previously described,
cus and pyramidal tract, but positive identification of the the VIM was visualized as a coronally oriented hypoin-
VIM has remained elusive.22,23 tense ovoid structure extending medially from the internal
Advanced MRI sequences may be better suited to iden- capsule.24 Using the fiber tracking module, DTI data were
tify the VIM.24 High-resolution proton density (PD) se- preprocessed (motion and distortion correction, genera-
quences seek to capture the most basic of MRI measure- tion of tensor image, fractional anisotropy, and apparent
ments, the apparent concentration of water protons in each diffusion maps), and deterministic tractography was per-
voxel. Tissue with a higher density of protons has a greater formed using the fiber assignment by continuous tracking
transverse component of magnetization and therefore a (FACT)42 and tensor deflection (TEND) algorithms.43
brighter signal.25 PD-weighted sequences have success- To isolate the DRTT, the contralateral dentate nucleus,
fully identified multiple subcortical regions, including the ipsilateral red nucleus, and VIM were used as seed re-
centromedian nucleus, globus pallidus, nucleus ventrocau- gions. The minimum fiber length was set to 30 mm, frac-
dalis parvocellularis externus, and VIM.18,26–29 tional anisotropy was set to 0.2, and maximum angulation
Multiple studies have also shown efficacious tremor con- was set to 50°. These thresholds were chosen based on
trol with targeting of nearby thalamic regions including the common practices in DBS surgical planning to effectively
zona incerta, posterior subthalamic nucleus, and radiation remove erroneous fibers and non–white matter voxels.34,44
prelemniscalis.11,30–32 These centers may together comprise Color-coded fractional anisotropy maps were used for vi-
an oscillatory network responsible for tremor generation,33 sual guidance.
and tractography studies have highlighted the DRTT as a
common route shared between these network hubs.34,35 In-
deed, studies have supported the use of targeting the DRTT Intersectional VIM/DRTT Targeting and Comparison With
for tremor control with DBS for Parkinson’s disease and Indirect Targeting
dystonia36,37 and MgFUS for ET.38–40 However, no study Using the AC-PC plane as a reference, DBS target-
has used a combination of PD-weighted sequences and the ing was performed by identifying the VIM on axial PD-
DRTT to target the VIM in patients with ET. In this study, weighted image slices, relative to the isolated DRTT.
we present a novel technique using optimized PD-weighted Adjustments to the target were made to ensure that the
sequences and tractography for direct “intersectional” tar- bottom DBS contact was in contact with both the DRTT
geting of both the VIM and DRTT in 2 patients with ET. and VIM. Once the DBS lead (Vercise Cartesia, Boston
Scientific Inc.) was placed using standard stereotactic
techniques, intraoperative clinical testing was performed
Methods to ensure improvement in tremors without side effects. Af-
Imaging Protocol ter the patient awakened from sedation, intraoperative test
Preoperative MRI examinations were performed on ei- stimulation was performed at 1–8+ at 130 Hz at 60-μsec
ther a 3-T GE Signa scanner or a 1.5-T Philips Achieva pulse width with increasing amplitudes until side effects
scanner using 8-channel head coils. In addition to standard such as hand and arm paresthesias or dysarthria was de-
T1- and T2-weighted sequences, optimized PD-weighted tected. The absence of dysarthria up to 4.5 mA with im-
and diffusion tensor imaging (DTI) sequences were ac- proved hand tremors at 1 mA was determined to be within
quired, with whole-brain coverage. The PD-weighted the therapeutic window. Planned coordinates derived from

J Neurosurg Volume 135 • September 2021 807

 
*40 *
Morrison et al.

TABLE 1. MRI protocol

Field Strength TR TE TI No. of Flip FOV Voxel No. of Acquisition
& Sequence Matrix (msec) (msec) (msec) ETL Averages Angle (cm) Size (mm) Slices Time (mins:sec)
1.5 T
  3D T1 TFE 192 × 178 20 3 — 38 1 30° 26.0 × 22.2 1.35 × 1.35 × 1.5 120 8:49
  Axial T2 TSE 384 × 262 3000 90 — 16 6 90° 26.0 × 22.2 0.68 × 0.99 × 2 30 8:42
  Axial PD TSE 220 × 180 3000 20 — 5 2 90° 22.0 × 18.0 1×1×2 24 7:18
3T
  3D T1 IR SPGR 256 × 256 8 3 450 1 1 11° 25.6 × 25.6 0.5 × 0.5 × 1 208 3:43
  Axial T2 FSE 320 × 260 5494 84 — 15 4 115° 17.6 × 17.6 0.4 × 0.4 × 2 39 3:13
  Axial PD FSE 320 × 320 3000 15 — 5 2 115° 19.2 × 19.2 0.4 × 0.4 × 3 26 3:90
  Axial DTI* 128 × 128 7765 71 — 1 1 90° 28.0 × 28.0 1.09 × 1.09 × 2 91 5:49
ETL = echo train length; FOV = field of view; FSE = fast spin echo; IR-SPGR = inversion recovery–spoiled gradient recalled echo; TFE = turbo field echo; TI = inversion
time; TSE = turbo spin echo.
* Fifty-five directions, b-value = 2000.

direct targeting were compared with those obtained with Results

indirect targeting based on consensus coordinates (10 mm The patient in case 1 was a 55-year-old left-handed man
lateral from the lateral wall of the third ventricle and 6 with 40 years of ET, starting initially in both hands, with
mm anterior to the PC).15 eventual progression to his head and voice. His ET had a
significant impact on activities of daily living, including an
DBS Lead Localization inability to independently feed and shave himself and open
After lead placement, patients underwent intraoperative jars. His symptoms were previously responsive to alcohol
CT scanning using O-arm II (Medtronic Inc.). The intra- and were subsequently refractory to several medications,
operative CT scan was fused with preoperative MRI, DBS including primidone, propranolol, gabapentin, and topira-
leads were detected using the lead detection module of the mate (Table 2). His preoperative imaging was performed at
Brainlab software, and segmentation was verified. Final 3 T. The patient in case 2 was a 72-year-old right-handed
lead locations were compared with coordinates obtained woman with 20 years of ET in both hands that rendered
using direct and indirect targeting techniques, based on her unable to independently cook, feed herself, sew, and
their respective locations at the AC-PC plane. apply makeup. Several medications had failed, including
propranolol, primidone, and clonazepam. Given that her
Neurological Evaluation of Tremor symptoms were subjectively worse in her right hand, the
The severity of pretreatment versus 3-month posttreat- patient first underwent unilateral implantation of a left-sid-
ment tremor was assessed by each patient’s movement ed VIM DBS device with standard indirect targeting prior
disorder neurologist using the Fahn-Tolosa-Marin tremor to implantation of a right-sided VIM DBS device 8 months
rating scale (TRS). The examination is divided into three later with the direct intersectional approach described in
sections—A, B, and C—which evaluate tremor location this study. Her DTI sequences were acquired at 3 T prior to
and severity, motor tasks and function, and disability re- the initial DBS surgery, and her preoperative imaging for
lated to tremor, respectively. Higher TRS scores indicate right-sided VIM DBS was performed at 1.5 T to minimize
overall greater tremor severity. the electrode artifact and per device safety guidelines.

TABLE 2. Patient characteristics and tremor scores

TRS Scores
Duration
Case Age of Sx Targeting Pretreatment Posttreatment % Improvement
No. (yrs) Sex (yrs) Side Method Side A B C Total A B C Total in Sx
Overall 24 16 24 64 10 4 8 22 68%
1 55 M 40 Bilat Direct Rt 8 8 — 16 5 4 — 9 44%
Direct Lt* 8 10 — 18 4 3 — 7 61%
Overall 6 12 11 29 3 3 2 8 72%
2 72 F 20 Staged (lt → rt) Indirect Rt* 2 6 — 8 2 1 — 3 63%
         
Direct Lt 3 6 — 9 1 2 — 3 67%
A = tremor location and severity; B = motor task and functions; C = disability; Sx = symptoms.
* Dominant hand.

808 J Neurosurg Volume 135 • September 2021

*40 *
 Morrison et al.

Both patients exhibited significant pretreatment tremors as tional targeting outperforming indirect targeting. First,
assessed using the TRS. The patient in case 1 had a total the indirect targeting method used in the first stage of the
TRS score of 64 (A, tremor location and severity: 24; B, patient’s left VIM resulted in electrode placement that was
motor task and functions: 16; C, disability: 24), and the pa- posterior and lateral to the DRTT (Fig. 3A, indirect tar-
tient in case 2 had a total TRS score of 29 (A: 6, B: 12, C: geting in orange). Optimal stimulation parameters utilized
11). Her TRS score for nondominant symptoms was 9. channels 2–4 (2 mm above the tip of the electrode), which
Preoperative MRI showed that superior visualization coincidentally placed the stimulation directly at the inter-
of the VIM for the patient in case 1 was achieved with PD- section of the DRTT and VIM (Table 4, Fig. 3B), thereby
weighted MRI as compared to T2-weighted MRI at 3 T compensating for the initial placement by indirect target-
(Fig. 1A). Notably, visualization of the VIM for the patient ing. Retrospective planning of the electrode trajectory us-
in case 2 was also superior at 1.5 T using PD-weighted ing direct intersectional targeting (Fig. 3B, direct target-
MRI (Fig. 1B) compared to T2 sequences. Reconstruction ing in green) would have centered the tip of the electrode
of the DRTT with image fusions, anatomical segmenta- directly at this intersection. For implantation of the right
tion, and fiber tracking was performed in under 30 minutes VIM, indirect targeting would have resulted in placement
(see Methods). Figure 2A and B shows the planned trajec- of the lead 0.8 mm lateral and 1 mm posterior to the right
tory of electrodes using direct intersectional targeting on VIM (Table 3, Fig. 3C, direct targeting in orange, indirect
a 3D multiplanar representation of the DRTT with the cor- targeting in yellow).
responding seed regions of interest. Final DRTT recon- Both patients experienced significant improvement in
structions with the planned electrode trajectories, overlaid tremor symptoms subjectively and when reevaluated with
onto PD-weighted AC-PC image slices, are shown in Fig. the TRS. The postoperative TRS score with DBS was 22
2C and D, respectively. for the patient in case 1 (A: 10, B: 4, C: 8; 68% improve-
Both patients underwent uncomplicated implantation ment from preoperatively) and 8 for the patient in case 2
of 8-contact directional leads (Vercise Cartesia). After (A: 3, B: 3, C: 2; 72% improvement from preoperatively;
lead localization using intraoperative CT fused with pre- Table 2). When stratifying by side, the TRS scores for the
operative MRI, we directly compared the AC-PC coordi- patient in case 1 improved by 44% and 61% for the right
nates based on indirect targeting (consensus coordinates), and left motor subscores, respectively. For the patient in
direct intersectional targeting (PD-VIM/DRTT), and final case 2, her right-sided symptoms (indirectly targeted, with
electrode location at the AC-PC plane based on direct tar- compensation using channels 2–4 stimulation) improved
geting followed by intraoperative validation (Fig. 2E and by 63%, and her left-sided symptoms (directly targeted)
F, Table 3). For the patient in case 1, the indirect coordi- improved by 67%. There was no clear correlation between
nates were −12.0, −8.1, 0 and 12.0, −8.1, 0 in the lateral, distance from direct targeting and symptom improvement
anteroposterior (AP), and vertical planes for the left and (R2 = 0.28, p = 0.47). The patient in case 1 experienced
right VIM, respectively (all coordinates given in lateral, slightly more effortful speech following DBS that was re-
AP, vertical planes hereafter). Direct coordinates for the versed when the stimulation was turned off. The patient in
left VIM were −12.5, −7.5, 0, and coordinates for the right case 2 did not experience any side effects following either
VIM were 12.5, −8.1, 0. For the patient in case 2, the indi- DBS surgeries.
rect coordinates for the left VIM were −13.0, −6.3, 0 and
13.0, −6.3, 0 for the right VIM. Because the left VIM had
been previously implanted with indirect coordinates, the Discussion
direct coordinates for the left VIM were obtained retro- Our study demonstrates a novel direct VIM targeting
spectively based on tractography from preoperative imag- approach using PD-weighted MRI and tractography of
ing and PD-weighted imaging prior to the second-side sur- the DRTT. Combined targeting of the VIM and DRTT
gery. The direct coordinates for her left VIM were −12.0, resulted in significant reductions in tremor. Notably, this
−4.0, 0, and the direct coordinates for her right VIM were intersectional approach functionally and anatomically
12.3, −5.3, 0. The difference between direct and indirect outperformed classic consensus coordinates in 1 patient.
coordinates for all four VIM targets ranged from −0.7 to 1 Our method added minimal time to preoperative plan-
in the lateral plane and 0 to 2.3 in the AP plane (Table 3). ning, was feasible in both a bilateral (case 1) and staged
Following intraoperative awake testing, the final elec- (case 2) fashion, and was compatible with 3-T (case 1) and
trode coordinates at the AC-PC plane for the patient in 1.5-T (case 2) MRI.
case 1 were −12.7, −5.7, 0 and 12.7, −6.1, 0 for the left and Case 2 highlights the small but potentially meaning-
right VIM, respectively. For the patient in case 2, the post- ful clinical differences that may arise from intersectional
operative coordinates were −12.8, −4.6, 0 for the left VIM targeting of the DRTT and VIM on PD-weighted MRI
and 12.2, −5.3, 0 for the right VIM. The distances from the compared to targeting with indirect coordinates. Indirect
electrode to the target at the AC-PC plane for all four elec- targeting of the left VIM in a prior procedure led to elec-
trodes were smaller when comparing the final electrode trode placement that necessitated the use of more superior
position to the planned direct target versus indirect target. channels (2–4) on the DBS electrode, coincidentally plac-
The distances between the target and final electrode po- ing stimulation within the DRTT (Fig. 3). Furthermore, an
sition using direct targeting were 0.1 to 2 mm, whereas indirect targeting approach for the right VIM on the index
the distances using indirect targeting were 1.3 to 2.5 mm procedure would have placed the electrode posterolateral
(Table 3). to the optimal VIM site. An optimal electrode trajectory
Case 2 highlighted two examples of direct intersec- with the direct intersectional method may result in faster

J Neurosurg Volume 135 • September 2021 809

 
*40 *
Morrison et al.

FIG. 1. Superior visualization of the VIM using PD-weighted MRI at 3 T (A) in the patient in case 1, who underwent bilateral DBS surgery, at 1.5 T (B) in
the patient in case 2, who underwent a staged DBS procedure, and at 3 T (C) in a third patient who underwent a staged DBS procedure. Axial views of
PD-weighted and T2-weighted images are shown with the VIM, DRTT, and internal capsule outlined in pink, purple, and green, respectively. Images in
the third patient have been included here to provide an additional case example of PD visualization. This patient did not have DTI data and is therefore
not otherwise mentioned in this study. Figure is available in color online only.

810 J Neurosurg Volume 135 • September 2021

*40 *
 Morrison et al.

FIG. 2. Planned electrode trajectories with direct intersectional targeting and final electrode placement. Tractographies of the bilateral DRTTs for the
patients in cases 1 (A) and 2 (B) based on manual segmentations of the motor cortex, VIM, red nucleus (RN), and dentate nucleus (DN) are shown.
The DRTTs are also shown mapped onto axial PD-weighted MR images with the electrode trajectories (C and D for cases 1 and 2, respectively), and
final electrode placements for cases 1 (E) and 2 (F) are illustrated by the reconstructed electrodes. The yellow arrow indicates the electrode-induced
susceptibility artifact as a result of the left VIM DBS implantation surgery 8 months earlier. Figure is available in color online only.

J Neurosurg Volume 135 • September 2021 811

 
*40 *
Morrison et al.

TABLE 3. Comparison of planned indirect and direct coordinates with postoperative electrode locations following direct targeting and
intraoperative awake testing

Side Planned Indirect Planned Direct Postop Electrode Location Distance From
Case of Coordinates (mm) Coordinates (mm) at the AC-PC Plane (mm) Electrode to Target (mm)
No. VIM Lat AP Vertical Lat AP Vertical Lat AP Vertical Indirect Target Direct Target
1 Lt −12.0 −8.1 0 −12.5 −7.5 0 −12.7 −5.7 0 2.5 1.8
Rt 12.0 −8.1 0 12.5 −8.1 0 12.7 −6.1 0 2.1 2.0
2 Lt −13.0 −6.3 0 −12.0* −4.0* 0* −12.8 −4.6 0 1.7 1
Rt 13.0 −6.3 0 12.3 −5.3 0 12.2 −5.3 0 1.3 0.1
* Retrospectively obtained based on tractography from preoperative imaging and PD-weighted imaging prior to second-side surgery.

awake validation intraoperatively, decrease the risk of side may be similar in some cases, there can be significant be-
effects from stimulation of nearby pathways (such as the tween-patient variations in structural anatomy.17–21 Previ-
internal capsule), minimize stimulation current, and may ous studies have used PD-weighted MRI and the DRTT in-
even facilitate asleep DBS surgery for ET. At a minimum, dependently in the planning of VIM DBS and MgFUS and
knowledge of the DRTT may decrease programming time reported reliable target identification and clinical improve-
by knowing which channels are closer to the intersection ment in ET.34,36–40 However, a combined direct targeting
of the VIM and DRTT. Prior to trials of intersectional tar- approach as shown here can further increase the surgeon’s
geting for ET DBS, larger cohorts will be needed to vali- ability to offer patient-specific approaches by providing
date repositioning attempts during intraoperative awake the DRTT as another directly visible anatomical landmark
validation, a shorter programming time in the outpatient to target. Furthermore, the DRTT itself has been suggested
setting, and stimulation current with intersectional target- to play an essential role in tremor-generating oscillatory
ing compared to that with indirect targeting. networks.33–35 By being able to directly identify both the
While coordinates of the indirect and direct targeting VIM and DRTT, further research will be able to determine

FIG. 3. Case 2. Illustrative functional and anatomical advantages of direct intersectional targeting. A: Indirect targeting used in
the first stage of the patient’s left (L) VIM resulted in electrode placement that was posterolateral to the DRTT (orange, indirect
targeting; purple, direct targeting; pink, DRTT). B: Direct targeting would have resulted in a trajectory that was closer to the final
electrode position (green, direct targeting; yellow, indirect targeting; purple, DBS lead). Optimal DBS programming utilized chan-
nels (Ch) 2–4, which directly stimulated the DRTT (inset). C: An indirect targeting approach would have placed the DBS electrode
posterolateral to the intersection of the VIM and DRTT (yellow, indirect targeting; orange, direct targeting; blue, DBS lead). R =
right. Figure is available in color online only.

812 J Neurosurg Volume 135 • September 2021

*40 *
 Morrison et al.

TABLE 4. Stimulation parameters   6. Schuurman PR, Bosch DA, Bossuyt PM, et al. A comparison of
continuous thalamic stimulation and thalamotomy for suppres-
Pulse sion of severe tremor. N Engl J Med. 2000;​342(7):​461–468.
Case Channel+/− Amplitude Width Frequency   7. Vaillancourt DE, Sturman MM, Verhagen Metman L, et al.
No. Side (% of total current) (mA) (µsec) (Hz) Deep brain stimulation of the VIM thalamic nucleus modifies
several features of essential tremor. Neurology. 2003;​61(7):​
1 Lt 1− (100%) 3.0 60 143 919–925.
Rt 9− (80%), 10− (20%) 4.0 80 143   8. Koller W, Pahwa R, Busenbark K, et al. High-frequency uni-
2 Lt 1−, 2+ (34%), 1.8 60 139 lateral thalamic stimulation in the treatment of essential and
3+ (33%), 4+ (33%) parkinsonian tremor. Ann Neurol. 1997;​42(3):​292–299.
  9. Paschen S, Forstenpointner J, Becktepe J, et al. Long-term
Rt 9− (100%) 1.7 90 139 efficacy of deep brain stimulation for essential tremor:​An
+/− denotes cathode and anode modes, respectively. observer-blinded study. Neurology. 2019;​92(12):​e1378–e1386.
10. Lozano AM, Lipsman N, Bergman H, et al. Deep brain
stimulation:​current challenges and future directions. Nat Rev
Neurol. 2019;​15(3):​148–160.
how stimulation of the VIM and DRTT, independently and 11. Benabid AL, Pollak P, Gervason C, et al. Long-term suppres-
sion of tremor by chronic stimulation of the ventral interme-
collectively, may affect tremor control, adaptation to stim- diate thalamic nucleus. Lancet. 1991;​337(8738):​403–406.
ulation, and the minimization of side effects. Patients who 12. Lozano AM. Vim thalamic stimulation for tremor. Arch Med
experience unsatisfactory ET control using standard indi- Res. 2000;​31(3):​266–269.
rect targeting methods may benefit from revision that tar- 13. Hirai T, Nagaseki Y, Wada H, et al. Cell sparse zones in the
gets both the VIM and DRTT. Practically, the acquisition ventrolateral thalamic mass in humans, monkeys and cats:​
of PD and DTI sequences requires less than 15 minutes of their special reference to kinesthetic neurons. Appl Neuro-
additional scanning time—and having more than one se- physiol. 1985;​48(1-6):​316–323.
14. Ilinsky IA, Kultas-Ilinsky K. Motor thalamic circuits in
quence available for direct targeting is valuable especially primates with emphasis on the area targeted in treatment of
in the event of motion artifacts that could deem some data movement disorders. Mov Disord. 2002;​17(suppl 3):​S9–S14.
less reliable. 15. Papavassiliou E, Rau G, Heath S, et al. Thalamic deep brain
Notable limitations to this study stem primarily from stimulation for essential tremor:​relation of lead location to
its small sample size, precluding our ability to find a re- outcome. Neurosurgery. 2004;​54(5):​1120–1130.
lationship between tremor control and electrode distance 16. Morishita T, Hilliard JD, Okun MS, et al. Postoperative lead
to the direct target, as well as differences in tremor con- migration in deep brain stimulation surgery:​Incidence, risk
factors, and clinical impact. PLoS One. 2017;​12(9):​e0183711.
trol between direct and indirect timing. For case 2, the 17. Nowinski WL, Belov D, Thirunavuukarasuu A, Benabid
use of DTI data obtained prior to stage 1 may also have AL. A probabilistic functional atlas of the VIM nucleus
confounded our targeting of the DRTT, though the pre- constructed from pre-, intra- and postoperative electrophysi-
operative stage 2 MRI did not show any gross anatomical ological and neuroimaging data acquired during the surgical
differences in the intervening 8 months. treatment of Parkinson’s disease patients. Stereotact Funct
Neurosurg. 2005;​83(5-6):​190–196.
18. O’Gorman RL, Shmueli K, Ashkan K, et al. Optimal MRI
Conclusions methods for direct stereotactic targeting of the subthalamic
PD-weighted MRI allows for direct visualization of nucleus and globus pallidus. Eur Radiol. 2011;​21(1):​130–136.
the VIM. Direct intersectional targeting of the VIM and 19. Tsolaki E, Downes A, Speier W, et al. The potential value of
probabilistic tractography-based for MR-guided focused ul-
DRTT is feasible and provides effective ET control. Com- trasound thalamotomy for essential tremor. Neuroimage Clin.
bined targeting of the VIM and DRTT may result in more 2018;​17:​1019–1027.
optimal DBS electrode placement. 20. Vayssiere N, Hemm S, Cif L, et al. Comparison of atlas- and
magnetic resonance imaging-based stereotactic targeting of
the globus pallidus internus in the performance of deep brain
References stimulation for treatment of dystonia. J Neurosurg. 2002;​
  1. Louis ED, Ferreira JJ. How common is the most common 96(4):​673–679.
adult movement disorder? Update on the worldwide 21. Tian Q, Wintermark M, Jeffrey Elias W, et al. Diffusion MRI
prevalence of essential tremor. Mov Disord. 2010;​25(5):​ tractography for improved transcranial MRI-guided focused
534–541. ultrasound thalamotomy targeting for essential tremor. Neu-
  2. Deuschl G, Bain P, Brin M. Consensus statement of the roimage Clin. 2018;​19:​572–580.
Movement Disorder Society on Tremor. Mov Disord. 1998;​ 22. Krishna V, Sammartino F, Agrawal P, et al. Prospective
13(suppl 3):​2–23. tractography-based targeting for improved safety of focused
  3. Bhatia KP, Bain P, Bajaj N, et al. Consensus statement on the ultrasound thalamotomy. Neurosurgery. 2019;​84(1):​160–168.
classification of tremors. From the Task Force on Tremor of 23. Sammartino F, Krishna V, King NK, et al. Tractography-based
the International Parkinson and Movement Disorder Society. ventral intermediate nucleus targeting:​novel methodology and
Mov Disord. 2018;​33(1):​75–87. intraoperative validation. Mov Disord. 2016;​31(8):​1217–1225.
  4. Zesiewicz TA, Elble R, Louis ED, et al. Practice parameter:​ 24. Gravbrot N, Saranathan M, Pouratian N, Kasoff WS. Ad-
therapies for essential tremor:​report of the Quality Standards vanced imaging and direct targeting of the motor thalamus
Subcommittee of the American Academy of Neurology. Neu- and dentato-rubro-thalamic tract for tremor:​a systematic
rology. 2005;​64(12):​2008–2020. review. Stereotact Funct Neurosurg. 2020;​98(4):​220–240.
  5. Deuschl G, Raethjen J, Hellriegel H, Elble R. Treatment of 25. Mezer A, Rokem A, Berman S, et al. Evaluating quantitative
patients with essential tremor. Lancet Neurol. 2011;​10(2):​ proton-density-mapping methods. Hum Brain Mapp. 2016;​
148–161. 37(10):​3623–3635.

J Neurosurg Volume 135 • September 2021 813

 
*40 *
Morrison et al.

26. Kanowski M, Voges J, Tempelmann C. Delineation of the 39. Chazen JL, Sarva H, Stieg PE, et al. Clinical improvement
nucleus centre median by proton density weighted magnetic associated with targeted interruption of the cerebellothalamic
resonance imaging at 3 T. Oper Neurosurg (Hagerstown). tract following MR-guided focused ultrasound for essential
2010;​66(suppl 1):​E121, E123. tremor. J Neurosurg. 2018;​129(2):​315–323.
27. Kawashima Y, Chen HJ, Takahashi A, et al. Application of 40. Miller TR, Zhuo J, Eisenberg HM, et al. Targeting of the den-
magnetic resonance imaging in functional stereotactic thala- tato-rubro-thalamic tract for MR-guided focused ultrasound
motomy for the evaluation of individual variations of the treatment of essential tremor. Neuroradiol J. 2019;​32(6):​
thalamus. Stereotact Funct Neurosurg. 1992;​58(1-4):​33–38. 401–407.
28. Spiegelmann R, Nissim O, Daniels D, et al. Stereotactic tar- 41. Vassal F, Coste J, Derost P, et al. Direct stereotactic targeting
geting of the ventrointermediate nucleus of the thalamus by of the ventrointermediate nucleus of the thalamus based on
direct visualization with high-field MRI. Stereotact Funct anatomic 1.5-T MRI mapping with a white matter attenuated
Neurosurg. 2006;​84(1):​19–23. inversion recovery (WAIR) sequence. Brain Stimul. 2012;​
29. Sidiropoulos C, Mubita L, Krstevska S, Schwalb JM. Suc- 5(4):​625–633.
cessful Vim targeting for mixed essential and parkinsonian 42. Mori S, Crain BJ, Chacko VP, van Zijl PCM. Three-dimen-
tremor using intraoperative MRI. J Neurol Sci. 2015;​358(1-2):​ sional tracking of axonal projections in the brain by magnetic
488–489. resonance imaging. Ann Neurol. 1999;​45(2):​265–269.
30. Plaha P, Javed S, Agombar D, et al. Bilateral caudal zona 43. Lazar M, Weinstein DM, Tsuruda JS, et al. White matter
incerta nucleus stimulation for essential tremor:​outcome and tractography using diffusion tensor deflection. Hum Brain
quality of life. J Neurol Neurosurg Psychiatry. 2011;​82(8):​ Mapp. 2003;​18(4):​306–321.
899–904. 44. Nowacki A, Schlaier J, Debove I, Pollo C. Validation of dif-
31. Plaha P, Patel NK, Gill SS. Stimulation of the subthalamic fusion tensor imaging tractography to visualize the dentato-
region for essential tremor. J Neurosurg. 2004;​101(1):​48–54. rubrothalamic tract for surgical planning. J Neurosurg. 2018;​
32. Hamel W, Herzog J, Kopper F, et al. Deep brain stimulation 130(1):​99–108.
in the subthalamic area is more effective than nucleus ventra-
lis intermedius stimulation for bilateral intention tremor. Acta
Neurochir (Wien). 2007;​149(8):​749–758. Disclosures
33. Klein JC, Barbe MT, Seifried C, et al. The tremor network Dr. Martin is a consultant for UniQure and receives support from
targeted by successful VIM deep brain stimulation in hu- ClearPoint Neuro for work outside of this study. Dr. Wang is a
mans. Neurology. 2012;​78(11):​787–795. consultant for Medtronic and Boston Scientific.
34. Coenen VA, Allert N, Mädler B. A role of diffusion tensor
imaging fiber tracking in deep brain stimulation surgery:​ Author Contributions
DBS of the dentato-rubro-thalamic tract (drt) for the treat-
ment of therapy-refractory tremor. Acta Neurochir (Wien). Conception and design: Wang, Morrison, Lee. Acquisition
2011;​153(8):​1579–1585. of data: Wang, Lee, Martin, Dietiker, Brown. Analysis and
35. Coenen VA, Mädler B, Schiffbauer H, et al. Individual fiber interpretation of data: Wang, Morrison, Lee. Drafting the article:
anatomy of the subthalamic region revealed with diffusion Wang, Morrison, Lee. Critically revising the article: Wang,
tensor imaging:​a concept to identify the deep brain stimula- Morrison, Lee. Reviewed submitted version of manuscript: all
tion target for tremor suppression. Neurosurgery. 2011;​68(4):​ authors. Approved the final version of the manuscript on behalf
1069–1076. of all authors: Wang. Administrative/technical/material support:
36. Coenen VA, Rijntjes M, Prokop T, et al. One-pass deep brain Dietiker, Brown. Study supervision: Wang.
stimulation of dentato-rubro-thalamic tract and subthalamic
nucleus for tremor-dominant or equivalent type Parkinson’s Correspondence
disease. Acta Neurochir (Wien). 2016;​158(4):​773–781. Doris D. Wang: Weill Institute for Neurosciences, University of
37. Sweet JA, Walter BL, Gunalan K, et al. Fiber tractography of California, San Francisco, CA. [email protected].
the axonal pathways linking the basal ganglia and cerebellum
in Parkinson disease:​implications for targeting in deep brain
stimulation. J Neurosurg. 2014;​120(4):​988–996.
38. Weidman EK, Kaplitt MG, Strybing K, Chazen JL. Repeat
magnetic resonance imaging-guided focused ultrasound
thalamotomy for recurrent essential tremor:​case report and
review of MRI findings. J Neurosurg. 2020;​132(1):​211–216.

814 J Neurosurg Volume 135 • September 2021

*40 *
 LABORATORY INVESTIGATION
J Neurosurg 135:815–827, 2021

Deciphering the frontostriatal circuitry through the fiber

dissection technique: direct structural evidence on the
morphology and axonal connectivity of the fronto-caudate
tract
*Spyridon Komaitis, MD, MSc,1,2,4,5 Christos Koutsarnakis, MD, MSc,1–5 Evgenia Lani, MD,1,4
Theodosis Kalamatianos, PhD,1,5 Evangelos Drosos, MD, MSc,1,2,5
Georgios P. Skandalakis, MD, MSc,1,4,7 Faidon Liakos, MD,1,4 Evangelia Liouta, PhD,1,5
Aristotelis V. Kalyvas, MD, MSc,1,4,6 and George Stranjalis, MD, PhD1,2,4,5
1
Athens Microneurosurgery Laboratory, Evangelismos Hospital, Athens; 2Department of Neurosurgery, Evangelismos Hospital,
National and Kapodistrian University of Athens, Greece; 3Edinburgh Microneurosurgery Education Laboratory, Department
of Clinical Neurosciences, Western General Hospital, Edinburgh, United Kingdom; 4Department of Anatomy, Medical School,
National and Kapodistrian University of Athens; 5Hellenic Center for Neurosurgical Research, “Petros Kokkalis,” Athens, Greece;
6
Division of Neurosurgery, Toronto Western Hospital, University Health Network, University of Toronto, Ontario, Canada; and
7
Department of Neurosurgery, Icahn School of Medicine at Mount Sinai, New York, New York

OBJECTIVE  The authors sought to investigate the very existence and map the topography, morphology, and axonal
connectivity of a thus far ill-defined subcortical pathway known as the fronto-caudate tract (FCT) since there is a paucity
of direct structural evidence regarding this pathway in the relevant literature.
METHODS  Twenty normal adult cadaveric formalin-fixed cerebral hemispheres (10 left and 10 right) were explored
through the fiber microdissection technique. Lateral to medial and medial to lateral dissections were carried out in a tan-
dem manner in all hemispheres. Attention was focused on the prefrontal area and central core since previous diffusion
tensor imaging studies have recorded the tract to reside in this territory.
RESULTS  In all cases, the authors readily identified the FCT as a fan-shaped pathway lying in the most medial layer of
the corona radiata and traveling across the subependymal plane before terminating on the superolateral margin of the
head and anterior part of the body of the caudate nucleus. The FCT could be adequately differentiated from adjacent
fiber tracts and was consistently recorded to terminate in Brodmann areas 8, 9, 10, and 11 (anterior pre–supplementary
motor area and the dorsolateral, frontopolar, and fronto-orbital prefrontal cortices). The authors were also able to divide
the tract into a ventral and a dorsal segment according to the respective topography and connectivity observed. Hemi-
spheric asymmetries were not observed, but instead the authors disclosed asymmetry within the FCT, with the ventral
segment always being thicker and bulkier than the dorsal one.
CONCLUSIONS  By using the fiber microdissection technique, the authors provide sound structural evidence on the
topography, morphology, and connectional anatomy of the FCT as a distinct part of a wider frontostriatal circuitry. The
findings are in line with the tract’s putative functional implications in high-order motor and behavioral processes and can
potentially inform current surgical practice in the fields of neuro-oncology and functional neurosurgery.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201287
KEYWORDS  fronto-caudate tract; frontostriatal; premotor; prefrontal; anatomy; functional neurosurgery

ABBREVIATIONS  ACC = anterior cingulate cortex; ADHD = attention deficit hyperactivity disorder; AF = arcuate fasciculus; ALIC = anterior limb of the internal capsule;
ATR = anterior thalamic radiation; BA = Brodmann area; CC = corpus callosum; CR = corona radiata; CRad = callosal radiation; CS = centrum semiovale; DBS = deep brain
stimulation; DLPFC = dorsolateral prefrontal cortex; DMPFC = dorsomedial prefrontal cortex; DTI = diffusion tensor imaging; DWI = diffusion-weighted imaging; FAT = frontal
aslant tract; FCT = fronto-caudate tract; FCTd = dorsal component of the FCT; FCTv = ventral segment of the FCT; FOC = fronto-orbital cortex; OPPFC = orbitopolar pre-
frontal cortex; SLF = superior longitudinal fasciculus; SMA = supplementary motor area; VCVS = ventral internal capsule/ventral striatum; VLPFC = ventrolateral prefrontal
cortex.
SUBMITTED  April 16, 2020.  ACCEPTED  July 13, 2020.
INCLUDE WHEN CITING  Published online January 1, 2021; DOI: 10.3171/2020.7.JNS201287.
* S.K. and C.K. contributed equally to this study.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  815

*40 *
Komaitis et al.

T he interest in frontostriatal connectivity is not new.

In the 19th century, Meynert et al. used the term
“corona radiata of the caudate nucleus” to refer to a
system of fibers connecting the frontal lobe with the cau-
Methods
Twenty normal adult cadaveric formalin-fixed cere-
bral hemispheres (10 right and 10 left) were treated with
Klinger’s preparation and were explored using the white
date nucleus in the human brain.1 Six years later (in 1893),
matter dissection technique and an operating microscope
Muratoff was able to trace a homologous bundle in the ca-
(Carl Zeiss OPMI). Prior to dissection, the surface anato-
nine brain.2 At this time, the term “subcallosal fasciculus”
my of the frontal area was recorded.
was coined. During the mid-20th century Yakovlev and
Focused lateromedial and mediolateral dissections on
Locke indicated similar connections between the prefron-
the prefrontal/premotor areas were employed in a tandem
tal cortex and the caudate nucleus in primates.3
manner in all specimens to map the FCT and illustrate its
Nevertheless, it was not until the late 20th century
spatial relationship with adjacent bundles such as the fron-
that significant progress toward a more thorough un-
tal aslant tract (FAT), anterior limb of the internal capsule
derstanding of corticostriatal connectivity was made.
Seminal animal studies by Alexander and Crutcher4 and (ALIC), anterior thalamic radiations (ATRs), corpus cal-
Webster5 offered valuable insights into the fundamental losum (CC), and callosal radiations (CRads). Since most
principles regulating the functional and structural inter- of these tracts converge tightly toward the central core,
connections between the frontal cortex and the striatum. we were very diligent in differentiating them by mapping
The presence of a direct and an indirect pathway was and recording their respective trajectories and termination
advocated, supporting the theory that the corticostria- patterns. To elucidate these crucial details, during the final
tal circuit is parcellated into subcomponents according steps of the dissection we detached the central core from
to anatomy and function. Indeed, Alexander was able the rest of the hemisphere to clearly delineate the different
to trace different frontostriatal loops mediating motor, fiber layers and their exact termination areas on the thala-
cognitive, and behavioral processes. This theory of a mus and caudate nucleus. In 5 hemispheres, the fibers of
“parallel functional architecture” remains a cornerstone the FCT along with their termination zone on the head and
in our perception of basal ganglia anatomy and function body of the caudate nucleus were dissected free from the
to this day.6 rest of the hemisphere to illustrate the intrinsic morphol-
The interest in the relationship of structure to func- ogy of the tract.
tion in the frontostriatal pathways has been revived dur- Our dissection tools were variously sized microfor-
ing recent decades, as various clinical entities, ranging ceps, microdissectors, and microscissors, which we used
from behavioral and psychiatric conditions to movement because they proved to be more precise in dissecting than
disorders, have been attributed to alterations in the white wooden spatulas. In each step of the process, we obtained
matter integrity of this subcortical network.7–12 In this multiple photos from various angles to document and il-
context, increased understanding of the anatomy of the lustrate our findings.
corticostriatal connections in the human brain may be
valuable in designing novel treatment strategies for vari- Results
ous pathologies. Nevertheless, the process of elucidating Surface Anatomy
this anatomy is arduous and challenging, since many of Surface landmarks on the lateral and medial cerebral
the basic techniques implemented in the study of animal aspect were recorded to act as reference points for future
brains cannot be applied in humans. Due to this fact, most dissections. Hence, the superior and inferior frontal sulci;
of the available data on human frontostriatal connectivity the precentral, central, and postcentral sulci; the superior,
are derived from diffusion tensor imaging (DTI) studies. middle, and inferior frontal gyri; the precentral and post-
Yet, the various diffusion-weighted imaging (DWI) proto- central gyri; the cingulate sulcus and gyrus; the medial as-
cols, although they can be performed quickly and in vivo, pect of the superior frontal gyrus; the CC; and the anterior
present serious inherent limitations, as proven by relevant commissure were all identified.
studies.13
Therefore, the amount as well as the quality of direct
structural evidence available regarding the anatomy and Stepwise Dissections
connectivity of the frontostriatal pathways is limited. In Lateral to Medial Dissections
addition, two further issues complicate the acquisition Upon removing the cortical gray matter and the under-
and dissemination of new anatomical information. First, lying U-fibers, we expose the superior longitudinal fascic-
the terminology used for the fronto-caudate tract (FCT) ulus (SLF)/arcuate fasciculus (AF) complex. Dissecting
and other frontostriatal tracts is inconsistent throughout away the longitudinal fibers of the SLF/AF complex helps
the literature, and second, very few studies have addressed to uncover the continuous fibers of the centrum semiovale
the FCT’s intrinsic anatomy. With these shortcomings in (CS; running above the level of the ventricular system) and
mind, we opted to investigate the topography, morpholo- corona radiata (CR; running at the level of the lateral ven-
gy, connectional anatomy, and asymmetry of the FCT. We tricles) (Fig. 1B). In this step, the FAT is also identified as
used fiber dissections since they provide direct anatomi- a distinct group of fibers running from the supplementary
cal evidence of high sensitivity and are used to validate motor area (SMA) toward the posterior part of the inferior
conflicting DTI results.13–15 To the best of our knowledge, frontal gyrus. In the region of the insula, upon dissect-
this is the first focused anatomical study in the English ing the fibers of the extreme capsule and claustrum, one
literature addressing this issue. encounters the gray substance of the putamen (Fig. 1C).

816 J Neurosurg Volume 135 • September 2021

 
*40 *
 Komaitis et al.

FIG. 1. Stepwise lateral to medial dissection of a left hemisphere. A: The superficial U-fibers are exposed. The silhouette of the main sulci is superim-
posed on the specimen for orientation purposes (dotted and dashed lines).  FIG. 1. (continued)→

J Neurosurg Volume 135 • September 2021 817

 
*40 *
Komaitis et al.

FIG. 1. Inset: The regional sulci and gyri are marked in the same specimen before removal of the cortical gray matter. The SMA and pre-SMA are
highlighted in red and blue, respectively. B: The U-fibers and anterior part of the SLF/AF complex have been removed. The fibers of the CR are demon-
strated. In the insular region, the claustrum and the fibers of the external capsule can be appreciated. C: Removing the external capsule and claustrum
in the region of the anterior insula reveals the gray matter of the putamen. D: Parts of the putamen, globus pallidus, and CR/internal capsule are dis-
sected to reveal the caudate nucleus. The internal globus pallidus is maintained to illustrate the trajectory of the fibers of the anterior limb of the internal
capsule that travel between the head of the caudate and the globus pallidus. The fibers of the FCT can be seen radiating from the prefrontal area toward
the caudate nucleus. Inset: Different colors are used to illustrate the various structures and tracts at this level. The FCT is highlighted in light blue,
the caudate in red, the internal capsule in dark green, the arcuate fasciculus in light green, and the globus pallidus in yellow. E: Removing the globus
pallidus at the midinsular level exposes part of the thalamus. The fibers corresponding to the ATR can be seen running lateral to the FCT, bypassing
the caudate nucleus and terminating at the thalamus. The genu of the internal capsule is also revealed between the globus pallidus and the thalamus.
F: The fibers of the ATR are removed and the FCT along with its termination zone on the caudate nucleus (head and body) is dissected free from the
rest of the hemisphere. Lower left inset: Morphology of the FCT. Advanced stage of a lateromedial dissection of a left hemisphere. The FCT along with
its termination zone on the head and body of the caudate nucleus is dissected free from the rest of the hemisphere. The hook-shaped silhouette of the
FCT as it reaches the caudate nucleus can be appreciated. Lower right inset: The silhouette of the FCT is superimposed on the lateral surface of the
same hemisphere. AF = arcuate fasciculus; ALIC = anterior limb of the internal capsule; ATR = anterior thalamic radiation; Caud = caudate nucleus; Cl =
claustrum; CR = corona radiata; CS = central sulcus; EC = extreme capsule; FCT = fronto-caudate tract; FCTv = ventral segment of the fronto-caudate
tract; FP = frontal pole; GnIC = genu of the internal capsule; GPi = internal globus pallidus; ICgn = genu of the internal capsule; IFS = inferior frontal
sulcus; PrC = precentral sulcus; preSMA = pre–supplementary motor area; PtC = postcentral sulcus; Pu = putamen; SFS = superior frontal sulcus; SMA
= supplementary motor area; Th = thalamus. Copyright Spyridon Komaitis. Used with permission. Figure is available in color online only.

The putamen is removed along with the part of the globus removing the cortex and superficial U-fibers up to the level
pallidus that overlies the thalamus (Fig. 1E). In addition, of the marginal ramus of the cingulate sulcus, we encoun-
the projecting fibers of the CS/CR are gradually dissected ter the superior arm of the cingulum (Fig. 2C). Then, the
until their deepest layer that resides in the subventricular anterior part of the cingulum and CC are dissected away
zone is reached. At this stage, a meticulous dissection and the anatomy of the frontal horn and caudate nucleus
technique is required to preserve the integrity of the very is appreciated (Fig. 2D). At this stage, the CRads are also
thin white matter that comprises the fronto-caudate and apparent. Upon removing the CRads, we reveal the fibers
fronto-thalamic projections. At the end of this step, the of the FCT seen to radiate from the head and body of the
ventral segment of the FCT (FCTv) is identified and seen caudate nucleus toward the frontal area (Fig. 2E). At this
to terminate on the head of the caudate nucleus (Fig. 1E). last step of the entire dissection process, the fibers of the
The FCTv exhibits a lazy S–shaped configuration, with FCT are the only fibers preserved in the frontal area. The
its fibers initially observed to travel in a relatively straight FCT and its respective termination zone on the caudate
direction and then seen to curve to adapt to the silhouette nucleus can be dissected free from the rest of the hemi-
of the head of the caudate nucleus before terminating at sphere (Fig. 3C and D). The stepwise medial to lateral dis-
this area. While the FCTv is nicely exposed at this stage, section process is presented in Video 1.
the dorsal component of the tract (FCTd) is covered by
fibers of the ATR, which are seen to pass slightly lateral Segmentation and Connectivity Pattern of the FCT
to the caudate nucleus in order to terminate to the thala- We consistently identified two discrete segments of the
mus (Fig. 1E). These fronto-thalamic fibers are carefully FCT. The ventral segment (FCTv) is formed from fibers
removed during the medial to lateral dissections to reveal recorded to originate from the frontal pole (Brodmann
the underlying fibers of the FCTd. In the last step, the fron- area [BA] 10), fronto-orbital region (BA 11), ventrolateral
tal horn is entered and the spatial relationship of the FCT prefrontal cortex (VLPFC; BA 47), and ventral part of the
with the intraventricular structures and the CC is demon- anterior cingulate cortex (ACC; BA 32) and terminate on
strated. The stepwise lateral to medial dissection process the head of the caudate nucleus (100% of the specimens).
is illustrated in Video 1. The dorsal segment (FCTd) receives fibers from the an-
VIDEO 1. Dissecting the FCT. Stepwise lateral to medial and medial terior part of the pre-SMA (BA 6), dorsolateral prefron-
to lateral dissections. AF = arcuate fasciculus; ALIC = anterior limb tal cortex (DLPFC; BA 8), dorsomedial prefrontal cortex
of the internal capsule; ATR = anterior thalamic radiation; Caud = (DMPFC; BA 9), and dorsal ACC (BA 32) and then trav-
caudate nucleus; Cl = claustrum; CR = corona radiata; CS = central els medial to the fibers of the ATR and terminates at the
sulcus; EC = extreme capsule; FCT = fronto-caudate tract; GPi = body of the caudate nucleus. The transition zone of FCTv
internal globus pallidus; IFG = inferior frontal gyrus; MFG = middle
frontal gyrus; PrC = precentral sulcus; PrCg = precentral gyrus; PtC
to FCTd corresponds to the transition area of the head to
= postcentral sulcus; PtCg = postcentral gyrus; Pu = putamen; SFG the body of the caudate, which is located at the level of
= superior frontal gyrus; Th = thalamus. Additional structures that the foramen of Monro. Further, the FCTv was consistently
can be seen but are not labled: BA = Brodmann area; CC = corpus seen to be thicker and bulkier than the FCTd. The segmen-
callosum; Cg = cingulate gyrus; Cgb = cingulum bundle; Cgr = cin- tation and termination pattern of the FCT is summarized
gulate radiations; Crad = callosal radiations; Fx = fornix. Copyright in Table 1. Additionally, the tract’s ventral and dorsal con-
Spyridon Komaitis. Used with permission. Click here to view. nectivities are illustrated in Fig. 3.
Medial to Lateral Dissections Spatial Relationships of the FCT
Medial to lateral dissections were carried out in a tan- The FCT is a deep-seated group of fibers that lies in
dem manner in all specimens in order to accurately under- the subventricular zone of the frontal horn, which is lo-
stand the morphology and topography of the FCT. After cated medial with respect to the CS/CR and lateral to the

818 J Neurosurg Volume 135 • September 2021

 
*40 *
 Komaitis et al.

FIG. 2. Stepwise medial to lateral dissection of a right hemisphere. A: The relevant superficial anatomy is marked with dotted and dashed lines. B: The
superficial U-fibers are exposed. The silhouette of the main sulci is superimposed on the specimen for orientation purposes.  FIG. 2. (continued)→

J Neurosurg Volume 135 • September 2021 819

 
*40 *
Komaitis et al.

FIG. 2. C: The superficial U-fibers of the frontal area are removed. The superior arm of the cingulum is evident. D: The rostrum, genu, and anterior part
of the body of the CC as well as the superior arm of the cingulum are dissected away. The intraventricular parts of the caudate nucleus and the CRads
are illustrated. E: The ependymal layer covering the caudate nucleus is carefully removed. The CRads are dissected in a superior to inferior direction to
reveal the fibers of the FCT and their termination zone on the caudate nucleus. Upper inset: The FCTv and FCTd are highlighted in light blue and dark
blue, respectively, with the first terminating on the head of the caudate nucleus and the second on the body of the caudate nucleus. The relevant BAs at
which the tracts terminate are highlighted. Middle inset: The silhouette of the FCT superimposed on the surface of the same specimen. Lower inset:
Advanced stage of a mediolateral dissection of a right hemisphere. The FCT along with its termination zone on the caudate nucleus and the structures
of the limbic lobe and diencephalon are dissected free from the rest of the hemisphere. The morphology of the tract is demonstrated. ac = anterior com-
missure; BA = Brodmann area; CC = corpus callosum; CCg = genu of the corpus callosum; CCs = splenium of the corpus callosum; Cdb = body of the
caudate nucleus; Cdh = head of the caudate nucleus; CF = calcarine fissure; Cg = cingulate gyrus; Cgb = body of the corpus callosum; Cp = choroid
plexus; CR = corona radiata; Crad = callosal radiations; Cs = cingulate sulcus; FCT = fronto-caudate tract; FCTd = dorsal component of the fronto-
caudate tract; FCTv = ventral component of the fronto-caudate tract; fm = foramen of Monro; Fx = fornix; IC = internal capsule; OC = optic chiasm; POS
= parieto-occipital sulcus; Pv = pulvinar; SP = septum pellucidum; SpS = subparietal sulcus; Th = thalamus. Copyright Spyridon Komaitis. Used with
permission. Figure is available in color online only.

FIG. 3. Dorsal and ventral connectivity of the FCT. A: Advanced stage of a lateromedial dissection of a right hemisphere. The morphology of the dorsal
terminations of the FCT is illustrated. Dashed lines with arrowheads indicate fiber trajectories. Inset: The fibers of the FCTd (dark blue) arising from the
body of the caudate nucleus can be seen radiating toward BA 8 and BA 9 (anterior pre-SMA and DLPFC). The fibers of the FCTv (light blue) arising from
the head of the caudate nucleus can be seen terminating toward BA 10 and BA 11. B: Advanced stage of a lateromedial dissection of a left hemisphere.
The dorsal connectivity of the FCT is again illustrated. Inset: The FCTd (dark blue) connecting the body of the caudate with the anterior pre-SMA (BA
8) and DLPFC (BA 9) and the FCTv (light blue) radiating from the head of the caudate toward the frontal pole (BA 10) and FOC (BA 11). C: Ventral con-
nectivity of the FCT and its relationship to the ATR. Oblique views of advanced stages of a lateromedial dissection of a left hemisphere. The central core
of the hemisphere along with the diencephalic and mesencephalic structures is dissected free from the rest of the hemisphere. The termination fibers
of the FCT on the head and body of the caudate have been preserved. At this step, the FCTv is seen to terminate at the head of the caudate nucleus,
while the FCTd is covered by the fibers of the ATR that pass laterally to reach the thalamus. Inset: The FCTv and ATR are highlighted with dark blue and
red, respectively. D: Ventral connectivity of the FCT. Oblique views of advanced stages of a lateromedial dissection of a left hemisphere. The fibers of
the ATR have been removed and the fibers of the FCTd are nicely illustrated to terminate into the body of the caudate nucleus. Lower inset: The FCTv
and FCTd are highlighted in dark blue and light blue, respectively. Upper inset: An arbitrary dashed line (1) passing through the anterior border of the
anterior perforated substance demarcates the transition between the FCTv and FCTd. A second dashed line (2) passing through the lateral geniculate
body defines the posterior limit of the FCTd and thus the posterior limit of the entire FCT. These lines were consistently recorded to delineate the afore-
mentioned boundaries in all studied specimens.  FIG. 3. (continued)→

820 J Neurosurg Volume 135 • September 2021

 
*40 *
 Komaitis et al.

FIG. 3. AlIC = anterior limb of the internal capsule; APS = anterior perforated substance; ATR = anterior thalamic radiations; BA = Brodmann area;
Caud = caudate nucleus; Cdb = body of the caudate nucleus; Cdh = head of the caudate nucleus; CR = corona radiata; FCTd = dorsal component of
the fronto-caudate tract; FCTdtz = dorsal component of the fronto-caudate tract termination zone; FCTv = ventral component of the fronto-caudate tract;
FCTvtz = ventral component of the fronto-caudate tract termination zone; GP = globus pallidus; IC = internal capsule; Lgb = lateral geniculate body; Mb
= midbrain; ON = olfactory nerve; OT = optic tract; SS = sagittal stratum; Th = thalamus. Copyright Spyridon Komaitis. Used with permission. Figure is
available in color online only.

CRads. At the level of the superior frontal gyrus, the fibers connectivity. More specifically, in a DTI study, Lehéricy
of the FCT travel medial in relation to the fibers of the and colleagues aimed to elucidate the SMA and pre-SMA
FAT. As mentioned, the FCT is documented to terminate connections to the human striatum.22 The authors advocate
on the superolateral margin of the caudate nucleus at the that the caudate nucleus holds strong connectivity with the
level of the CR/ALIC transition. Therefore, the differen- pre-SMA, while the putamen is functionally relevant to the
tiation between fibers of FCT and ALIC is accurate and SMA and motor cortices. This pre-SMA/caudate stream
reliable because in contrast to the FCT, the fibers of the is further implicated by the authors to subserve motor
anterior limb bypass the basal ganglia and course between preparation and sequencing. Later, Kamali and colleagues
the head of the caudate and the globus pallidus. Finally, as coined the term “prefronto-caudate pathway” to describe
already stated, the ATR lies just lateral to the FCTd. The a group of fibers seen to arise from the prefrontal cortex
correlative anatomy of the FCT with respect to the adja- and curve around the head of the caudate nucleus before
cent fiber tracts is illustrated in Fig. 4. terminating at the thalamus.17 The authors found the so-
called prefronto-caudate tract to travel medially in relation
to the ATR and postulated its functional implications in
Discussion a broad spectrum of disorders, including schizophrenia,
In the neuroscientific literature, the existence of wide Parkinson disease, and Huntington disease. One of the
frontostriatal circuitry and its implications in higher motor very few studies providing direct structural evidence was
and cognitive processes have been well appreciated. The that by Rigoard and colleagues.20 These authors employed
caudate nucleus and putamen represent the main input of focused fiber dissections to explore the anatomy of the hu-
the basal ganglia and have been proven to exhibit strong man accumbofrontal fasciculus and proposed three types
connections with the prefrontal area in both animals and of connections between the fronto-orbital cortex (FOC)
humans. Particularly, the head of the caudate constitutes and the striatum: the accumbofrontal fasciculus connect-
the main target of afferent fibers stemming from both pre- ing the medial FOC with the nucleus accumbens and the
frontal and frontopolar cortices.4,5,16–21 Many animal and intermediate and the lateral fasciculi connecting the medi-
human studies have paved the way toward a more com- al and lateral FOC, respectively, with the head of the cau-
prehensive understanding of this frontostriatal complex date nucleus. In an elegant study, Kinoshita and colleagues
and have offered valuable insights into the topography combined evidence from DTI and awake subcortical map-
and morphology of its different subcomponents. However, ping and provided useful insights on the structural mor-
most of the existing data in humans derive from sophisti- phology, connectivity, and function of the pre-SMA/cau-
cated DWI protocols, whereas there seems to be a paucity date pathway.23 The term “frontostriatal tract” was used
of direct structural evidence that is much needed to shape to describe a group of fibers seen to lie medial to the FAT
a more accurate understanding of these pathways. In ad- and to connect the pre-SMA with the anterior part of the
dition, the nomenclature used to describe the subcompo- caudate nucleus. The authors further advocated the view
nents of this network is inconstant, thus adding further of this pathway as an integral part of a wider “negative
perplexity. motor network,” because its stimulation elicited inhibition
Indeed, in Table 2 we provide a review of the literature of motor initiation in both hemispheres and verbal fluency
focusing on the structural evidence of the fronto-caudate deficits in the dominant hemisphere. The topography of

TABLE 1. Connectivity and segmentation of the FCT

Segment Trajectory Ventral Termination Dorsal Termination Functional Area % (no.) (n = 10) Rt/Lt Asymmetry
FCTd fibers arise from pre-SMA BA 6 (anterior) Premotor cortex/pre-SMA 100% (10)
& DLPFC, follow straight BA 8 DLPFC 100% (10)
trajectory & finally exhibit Body of caudate BA 9 DLPFC & DMPFC 100% (10)
FCTd Not present
hook-like trajectory before nucleus
BA 32 (dorsal) ACC 100% (10)
terminating on body of
caudate nucleus
FCTv fibers arise from frontal BA 10, BA 11, BA 47 FPPFC 100% (10)
pole, FOC, & VLPFC & follow Head of caudate BA 11 FOC 100% (10)
FCTv Not present
straight trajectory toward nucleus BA 47 VLPFC 100% (10)
head of caudate nucleus BA 32 (ventral) ACC 100% (10)
FPPFC = frontopolar prefrontal cortex.

J Neurosurg Volume 135 • September 2021 821

*40 *
Komaitis et al.

FIG. 4. Spatial relationship of the FCT with adjacent structures. Progressive lateromedial dissection of a right hemisphere. In the superior view (upper
left), the different fiber layers of the frontal white matter can be appreciated. The respective planes of the CS (above the level of the lateral ventricle), CR
(at the level of the lateral ventricle), and internal capsule (below the level of the lateral ventricle) are marked with dashed lines. At the level of the CS and
CR the different fiber pathways cannot be differentiated. Below the level of the lateral ventricle, the different tracts can be distinguished based on their
termination areas. Different colors are used for the various bundles: yellow for the FAT, green for the external capsule, red for the internal capsule, and
blue for the FCT. The frontal horn is entered to give a perspective of depth to the reader. The FCT forms the inner layer of the CS/CR white matter before
entering the ventricle. Upper inset: Lateral view of the same hemisphere without colored highlights.  FIG. 4. (continued)→

822 J Neurosurg Volume 135 • September 2021

 
*40 *
 Komaitis et al.

FIG. 4. Middle inset: Close-up of the frontal area. Arrows indicate the relative trajectory of the different tracts within the frontal white matter. The inferior
view (lower left) illustrates the relationship of the FCT with respect to the CC and CRads. The dorsal segment of the FCT has been removed and the
frontal horn is entered. The FCT lies lateral to the fibers of the CRads and medial to the fibers of the ATR and ALIC. Lower inset: The CC and CRads
are highlighted in green, the FCT in blue, and the ATR in red. AF = arcuate fasciculus; ALIC = anterior limb of the internal capsule; ATR = anterior tha-
lamic radiations; Caud = caudate nucleus; CC = corpus callosum; Cdb = body of the caudate nucleus; Cdh = head of the caudate nucleus; CR = corona
radiata; Crad = callosal radiations; CS = centrum semiovale; ExC = external capsule; FAT = frontal aslant tract; FCT = fronto-caudate tract; FCTd = dor-
sal component of the fronto-caudate tract; FCTv = ventral component of the fronto-caudate tract; Fh = frontal horn; GP = globus pallidus; IC = internal
capsule; Ins = insula; Pop = pars opercularis; SLF = superior longitudinal fasciculus. Copyright Spyridon Komaitis. Used with permission. Figure is avail-
able in color online only.

the frontostriatal tract proposed in the paper by Kinoshita to a wider termination area than previously thought, in-
et al. was further supported by Bozkurt and colleagues in cluding the dorsolateral, fronto-orbital, and ventrolateral
a DTI and fiber dissection study.24 Again, the frontostri- prefrontal cortices.25–27 Similarly, accumulating evidence
atal tract was identified as a fiber bundle connecting the supports the theory that the FCT is also heavily implicated
pre-SMA with the caudate nucleus. However, in contrast in behavioral symptoms, including apathy in patients with
to this anatomofunctional framework, recent DTI studies attention deficit hyperactivity disorder (ADHD).
support the notion that the fronto-caudate pathway extends It is therefore evident that the nomenclature used across

TABLE 2. DTI and dissection studies offering structural evidence on the morphology and connectivity of the FCT
Authors Method of Segmentation
& Year Investigation Terminology Connectivity/Segmentation Functional Comment
Lehéricy et DTI Frontostriatal Caudate nucleus found to be more densely connected w/ Motor preparation
al., 200422 tract ipsilateral pre-SMA than SMA & MC
Kamali et DTI Prefronto- Fibers originate from prefrontal cortex, pass through Potential clinical applications in assessment
al., 201017 caudate head of caudate nucleus (medially to ATR fibers), & of schizophrenia & other cognitive disor-
pathway terminate at thalamus ders as well as PD, HD, & MS
Rigoard et Dissection Frontostriatal 3 types of projections arising from OPFC: accumbofron- NM
al., 201120 pathway tal fasciculus: connects most medial part of OMPFC
w/ nucleus accumbens; intermediate fasciculus:
remaining part of OMPFC to inferomedial part of head
of caudate nucleus; & lateral fasciculus: orbitolateral
prefrontal cortex projection fibers to caudate nucleus
Catani et DTI & dis- Frontostriatal Connects premotor cortex & frontal pole w/ head of Connections btwn caudate nucleus & IFG
al., 201228 section tract caudate nucleus participate in syntactic processing
Shang et DTI Frontostriatal 4 connections: dorsolateral caudate, medial prefronto- Children w/ ADHD had significantly lower
al., 201326 tract caudate, orbitofronto-caudate, & ventrolateral caudate GFA values in frontostriatal tracts
Lin et al., DTI Frontostriatal 4 connections: caudate-DLPFC, caudate-VLPFC, Youths w/ ADHD had significantly lower GFA
201425 tract caudate-MPFC, & caudate-FOC values than controls in 4 pairs of frontostri-
atal tracts
Kinoshita et DTI & map- Frontostriatal Connects anterior part of caudate w/ pre-SMA (in pre- Stimulation of FST in both sides resulted in
al., 201523 ping tract SMA intermingles w/ FAT fibers, w/ latter found more motor initiation inhibition; lt FST stimulation
anterior & lateral) linked to verbal fluency deficits
Chiang et DTI Frontostriatal Three connections: caudate-VLPFC, caudate-DLPFC, & ADHD pts had statistically significantly lower
al., 201527 tract caudate-FOC tracts GFA in 3 lt FSTs than controls; GFA values
of rt caudate–VLPFC & bilateral caudate–
DLPFCs negatively correlated w/ inatten-
tion symptoms in youths w/ ADHD
Bozkurt et DTI & dis- Frontostriatal Connects pre-SMA w/ head of caudate nucleus NM
al., 201624 section tract
De Paepe DTI Fronto- 2 types of fronto-caudate connections: pre-SMA–cau- HD pts w/ elevated MD in lt DLPFC–caudate
et al., 201912 caudate date, DLPFC-caudate nucleus white matter presented w/ higher
connections levels of cognitive apathy; rt FST also ex-
hibited increased MD associated w/ higher
cognitive apathy levels
FST = frontostriatal tract; GFA = generalized fractional anisotropy; HD = Huntington disease; IFG = inferior frontal gyrus; MC = motor cortex; MD = mean diffusiv-
ity; MPFC = medial prefrontal cortex; MS = multiple sclerosis; NM = not mentioned; OMPFC = orbitomedial prefrontal cortex; OPFC = orbital prefrontal cortex; PD =
Parkinson disease.

J Neurosurg Volume 135 • September 2021 823

*40 *
Komaitis et al.

studies is variable since some authors use the term “fron- ment always being thicker and bulkier than the dorsal one.
to-caudate tract” or “frontostriatal tract” to describe the One can therefore hypothesize that the caudate nucleus ex-
connectivity between the pre-SMA22–24 and caudate, while hibits stronger connectivity with the frontopolar and fron-
others use the same term to include additional fibers origi- to-orbital cortices than with the pre-SMA and DLPFC.
nating from the frontal pole and the dorsolateral, ventro-
lateral, and fronto-orbital cortices.20,25–28 This lack of con- Putative Functional Role of the FCT
sistent terminology paired with the diverse structural data The results of the current study lend support to tradi-
met among different research groups adds confusion on tional theories on the functional significance of prefronto-
the anatomy and functional significance of the FCT. caudate connectivity. Alexander describes three discrete
In an effort to elucidate this controversial issue, we used frontostriatal circuits: a sensorimotor circuit processing
focused fiber dissections with the aim to offer sound evi- information related to somatomotor and oculomotor func-
dence on the anatomy and essential characteristics of the tion, an association (or cognitive) circuit connecting the
FCT. We intentionally employed the white matter dissec- pre-SMA and DLPFC to the striatum, and a limbic circuit
tion technique as our basic method of investigation since it connecting the FOC to the striatum.4 The association cir-
is documented to yield direct structural data of high accu- cuit is believed to orchestrate cognitive motor functions,
racy and, as such, is currently used as one of the gold stan- such as visuomotor association, sequencing, and motor
dards for validating DWI results. It needs to be stressed learning. Awake stimulation evidence also supports the
that although the DWI tractographic technique is a fast and notion that the association circuit is heavily involved in
noninvasive method that allows for real-time fiber dissec- motor initiation and verbal fluency.23 The limbic circuit in
tions, it is unfortunately prone to erroneous results, mainly turn is implicated in behavioral inhibition, decision-mak-
due to the effect of noisy peaks and ambiguous “crossing ing, and motivation. In this framework, the frontostriatal
and kissing” configurations of fiber populations. There- loop has been linked with a wide spectrum of motor and
fore, it has been repeatedly emphasized that even under behavioral clinical entities, like autism, Parkinson disease,
ideal conditions the accuracy of such an indirect method Huntington disease, schizophrenia, and ADHD.7,8,12,30 In-
like the DWI is inherently limited and further validation deed, our proposed connectivity and segmentation pattern
is necessary.13,29 offers a sound structural basis for Alexander’s functional
Hence, by using fiber dissections we were able to con- paradigm. In that respect, the dorsal FCT seems to equal
sistently identify a deep-seated fiber tract seen to connect the association stream, while the ventral FCT is homolo-
a wide area of the prefrontal cortex with the head and an- gous to the limbic stream.
terior part of the body of the caudate nucleus. This tract, Overall, the FCT is documented to participate in the
for which we use the name “fronto-caudate tract” after functional connectivity of the caudate nucleus with pre-
its connectivity pattern, exhibits a fan-shaped configura- motor and prefrontal areas. Hence, it can be conceptual-
tion with wide axonal terminations recorded toward the ized as a pathway bridging cognitive and motor behaviors.
DLPFC, VLPFC, pre-SMA, orbitopolar prefrontal cortex The caudate nucleus has been advocated to participate
(OPPFC), and FOC. We were also able to parcellate and in a plethora of motor and nonmotor processes, such as
differentiate the fibers radiating from the pre-SMA and procedural and associative learning and motor initiation
DLPFC from those coming from the frontopolar, fronto- and/or inhibition.21 Likewise, the premotor cortex orches-
orbital, and ventrolateral prefrontal cortices. The former trates motor planning, sequencing, and modification by
group of fibers was invariably observed to end posterior integrating internal cognitive and external sensory cues.31
to the point marked by the transition of the head to the The SMA/pre-SMA complex, in turn, controls motor se-
body of the caudate nucleus, i.e., a point that corresponds quencing, internal generated movement, locomotion, and
to the foramen of Monro, while the latter group terminated postural stability and also plays a key role in motor igni-
anterior to this point. In this way, we could divide the tract tion during volitional movement.32 Finally, the dorsal pre-
into two distinct segments, i.e., a dorsal one (FCTd) con- frontal, fronto-orbital, and frontal pole cortices are linked
necting the pre-SMA and DLPFC with the anterior part of to a wide spectrum of cognitive and behavioral functions,
the body of the caudate, and a ventral one (FCTv) connect- including execution, attention, decision-making, complex
ing the OFPFC, OPPFC, and VLPFC with the head of the behavioral planning, and social control.33 Since one way to
caudate nucleus. This structural segmentation pattern is in infer the function of subcortical pathways is to map their
agreement with previous studies advocating a functional axonal connectivity, it can be postulated that the FCT is
distinction between a limbic and a cognitive frontostriatal implicated in integrating high-order behaviors such as mo-
stream. tor and speech initiation, inhibition and sequencing, pos-
Further, our findings on the topography and spatial rela- tural and gait stability, locomotion, attention preservation
tionships of the FCT are in line with the relevant literature. and shift, reward/reinforcement processes, planning, and
The FCT was always encountered to reside lateral to the motivation. It is clear, however, that more clinical data are
CRads and medial to the fibers of the CR/internal cap- needed for a thorough understanding of the tract’s anato-
sule. Interestingly, the FCTv was seen to lie medial to the mofunctional significance.
anterior limb of the internal capsule, while the FCTd was
encountered medial to the fibers of the ATR that bypass Role of the FCT in Modern Neurosurgery: Implications,
the caudate nucleus to reach the thalamus. Left to right Applications, and Complications
asymmetries were not observed, but instead we recorded The role of the ventral internal capsule/ventral stria-
an asymmetry within the FCT itself, with the ventral seg- tum (VCVS) as an effective and feasible target for deep
824 J Neurosurg Volume 135 • September 2021
 
*40 *
 Komaitis et al.

brain stimulation (DBS) in patients with depression- and spatial relationships are maintained and can be explored.
obsessive compulsive disorder–related conditions has For these reasons, the fiber dissection technique is one
been thoroughly addressed by recent well-designed clini- of the gold-standard direct anatomical methods used to
cal studies.34–36 This surgical application arises from the validate indirect structural data coming from DWI trac-
hypothesis that DBS may act indirectly on the cortex, by tographic protocols.13 More specifically, in the case of the
intervening with the underlying white matter.37 Because FCT, the ability of tractography to readily differentiate
of the wide cortical distribution pattern of fiber pathways, this bundle from the adjacent ATR and ALIC is inherently
stimulating a confined area of white matter can puta- limited, and that is what particularly accounts for the dif-
tively affect multiple cortical territories. In this regard, ficulty seen in previous tractographic studies to offer con-
by stimulating the VCVS one could modify the activity sistent structural evidence on the morphology, topography,
of the fronto-orbital and medial prefrontal cortices, which and connectivity of the FCT.
are believed to participate in cognitive processes such as The fiber microdissection method is, however, an ex-
that of behavioral shaping driven by environmental cues. pensive, time-consuming, operator-dependent in vitro
Hence, a better understanding of the spatial distribution of technique. The spatial resolution of the data provided is
frontal fibers and their respective termination areas could lower in comparison with histology, optical coherence to-
contribute to a more effective and sophisticated DBS plan- mography, and polarized light imaging, and there are also
ning. In this framework, the confined area in which the limitations when simultaneously exploring fiber tracts
FCT converges either to the head of the caudate as the with intermingling perpendicular trajectories because the
FCTv or to the transition of the caudate head and body proper dissection of the one can result in the destruction
as the FCTd may potentially offer novel DBS targets for of the other.44,45
motor or psychiatric disorders. This also holds true for the
newly introduced stereotactic ablative procedures used in Conclusions
modern functional neurosurgery.
In modern glioma surgery, awareness should be raised Literature on the structure of the human fronto-cau-
with regard to the implication of the FCT in what is date tract (FCT) has been not only scarce but also vague
known as the “negative motor modulatory network.”38 and at times inconsistent. With the fiber microdissection
Evidence from awake brain mapping studies has linked technique in our armamentarium, we were able to pro-
the stimulation of the frontostriatal circuitry with intraop- vide direct evidence of high accuracy in our analysis of
erative unilateral or bilateral movement disorders for both the topography, morphology, and axonal connectivity of
hemispheres and verbal fluency deficits for the dominant the FCT. We therefore consistently recorded the FCT as
one.23 Nevertheless, the risk of a permanent and complete a discrete group of fibers stemming from a wide cortical
SMA syndrome (contralateral akinesia and mutism) is not area, including the pre-SMA, prefrontal cortex, and fron-
high when resecting or transgressing negative motor ar- tal pole, and terminating at the head and anterior part of
eas; indeed, deficits in the control of fine motor behaviors, the body of the caudate nucleus. The tract could be divided
bimanual coordination, and dexterity may be present and into two segments—dorsal and ventral—according to the
should be carefully balanced in preoperative planning respective connectivity pattern observed. These findings
and decision-making.39,40 A tailored approach to achieve seem to be in agreement with the putative functional sig-
the optimal onco-functional balance for a given patient nificance attributed to the frontostriatal circuitry in the
taking into consideration oncological, functional, oc- neuroscientific literature and can potentially inform cur-
cupational, and quality-of-life characteristics along with rent surgical practice in the area of neuro-oncology and
the patient’s personal will and consent is the mainstay functional neurosurgery.
of modern neuro-oncological treatment.41 This becomes
even more relevant with regard to the emerging concept References
of “supratotal resections” applied in both low- and high-   1. Meynert T, Sachs B. Psychiatry:​A Clinical Treatise on Dis-
grade lesions to extend progression-free and overall sur- eases of the Fore-Brain Based Upon a Study of Its Structure,
vival, for which very balanced decisions along the afore- Functions, and Nutrition. Pt. I. G.P. Putnam’s Sons;​1885.
mentioned lines should be carefully made by patients and   2. Muratoff V. Secundare Degenerationen nach Durchschnei-
their physicians. dung des Balkens. Neurologisches Centralblatt. 1893;​12:​
714–729.
Study Strengths and Limitations   3. Yakovlev PI, Locke S. Limbic nuclei of thalamus and connec-
tions of limbic cortex. III. Corticocortical connections of the
Klingler’s preparation entails the fixation of cerebral anterior cingulate gyrus, the cingulum, and the subcallosal
hemispheres in a formalin solution followed by a freeze- bundle in monkey. Arch Neurol. 1961;​5:​364–400.
thaw process.42 The ice crystals that form during the freez-   4. Alexander GE, Crutcher MD. Functional architecture of
ing process separate the white matter fibers apart, and basal ganglia circuits:​neural substrates of parallel process-
therefore one can subsequently identify and dissect them ing. Trends Neurosci. 1990;​13(7):​266–271.
in the setting of a microneurosurgical laboratory. As re-   5. Webster KE. The cortico-striatal projection in the cat. J Anat.
1965;​99:​329–337.
cently documented by Zemmoura and colleagues, this   6. Graff-Radford J, Williams L, Jones DT, Benarroch EE. Cau-
procedure preserves the structural integrity of the nerve date nucleus as a component of networks controlling behav-
axons, and therefore the direct anatomical evidence pro- ior. Neurology. 2017;​89(21):​2192–2197.
vided is of high sensitivity and accuracy.43 Furthermore,   7. Zgaljardic DJ, Borod JC, Foldi NS, Mattis P. A review of the
the 3D architecture of the subcortical pathways and their cognitive and behavioral sequelae of Parkinson’s disease:​

J Neurosurg Volume 135 • September 2021 825

*40 *
Komaitis et al.

relationship to frontostriatal circuitry. Cogn Behav Neurol. 26. Shang CY, Wu YH, Gau SS, Tseng WY. Disturbed micro-
2003;​16(4):​193–210. structural integrity of the frontostriatal fiber pathways and
  8. Schneider MF, Krick CM, Retz W, et al. Impairment of executive dysfunction in children with attention deficit hyper-
fronto-striatal and parietal cerebral networks correlates with activity disorder. Psychol Med. 2013;43(5):1093–1107.
​ ​
attention deficit hyperactivity disorder (ADHD) psychopa- 27. Chiang HL, Chen YJ, Lo YC, et al. Altered white matter tract
thology in adults—a functional magnetic resonance imaging property related to impaired focused attention, sustained
(fMRI) study. Psychiatry Res. 2010;​183(1):​75–84. attention, cognitive impulsivity and vigilance in attention-
  9. Rothemund Y, Buchwald C, Georgiewa P, et al. Compulsivity deficit/hyperactivity disorder. J Psychiatry Neurosci. 2015;​
predicts fronto striatal activation in severely anorectic indi- 40(5):325–335.
​
viduals. Neuroscience. 2011;​197:​242–250. 28. Catani M, Dell’acqua F, Vergani F, et al. Short frontal lobe
10. Quidé Y, Morris RW, Shepherd AM, et al. Task-related fron- connections of the human brain. Cortex. 2012;48(2):273–291.
​ ​
to-striatal functional connectivity during working memory 29. Maier-Hein KH, Neher PF, Houde JC, et al. The challenge of
performance in schizophrenia. Schizophr Res. 2013;​150(2-3):​ mapping the human connectome based on diffusion tractog-
468–475. raphy. Nat Commun. 2017;8(1):1349.
​ ​
11. Nakamae T, Sakai Y, Abe Y, et al. Altered fronto-striatal 30. Langen M, Leemans A, Johnston P, et al. Fronto-striatal
fiber topography and connectivity in obsessive-compulsive circuitry and inhibitory control in autism:​findings from
disorder. PLoS One. 2014;​9(11):​e112075. diffusion tensor imaging tractography. Cortex. 2012;48(2):​ ​
12. De Paepe AE, Sierpowska J, Garcia-Gorro C, et al. White 183–193.
matter cortico-striatal tracts predict apathy subtypes in Hun- 31. Schubotz RI, von Cramon DY. Functional-anatomical con-
tington’s disease. Neuroimage Clin. 2019;​24:​101965. cepts of human premotor cortex:​evidence from fMRI and
13. Thomas C, Ye FQ, Irfanoglu MO, et al. Anatomical accuracy PET studies. Neuroimage. 2003;20(suppl
​ 1):S120–S131.
​
of brain connections derived from diffusion MRI tractogra- 32. Toyokura M, Muro I, Komiya T, Obara M. Activation of pre-
phy is inherently limited. Proc Natl Acad Sci U S A. 2014;​ supplementary motor area (SMA) and SMA proper during
111(46):​16574–16579. unimanual and bimanual complex sequences:​an analysis us-
14. Komaitis S, Kalyvas AV, Skandalakis GP, et al. The frontal ing functional magnetic resonance imaging. J Neuroimaging.
longitudinal system as revealed through the fiber microdis- 2002;12(2):172–178.
​ ​
section technique:​structural evidence underpinning the 33. Otani S. Prefrontal cortex function, quasi-physiological
direct connectivity of the prefrontal-premotor circuitry. J stimuli, and synaptic plasticity. J Physiol Paris. 2003;97(4-6):
​ ​
Neurosurg. 2020;​133(5):​1503–1515. 423–430.
15. Koutsarnakis C, Kalyvas AV, Skandalakis GP, et al. Sledge 34. Senova S, Clair AH, Palfi S, et al. Deep brain stimulation for
runner fasciculus:​anatomic architecture and tractographic refractory obsessive-compulsive disorder:​towards an indi-
morphology. Brain Struct Funct. 2019;​224(3):​1051–1066. vidualized approach. Front Psychiatry. 2019;10:905.
​ ​
16. Hattori T, McGeer EG, McGeer PL. Fine structural analysis 35. Denys D, Graat I, Mocking R, et al. Efficacy of deep brain
of the cortico-striatal pathway. J Comp Neurol. 1979;​185(2):​ stimulation of the ventral anterior limb of the internal cap-
347–353. sule for refractory obsessive-compulsive disorder:​a clinical
17. Kamali A, Kramer LA, Hasan KM. Feasibility of prefronto- cohort of 70 patients. Am J Psychiatry. 2020;177(3):265–271.
​ ​
caudate pathway tractography using high resolution diffusion 36. van der Wal JM, Bergfeld IO, Lok A, et al. Long-term deep
tensor tractography data at 3T. J Neurosci Methods. 2010;​ brain stimulation of the ventral anterior limb of the internal
191(2):​249–254. capsule for treatment-resistant depression. J Neurol Neuro-
18. Leh SE, Ptito A, Chakravarty MM, Strafella AP. Fronto-stri- surg Psychiatry. 2020;91(2):189–195.
​ ​
atal connections in the human brain:​a probabilistic diffusion 37. Widge AS, Zorowitz S, Basu I, et al. Deep brain stimulation
tractography study. Neurosci Lett. 2007;​419(2):​113–118. of the internal capsule enhances human cognitive control and
19. Puglisi G, Howells H, Sciortino T, et al. Frontal pathways prefrontal cortex function. Nat Commun. 2019;10(1):1536.
​ ​
in cognitive control:​direct evidence from intraoperative 38. Rech F, Herbet G, Gaudeau Y, et al. A probabilistic map
stimulation and diffusion tractography. Brain. 2019;​142(8):​ of negative motor areas of the upper limb and face:​a brain
2451–2465. stimulation study. Brain. 2019;142(4):952–965.
​ ​
20. Rigoard P, Buffenoir K, Jaafari N, et al. The accumbofrontal 39. Vilasboas T, Herbet G, Duffau H. Challenging the myth of
fasciculus in the human brain:​a microsurgical anatomical right nondominant hemisphere:​lessons from corticosubcorti-
study. Neurosurgery. 2011;​68(4):​1102–1111. cal stimulation mapping in awake surgery and surgical impli-
21. Robinson JL, Laird AR, Glahn DC, et al. The functional cations. World Neurosurg. 2017;103:449–456.
​ ​
connectivity of the human caudate:​an application of meta- 40. Rech F, Herbet G, Moritz-Gasser S, Duffau H. Somatotopic
analytic connectivity modeling with behavioral filtering. organization of the white matter tracts underpinning mo-
Neuroimage. 2012;​60(1):​117–129. tor control in humans:​an electrical stimulation study. Brain
22. Lehéricy S, Ducros M, Krainik A, et al. 3-D diffusion tensor Struct Funct. 2016;221(7):3743–3753.
​ ​
axonal tracking shows distinct SMA and pre-SMA projec- 41. Rech F, Duffau H, Pinelli C, et al. Intraoperative identifica-
tions to the human striatum. Cereb Cortex. 2004;​14(12):​ tion of the negative motor network during awake surgery to
1302–1309. prevent deficit following brain resection in premotor regions.
23. Kinoshita M, de Champfleur NM, Deverdun J, et al. Role of Neurochirurgie. 2017;63(3):235–242.
​ ​
fronto-striatal tract and frontal aslant tract in movement and 42. Klingler J. Erleichterung der makroskopischen Präparation
speech:​an axonal mapping study. Brain Struct Funct. 2015;​ des Gehirn durch den Gefrierprozess. Schweiz Arch Neurol
220(6):​3399–3412. Psychaitr. 1935;36:247–256.
​ ​
24. Bozkurt B, Yagmurlu K, Middlebrooks EH, et al. Microsurgi- 43. Zemmoura I, Blanchard E, Raynal PI, et al. How Klingler’s
cal and tractographic anatomy of the supplementary motor dissection permits exploration of brain structural connectiv-
area complex in humans. World Neurosurg. 2016;​95:​99–107. ity? An electron microscopy study of human white matter.
25. Lin HY, Gau SS, Huang-Gu SL, et al. Neural substrates of Brain Struct Funct. 2016;221(5):2477–2486.
​ ​
behavioral variability in attention deficit hyperactivity disor- 44. Dammers J, Axer M, Grässel D, et al. Signal enhancement in
der:​based on ex-Gaussian reaction time distribution and dif- polarized light imaging by means of independent component
fusion spectrum imaging tractography. Psychol Med. 2014;​ ​ ​
analysis. Neuroimage. 2010;49(2):1241–1248.
44(8):​1751–1764. 45. Magnain C, Augustinack JC, Reuter M, et al. Blockface

826 J Neurosurg Volume 135 • September 2021

 
*40 *
 Komaitis et al.

histology with optical coherence tomography:​a comparison Supplemental Information

with Nissl staining. Neuroimage. 2014;​84:​524–533. Videos
Video 1. https://vimeo.com/448501596.
Disclosures
Correspondence
The authors report no conflict of interest concerning the materi-
als or methods used in this study or the findings specified in this Christos Koutsarnakis: Evangelismos Hospital, National and Kap-
paper. odistrian University of Athens, Greece. [email protected].

Author Contributions
Conception and design: Komaitis, Koutsarnakis. Acquisition of
data: Komaitis, Koutsarnakis, Lani. Analysis and interpretation
of data: all authors. Drafting the article: Koutsarnakis,
Komaitis. Critically revising the article: Koutsarnakis, Komaitis,
Kalamatianos, Kalyvas, Stranjalis. Reviewed submitted version
of manuscript: Koutsarnakis, Komaitis. Approved the final
version of the manuscript on behalf of all authors: Koutsarnakis.
Administrative/technical/material support: Stranjalis. Study
supervision: Koutsarnakis, Stranjalis.

J Neurosurg Volume 135 • September 2021 827

*40 *
 CLINICAL ARTICLE
J Neurosurg 135:828–834, 2021

Superselective pseudocontinuous arterial spin labeling in

patients with meningioma: utility in prediction of feeding
arteries and preoperative embolization feasibility
Dong Hyun Yoo, MD,1 Chul-Ho Sohn, MD, PhD,1 Young Dae Cho, MD, PhD,1
Hyun-Seung Kang, MD, PhD,2 Chul-Kee Park, MD, PhD,2 Jin Wook Kim, MD, PhD,2 and
Jae Hyoung Kim, MD, PhD3

Departments of 1Radiology and 2Neurosurgery, Seoul National University Hospital, Seoul; and 3Department of Radiology,
Seoul National University Bundang Hospital, Gyeonggi-do, Republic of Korea

OBJECTIVE  Superselective pseudocontinuous arterial spin labeling (ss-pCASL) is an MRI technique in which individual
vessels are labeled to trace their perfusion territories. In this study, the authors assessed its merit in defining feeding
vessels and gauging preoperative embolization feasibility for patients with meningioma, using digital subtraction angiog-
raphy (DSA) as the reference method.
METHODS  Thirty-one consecutive patients with meningiomas were prospectively recruited, each undergoing DSA
(and embolization, if feasible) before resection. All ss-pCASL imaging studies were performed 1 day prior to DSA. Two
neuroradiologists independently reviewed ss-pCASL images, rating the contribution of each labeled vessel to tumor
blood supply as none, minor, or major. Two neuroradiologists also gauged the feasibility of embolization in each patient,
based on ss-pCASL images. Interobserver and intermodality agreement were determined using Cohen’s kappa statistic.
The diagnostic performance of ss-pCASL was assessed in terms of discerning tumor blood supply and the potential for
embolization.
RESULTS  Interobserver agreement in the rating of blood supply by ss-pCASL was very good (κ = 0.817, 95% CI
0.771–0.863), and intermodality agreement (consensus ss-pCASL readings vs DSA findings) was good (κ = 0.688, 95%
CI 0.632–0.744). In delineating tumor blood supply, ss-pCASL showed high sensitivity (87.1%) and specificity (87.2%).
The positive and negative predictive values for embolization feasibility were 85.2% and 100%, respectively.
CONCLUSIONS  In patients with meningiomas, feeding vessels are reliably predicted by ss-pCASL. This noninvasive
approach, involving no iodinated contrast or radiation exposure, is particularly beneficial if there are no prospects of em-
bolization.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201915
KEYWORDS  arterial spin labeling; digital subtraction angiography; embolization; meningioma; territorial imaging;
oncology

M
eningioma is the most frequently diagnosed pri- chances of complete resection and limits intraoperative
mary intracranial tumor.1 Most are hypervascu- blood loss.4–6 However, not all feeding arteries are safe and
lar, the usual feeding arteries being dural branch- suitable for this purpose. Those arising from the ICA or
es of the external carotid artery (ECA) such as the middle VA carry a substantial risk of cerebral ischemic complica-
meningeal artery (MMA). Feeding arteries also originate tions, and embolization of superficial arteries (i.e., super-
from dural or pial branches of the internal carotid artery ficial temporal artery [STA] or occipital artery [OA]) may
(ICA) or vertebral artery (VA).2 Although some have ques- result in scalp necrosis.7,8 Digital subtraction angiography
tioned the benefit of preoperative embolization,3 depriving (DSA) remains the gold standard modality for the delinea-
tumors of blood supply through embolization increases the tion of feeding arteries, and only upon its completion is

ABBREVIATIONS  ASL = arterial spin labeling; DSA = digital subtraction angiography; ECA = external carotid artery; ICA = internal carotid artery; MMA = middle meningeal
artery; NPV = negative predictive value; OA = occipital artery; PPV = positive predictive value; SNR = signal-to-noise ratio; ss-pCASL = superselective pseudocontinuous
ASL; STA = superficial temporal artery; TOF-MRA = time-of-flight magnetic resonance angiography; VA = vertebral artery.
SUBMITTED  May 22, 2020.  ACCEPTED  July 1, 2020.
INCLUDE WHEN CITING  Published online November 13, 2020; DOI: 10.3171/2020.7.JNS201915.

828 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*41 *
 Yoo et al.

the prospect of embolization ascertained. However, DSA per patient. The images were acquired in real time, and no
is an invasive procedure involving iodinated contrast and postprocessing on a dedicated workstation was required.
radiation exposure, carrying its own risks.9
Arterial spin labeling (ASL) is an entirely noninva- DSA and Embolization
sive MRI technique used to assess cerebral circulation by After puncture of the femoral artery and insertion of
magnetically labeling inflowing blood.10 The more recent a 5-Fr arterial sheath, the VA on the dominant (larger) or
introduction of vessel-selective ASL further enables the lesion (in the case of a posterior fossa tumor) side, both
tracing of single-vessel perfusion territories by labeling in- ICAs, and both ECAs were selectively catheterized. If, in
dividual arteries.11–13 A number of studies have shown the the presence of a hypervascular tumor stain, embolization
merit of this technique in certain cerebrovascular disorders was deemed safe and productive, a microcatheter was ad-
such as arteriovenous malformations and carotid artery vanced to the presumptive feeding artery. Angiography
stenosis,14,15 but the few efforts to date exploring its appli- was then performed via a microcatheter to replicate the
cability to meningioma have lacked full comparisons with tumor blush and exclude a potentially dangerous anasto-
DSA.16–18 The present study provides a comparative analy- mosis. Polyvinyl alcohol particles were used for emboliza-
sis of superselective pseudocontinuous ASL (ss-pCASL)12 tion.
and DSA in patients with meningiomas, each used to de-
fine feeding vessels and gauge preoperative embolization
feasibility. We hypothesized that ss-pCASL would reliably Analysis of ss-pCASL Technique
predict feeding vasculature of meningiomas. Two readers (D.H.Y. and C.H.S.) experienced in ASL
imaging and interventional neuroradiology (5 and 15
years, respectively) and blinded to DSA images indepen-
Methods dently reviewed the ss-pCASL images. The images from
Patient Population each patient were thoroughly anonymized, and then the
This study was approved by the Seoul National Univer- images were classified and loaded in separate files with
sity Hospital IRB, with all participating patients providing a description on which vessel was labeled. Initially, each
written informed consent. We prospectively recruited con- labeled vessel was rated according to the extent of tumor
secutive patients who met the following criteria: 1) age > blood supply as none, minor, or major. Minor supply ves-
18 years, 2) intracranial meningioma suspected by imag- sels were those with ASL signals so inconsequential that
ing study, and 3) DSA (with embolization, if feasible) and embolization, even if technically feasible, would not ef-
subsequent resection scheduled at Seoul National Univer- fectively achieve tumor devascularization. Interobserver
sity Hospital. In our institution, not all patients with me- agreement was calculated using Cohen’s kappa statistic,
ningioma were consulted for DSA. The need for DSA and interpreted as follows: < 0.20 = poor, 0.21–0.40 = fair,
embolization was defined by the neurosurgeons in charge 0.41–0.60 = moderate, 0.61–0.80 = good, 0.81–0.90 =
of the surgery. Exclusion criteria were as follows: 1) unin- very good, and > 0.90 = excellent. Final decisions on dis-
terpretable MRI data due to artifacts, 2) previous history crepant ratings were reached by consensus. In each study
of open cranial surgery, 3) neurological disorders (aside patient, image quality was also graded as fair or good by
from meningioma), or 4) histological evidence of tumors consensus of both readers. Using consensus readings of
other than meningioma. ss-pCASL images, the two readers then predicted the fea-
sibility of embolization in each patient. When major sup-
MRI Studies ply was identified on ECA-labeled images of one or both
All ss-pCASL imaging was performed 1 day prior to sides, the tumor was judged suitable for embolization.
DSA using an Ingenia 3.0T MRI scanner (Philips Health-
care). Labeling areas were planned using time-of-flight Comparison With DSA Studies
magnetic resonance angiography (TOF-MRA), extend- All tumor signals from labeled vessels were validated
ing from the carotid artery bifurcation to the base of the against the hypervascularity manifested in corresponding
skull. Potential labeling targets were the 5 major cervical DSA views. DSA images were reviewed, blinded to ss-
arteries: both ICAs, both ECAs, and dominant- or lesion- pCASL image, and angiographic vascularity was rated in
side VA. However, vessels were selectively labeled for ss- the same manner (none, minor, or major supply) as in ss-
pCASL in each patient according to tumor location. At pCASL images. Intermodality agreement was also calcu-
the investigator’s discretion, vessels with little possibility lated using Cohen’s kappa. The diagnostic performance of
of supplying the tumor were not labeled for image acqui- ss-pCASL in delineating tumor blood supply was calcu-
sition. For example, right ICA–labeled images were not lated on a per vessel basis, whereas the gauging of emboli-
pursued in a patient with left temporal convexity menin- zation was calculated on a per patient basis. All statistical
gioma. All ss-pCASL images were acquired using the fol- computations relied on standard software (SPSS version
lowing parameters: field of view 240 × 240 × 96 mm3, 25, IBM Corp.).
voxel size 2.75 × 2.75 × 6 mm3, flip angle 90°, single-shot
echo-planar imaging acquisition (TR/TE = 4265/9.7 msec,
number of excitations = 1), labeling duration 1.8 seconds, Results
postlabeling delay 2.0 seconds, and labeled image slices = Between September 2018 and September 2019, 34
16. The scan time was approximately 3 minutes per vessel, patients were prospectively enrolled for this study. Two
with a maximum of approximately 15 minutes required were poorly cooperative and were excluded due to unin-

J Neurosurg Volume 135 • September 2021 829

 
*41 *
Yoo et al.

TABLE 1. Characteristics of the tumors (N = 31) in the study TABLE 2. Intermodality agreement for rating blood supply
population
DSA
Characteristic Value (%) Rating None Minor Major Total
Tumor diameter (mm) ss-pCASL
  Mean max ± SD 50.1 ± 13.4  None 41 6 3 50
 Range 18–72  Minor 5 6 3 14
Location of the tumor  Major 1 4 48 53
  Cerebral convexity 20 (64.5)  Total 47 16 54
 Parasagittal 4 (12.9)
Intermodality agreement was good (κ = 0.688, 95% CI 0.632–0.744). Sensitiv-
 Falx 2 (6.5) ity was 87.1% (95% CI 77.0%–94.0%) and specificity was 87.2% (95% CI
  Sphenoid ridge 2 (6.5) 74.3%–95.2%).
  Planum sphenoidale 2 (6.5)
 Petroclival 1 (3.2)
Clinical symptoms “feasible” in 27 and “infeasible” in 4 patients. All 4 pa-
 Hemiparesis 5 (16.1) tients considered unfit underwent DSA only (negative pre-
 Paresthesia 2 (6.5) dictive value [NPV] = 100%; Supplementary Figure), with
  Visual symptom 4 (12.9) 23 of 27 patients judged as suitable proceeding to preop-
 Seizure 2 (6.5)
erative embolization (positive predictive value [PPV] =
85.2%, 95% CI 83.1%–87.1%). Tumor embolization proved
  Cognitive dysfunction 4 (12.9) unacceptable in 3 patients whose main feeders, despite
 Headache 6 (19.3) originating from the ECA, were either the STA or OA. In
 Others 2 (6.5) 1 patient, although bilateral MMAs were supplying the tu-
 Asymptomatic 6 (19.3) mor, embolization was overruled due to small-sized feed-
Histologic classification ing arteries and the faintness of tumor blush. There were
  WHO grade I 24 (77.4)
no adverse events associated with DSA or embolization.
Representative images from illustrative cases are shown
  WHO grade II 7 (22.6) in Figs. 1–3.

Discussion
terpretable ss-pCASL images. The other 32 were treated To our knowledge, this study is the first to use DSA
surgically, with 1 subsequently receiving a diagnosis of studies to fully corroborate vessel-selective ASL images
glioblastoma. Therefore, 31 patients (18 women and 13 of patients with meningiomas. Our data showed good in-
men, mean age 58.9 ± 13.4 years, range 26–81 years) with termodality agreement (κ = 0.688), with ss-pCASL per-
histologically proven meningioma were eligible for study. forming well (sensitivity = 87.1%, specificity = 87.2%) in
Characteristics of the tumors in this study population are delineating tumor blood supply and in gauging emboliza-
summarized in Table 1. tion feasibility (PPV = 85.2%, NPV = 100%).
Of the 155 vessels (i.e., 5 major cervical arteries in Previous studies involving intracranial tumor diag-
each of 31 patients) eligible for labeling, only 117 (21 right nostics have readily demonstrated the utility of ASL.19–21
ICAs, 27 left ICAs, 30 right ECAs, 31 left ECAs, and 8 In patients with meningiomas, ASL images were found
VAs) were actually labeled during ss-pCASL imaging and to reflect vascular density and facilitate histopathological
served for the final analysis. Among the rest, 36 were de- differentiation.22–24 Other researchers have also reported
void of any blood supply to the tumor on DSA, and angio- that ASL helped define angiographic vascularity, thereby
grams of the right ICA in 2 patients showed minor blood aiding in the selection of candidates for preoperative em-
flow to contralateral convexity meningiomas. bolization.25,26 Conventional ASL has its own merits as
Interobserver agreement in rating of blood supply on it produces more consistent images with less acquisition
ss-pCASL was very good (κ = 0.817, 95% CI 0.771–0.863). time and less artifact because there is no need for labeling
Consensus readings of ss-pCASL identified 16 major and of selected arteries. In addition, quantitative measurement
6 minor supply vessels originating from the ICA (either of cerebral blood flow is possible on conventional ASL
side) or VA, whereas 37 major and 8 minor supply vessels images. However, conventional ASL images acquired by
originated from the ECA (either side). Image quality was nonselective labeling in the cervical plane do not provide
graded as good and fair in 19 and 12 patients, respectively. information on the origins of feeding arteries. Even if hy-
Intermodality agreement (consensus ss-pCASL readings pervascularity is documented on conventional ASL, those
vs DSA findings) was good (κ = 0.688, 95% CI 0.632– with dominant pial supply would not be amenable to em-
0.744; Table 2). High sensitivity (87.1%, 95% CI 77.0%– bolization. The definite advantage of ss-pCASL lies in its
94.0%) and specificity (87.2%, 95% CI 74.3%–95.2%) ability to distinguish the origin of feeding arteries, which
were observed in defining tumor blood supply (minor or may assist in selecting suitable candidates for effective
major) by ss-pCASL. embolization. Even in circumstances in which emboliza-
Consensus decisions on tumor embolization yielded tion is not planned, neurosurgeons can obtain preoperative

830 J Neurosurg Volume 135 • September 2021

 
*41 *
 Yoo et al.

FIG. 1. Images obtained in a 35-year-old woman presenting with headache. A: Contrast-enhanced T1-weighted MR image of an
anterior falcine meningioma (maximum diameter, 70 mm). B: Left ICA–labeled ss-pCASL image of the posterior part of the tumor
showing hyperperfusion (arrows). C: DSA of left ICA (lateral projection), hypervascular stain (arrows) at corresponding area of
tumor. D: Left ECA–labeled ss-pCASL image of the anterolateral part of the tumor showing hyperperfusion (arrows). E and F: DSA
of left ECA (lateral projection) and selective DSA of left MMA (anterior projection), hypervascular stain (arrows) at corresponding
area of tumor. G: Right ECA–labeled ss-pCASL image of the posteromedial part of the tumor showing hyperperfusion (arrows).
H and I: DSA of right ECA (lateral projection) and selective DSA of right MMA (anterior projection), hypervascular stain (arrows) at
corresponding area of tumor. Bilateral MMA embolizations were performed.

FIG. 2. Images from a 49-year-old woman presenting with seizure. A: Contrast-enhanced T1-weighted MR image of a left frontal
convexity meningioma (maximum diameter, 42 mm). B: Left ICA–labeled ss-pCASL image of the anteroinferior part of the tumor
showing hyperperfusion (arrows). C: DSA of the left ICA (lateral projection), hypervascular stain (arrows) from the ethmoidal
branch of the ophthalmic artery at the corresponding area of tumor. D: Left ECA–labeled ss-pCASL image of the anterosuperior
part of the tumor showing hyperperfusion (arrows), rated as major supply. E: DSA of left ECA (lateral projection), hypervascular
stain (arrows) from the STA at corresponding area of the tumor. F: DSA (more delayed phase), hypervascular stain (arrows) shifted
to the posterior aspect of the tumor. Embolization was abandoned to avoid skin necrosis after STA occlusion.

J Neurosurg Volume 135 • September 2021 831

 
*41 *
Yoo et al.

FIG. 3. Imaging in a 76-year-old woman presenting with diminished visual acuity. A: Contrast-enhanced T1-weighted MR image of
a right planum sphenoidale meningioma (maximum diameter, 38 mm). B: Right ICA–labeled ss-pCASL image of most of the tumor
hyperperfused (arrows). C: Right ECA–labeled ss-pCASL image with limited focal hyperperfusion of the tumor (arrows), rated as
minor supply. ICA-predominant vascular supply deemed unfit for embolization. D and E: DSA of right ICA showing hypervascular
tumor stain (arrows) from a recurrent meningeal branch of the ophthalmic artery. F: DSA of right ECA (anterior projection) with
focal hypervascular stain (arrows), presumably from an artery of the foramen rotundum. Embolization was not performed because
arterial supply from the ECA branch was very small.

information on feeding vasculature, the presence and pro- agreement was good (κ = 0.772, 95% CI 0.709–0.835) in
portion of pial supply, without performing DSA. 19 patients with good image quality, whereas it was mod-
The regional perfusion imaging of ss-pCASL is erate (κ = 0.551, 95% CI 0.450–0.602) in 12 patients with
achieved through circular labeling of single arteries. fair image quality. These data imply that optimal imag-
These labeling spots, generated by rotating gradients with ing is crucial for the accurate definition of vessels feeding
phase changes in radiofrequency pulses of pCASL label- meningiomas. We were therefore compelled to exclude a
ing trains, are flexibly positioned, adapting to individual pair of patients who constantly moved, undermining our
vessel geometries.12,15 As in previous studies,15,18 we suc- repeated attempts at labeled image acquisition.
cessfully placed labeling spots on ICAs, ECAs, and the Apart from reduced image quality through movement,
VA without much difficulty, even in patients with tortuous other factors were contemplated to explain discrepant out-
vessels. However, labeling is reliant on preacquired TOF- comes of ss-pCASL and DSA studies. As shown in Fig. 2,
MRA images, so even slight interim shifts in patient po- insufficient postlabeling delays may falsely reduce areas
sitions (between TOF-MRA and labeling) may interfere, of true hypervacularity in ASL images, fueling underes-
leading to partial or no overlay of targeted arterial labeling timates. Discrepancies may similarly arise when estimat-
spots. Inefficient arterial labeling of this sort results in a ing visible tumor sizes and rating tumor blood supplies in
suboptimal signal-to-noise ratio (SNR). Moreover, images each modality. Unlike biplanar 2D DSA projections, ss-
may be corrupted by signal contamination from nontar- pCASL images are analyzed via scrolling axial views, so
geted arteries, given the close proximity of the ICA and perceived sizes may differ. Upon retrospective review of
ECA. In brief, improper labeling may falsely depict tumor discordant ratings, departures typically arose when decid-
vascular supply, causing over- or underestimates. ing if smaller fractions of visible tumors denoted minor or
In our patients, there were 12 underestimates (lower major arterial supply.
rating) and 10 overestimates (higher rating) of tumor vas- According to our data, ss-pCASL is a conclusive means
cular supply by ss-pCASL compared to DSA (Table 2). of excluding tumors that are not amenable to emboliza-
Underestimation was, in part, due to low SNRs obscur- tion. However, embolization was abandoned in 4 of 27
ing fringe hyperperfusion of tumors (especially in ICA- meningiomas deemed acceptable for embolization by
labeled images), viewed as minor supply during DSA. ss-pCASL. The chief culprits were tumors primarily sup-
Background noise or signal contamination also mimicked plied by scalp arteries through transosseous branches. A
tumor vascularity on occasion, causing overestimation. previous study has shown that primary feeder vessels are
Signal artifact induced by eyeball movement was yet an- reliably identified by unenhanced MRA in 3D.28 Hence,
other impediment to image interpretation.27 Intermodality adjunctive MRA may have additive value in gauging the

832 J Neurosurg Volume 135 • September 2021

 
*41 *
 Yoo et al.

feasibility of embolization. It is also worth noting that our preoperative meningioma embolization? Neurosurgery.
policy of leaving pial feeders or superficial arteries un- 2000;47(6):1306–1312.
touched for fear of complications is not a standard of care.   4. Dowd CF, Halbach VV, Higashida RT. Meningiomas: the role
of preoperative angiography and embolization. Neurosurg
Other dedicated colleagues have certainly performed oth- Focus. 2003;15(1):E10.
erwise and described their work.6,29   5. Shah AH, Patel N, Raper DM, et al. The role of preoperative
Several limitations of our study must be addressed. Due embolization for intracranial meningiomas. J Neurosurg.
to limited resources and our reluctance to tax patient com- 2013;119(2):364–372.
pliance, we did not acquire labeled images of all 5 major   6. Raper DM, Starke RM, Henderson F Jr, et al. Preoperative
cervical arteries in every study patient. Considering the embolization of intracranial meningiomas: efficacy, technical
high specificity (87.2%) shown by ss-pCASL, inclusion of considerations, and complications. AJNR Am J Neuroradiol.
all vessels would likely have strengthened our results. As a 2014;35(9):1798–1804.
  7. Adler JR, Upton J, Wallman J, Winston KR. Management
supplemental exercise, we designated all unlabeled arter- and prevention of necrosis of the scalp after embolization and
ies as nonsupply vessels, to reflect combined performance surgery for meningioma. Surg Neurol. 1986;25(4):357–360.
of ss-pCASL and investigator input. Again, intermodality   8. Dubel GJ, Ahn SH, Soares GM. Contemporary endovascular
agreement was good (κ = 0.727, 95% CI 0.679–0.775), with embolotherapy for meningioma. Semin Intervent Radiol.
high sensitivity (84.7%) and specificity (92.8%) for defin- 2013;30(3):263–277.
ing tumor blood supply.   9. Willinsky RA, Taylor SM, TerBrugge K, et al. Neurologic
Second, in the absence of volumetric computations, our complications of cerebral angiography: prospective analysis
of 2,899 procedures and review of the literature. Radiology.
subjective and inconsistent visual rating of minor versus 2003;227(2):522–528.
major tumor supply was problematic. Lastly, the notion of 10. Haller S, Zaharchuk G, Thomas DL, et al. Arterial spin label-
DSA as the gold standard is also subject to debate. Vari- ing perfusion of the brain: emerging clinical applications.
ous factors may be attributable to acquisition of subopti- Radiology. 2016;281(2):337–356.
mal DSA images. At times, small areas of subtle tumor 11. Wong EC. Vessel-encoded arterial spin-labeling using pseu-
hypervascularity were not readily discernible from brain docontinuous tagging. Magn Reson Med. 2007;58(6):1086–
parenchymal stain or overlapping vessels on 2D images. 1091.
12. Helle M, Norris DG, Rüfer S, et al. Superselective pseu-
These matters were exacerbated by patient movement or docontinuous arterial spin labeling. Magn Reson Med.
catheter-induced spasm of targeted arteries. In reviewing 2010;64(3):777–786.
a few supposed false-positive cases, suspicious tumor vas- 13. Lindner T, Helle M, Jansen O. A short introduction to arterial
cularity (previously unnoticed in blind readings) was ob- spin labeling and its application to flow territory mapping.
served on DSA images. Clin Neuroradiol. 2015;25(suppl 2):211–218.
Despite the above shortcomings, ss-pCASL has true 14. Yu SL, Wang R, Wang R, et al. Accuracy of vessel-encoded
potential as a viable substitute for DSA in patients with pseudocontinuous arterial spin-labeling in identification of
meningiomas. Future directions may include superselec- feeding arteries in patients with intracranial arteriovenous
malformations. AJNR Am J Neuroradiol. 2014;35(1):65–71.
tive labeling of smaller arterial branches such as the MMA 15. Richter V, Helle M, van Osch MJ, et al. MR imaging of in-
or STA. Further research incorporating conventional ASL dividual perfusion reorganization using superselective pseu-
and MRA with ss-pCASL may also help better assess the docontinuous arterial spin-labeling in patients with complex
feasibility of embolization in this setting. extracranial steno-occlusive disease. AJNR Am J Neurora-
diol. 2017;38(4):703–711.
16. Sasao A, Hirai T, Nishimura S, et al. Assessment of vascular
Conclusions supply of hypervascular extra-axial brain tumors with 3T
Our findings suggest that vessels feeding meningiomas MR regional perfusion imaging. AJNR Am J Neuroradiol.
are reliably delineated using the ss-pCASL technique. 2010;31(3):554–558.
Furthermore, the screening of meningiomas by ss-pCASL 17. Lu Y, Luan S, Liu L, et al. Evaluation of the applicabil-
ity of territorial arterial spin labeling in meningiomas for
seems particularly beneficial if embolization is not plau- presurgical assessments compared with 3-dimensional
sible. In such circumstances, DSA may be entirely avoided time-of-flight magnetic resonance angiography. Eur Radiol.
or limited to selected vessels, reducing radiation exposure, 2017;27(10):4072–4081.
use of contrast agents, and procedural risks. 18. Jensen-Kondering U, Helle M, Lindner T, et al. Non-invasive
qualitative and semiquantitative presurgical investigation
of the feeding vasculature to intracranial meningiomas
Acknowledgments using superselective arterial spin labeling. PLoS One.
The Seoul National University Hospital (SNUH) Medical 2019;14(4):e0215145.
Research Collaborating Center contributed to the review of statis- 19. Warmuth C, Gunther M, Zimmer C. Quantification of blood
tical analysis. This study was supported by grant no. 0420180770 flow in brain tumors: comparison of arterial spin labeling
from the SNUH Research Fund. and dynamic susceptibility-weighted contrast-enhanced MR
imaging. Radiology. 2003;228(2):523–532.
20. Noguchi T, Yoshiura T, Hiwatashi A, et al. Perfusion imaging
References of brain tumors using arterial spin-labeling: correlation with
  1. Wiemels J, Wrensch M, Claus EB. Epidemiology and etiol- histopathologic vascular density. AJNR Am J Neuroradiol.
ogy of meningioma. J Neurooncol. 2010;99(3):307–314. 2008;29(4):688–693.
  2. Bitzer M, Wöckel L, Luft AR, et al. The importance of pial 21. Kong L, Chen H, Yang Y, Chen L. A meta-analysis of arterial
blood supply to the development of peritumoral brain edema spin labelling perfusion values for the prediction of glioma
in meningiomas. J Neurosurg. 1997;87(3):368–373. grade. Clin Radiol. 2017;72(3):255–261.
  3. Bendszus M, Rao G, Burger R, et al. Is there a benefit of 22. Kimura H, Takeuchi H, Koshimoto Y, et al. Perfusion

J Neurosurg Volume 135 • September 2021 833

 
*41 *
Yoo et al.

imaging of meningioma by using continuous arterial spin- Author Contributions

labeling: comparison with dynamic susceptibility-weighted Conception and design: Sohn, Yoo, JH Kim. Acquisition of data:
contrast-enhanced MR images and histopathologic features. Yoo, Cho, Kang, Park, JW Kim. Analysis and interpretation of
AJNR Am J Neuroradiol. 2006;27(1):85–93. data: Sohn, Yoo, Cho, Kang. Drafting the article: Yoo. Critically
23. Koizumi S, Sakai N, Kawaji H, et al. Pseudo-continuous arte- revising the article: Sohn, Cho, Kang, Park, JW Kim, JH Kim.
rial spin labeling reflects vascular density and differentiates Reviewed submitted version of manuscript: all authors. Approved
angiomatous meningiomas from non-angiomatous meningio- the final version of the manuscript on behalf of all authors: Sohn.
mas. J Neurooncol. 2015;121(3):549–556. Statistical analysis: Yoo. Administrative/technical/material sup-
24. Lu Y, Xiong J, Yin B, et al. The role of three-dimensional port: Park, JW Kim. Study supervision: Sohn, JH Kim.
pseudo-continuous arterial spin labelling in grading and
differentiating histological subgroups of meningiomas. Clin Supplemental Information
Radiol. 2018;73(2):176–184.
25. Yoo RE, Yun TJ, Cho YD, et al. Utility of arterial spin la- Online-Only Content
beling perfusion magnetic resonance imaging in prediction Supplemental material is available with the online version of the
of angiographic vascularity of meningiomas. J Neurosurg. article.
2016;125(3):536–543. Supplementary Figure. https://thejns.org/doi/suppl/10.3171/​
26. Mayercik V, Ma M, Holdsworth S, et al. Arterial spin-label- 2020.7.JNS201915.
ing MRI identifies hypervascular meningiomas. AJR Am J
Roentgenol. 2019;213(5):1124–1128. Correspondence
27. Wood ML, Henkelman RM. MR image artifacts from peri- Chul-Ho Sohn: Seoul National University Hospital, Seoul, Korea.
odic motion. Med Phys. 1985;12(2):143–151. [email protected].
28. Uetani H, Akter M, Hirai T, et al. Can 3T MR angiogra-
phy replace DSA for the identification of arteries feed-
ing intracranial meningiomas? AJNR Am J Neuroradiol.
2013;34(4):765–772.
29. Hirohata M, Abe T, Fujimura N, et al. Preoperative embo-
lization of brain tumor with pial artery or dural branch of
internal carotid artery as feeding artery. Interv Neuroradiol.
2006;12(1)(suppl 1):246–251.

Disclosures
The authors report no conflict of interest concerning the materi-
als or methods used in this study or the findings specified in this
paper.

834 J Neurosurg Volume 135 • September 2021

 
*41 *
 CLINICAL ARTICLE
J Neurosurg 135:835–844, 2021

Clinical and genomic factors associated with seizures in

meningiomas
*Trisha P. Gupte, BS,1,2 Chang Li, BS,1–4 Lan Jin, PhD,1,2,5,6 Kanat Yalcin, MD,1,2
Mark W. Youngblood, MD, PhD,7 Danielle F. Miyagishima, BA,1,2 Ketu Mishra-Gorur, PhD,1,2
Amy Y. Zhao, BS,1,2 Joseph Antonios, MD, PhD,1,2 Anita Huttner, MD,2,8 Declan McGuone, MD,2,8
Nicholas A. Blondin, MD,2,9 Joseph N. Contessa, MD, PhD,2,10 Yawei Zhang, PhD,5,6
Robert K. Fulbright, MD,2,11 Murat Gunel, MD,1,2,12 Zeynep Erson-Omay, PhD,1,2 and
Jennifer Moliterno, MD1,2

Departments of 1Neurosurgery, 5Surgery, 8Pathology, 9Clinical Neurology, 10Therapeutic Radiology and Pharmacology,
11
Radiology and Biomedical Imaging, and 12Genetics, Yale School of Medicine, New Haven; 2Yale Brain Tumor Center,
Smilow Cancer Hospital, New Haven, Connecticut; 3Department of Neurosurgery, Xiangya Hospital, Central South University,
Changsha; 4The Third Xiangya Hospital, Central South University, Changsha, Hunan, People’s Republic of China; 6Department
of Environmental Health Sciences, Yale School of Public Health, New Haven, Connecticut; and 7Department of Neurological
Surgery, Northwestern University, Chicago, Illinois

OBJECTIVE  The association of seizures with meningiomas is poorly understood. Moreover, any relationship between
seizures and the underlying meningioma genomic subgroup has not been studied. Herein, the authors report on their
experience with identifying clinical and genomic factors associated with preoperative and postoperative seizure presen-
tation in meningioma patients.
METHODS  Clinical and genomic sequencing data on 394 patients surgically treated for meningioma at Yale New Haven
Hospital were reviewed. Correlations between clinical, histological, or genomic variables and the occurrence of preop-
erative and postoperative seizures were analyzed. Logistic regression models were developed for assessing multiple risk
factors for pre- and postoperative seizures. Mediation analyses were also conducted to investigate the causal pathways
between genomic subgroups and seizures.
RESULTS  Seventeen percent of the cohort had presented with preoperative seizures. In a univariate analysis, patients
with preoperative seizures were more likely to have tumors with a somatic NF2 mutation (p = 0.020), WHO grade II or III
tumor (p = 0.029), atypical histology (p = 0.004), edema (p < 0.001), brain invasion (p = 0.009), and worse progression-
free survival (HR 2.68, 95% CI 1.30–5.50). In a multivariate analysis, edema (OR 3.11, 95% CI 1.46–6.65, p = 0.003)
and atypical histology (OR 2.00, 95% CI 1.03–3.90, p = 0.041) were positive predictors of preoperative seizures, while
genomic subgroup was not, such that the effect of an NF2 mutation was indirectly mediated through atypical histology
and edema (p = 0.012). Seizure freedom was achieved in 83.3% of the cohort, and only 20.8% of the seizure-free pa-
tients, who were more likely to have undergone gross-total resection (p = 0.031), were able to discontinue antiepileptic
drug use postoperatively. Preoperative seizures (OR 3.54, 95% CI 1.37–9.12, p = 0.009), recurrent tumors (OR 2.89,
95% CI 1.08–7.74, p = 0.035), and tumors requiring postoperative radiation (OR 2.82, 95% CI 1.09–7.33, p = 0.033) were
significant predictors of postoperative seizures in a multivariate analysis.
CONCLUSIONS  Seizures are relatively common at meningioma presentation. While NF2-mutated tumors are sig-
nificantly associated with preoperative seizures, the association appears to be mediated through edema and atypical
histology. Patients who undergo radiation and/or have a recurrence are at risk for postoperative seizures, regardless of
the extent of resection. Preoperative seizures may indeed portend a more potentially aggressive molecular entity and
challenging clinical course with a higher risk of recurrence.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201042
KEYWORDS  meningiomas; seizures; genomics; oncology

ABBREVIATIONS  AED = antiepileptic drug; CNV = copy number variation; EOR = extent of resection; GTR = gross-total resection; LOH = loss of heterozygosity; PFS =
progression-free survival; PTBE = peritumoral brain edema; STR = subtotal resection; WES = whole exome sequencing.
SUBMITTED  April 1, 2020.  ACCEPTED  July 7, 2020.
INCLUDE WHEN CITING  Published online December 4, 2020; DOI: 10.3171/2020.7.JNS201042.
* T.P.G. and C.L. contributed equally to this work.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  835

*42 *
Gupte et al.

M eningiomas are the most common central nervous

system tumors, and approximately 30%–40% of
patients with meningioma have seizures.1,2 Un-
controlled seizures can significantly affect the quality of
on findings previously relating the Ki-67 index to menin-
gioma recurrence.17 The presence of brain invasion was
also recorded from available pathology reports. After
2016, a neuropathologist retrospectively reviewed cases
life of patients with brain tumors and result in cognitive with brain invasion and changed the previously assigned
deterioration.3–6 While seizure freedom can be achieved WHO grade to grade II if brain invasion was present.
in approximately 80% of meningioma patients after sur-
gery, some can experience a recurrence, and nearly 20% Tumor Volume
of meningioma patients without preoperative seizures can Preoperative MRI or CT imaging in all available cases
develop seizures postoperatively.3,5,7,8 Given that worse pa- was utilized to record the maximal coronal (a), axial (b),
tient outcomes have been associated with the presence of and sagittal (c) diameters of the tumor. Tumor volume was
preoperative seizures, understanding the associated risk calculated using the ellipsoid formula: (4/3)π × (a/2) ×
factors is critical in treating patients.3–5 (b/2) × (c/2).
Several factors related to seizure formation in patients
with meningioma have been proposed. The risk of sei- Tumor Location
zures is associated with tumor location, as seizures tend Tumor location was assessed by board-certified neu-
to be more common with supratentorial tumors that have roradiologists based on preoperative MRI or CT imaging
cortical proximity than with lesions with posterior fossa or by descriptions in operative notes when imaging was
or skull base locations.7,9–12 The WHO grade has been not available. We included only those tumors in which
identified as an independent predictor of preoperative location could be determined with complete confidence.
seizures in meningiomas, and in grade II or III tumors, Though infratentorial meningiomas would not likely be
brain invasion appears to play an important role in seizure associated with seizures, all consecutive meningiomas
formation.13 Similarly, peritumoral brain edema (PTBE) were included in our analysis. Tumor location was clas-
has also been shown to potentially contribute to seizures sified as “skull base” or “non–skull base,” “midline” or
in meningiomas and may be related to persistent seizures “non-midline,” and “anterior” or “posterior.”
after resection.5,13 Potentially distinct PTBE of brain-inva-
sive meningiomas has been recently studied in particular, Peritumoral Brain Edema
as cortical invasion and subsequent corresponding edema
PTBE was assessed using preoperative FLAIR MRI
may facilitate seizures in these tumors.13,14 While genetic
features such as isocitrate dehydrogenase 1 (IDH1) and 2 sequences or T2-weighted sequences when FLAIR se-
(IDH2) mutations have been suggested as pertinent risk quences were not available. The presence of PTBE was re-
factors for seizures in low-grade gliomas,3,5,15,16 the ge- corded as “yes” or “no” and was used in all analyses. The
nomic background of meningiomas has not been corre- extent of PTBE was further delineated on a graded scale
lated with seizure risk. as not present (“N”), smaller than the tumor’s estimated
To better elicit the potential underlying mechanism volume (“S”), the same size as the tumor’s estimated vol-
and the correlation between genetic drivers and seizures ume (“T”), or larger than the tumor’s volume (“L”).
in meningiomas, we retrospectively reviewed a large se-
Extent of Resection
ries of meningiomas and here describe the clinical and
genomic correlations. Extent of resection (EOR) was determined from the op-
erative reports, as well as with postoperative radiographic
confirmation. EOR was characterized using the Simpson
Methods grading scale such that resections with Simpson grades I,
Clinical Data Acquisition II, or III were classified as gross-total resection (GTR) and
Electronic medical records for patients with meningio- resections with Simpson grade IV were classified as sub-
mas who had undergone resection with targeted or whole total resection (STR). There were no biopsies in our series.
exome sequencing (WES) at the Yale New Haven Hos-
pital and Yale Brain Tumor Center were retrospectively Seizures and Antiepileptic Drug Use
reviewed. The cohort included a consecutive series of pa- Charts were reviewed for seizure history, and patients
tients, age 18 years and older, with a tumor histologically were considered to have had pre- and postoperative sei-
confirmed as a meningioma for which genomic sequenc- zures if the occurrence of seizures was temporally related
ing had been performed. Patient demographics, treatment to diagnosis and surgery, respectively, and believed to be
history, and tumor-specific characteristics were recorded a result of the tumor. Antiepileptic drug (AED) use, type
and are described below. of medications administered, and whether patients were
taking AEDs at the time of data collection was document-
Tumor Pathology: WHO Grade, Histological Subtype, Ki-67, and ed. Perioperative AED administration, defined as loading
Brain Invasion the patient with an AED, usually levetiracetam (Keppra),
The histological subtype and WHO grade for all tu- at the time of surgery and continuing peri- and postop-
mors were determined by board-certified neuropatholo- eratively for 10–14 days was recorded. Seizure freedom,
gists in accordance with WHO guidelines. Ki-67 expres- defined as the absence of seizure, was assessed from the
sion was recorded from available pathology reports and time of surgery until the time of data collection. While
classified as < 5% (low) or ≥ 5% (high) with cutoffs based seizure freedom was considered regardless of AED use,

836 J Neurosurg Volume 135 • September 2021

*42 *
 Gupte et al.

we documented whether seizure freedom was in the set- developed using the lavaan package in R (R Foundation
ting of discontinuation of AEDs. for Statistical Computing). All p values ≤ 0.05 were con-
sidered statistically significant.
Genomic Characterization of Meningiomas
IRB approval from Yale University for this study’s Results
methods, along with informed consent from all of the in- Patient Characteristics
cluded patients, was obtained. Samples were submitted for Of the 394 patients who had appropriate data for the
WES or targeted sequencing to identify known genomic aforementioned variables and were included in this study,
drivers. The methods involving the targeted sequencing 268 were female (68.0%), and the overall median age at
and annotation of genomic subgroups have been previ- diagnosis was 58 years (Fig. 1B). Median follow-up time
ously reported.18,19 Using these methods, we classified ge- was 21.12 months (range 15.48–26.88 months). Sixty-six
nomic subgroups based on somatic alterations in the fol- patients (16.8%) presented with preoperative seizures, and
lowing genes: 1) NF2 (and/or Chr22q loss); 2) SMARCB1 24 patients (6.1%) experienced postoperative seizures. Ul-
with or without NF2/Chr22; 3) KLF4K409Q with or without timately, seizure freedom was achieved in 83.3% of the
mutant TRAF7; 4) PI3K signaling, with or without mu- preoperative seizure cohort. Eleven patients with preoper-
tant TRAF7; 5) Hedgehog signaling; 6) POLR2AQ403 K or ative seizures continued to exhibit postoperative seizures.
POLR2AL438_H439del; or 7) TRAF7 without KLF4K409Q or
PI3K signaling mutations (Fig. 1A). All other samples were Role of Prophylactic AED Use in Pre- and Postoperative
classified as “mutation unknown” and were excluded from Seizures
analyses involving genomic variables. These groups here-
in will be referred to as “NF2,” “SMARCB1,” “KLF4,” Of the 63 patients with preoperative seizures and avail-
“PI3K,” “HH,” “POLR2A,” and “TRAF7-alone,” respec- able information on AED usage, all were managed with
tively. AEDs preoperatively, as expected (Table 1). At the time of
For comparing mutation count load and percent of data collection, 79.2% of the patients in whom seizure free-
genome alteration by copy number variations (CNVs) or dom had been achieved were still actively taking AEDs.
The remaining 20.8% who were able to discontinue AED
loss of heterozygosity (LOH), a total of 270 samples, 47
use postoperatively were more likely to have undergone a
(17.4%) from patients with preoperative seizures and 223 GTR (p = 0.031), with the majority having WHO grade I
(82.6%) from patients with no preoperative seizures, were tumors (90.9%) and primary tumors (81.8%). Periopera-
analyzed. tive AED administration was used in 43.7% of the patients
without preoperative seizures. Of the 24 patients with
Statistical Analysis postoperative seizures, 11 had received AEDs preopera-
Statistical analysis was performed using standard sta- tively when they had presented with preoperative seizures.
tistical software (IBM SPSS Statistics, version 24, IBM The remaining 13 patients all received AEDs periopera-
Corp.). All analyses were reaffirmed by an independent tively; despite this, they had postoperative seizures. At the
statistician. Univariate analyses were performed to test time of data collection for this study, 95.8% of postopera-
clinical, histological, and genomic correlations with pre- tive seizure patients were still actively taking AEDs.
and postoperative seizures. Fisher’s exact test and the
1-way ANOVA test were used for categorical and con- Factors Associated With Preoperative Seizures
tinuous variables, respectively. For multivariate analyses, The results from univariate analyses on associations
logistic regression models were developed, and backward between clinical or histological parameters and preop-
stepwise model selection (by the Akaike information cri- erative seizures are summarized in Table 2. Patients with
terion) was performed to select final predictors from all preoperative seizures were more likely to be male (p =
variables that were significantly associated with pre- and 0.014) and have tumors with WHO grade II or III (p =
postoperative seizures in univariate analyses. To visual- 0.029) in a non–skull base location (p = 0.012). They were
ly depict the survival and log-rank statistic, the Kaplan- also more likely to have tumors with PTBE (p < 0.001),
Meier survival analysis (log-rank test) was used. The Cox brain invasion (p = 0.009), elevated Ki-67 expression (i.e.,
proportional-hazards model was used to examine as- ≥ 5%; p = 0.012), atypical histology (p = 0.004), and a so-
sociations between pre- and postoperative seizures with matic NF2 driver mutation (p = 0.020). NF2 was the only
progression-free survival (PFS). To compare the mutation genomic subgroup significantly associated with preop-
count and percent of genome alteration by CNV/LOH, the erative seizures across all tumor grades, whereas among
Welch two-sample t-test was conducted. WHO grade I meningiomas only, no genomic subgroups
To better understand relevant factors potentially associ- were significantly associated with preoperative seizures
ated with the occurrence of preoperative seizures, a series (Fig. 2A and C, Supplemental Tables 1–2). In further
of univariate analyses were conducted to test relationships analysis, there was no statistically significant difference
between hypothesized variables. Next, mediation analyses in the mutation count or percent of genome alteration by
were performed to investigate the causal impacts of cer- LOH events in cases with preoperative seizures compared
tain genomic subgroups on preoperative seizures. Serial to those without. Comparison of the percent of genome
multiple mediator models were utilized to investigate the alteration by CNV events revealed no significance (p =
direct and indirect pathways with atypical histology, ede- 0.073), although samples from patients with preopera-
ma, and brain invasion as mediators. These models were tive seizures had CNV events in 13.0% of their genomes,

J Neurosurg Volume 135 • September 2021 837

 
*42 *
Gupte et al.

FIG. 1. A: Meningioma driver mutations and pathways.20,35,36 B: Characteristics of the study cohort. Figure is available in color
online only.

838 J Neurosurg Volume 135 • September 2021

*42 *
 Gupte et al.

TABLE 1. AED use in patients with pre- and postoperative TABLE 2. Univariate analysis: clinical and histological
seizures correlations with preoperative seizures
Preop Patients w/ Preop Patients w/ Postop No Preop Preop
AED Seizure (n = 63) Seizure (n = 11) Variable Seizures Seizures p Value
Levetiracetam 49 (78%) 7 (64%) Sex
Valproic acid 4 (6%) 1 (9%)  M 96 (29.3) 30 (45.5) 0.014
Phenytoin 2 (3%) —  F 232 (70.7) 36 (54.5) 
Lacosamide 1 (2%) — Median age in yrs (SD) 57.6 (13.1) 58.2 (12.2) 0.430
Lorazepam — 1 (9%) Median tumor vol in cm3 (SD) 21.3 (40.0) 26.6 (41.6) 0.471
Combinatorial therapy 7 (11%) 2 (18%) WHO grade
Only patients who took AEDs preoperatively were included in this table.  I 228 (71.0) 37 (56.9) 0.029
  II & III 93 (29.0) 28 (43.1)
Tumor location
whereas samples from patients without seizures had only   Skull base 138 (43.3) 17 (26.2) 0.012
7.6% of their genomes altered by CNV events (Supple-   Non–skull base 181 (56.7) 48 (73.8)
mental Fig. 1). In a multivariate analysis, PTBE (OR 3.11, Edema
95% CI 1.46–6.65, p = 0.003) and atypical histology (OR  Present 167 (56.8) 50 (80.6) <0.001
2.00, 95% CI 1.03–3.90, p = 0.041) emerged as positive  Absent 127 (43.2) 12 (19.4)
predictors of preoperative seizures (Fig. 3A, Supplemental Brain invasion
Table 3). Patients with preoperative seizures also exhibited
 Present 35 (26.5) 18 (50.0) 0.009
overall worse PFS (HR 2.68, 95% CI 1.30–5.50; Fig. 3C).
 Absent 97 (73.5) 18 (50.0)
Seizure Formation Analyses Ki-67 expression
In univariate analyses, NF2-mutated tumors were found   Low (<5%) 131 (53.0) 18 (33.9)  0.012
to be significantly correlated with atypical histology (p <   High (≥5%) 116 (47.0) 37 (66.1)
0.001) but not with PTBE (p = 0.99); atypical histology was Histology
significantly correlated with both brain invasion (p < 0.001)  Atypical 71 (21.6) 26 (39.4) 0.004
and PTBE (p < 0.001); and brain invasion and PTBE were  Non-atypical 257 (78.4) 40 (60.6)
also significantly correlated (p < 0.001). Mediation analysis
revealed a significant indirect effect of the NF2 genomic Values are expressed as number (%), unless indicated otherwise. Boldface
subgroup on preoperative seizures through the sequential type indicates statistical significance.
mediators of atypical histology and PTBE (p = 0.012; Fig.
4). The corresponding direct effect was not significant (p
= 0.114). The indirect pathway through atypical histology
and brain invasion was not significant (p = 0.064). associated with postoperative seizures among WHO grade
I meningiomas (p = 0.026; Fig. 2D, Supplemental Table
2). In a multivariate analysis, tumor recurrence (OR 2.89,
Factors Associated With Postoperative Seizures
95% CI 1.08–7.74, p = 0.035), postoperative radiation (OR
The results from univariate analyses of associations be- 2.82, 95% CI 1.09–7.33, p = 0.033), and preoperative sei-
tween clinical or histological parameters and postoperative zures (OR 3.54, 95% CI 1.37–9.12, p = 0.009) all emerged
seizures are summarized in Table 3. Patients with postop- as positive predictors of postoperative seizures (Fig. 3B,
erative seizures were more likely to present with preopera- Supplemental Table 4). Postoperative seizures were not
tive seizures (p = 0.001). They were also more likely to significantly associated with overall worse PFS (HR 2.26,
have recurrent tumors (p = 0.009) with atypical histology 95% CI 0.93–5.45, p = 0.063; Fig. 3D). No patients with
(p = 0.024) and PTBE (p = 0.020), which required postop- postoperative seizures died during this study.
erative radiation (p = 0.013). Of the patients who received
postoperative radiation, 2 underwent Gamma Knife ra-
diosurgery while 9 received standard fractionated radio- Discussion
therapy. In all cases, patients presented with postoperative To our knowledge, this is the first study to describe sig-
seizures before receiving postoperative radiation. Notably, nificant correlations between seizures and known genom-
EOR (p = 0.830) and brain invasion (p = 1.000) were not ic markers in meningiomas. The occurrence of seizures
found to be significantly correlated with postoperative sei- at the time of presentation appears to signal a potentially
zures. For EOR in particular, there was also no observ- more challenging clinical entity, as it is more likely to be
able difference between STR and GTR in the preopera- associated with atypical histology and a corresponding so-
tive seizure patients who had seizures postoperatively (p matic driver mutation in NF2, more commonly seen with
= 0.490). There were no significant genomic correlations aggressive meningiomas.19 Taken together, these findings
for postoperative seizures (Fig. 2B, Supplemental Table 1). suggest that preoperative seizures may portend a more
However, when stratifying by tumor grade, the Hedgehog challenging clinical course with an associated worse PFS.
pathway–mutated tumors were found to be significantly Seizures are a relatively common presentation of me-

J Neurosurg Volume 135 • September 2021 839

 
*42 *
Gupte et al.

FIG. 2. Association of genomic subgroups with pre- and postoperative seizures. A: The NF2 subgroup was significantly associated
with preoperative seizures. B: No significant associations between genomic subgroups and postoperative seizures were found.
C: No significant associations between genomic subgroups and preoperative seizures were found among grade I tumors. D: The
HH subgroup was significantly associated with postoperative seizures among WHO grade I tumors. Figure is available in color
online only.

ningioma patients, perhaps surprisingly so given the extra- atypical histology and tumor locations in the convexity.18
axial location of the meningiomas. Similar to others, we In our current study, we add that patients who present with
found that patients with preoperative seizures were more preoperative seizures are more likely to have a tumor in a
likely to have tumors with a higher grade, atypical his- non–skull base location, as well as an associated under-
tology, elevated Ki-67 expression, PTBE, and brain inva- lying somatic NF2 mutation, which is in line with these
sion.4,13 The finding that seizure patients were more likely reports. While a mutation in the NF2 gene may not lead to
to have tumors in the convexity fits well with the known seizure occurrence itself, its association with more aggres-
mechanistic nature of seizures to arise from cortical irrita- sive characteristics may be responsible.
tion and corresponds nicely with our previously defined The underlying mechanism behind seizure formation in
location for NF2-mutated tumors along the convexity.18 meningioma patients has yet to be uncovered. When Hess
Given our previous findings and those of others, cor- et al. investigated the role that brain invasion might play
relations between the NF2 genomic subgroup and tumor- in the risk of seizures in meningiomas, they found cor-
specific characteristics suggest that NF2-mutated tumors relations among PTBE, brain invasion, and preoperative
are associated with more aggressive meningioma forma- seizures but concluded that the pathogenesis of PTBE in
tion. Clark et al. was one of the first to publish findings on more invasive meningiomas was unclear.13 Our ability to
7 distinct and novel genomic subgroups in meningiomas: correlate genomic data with the occurrence of seizures
NF2, POLR2A, SMARCB1, TRAF7, KLF4, HH, and mu- in meningioma patients for the first time may be an im-
tation unknown.20 Harmancı et al. further explained the portant step in understanding the underlying mechanism.
biological pathways underlying more aggressive atypi- We found that the NF2 genomic subgroup was not signifi-
cal de novo meningioma formation and found the NF2 cant in the multivariate analysis for preoperative seizures,
subgroup to be strongly correlated with these tumors.19 which underscores that a driver mutation is often unlikely
More recently, Youngblood et al. examined these genomic to directly result in a clinical manifestation itself. Through
subgroups in relation to clinical correlates and found the mediation analysis, however, we found that the correlation
NF2 genomic subgroup to be significantly associated with between the NF2 genomic subgroup and preoperative sei-
840 J Neurosurg Volume 135 • September 2021

*42 *
 Gupte et al.

FIG. 3. Association of clinical features and NF2 status with pre- and postoperative seizures. A: Preoperative seizures were signifi-
cantly associated with PTBE and atypical histology. B: Postoperative seizures were significantly associated with tumor recurrence,
postoperative radiation, and preoperative seizures. C: Patients with preoperative seizures demonstrated significantly worse PFS
compared to those without preoperative seizures. D: Patients with postoperative seizures did not have a significantly worse PFS
compared to those without postoperative seizures. Figure is available in color online only.

zures was significantly mediated by atypical histology and Aside from correlations with tumor genomic sub-
PTBE in a causal pathway. An indirect pathway involving groups, the other factors we found to be associated with
atypical histology and brain invasion was not found to be preoperative seizures were mostly in accordance with the
significant (p = 0.064). Additionally, although the analysis literature. When Baumgarten et al. conducted a focused
did not reach significance (p = 0.073), we observed anoth- review on seizures in meningiomas, they described sei-
er feature associated with atypical tumors, namely an in- zure patients as more likely to have PTBE and a non–skull
crease in the percent of genomes altered in the group with base tumor localization.21 Although larger tumor size was
preoperative seizures, compared with that in the group also listed as a significant risk factor, we along with oth-
without seizures (13.0% vs 7.6%, respectively). ers did not find an association between tumor volume and
Other known genomic drivers, such as KLF4 mutation, preoperative seizures.3,22,23 Interestingly, of the 20 studies
have been shown to be associated with a secretory his- Baumgarten et al. included in their systematic review, only
tology and subsequently with PTBE as well.18 However, one described tumor histological subtypes in patients with
our finding that the KLF4 genomic subgroup was not sig- preoperative seizure.24 We are therefore one of the few to
nificant in relation to seizures leads us to posit that a more report a significant association between atypical menin-
benign PTBE formation pathway may exist, different from giomas and preoperative seizures.4,13 Furthermore, a need
that which occurs with more aggressive meningiomas, for future studies to examine preoperative seizures in re-
and one that does not lead to preoperative seizures. We lation to tumor histology would be useful. However, this
hypothesize that perhaps brain invasion is the difference could very well be superseded by WES and genomics data
between these groups with regard to seizures, and further as they are more commonly used in clinical care.
clarification is needed. When identifying risk factors associated with postoper-
J Neurosurg Volume 135 • September 2021 841
 
*42 *
Gupte et al.

TABLE 3. Univariate analysis: clinical and histological

correlations with postoperative seizures
No Postop Postop
Variable Seizures Seizures p Value
Preop seizures  
 Present 55 (14.9) 11 (45.8) 0.001
 Absent 315 (85.1) 13 (54.2)  
WHO grade    
 I 253 (69.9) 12 (50.0) 0.066
FIG. 4. Mediation analysis investigating the causal pathways between   II & III 109 (30.1) 12 (50.0)
genomic subgroups and preoperative seizures. A significant indirect ef- Tumor location  
fect instead of a direct effect of the NF2 subgroup on preoperative sei-   Skull base 144 (40.0) 11 (45.8) 0.668
zures through the sequential mediators of atypical histology and PTBE
was revealed. An indirect effect of the NF2 subgroup on preoperative   Non–skull base 216 (60.0) 13 (54.2)  
seizures through the sequential mediators of atypical histology and brain Edema  
invasion was suggested, but not signifiicant.  Present 199 (59.4) 18 (85.7) 0.020
 Absent 136 (40.6) 3 (14.3)  
Brain invasion  
 Present 50 (31.6) 3 (30.0) 0.99
ative seizures, we found that patients were more likely to
have recurrent tumors with a higher grade, atypical histolo-  Absent 108 (68.4) 7 (70.0)  
gy, and PTBE, which required treatment with postoperative Histology  
radiation. Prior studies have shown similar results with re-  Atypical 86 (23.2) 11 (45.8) 0.024
gard to PTBE5,25,26 and tumor recurrence.4,27 Notably, while  Non-atypical 284 (76.8) 13 (54.2)  
we along with others did not find EOR to be a significant Recurrence  
predictor,3,14,24 Islim et al. did find a positive correlation be-
  Recurrent tumor 54 (15.1) 9 (37.5) 0.009
tween greater EOR and seizure freedom in patients.25 With
regard to radiation, because all patients who underwent   Primary tumor 304 (84.9) 15 (62.5)  
postoperative radiation had experienced postoperative sei- Postop radiation  
zures beforehand, it would appear that the correlation we  Yes 67 (18.1) 10 (41.7) 0.013
observed has little to do with the potential of radiation to  No 303 (81.9) 14 (58.3)  
induce seizures postoperatively. Instead, our finding sug- EOR  
gests that postoperative seizure patients are more likely
 STR 158 (48.3) 12 (52.2) 0.830
to have aggressive tumor features like atypical histology,
which are more typically radiated at our institution. Addi-  GTR 169 (51.7) 11 (47.8)  
tionally, while our sample size was too small to discern any Values are expressed as number (%), unless indicated otherwise. Boldface
difference between radiation modalities, the potential for type indicates statistical significance.
either form of radiation to be associated with postoperative
seizures warrants further examination. Taken together with
the finding that tumor recurrence also emerged as a predic-
tor of postoperative seizures in the multivariate analysis, tumor locations. Thus, the underlying molecular makeup of
it appears that patients with postoperative seizures have a the tumor and perhaps the degree and effect of associated
greater likelihood of experiencing an overall more compli- edema, rather than its location, may be more important for
cated clinical course. Interestingly, however, the findings explaining the persistence of seizures after resection.
we observed with postoperative radiation and tumor recur- While AEDs are commonly used and can control most
rence were independent of tumor grade, which could per- seizures in meningiomas, some seizures can be pharma-
haps be better explained by underlying biology. coresistant, requiring multiple agents.6,16 Indeed, approxi-
When genomic correlations were tested within tumor mately 11% of preoperative seizure patients in our cohort
grade subgroups for both pre- and postoperative seizures, had received combinatorial AED therapy. Nonetheless, we
analyses yielded overwhelmingly nonsignificant findings, observed a seizure freedom rate of 83.3%, which is notice-
as they were likely underpowered given the relatively small ably higher than what has been reported in other studies,
sample sizes within each tumor grade subgroup. In the case with reported rates more often in the 60%–70% range.4,29
of postoperative seizures, however, Hedgehog-mutated However, the majority of pre- and postoperative seizure
tumors were found to be significantly associated among patients (79.2% and 95.8%, respectively) were still actively
grade I meningiomas only. It has been well established that taking their AED medications at the time of data col-
Hedgehog-mutated tumors more typically occur along the lection. Through an analysis of the seizure-free patients
olfactory groove in the skull base18 and can often be seen who were able to discontinue AED use postoperatively,
with bilateral frontal lobe edema.28 Unlike our finding with we found that they were more likely to have undergone a
preoperative seizures, we did not observe a significant rela- GTR. This finding lends further support to the view that
tionship between postoperative seizures and non–skull base a greater EOR can help alleviate associated symptoms,
842 J Neurosurg Volume 135 • September 2021

*42 *
 Gupte et al.

such as seizures,30,31 as well as offer the potential benefits References

of a decreased recurrence risk. The usefulness of periop-   1. Englot DJ, Chang EF, Vecht CJ. Epilepsy and brain tumors.
erative AED administration, on the other hand, appears Handb Clin Neurol. 2016;​134:​267–285.
less beneficial. Although nearly half of the cohort without   2. Xue H, Sveinsson O, Tomson T, Mathiesen T. Intracranial
preoperative seizures received perioperative levetiracetam meningiomas and seizures:​a review of the literature. Acta
(Keppra), its use was not a protective factor against the oc- Neurochir (Wien). 2015;​157(9):​1541–1548.
currence of postoperative seizures. When considering the   3. Chaichana KL, Pendleton C, Zaidi H, et al. Seizure control
for patients undergoing meningioma surgery. World Neuro-
cost and side effects, most studies have concluded that the surg. 2013;​79(3-4):​515–524.
overall benefit of the routine application of AEDs to me-   4. Chozick BS, Reinert SE, Greenblatt SH. Incidence of sei-
ningioma patients remains unclear.2,8,32 zures after surgery for supratentorial meningiomas:​a modern
Study limitations included the retrospective nature of analysis. J Neurosurg. 1996;​84(3):​382–386.
patient data collection and the limited data for important   5. Lieu AS, Howng SL. Intracranial meningiomas and epilepsy:​
histological variables, primarily for evidence of brain incidence, prognosis and influencing factors. Epilepsy Res.
invasion. After 2016, our neuropathologists retrospec- 2000;​38(1):​45–52.
  6. Taphoorn MJ, Klein M. Cognitive deficits in adult patients
tively reviewed cases with evidence of brain invasion and with brain tumours. Lancet Neurol. 2004;​3(3):​159–168.
changed these cases from tumor grade I to grade II. Prior   7. van Breemen MS, Wilms EB, Vecht CJ. Epilepsy in patients
to 2016, it is possible that some cases were not reevalu- with brain tumours:​epidemiology, mechanisms, and manage-
ated accordingly. Several cases prior to 2016 also had no ment. Lancet Neurol. 2007;​6(5):​421–430.
mention of brain invasion or instead had evidence of brain   8. Wirsching HG, Morel C, Gmür C, et al. Predicting outcome
invasion marked as “unknown.” Thus, comprehensive of epilepsy after meningioma resection. Neuro Oncol. 2016;​
analyses of brain invasion in relation to other key variables 18(7):​1002–1010.
  9. DeAngelis LM. Brain tumors. N Engl J Med. 2001;​344(2):​
that we hypothesized may be involved in a more aggres- 114–123.
sive PTBE formation pathway were likely underpowered. 10. Kargiotis O, Markoula S, Kyritsis AP. Epilepsy in the cancer
Other studies have also cited difficulties in reporting on patient. Cancer Chemother Pharmacol. 2011;​67(3):​489–501.
brain invasion.33,34 When Brokinkel et al. conducted a re- 11. Luyken C, Blümcke I, Fimmers R, et al. The spectrum of
cent systematic review on brain invasion in meningiomas, long-term epilepsy-associated tumors:​long-term seizure and
they concluded that the varying techniques for identifying tumor outcome and neurosurgical aspects. Epilepsia. 2003;​
brain invasion among researchers have made it difficult 44(6):​822–830.
12. Sirven JI, Wingerchuk DM, Drazkowski JF, et al. Seizure
to consolidate findings.14 Further work to standardize the prophylaxis in patients with brain tumors:​a meta-analysis.
identification of brain invasion is necessary in order to bet- Mayo Clin Proc. 2004;​79(12):​1489–1494.
ter understand this key variable. Finally, relatively small 13. Hess K, Spille DC, Adeli A, et al. Brain invasion and the risk
sample sizes for each genomic subgroup could account for of seizures in patients with meningioma. J Neurosurg. 2018;​
observing certain trends, but not significant results. 130(3):​789–796.
14. Brokinkel B, Hess K, Mawrin C. Brain invasion in meningio-
mas-clinical considerations and impact of neuropathological
Conclusions evaluation:​a systematic review. Neuro Oncol. 2017;​19(10):​
Patients with meningiomas who present with seizures 1298–1307.
may indeed have a more aggressive meningioma and tend 15. Stockhammer F, Misch M, Helms HJ, et al. IDH1/2 muta-
tions in WHO grade II astrocytomas associated with local-
to experience a more complicated clinical course. While ization and seizure as the initial symptom. Seizure. 2012;​
patients with preoperative seizures may be more likely to 21(3):​194–197.
have recurrent tumors, pursuing a more extensive resec- 16. Vecht CJ, Kerkhof M, Duran-Pena A. Seizure prognosis in
tion may help in weaning them off of AED use postop- brain tumors:​new insights and evidence-based management.
eratively. Patients who undergo radiation and/or have a Oncologist. 2014;​19(7):​751–759.
recurrence are at risk for postoperative seizures, regard- 17. Takahashi JA, Ueba T, Hashimoto N, et al. The combination
of mitotic and Ki-67 indices as a useful method for predict-
less of the amount of residual tumor left after resection. ing short-term recurrence of meningiomas. Surg Neurol.
Although further work is needed, our ability to correlate 2004;​61(2):​149–156.
a mutation in the NF2 gene with the occurrence of preop- 18. Youngblood MW, Duran D, Montejo JD, et al. Correlations
erative seizures in meningioma patients may be the first between genomic subgroup and clinical features in a cohort
step to establishing a more in-depth understanding of the of more than 3000 meningiomas. J Neurosurg. 2020;​133(5):​
relationship between seizures and meningiomas. For in- 1345-1354.
stitutions without WES capabilities, encountering patients 19. Harmancı AS, Youngblood MW, Clark VE, et al. Integrated
genomic analyses of de novo pathways underlying atypical
with meningiomas and preoperative seizures, along with a meningiomas. Nat Commun. 2017;​8:​14433.
convexity location and PTBE, may signal a tumor with an 20. Clark VE, Erson-Omay EZ, Serin A, et al. Genomic analy-
underlying NF2 mutation and one that may be more likely sis of non-NF2 meningiomas reveals mutations in TRAF7,
to have characteristics of a more aggressive meningioma. KLF4, AKT1, and SMO. Science. 2013;​339(6123):​1077–1080.
21. Baumgarten P, Sarlak M, Baumgarten G, et al. Focused re-
view on seizures caused by meningiomas. Epilepsy Behav.
Acknowledgments 2018;​88:​146–151.
We are grateful to the patients who contributed to this study. 22. Islim AI, Mohan M, Moon RDC, et al. Incidental intracra-
This study was supported by the Gregory M. Kiez and Mehment nial meningiomas:​a systematic review and meta-analysis of
Kutman Foundation. This study was also supported by the Con- prognostic factors and outcomes. J Neurooncol. 2019;​142(2):​
necticut Brain Tumor Alliance. 211–221.

J Neurosurg Volume 135 • September 2021 843

 
*42 *
Gupte et al.

23. Kawaguchi T, Kameyama S, Tanaka R. Peritumoral edema 35. Clark VE, Harmancı AS, Bai H, et al. Recurrent somatic mu-
and seizure in patients with cerebral convexity and parasag- tations in POLR2A define a distinct subset of meningiomas.
ittal meningiomas. Neurol Med Chir (Tokyo). 1996;​36(8):​ Nat Genet. 2016;​48(10):​1253–1259.
568–574. 36. Smith MJ, O’Sullivan J, Bhaskar SS, et al. Loss-of-function
24. Skardelly M, Rother C, Noell S, et al. Risk factors of preop- mutations in SMARCE1 cause an inherited disorder of mul-
erative and early postoperative seizures in patients with me- tiple spinal meningiomas. Nat Genet. 2013;​45(3):​295–298.
ningioma:​a retrospective single-center cohort study. World
Neurosurg. 2017;​97:​538–546.
25. Islim AI, Ali A, Bagchi A, et al. Postoperative seizures in Disclosures
meningioma patients:​improving patient selection for antiepi- The authors report no conflict of interest concerning the materi-
leptic drug therapy. J Neurooncol. 2018;​140(1):​123–134. als or methods used in this study or the findings specified in this
26. Zhang BO, Wang D, Guo Y, Yu J. Clinical multifactorial paper.
analysis of early postoperative seizures in elderly patients
following meningioma resection. Mol Clin Oncol. 2015;​3(3):​ Author Contributions
501–505.
27. Das RR, Artsy E, Hurwitz S, et al. Outcomes after discon- Conception and design: Moliterno, Gupte, Li. Acquisition of
tinuation of antiepileptic drugs after surgery in patients with data: Moliterno, Gupte, Yalcin, Youngblood, Miyagishima,
low grade brain tumors and meningiomas. J Neurooncol. Mishra-Gorur, Zhao, Huttner, McGuone, Blondin, Contessa,
2012;​107(3):​565–570. Fulbright, Erson-Omay. Analysis and interpretation of data:
28. Boetto J, Bielle F, Sanson M, et al. SMO mutation status de- Moliterno, Gupte, Li, Jin, Yalcin, Erson-Omay. Drafting
fines a distinct and frequent molecular subgroup in olfactory the article: Moliterno, Gupte, Li, Jin, Yalcin, Erson-Omay.
groove meningiomas. Neuro Oncol. 2017;​19(3):​345–351. Critically revising the article: Moliterno, Gupte, Li, Jin,
29. Zheng Z, Chen P, Fu W, et al. Early and late postoperative Yalcin, Miyagishima, Mishra-Gorur, Zhao, Antonios, Gunel,
seizure outcome in 97 patients with supratentorial menin- Erson-Omay. Reviewed submitted version of manuscript: all
gioma and preoperative seizures:​a retrospective study. J authors. Approved the final version of the manuscript on behalf of
Neurooncol. 2013;​114(1):​101–109. all authors: Moliterno. Statistical analysis: Gupte, Li, Jin, Yalcin,
30. Nanda A, Bir SC, Maiti TK, et al. Relevance of Simpson Erson-Omay.
grading system and recurrence-free survival after surgery for
World Health Organization Grade I meningioma. J Neuro- Supplemental Information
surg. 2017;​126(1):​201–211. Online-Only Content
31. Gousias K, Schramm J, Simon M. The Simpson grading Supplemental material is available with the online version of the
revisited:​aggressive surgery and its place in modern menin- article.
gioma management. J Neurosurg. 2016;​125(3):​551–560. Supplemental Tables and Fig. https://thejns.org/doi/suppl/​
32. Sughrue ME, Rutkowski MJ, Chang EF, et al. Postoperative 10.3171/​2020.7.JNS201042.
seizures following the resection of convexity meningiomas:​
are prophylactic anticonvulsants indicated? Clinical article. J Correspondence
Neurosurg. 2011;​114(3):​705–709.
Jennifer Moliterno: Yale School of Medicine, New Haven, CT.
33. Spille DC, Heß K, Sauerland C, et al. Brain invasion in me-
[email protected].
ningiomas:​incidence and correlations with clinical variables
and prognosis. World Neurosurg. 2016;​93:​346–354.
34. Backer-Grøndahl T, Moen BH, Arnli MB, et al. Immunohis-
tochemical characterization of brain-invasive meningiomas.
Int J Clin Exp Pathol. 2014;​7(10):​7206–7219.

844 J Neurosurg Volume 135 • September 2021

*42 *
 LABORATORY INVESTIGATION
J Neurosurg 135:845–854, 2021

Efficacy of decitabine in malignant meningioma cells:

relation to promoter demethylation of distinct tumor
suppressor and oncogenes and independence from TERT
*Louise Stögbauer,1 Christian Thomas, Dr med,2 Andrea Wagner,1 Nils Warneke, Dr med,1
Eva Christine Bunk, Dr med, PhD,1 Oliver Grauer, Dr med, PhD,3 Julian Canisius,1
Werner Paulus, Dr med,2 Walter Stummer, Dr med,1 Volker Senner, Dr rer nat,2 and
Benjamin Brokinkel, Dr med1
1
Department of Neurosurgery, 2Institute of Neuropathology, and 3Department of Neurology, University Hospital Münster, North
Rhine-Westphalia, Germany

OBJECTIVE  Chemotherapeutic options for meningiomas refractory to surgery or irradiation are largely unknown. Hu-
man telomerase reverse transcriptase (hTERT) promoter methylation with subsequent TERT expression and telomerase
activity, key features in oncogenesis, are found in most high-grade meningiomas. Therefore, the authors investigated the
impact of the demethylating agent decitabine (5-aza-2ʹ-deoxycytidine) on survival and DNA methylation in meningioma
cells.
METHODS  hTERT promoter methylation, telomerase activity, TERT expression, and cell viability and proliferation were
investigated prior to and after incubation with decitabine in two benign (HBL-52 and Ben-Men 1) and one malignant
(IOMM-Lee) meningioma cell line. The global effects of decitabine on DNA methylation were additionally explored with
DNA methylation profiling.
RESULTS  High levels of TERT expression, telomerase activity, and hTERT promoter methylation were found in IOMM-
Lee and Ben-Men 1 but not in HBL-52 cells. Decitabine induced a dose-dependent significant decrease of proliferation
and viability after incubation with doses from 1 to 10 µM in IOMM-Lee but not in HBL-52 or Ben-Men 1 cells. However,
effects in IOMM-Lee cells were not related to TERT expression, telomerase activity, or hTERT promoter methylation.
Genome-wide methylation analyses revealed distinct demethylation of 14 DNA regions after drug administration in the
decitabine-sensitive IOMM-Lee but not in the decitabine-resistant HBL-52 cells. Differentially methylated regions cov-
ered promoter regions of 11 genes, including several oncogenes and tumor suppressor genes that to the authors’ knowl-
edge have not yet been described in meningiomas.
CONCLUSIONS  Decitabine decreases proliferation and viability in high-grade but not in benign meningioma cell lines.
The effects of decitabine are TERT independent but related to DNA methylation changes of promoters of distinct tumor
suppressor genes and oncogenes.
https://thejns.org/abs/10.3171/2020.7.JNS193097

M
KEYWORDS  decitabine; human telomerase reverse transcriptase; hTERT; meningioma; methylation; oncology

eningiomas are the most common primary in- Telomerase provides telomere lengthening and there-
tracranial tumors and are usually treated by mi- fore hinders cell senescence and apoptosis in most ma-
crosurgical resection and/or radiation therapy. lignant tumors.2 Telomerase reverse transcriptase (TERT)
Despite gross-total removal, recurrence is common, and expression and telomerase activity have been shown in up
rates of postoperative relapse distinctly increase with ris- to 70% of meningiomas and increase with rising WHO
ing WHO grade. However, to date further effective treat- grade.3–7 Previous series reported human TERT (hTERT)
ment options, e.g., chemotherapy, are lacking for tumors promoter mutations in a subset of meningiomas (< 15%)
refractory to surgical treatment or irradiation.1 with both worse prognosis and increased TERT expres-

ABBREVIATIONS  ALT = alternative lengthening of the telomeres; decitabine = 5-aza-2ʹ-deoxycytidine; DMR = differentially methylated region; hTERT = human TERT; MS-
PCR = methylation-specific PCR; MTT = 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; PCR = polymerase chain reaction; qRT-PCR = quantitative real-time
PCR; TERT = telomerase reverse transcriptase.
SUBMITTED  November 19, 2019.  ACCEPTED  July 6, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS193097.
* V.S. and B.B. contributed equally to this work.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  845

*42 *
Stögbauer et al.

sion.8–11 Aside from mutations, methylation of the hTERT were conducted both for characterization and to confirm
promoter was found in over 50% of meningiomas in gen- the individual profile of the cell line; furthermore, poly-
eral and in over 90% of anaplastic lesions and is associated merase chain reaction (PCR)–based tests for mycoplasma
with TERT expression and worse prognosis.4,11,12 Recently, contamination were conducted and periodically repeated
Spiegl-Kreinecker et al. initially reported the efficiency (Venor GeM Classic, Minerva Biolabs). Twenty-four hours
of targeted chemotherapy in hTERT-mutated meningio- after seeding, the medium was changed and decitabine
mas in vivo.13 However, because of the low frequency of was administered at various concentrations (0 [control], 1,
hTERT promoter mutations,10 presumably only a very 2, 5, and 10 µM) as described previously.24
small subgroup of meningioma patients could benefit from
future targeted drugs. In contrast, hTERT promoter meth- DNA Extraction and hTERT Promotor Methylation
ylation is found in up to 90% of meningiomas and, simi- Analysis
lar to findings in childhood brain tumors and other neo- Promoter methylation was determined by methylation-
plasms, hTERT promoter methylation in meningiomas has specific PCR (MS-PCR). Genomic DNA was isolated
been found to be associated with hTERT upregulation.2,14 (Maxwell 16 LEV blood DNA purification kit AS1290,
Although the underlying mechanisms of upregulation due Promega) from the cell lines, both untreated and after
to hypermethylation are not yet fully understood, targeted treatment with decitabine (1 × 106 cells, 10 µM, 72 hours),
therapy aimed at inhibiting this upregulation may help and then immediately subjected to bisulfide conversion
a considerable portion of meningioma patients. Further using a commercial kit according to the manufacturer’s
studies reported methylation of distinct tumor suppressor instructions (EZ DNA Methylation-Gold-Kit, Zymo Re-
genes, such as BRCA1 and DLC1, and other cancer-related search). DNA was then amplified by MS-PCR and visual-
genes in meningiomas.15–18 Correspondingly, genome- ized by gel electrophoresis using Agilent 2200 Tape Sta-
wide DNA analyses identified distinct methylation groups tion (Agilent Technologies) as described previously.4
to be strongly associated with tumor recurrence,19 further The following primer pairs were used: forward primer,
underlining the impact of epigenetic alterations such as 5ʹ-GGGAAGTGTTGTAGGGAGGTATTT-3ʹ (methylat-
hypermethylation on both tumorigenesis and prognosis in ed and unmethylated) and reverse primers 5ʹ-CGTACG- ­
meningiomas. ACGACCCTTTAACCG-3ʹ
­ (methylated) and 5ʹ-CAT-
Decitabine (5-aza-2ʹ-deoxycytidine) is a hypomethyl- ACAACAACCCTTTAACCA-3ʹ
­­ ­­ (unmethylated). MS-
ating agent used for the treatment of myeloid malignan- PCR was repeated in biologically independent triplicates.
cies. The antioncogenetic effect of the drug is basically
induced by global DNA demethylation, including promot- Telomerase Activity
ers of tumor suppressor and oncogenes.20 Correspond-
ingly, decitabine was shown to induce demethylation of Telomerase activity was assessed using the Telo
the hTERT promoter with a subsequent decrease of TERT TAGGG telomerase PCR enzyme-linked immunosorbent
expression, telomerase activity, and apoptosis in chronic assay (ELISA) kit (Roche), which is based on the Telo-
myeloid leukemia (CML) cells.21 Similar effects were re- meric Repeat Amplification Protocol assay, and subse-
ported after decitabine exposition of sarcoma cell lines,22 quently nonradioactive ELISA detection. The experiment
while corresponding investigations in meningiomas are was conducted with 2 × 105 cells from each cell line both
lacking. before and after exposure to decitabine (10 µM, 48 hours)
In this study, we investigated the impacts of decitabine and with strict adherence to the kit’s instructions. Posi-
on cell proliferation and viability and on hTERT promoter tive and negative controls were selected according to the
methylation, telomerase activity, and TERT expression in manufacturer’s instructions. In detail, negative controls
meningioma cells. In addition, we performed comparative were created by incubation of the meningioma cells with
genome-wide DNA methylation analyses to delineate the RNase, DNase free, destroying any telomerase activity.
hypomethylating effects of decitabine on additional onco- Immortalized telomerase-expressing kidney cells were
genes and tumor suppressor genes. used as positive controls.

TERT Expression
Methods TERT expression was analyzed by both quantitative
Meningioma Cell Lines and Demethylation Treatment real-time PCR (qRT-PCR) and immunohistochemical
Three established human meningioma cell lines (HBL- staining. RNA was isolated using the Maxwell 16 LEV
52, benign; Ben-Men 1, benign; and IOMM-Lee, malig- simplyRNA tissue kit (Promega) according to the man-
nant) were cultured and processed under standard condi- ufacturer’s protocol (1 × 106 cells, untreated vs treated
tions at 37°C and 5% CO2. While the HBL-52 line showed with 10 µM decitabine for 72 hours). cDNA was gener-
a limited number of cell divisions until cells reached ated using a high-capacity cDNA reverse transcription
senescence, the second benign cell line, Ben-Men 1, ex- kit (Applied Biosystems) following the manufacturer’s
hibited permanent cell division due to hTERT-mediated instructions. qRT-PCR was performed in triplicate using
immortalization.23 Cells were cultured in DMEM (high TaqMan gene expression assays (Life Technologies) for
glucose) supplemented with 10% fetal calf serum, 100 U/ monoplex reactions according to standard protocols and
ml penicillin, and 0.01 mg/ml streptomycin (all cell cul- using the primers hTERT (Hs00972656_m1) and GAPDH
ture reagents from Sigma). Cells were subcultivated until (Hs02758991_g1). The ΔΔCt method was used to analyze
they reached confluency. Short tandem repeat analyses the transcription levels of mRNA.

846 J Neurosurg Volume 135 • September 2021

*42 *
 Stögbauer et al.

Immunohistochemical staining was performed after Time-Lapse Microscopy

the cells were seeded in a 6-well plate (1 × 105/2 ml/well), Time-lapse microscopy of cell cultures was done using
incubated for 24 hours, and then exposed to decitabine the live cell movie analyzer JuLI Br (NanoEnTek). The
treatment for a further 48 hours. A polyclonal rabbit anti- system is equipped with two scope units (×4 objectives)
body against hTERT (AP07776PU-N, dilution: 1:200, Ac- placed in an incubator under standard conditions (37°C,
ris Antibodies GmbH) was used according to the manu- 5% CO2, 95% humidity); 1 × 105 cells were seeded in
facturer’s instructions in combination with a biotinylated 3.5-cm culture dishes and cultivated with or without 10
goat anti-rabbit secondary antibody and visualized with µM decitabine. Observation was done over a period of 48
a standardized AEC (3-amino-9-ethylcarbazol) stain- hours, with pictures taken every 10 minutes.
ing method (Vectastain ABC kit, Vector Laboratories).
hTERT expression was evaluated as “absent” or “present” DNA Methylation Analyses
(staining in ≥ 5% of the analyzed tumor cells) in dichoto- DNA methylation profiles of IOMM-Lee and HBL-52
mous analyses by two independent observers (L.S. and cells were analyzed using the Infinium MethylationEPIC
V.S.). BeadChip kit (Illumina) according to the manufacturer’s
instructions and repeated independently in biological tripli-
Expression of ATRX, DAXX, TRIM58, LMO3, and MAL2 cates. For each replicate, a total of 500 ng of genomic DNA
Immunohistochemical staining for ATRX and DAXX was extracted and subjected to bisulfide conversion using
expression was performed to detect potential mechanisms the ZymoResearch EZ DNA methylation kit (Zymo Re-
of alternative lengthening of the telomeres (ALT) using search Corp.). On-chip quality metrics of all samples were
polyclonal rabbit antibodies (anti-ATRX, 1:300 dilution, carefully controlled. Raw IDAT files were loaded into the R
catalog no. HPA001906, and anti-DAXX, 1:100 dilu- environment using the minfi package (version 1.28.0).26 Af-
tion, catalog no. HPA008736, both Sigma-Aldrich). Sub- ter quantile normalization, unsupervised hierarchical clus-
sequently, the standard streptavidin-biotin-peroxidase tering was performed using the beta values of the 1000 most
method and diaminobenzidine were used for analysis ac- variably methylated probes as measured by standard devia-
cording to the manufacturer’s instructions. Loss of ATRX tion. Samples were clustered using Euclidean distance as the
and DAXX expression was evaluated by two independent distance measure and Ward’s linkage as the clustering meth-
observers (L.S. and V.S.) as described previously.25 Im- od (x-axis). Methylation probes were reordered by hierarchi-
munohistochemical staining of TRIM58 (anti-TRIM58, cal clustering using Euclidean distance and Ward’s linkage
1:200 dilution, catalog no. HPA023637, Sigma Aldrich), as the clustering method (y-axis). Differentially methylated
LMO3 (anti-LMO3, 1:100 dilution, catalog no. 230490, regions (DMRs) were identified using DMRcate.27
Abcam plc), and MAL2 (anti-Mal2, 1:100 dilution, catalog
no. ab217919, Abcam plc) was performed using an auto- Statistical Analyses
matic immunostainer (Link48, DAKO) and classified as Calculations were performed using standard statisti-
either absent or present. cal software (R Statistical Environment, R Foundation),
and data are described using standard statistics, with me-
Cell Viability and Proliferation dian and range for continuous variables and absolute and
To investigate the impact of decitabine treatment on relative frequencies for categorical variables. Fisher’s ex-
cell viability, a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl- act and Mann-Whitney U-tests were performed to com-
tetrazolium bromide (MTT) assay was performed. Cells pare categorical and continuous variables, respectively.
were seeded into four 96-well plates in 12 technical rep- A p value < 0.05 was considered statistically significant
licates (1 × 104/200 µl/well) and incubated at 37° and 5% throughout the whole analyses. All reported p values are
CO2. After 24 hours, the drug was administered in con- two-sided. According to legal regulations from the local
centrations from 0 to 10 µM and the cell cultures were left ethics committee for studies exclusively involving estab-
untouched for various times of exposure of 24, 48, or 72 lished immortalized cell lines but neither humans nor ani-
hours. A plate analyzed before treatment (0 hours) served mals, no ethical approval was necessary for this study.
as a baseline reference. For the quantification of mitochon-
drial metabolism, the incubation medium was replaced Results
with a 100-µl MTT solution (0.5 mg/ml) and incubated for RT-PCR showed hTERT expression in the untreated
75 minutes at 37°C and 5% CO2. The MTT solution was malignant IOMM-Lee and in the benign hTERT-mediated
then discarded and 200 µl of propan-2-ol was added to immortalized Ben-Men 1 cell lines. In contrast, hTERT
lyse the cells and dissolve the released formazan crystals. expression was absent in the benign HBL-52 cells (Fig.
The absorption was measured at 570 nm using the Glo- 1, left). These findings were also confirmed by immuno-
Max Reader (Promega). All experiments were repeated in histochemical analyses (Supplemental Fig. 1). In line with
biologically independent triplicates. these findings, MS-PCR revealed a strong hTERT pro-
Cell proliferation was investigated by standard immu- moter methylation in IOMM-Lee and Ben-Men 1 cells,
nofluorescence staining using the monoclonal mouse anti- while methylation was barely detected in HBL-52 cells
body MIB1 against Ki-67 (M7240, dilution 1:100, DAKO) (Supplemental Fig. 2). Correspondingly, telomerase ac-
in combination with an Alexa Fluor 488–labeled second- tivity was absent in HBL-52 but was strong in both the
ary goat anti-mouse antibody (A-11029, dilution 1:400, IOMM-Lee and Ben-Men 1 cell lines (Fig. 1, right). On
Invitrogen). the other hand, immunohistochemical staining generally

J Neurosurg Volume 135 • September 2021 847

 
*42 *
Stögbauer et al.

FIG. 1. TERT expression (left) and telomerase activity (right) in different meningioma cell lines prior to and after decitabine treat-
ment. Real-time PCR shows hTERT expression in Ben-Men 1 and IOMM-Lee but not in HBL-52 cells. Although the levels of ex-
pression seemed to increase after decitabine treatment in IOMM-Lee and Ben-Men 1, this difference did not reach significance in
statistical analyses (p = 0.05 and p = 0.077, respectively). TERT expression was not detectable in HBL-52 cells (left). Correspond-
ingly, the Telomeric Repeat Amplification Protocol assay revealed distinct telomerase activity in IOMM-Lee and Ben-Men 1 but not
in HBL-52 cells (right). There were no statistically significant changes in activity following decitabine treatment (10 µM). - = before
decitabine treatment; + = after decitabine treatment; Dec = decitabine; n.d. = not detectable; n.s. = not statistically significant.

revealed both ATRX and DAXX maintenance in all cell activity, or hTERT expression were found after decitabine
lines, indicating the absence of ALT in all analyzed me- treatment in Ben-Men 1 cells. Moreover, despite dis-
ningioma cell lines (Supplemental Fig. 3). tinctly reducing viability, decitabine did not significantly
alter hTERT promoter methylation, hTERT expression,
Decitabine Impacts Viability and Proliferation but Not or telomerase activity in IOMM-Lee cells. As expected,
TERT-Related Alterations in High-Grade Meningioma Cell hTERT-related parameters did not change after decitabine
Lines exposition in HBL-52 cells (Figs. 1–3 and Supplemental
In IOMM-Lee cells, reduced viability was found at 48 Fig. 2).
hours following incubation with higher decitabine doses (5
and 10 µM). The effects of decitabine on viability increased Effects of Decitabine in Malignant Meningioma Cells Are
and were also clearly detectable after incubation with lower Associated With Demethylation of Promoters of Distinct
concentrations (1 and 2 µM) after 72 hours. At that time, Tumor Suppressors and Oncogenes
however, viability was no longer related to the decitabine Given the lack of effect on methylation of the hTERT
dose (Fig. 2). Similarly, the Ki-67 labeling index decreased promoter, explorative genome-wide DNA methylation
after decitabine treatment (p = 0.008, 48 hours after treat- analyses prior to and 72 hours after decitabine exposi-
ment with 10 µM; Fig. 3 and Supplemental Fig. 3). Corre- tion (10 µM) were conducted in HBL-52 and IOMM-Lee
spondingly, in time-lapse microscopy of IOMM-Lee cells, cell lines. In Ben-Men 1 cells, hTERT-mediated immor-
proliferation was distinctly decreased in decitabine-treated talization and the effects of decitabine on viability and
compared with untreated cells (Video 1). proliferation were not investigated. As expected from
VIDEO 1. Time-lapse microscopy of IOMM-Lee cells over 48 hours. the MS-PCR, CpGs targeting the hTERT promotor re-
Proliferation was distinctly lower in meningioma cells treated with gion and gene did not show changes in methylation levels
decitabine (right, 10 µM) than in untreated cells (left). Copyright upon decitabine treatment (Fig. 4A). Density plots of the
Benjamin Brokinkel. Used with permission. Click here to view. methylation beta values showed a bimodal distribution
In contrast, no effects of decitabine on viability or pro- across all samples. While there were no global changes
liferation were found in Ben-Men 1 (p = 0.10) or HBL-52 in HBL-52 cells after drug treatment, decitabine induced
(p = 0.143) cell lines (Fig. 3) even after incubation at doses a shift of the high methylation peak (average beta values
up to 10 µM. ≥ 0.75) to lower methylation values in IOMM-Lee cells,
To explore the effects of decitabine on the hTERT indicating a drug-induced demethylation (Fig. 4B, arrow).
promoter, hTERT methylation and expression as well as Unsupervised hierarchical clustering clearly separated
telomerase activity were analyzed in all 3 cell lines prior decitabine-treated IOMM-Lee replicates from their un-
to and 72 hours after decitabine administration. In line treated counterparts, whereas HBL-52 replicates did not
with the lack of efficacy on viability, no statistically signif- cluster together according to their treatment status (Fig.
icant changes of hTERT promoter methylation, telomerase 4C). In line with this finding, comparison of decitabine-

848 J Neurosurg Volume 135 • September 2021

*42 *
 Stögbauer et al.

FIG. 2. MTT-based meningioma cell viability assays beginning 24 hours after treatment with decitabine at different concentrations.
IOMM-Lee cells (A) showed significantly reduced viability (***p < 0.001) compared with the untreated control group. No significant
effects of decitabine treatment were seen in Ben-Men 1 (B) and HBL-52 (C) cells.

treated with control HBL-52 cells showed no significant

DMRs. In contrast, DMR analysis in IOMM-Lee re-
vealed 14 significantly differentially methylated regions
following decitabine treatment (Table 1). Among those,
13 DMRs showed lower and 1 DMR (NKX-6-2-001)
showed higher DNA methylation levels in decitabine-
treated cells. Demethylated regions cover overlapping
promoter regions (± 2000 bp from the transcription start
site) of 11 genes, including TRIM58, FAM84B, ELOVL2,
SLC4A10, MAL2, LMO3, PEX5L, DIO3, PHYHIPL,
ELAVL2, and ARL4A.

Expression of TRIM58, MAL2, and LMO3 in Meningiomas

According to our results from DNA methylation analy-
ses, the expression of selected genes (TRIM58, LMO3, and FIG. 3. Proliferative activity following decitabine treatment. Immu-
MAL2) found to be demethylated by decitabine was inves- nofluorescence revealed a distinctly reduced Ki-67 labeling index in
tigated by immunohistochemical staining. In 10 samples IOMM-Lee cells after decitabine treatment (*p = 0.008). In contrast, the
proliferation index did not significantly change 48 hours after decitabine
from formalin-fixed paraffin-embedded meningioma administration (10 µM) in Ben-Men 1 (p = 0.10) or HBL-52 (p = 0.143)
samples, the expression of TRIM58, LMO3, and MAL2 cells. Representative immunofluorescence imaging is shown in Supple-
was found in 9, 10, and 8 samples, respectively (Supple- mental Fig. 3.
mental Fig. 4).

J Neurosurg Volume 135 • September 2021 849

 
*42 *
Stögbauer et al.

FIG. 4. Results of the DNA methylation analyses. Methylation profiling confirmed similar DNA methylation levels of the hTERT promoter and gene,
located on chromosome 5, before and after decitabine treatment in IOMM-Lee (A) and HBL-52 (not shown) cells.  FIG. 4. (continued) →

850 J Neurosurg Volume 135 • September 2021

*42 *
 Stögbauer et al.

FIG. 4. Density plots of beta methylation levels revealed slight methylation changes in the high methylation peak (difference
between the treatment and the control peaks, arrow) of IOMM-Lee but not of HBL-52 cells (B). Heatmaps showing unsupervised
hierarchical clustering of IOMM-Lee and HBL-52 cells using the 1.000 most differentially methylated probes across each data
set. The relative DNA methylation levels are shown in red (high) and blue (low). Decitabine treatment in IOMM-Lee and HBL-52
meningioma cell lines revealed slight demethylation after decitabine treatment in IOMM-Lee but not in HBL-52 cells (C, analysis in
biological triplets; beta values: mean absolute deviation).

Discussion Lee meningioma cells. While the onset of efficacy was

TERT Alterations In Vivo Correspond to In Vitro earlier after treatment with higher decitabine doses, this
Observations dose-dependent relation dissolved at 72 hours after incu-
bation. Moreover, in both benign meningioma cell lines
Corresponding to findings in vivo, hTERT promoter
no effects of decitabine on either viability or proliferation
methylation, hTERT expression, and telomerase activity
were found in malignant IOMM-Lee cell lines but were were detectable. Considering similar observations from
distinctly lower in the benign HBL-52 cells, while ALT previous studies investigating the effects of chemotherapy
was absent throughout all cell lines.4,5,11,16,25 In contrast, in meningioma cell lines,28 decreased proliferative activity
hTERT promoter methylation as well as high levels of might significantly contribute to reduced chemosensitiv-
hTERT expression and telomerase activity were also found ity in benign cells compared with high-grade meningioma
in the benign Ben-Men 1 cells, presumably reflecting the cells. However, our results strongly encourage further in-
TERT-mediated immortalization.23 Although methylation vestigations into the efficiency of decitabine for the treat-
is usually considered to induce genome silencing, TERT ment of high-grade meningiomas in animal models and
upregulation by promoter methylation has been reported primary meningioma cell lines. Moreover, the molecular
in meningiomas and pediatric brain tumors as well as mechanisms underlying the impact of decitabine on viabil-
other malignancies.2,4,14 Underlying mechanisms remain ity and proliferation in meningiomas remain to be deter-
largely unexplored but might include 3D changes in the mined. Whereas decitabine was shown to induce hTERT
conformation of chromatin or methylation-induced block- promoter demethylation and apoptosis in CML cells,21 no
age of repressor binding sites.2 These observations are also effects of decitabine on TERT expression, telomerase ac-
in line with findings from Maes et al. showing strong cor- tivity, or hTERT promoter methylation were found in any
relations between in vivo and in vitro hTERT expression of the analyzed meningioma cell lines in our study. The
and telomerase activity in primary meningioma cells.7 lack of efficacy in Ben-Men 1 cells presumably results
from additional constitutive TERT upregulation by the
Decitabine Impacts Viability and Proliferation but Not recombinant constructed promoter, which appears not to
TERT-Related Alterations in High-Grade Meningioma Cell be affected by decitabine-induced demethylation. Another
Lines possible explanation might be additional mechanisms of
Remarkably, we observed a distinct impact of decitabine TERT regulation in Ben-Men 1 cells, e.g., by promoter
treatment on both proliferation and viability in IOMM- mutations, which have not been investigated in our study.

TABLE 1. Decitabine-induced differentially methylated DNA regions

No. Chromosome Start End Width  No. of CpGs Overlapping Promoter Gene Involved in Malignancy (references)
1 Chr1 248020692 248021163 472 5 TRIM58-001 Lung,17 colorectal23
2 Chr8 127568650 127569677 1028 11 FAM84B-001 Prostate42
3 Chr6 11044541 11044558 18 3 ELOVL2-001 Melanoma8
4 Chr2 162280341 162280741 401 4 SLC4A10-008 NA
5 Chr8 120220410 120220740 331 5 MAL2-201 Breast, colorectal, pancreatic,9 meningioma12
3 22

6 Chr5 81570383 81570494 112 2 NA NA

7 Chr12 16758465 16758943 479 8 LMO3-201 Meningioma,38 gastric,35 hepatic7
8 Chr3 179754239 179755696 1458 12 PEX5L-005 NA
9 Chr14 102026812 102026953 142 4 DIO3-201 Lung,29 prostate44
10 Chr10 60935910 60936925 1016 16 PHYHIPL-002 NA
11 Chr9 23821428 23822002 575 10 ELAVL2-002 NA
12 Chr7 12725802 12726431 630 8 ARL4A-002 NA
13 Chr2 114737046 114737730 685 7 AC010982.1-001 NA
14 Chr10 134599984 134600357 374 4 NKX6-2-001 Bladder19
NA = not available.
Genome-wide DNA methylation analyses of IOMM-Lee cells after decitabine treatment revealed 14 differentially methylated regions. Six of 13 demethylated regions
were located in promoter regions of previously described tumor suppressor and oncogenes. Among those differentially methylated regions, to our knowledge only 2
(LMO3 and MAL2) have not yet been investigated in meningiomas.

J Neurosurg Volume 135 • September 2021 851

 
*42 *
Stögbauer et al.

Effects of Decitabine in Malignant Meningioma Cells been obscured. Finally, all investigations were only per-
Are Associated With Demethylation of Distinct Tumor formed in established immortalized but not in primary
Suppressor and Oncogenes meningioma cells, so the transferability of our results to in
In a previous analysis on IOMM-Lee cell lines, Liu et al. vivo conditions is limited and requires both further labora-
revealed upregulation in 2 of 9 selected oncogenes (6-meth- tory and clinical investigations.
ylguanine-DNA methyltransferase [MGMT] and death-
associated protein kinase [DAPK1]) following decitabine Conclusions
treatment.24 Given the variety of demethylation effects in
different malignancies,20,22 we subsequently conducted ex- In conclusion, we showed a distinct efficacy of decitabine
plorative genome-wide DNA methylation analyses prior to in malignant IOMM-Lee but not in benign meningioma cell
and following decitabine treatment. Similarly to cell vi- lines. While effects on viability were not related to hTERT,
ability and proliferation, DNA methylation did not signifi- we identified several oncogenes and tumor suppressor genes
cantly change following treatment in HBL-52 cell lines. to be hypomethylated after decitabine treatment. Hence,
In contrast, 850 K methylation analyses of IOMM-Lee aside from providing potential molecular mechanisms un-
revealed hypomethylated promoter regions of 11 genes derlying decitabine efficacy, our results strongly encourage
following decitabine treatment (Table 1). Given the slight future studies investigating decitabine in high-grade menin-
effect on global DNA methylation, this finding suggests giomas in primary cell cultures and animal models.
a specific demethylation caused by decitabine treatment.
In fact, similar findings of rather specific DNA demeth- Acknowledgments
ylations were found by Hagemann et al. when treating co- This work (B.B.) was supported by the Innovative Medical
lon cancer and leukemia cells with decitabine. Regarding Research Fund of the University of Münster Medical School
their results and the findings from Grandjenette et al., it is (grant BR221703). C.T. is supported by the DFG (TH 2345/1-1).
even more remarkable that decitabine apparently induces We cordially thank Christian Mawrin and Elmar Kirches for
demethylation in different promoter regions depending on generously providing the IOMM-Lee cell line and their helpful
support.
the treated malignancy.21,29
Several of the promoter regions hypomethylated follow-
ing decitabine treatment have been described in the onco- References
genesis of nonneural malignancies, including lung, gastric,   1. Perry A, Louis DN, von Deimling A, et al. Meningiomas. In:​
pancreatic, colorectal, and hepatic cancer, and in malig- Louis DN, Ohgaki H, Wiestler OD, et al, eds. WHO Clas-
nant melanoma.30–40 Remarkably, a number of phase I or II sification of Tumors of the Central Nervous System. Interna-
studies as well as in vitro analyses reported the efficacy of tional Agency on Cancer Research;​2016:232–245.
​
decitabine in these lesions, e.g., in malignant melanoma as   2. Castelo-Branco P, Choufani S, Mack S, et al. Methylation of
the TERT promoter and risk stratification of childhood brain
well as prostate and gastric cancer.41–43 While generally hy- tumours:​an integrative genomic and molecular study. Lancet
pomethylating DNA, decitabine also induced CpG hyper- ​ ​
Oncol. 2013;14(6):534–542.
methylation of the promoter region of NKX6-2, described   3. Cabuy E, de Ridder L. Telomerase activity and expression of
in the oncogenesis of bladder cancer.44 In meningiomas, telomerase reverse transcriptase correlated with cell prolif-
LMO3 expression was found to be higher in atypical than eration in meningiomas and malignant brain tumors in vivo.
in benign tumors.45 While correlations with prognosis have ​ ​
Virchows Arch. 2001;439(2):176–184.
not yet been investigated, Gao et al. reported downregula-   4. Fürtjes G, Köchling M, Peetz-Dienhart S, et al. hTERT
tion of MAL2 expression by de novo DNA methylation in promoter methylation in meningiomas and central nervous
hemangiopericytomas. J Neurooncol. 2016;130(1):79–87.
​ ​
malignant meningiomas.16 In our own explorative immuno-   5. Kalala JP, Maes L, Vandenbroecke C, de Ridder L. The
histochemical analyses, we revealed distinct expressions of hTERT protein as a marker for malignancy in meningiomas.
TRIM58, MAL2, and LMO3 in 9, 8, and 10, respectively, Oncol Rep. 2005;13(2):273–277.
​ ​
of 10 formalin-fixed paraffin-embedded meningioma sam-   6. Langford LA, Piatyszek MA, Xu R, et al. Telomerase activ-
ples. Hence, the roles of the identified tumor suppressor ity in ordinary meningiomas predicts poor outcome. Hum
and oncogenes in tumorigenesis and prognosis in menin- Pathol. 1997;28(4):416–420.
​ ​
giomas remain to be determined. Moreover, the effects of   7. Maes L, Kalala JP, Cornelissen R, de Ridder L. Telomerase
decitabine-induced promoter demethylation on subsequent activity and hTERT protein expression in meningiomas:​an
analysis in vivo versus in vitro. Anticancer Res. 2006;26(3B):
​ ​
gene expression should be further elucidated. 2295–2300.
The authors are aware of some limitations of the study.   8. Koelsche C, Sahm F, Capper D, et al. Distribution of TERT
As already mentioned, investigations of the effects of de- promoter mutations in pediatric and adult tumors of the ner-
methylation on expression of the described genes were not vous system. Acta Neuropathol. 2013;126(6):907–915.
​ ​
performed and the roles of the described differentially   9. Sahm F, Schrimpf D, Olar A, et al. TERT promoter mutations
methylated genes in both prognosis and tumorigenesis and risk of recurrence in meningioma. J Natl Cancer Inst.
in meningiomas are largely unknown. Similarly to previ- 2015;108(5):djv377.
​ ​
ous studies,21,46 in the present study we observed a clear 10. Lu VM, Goyal A, Lee A, et al. The prognostic significance
of TERT promoter mutations in meningioma:​a systematic
impact and therefore stopped analyses 72 hours following review and meta-analysis. J Neurooncol. 2019;142(1):1–10.
​ ​
decitabine treatment, while other groups performed analy- 11. Stögbauer L, Stummer W, Senner V, Brokinkel B. Telomerase
ses later after drug administration.24 Thus, potential later activity, TERT expression, hTERT promoter alterations, and
effects of decitabine on DNA methylation and viability, alternative lengthening of the telomeres (ALT) in meningio-
e.g., in the HBL-52 or Ben-Men 1 cell lines, might have ​ ​
mas—a systematic review. Neurosurg Rev. 2020;43(3):903–910.

852 J Neurosurg Volume 135 • September 2021

*42 *
 Stögbauer et al.

12. Peyre M, Gauchotte G, Giry M, et al. De novo and 31. Cheng Y, Hou T, Ping J, et al. LMO3 promotes hepatocel-
secondary anaplastic meningiomas:​a study of clinical and lular carcinoma invasion, metastasis and anoikis inhibition
histomolecular prognostic factors. Neuro Oncol. 2018;​20(8):​ by directly interacting with LATS1 and suppressing Hippo
1113–1121. signaling. J Exp Clin Cancer Res. 2018;​37(1):​228.
13. Spiegl-Kreinecker S, Lötsch D, Neumayer K, et al. TERT 32. Dai W, Liu H, Xu X, et al. Genetic variants in ELOVL2 and
promoter mutations are associated with poor prognosis and HSD17B12 predict melanoma-specific survival. Int J Cancer.
cell immortalization in meningioma. Neuro Oncol. 2018;​ 2019;​145(10):​2619–2628.
20(12):​1584–1593. 33. Eguchi D, Ohuchida K, Kozono S, et al. MAL2 expression
14. Leão R, Apolónio JD, Lee D, et al. Mechanisms of human predicts distant metastasis and short survival in pancreatic
telomerase reverse transcriptase (hTERT) regulation:​clinical cancer. Surgery. 2013;​154(3):​573–582.
impacts in cancer. J Biomed Sci. 2018;​25(1):​22. 34. Kajiura K, Masuda K, Naruto T, et al. Frequent silencing
15. Bujko M, Kober P, Rusetska N, et al. Aberrant DNA methyla- of the candidate tumor suppressor TRIM58 by promoter
tion of alternative promoter of DLC1 isoform 1 in meningio- methylation in early-stage lung adenocarcinoma. Oncotarget.
mas. J Neurooncol. 2016;​130(3):​473–484. 2017;​8(2):​2890–2905.
16. Gao F, Shi L, Russin J, et al. DNA methylation in the malig- 35. Li J, Li Y, Liu H, et al. The four-transmembrane protein
nant transformation of meningiomas. PLoS One. 2013;​8(1):​ MAL2 and tumor protein D52 (TPD52) are highly expressed
e54114. in colorectal cancer and correlated with poor prognosis.
17. Lombardi IAS, Faria MHG, Rabenhorst SHB, et al. Hyper- PLoS One. 2017;​12(5):​e0178515.
methylation of BRCA1 gene in meningioma in elderly males. 36. Liu M, Zhang X, Cai J, et al. Downregulation of TRIM58
Anticancer Res. 2018;​38(5):​2819–2822. expression is associated with a poor patient outcome and
18. Majchrzak-Celińska A, Paluszczak J, Szalata M, et al. DNA enhances colorectal cancer cell invasion. Oncol Rep. 2018;​
methylation analysis of benign and atypical meningiomas:​ 40(3):​1251–1260.
correlation between RUNX3 methylation and WHO grade. J 37. Molina-Pinelo S, Salinas A, Moreno-Mata N, et al. Impact
Cancer Res Clin Oncol. 2015;​141(9):​1593–1601. of DLK1-DIO3 imprinted cluster hypomethylation in smoker
19. Sahm F, Schrimpf D, Stichel D, et al. DNA methylation- patients with lung cancer. Oncotarget. 2016;​9(4):​4395–4410.
based classification and grading system for meningioma:​a 38. Qiu YS, Jiang NN, Zhou Y, et al. LMO3 promotes gastric
multicentre, retrospective analysis. Lancet Oncol. 2017;​18(5):​ cancer cell invasion and proliferation through Akt-mTOR
682–694. and Akt-GSK3β signaling. Int J Mol Med. 2018;​41(5):​2755–
20. He PF, Zhou JD, Yao DM, et al. Efficacy and safety of 2763.
decitabine in treatment of elderly patients with acute myeloid 39. Wong N, Gu Y, Kapoor A, et al. Upregulation of FAM84B
leukemia:​a systematic review and meta-analysis. Oncotarget. during prostate cancer progression. Oncotarget. 2017;​8(12):​
2017;​8(25):​41498–41507. 19218–19235.
21. Grandjenette C, Schnekenburger M, Karius T, et al. 5-aza- 40. Zhang H, Zheng J, Shen H, et al. Curcumin suppresses in
2ʹ-deoxycytidine-mediated c-myc Down-regulation triggers vitro proliferation and invasion of human prostate cancer
telomere-dependent senescence by regulating human telom- stem cells by modulating DLK1-DIO3 imprinted gene cluster
erase reverse transcriptase in chronic myeloid leukemia. microRNAs. Genet Test Mol Biomarkers. 2018;​22(1):​43–50.
Neoplasia. 2014;​16(6):​511–528. 41. Nakamura M, Nishikawa J, Saito M, et al. Decitabine inhibits
22. Hamm CA, Xie H, Costa FF, et al. Global demethylation tumor cell proliferation and up-regulates e-cadherin expres-
of rat chondrosarcoma cells after treatment with 5-aza-2ʹ- sion in Epstein-Barr virus-associated gastric cancer. J Med
deoxycytidine results in increased tumorigenicity. PLoS One. Virol. 2017;​89(3):​508–517.
2009;​4(12):​e8340. 42. Thibault A, Figg WD, Bergan RC, et al. A phase II study of
23. Püttmann S, Senner V, Braune S, et al. Establishment of a 5-aza-2ʹdeoxycytidine (decitabine) in hormone independent
benign meningioma cell line by hTERT-mediated immortal- metastatic (D2) prostate cancer. Tumori. 1998;​84(1):​87–89.
ization. Lab Invest. 2005;​85(9):​1163–1171. 43. Xia C, Leon-Ferre R, Laux D, et al. Treatment of resistant
24. Liu Y, Pang JC, Dong S, et al. Aberrant CpG island hyper- metastatic melanoma using sequential epigenetic therapy
methylation profile is associated with atypical and anaplastic (decitabine and panobinostat) combined with chemotherapy
meningiomas. Hum Pathol. 2005;​36(4):​416–425. (temozolomide). Cancer Chemother Pharmacol. 2014;​74(4):​
25. Fürtjes G, Heß K, Wagner A, et al. Retention of ATRX and 691–697.
DAXX expression in meningiomas. Letter. Neurosurgery. 44. Kitchen MO, Bryan RT, Haworth KE, et al. Methylation of
2017;​81(2):​E22–E23. HOXA9 and ISL1 predicts patient outcome in high-grade
26. Aryee MJ, Jaffe AE, Corrada-Bravo H, et al. Minfi:​a flexible non-invasive bladder cancer. PLoS One. 2015;​10(9):​e0137003.
and comprehensive Bioconductor package for the analysis 45. Serna E, Morales JM, Mata M, et al. Gene expression profiles
of Infinium DNA methylation microarrays. Bioinformatics. of metabolic aggressiveness and tumor recurrence in benign
2014;​30(10):​1363–1369. meningioma. PLoS One. 2013;​8(6):​e67291.
27. Peters TJ, Buckley MJ, Statham AL, et al. De novo identi- 46. Pettigrew KA, Armstrong RN, Colyer HA, et al. Differen-
fication of differentially methylated regions in the human tial TERT promoter methylation and response to 5-aza-2ʹ-
genome. Epigenetics Chromatin. 2015;​8:​6. deoxycytidine in acute myeloid leukemia cell lines:​TERT
expression, telomerase activity, telomere length, and cell
28. Andrae N, Kirches E, Hartig R, et al. Sunitinib targets
death. Genes Chromosomes Cancer. 2012;​51(8):​768–780.
PDGF-receptor and Flt3 and reduces survival and migra-
tion of human meningioma cells. Eur J Cancer. 2012;​48(12):​
1831–1841.
29. Hagemann S, Heil O, Lyko F, Brueckner B. Azacytidine and Disclosures
decitabine induce gene-specific and non-random DNA de- The authors report no conflict of interest concerning the materi-
methylation in human cancer cell lines. PLoS One. 2011;​6(3):​ als or methods used in this study or the findings specified in this
e17388. paper.
30. Bhandari A, Shen Y, Sindan N, et al. MAL2 promotes prolif-
eration, migration, and invasion through regulating epithelial- Author Contributions
mesenchymal transition in breast cancer cell lines. Biochem Conception and design: Brokinkel, Senner. Acquisition of data:
Biophys Res Commun. 2018;​504(2):​434–439. Stögbauer, Thomas, Wagner, Warneke, Bunk, Canisius. Analysis

J Neurosurg Volume 135 • September 2021 853

 
*42 *
Stögbauer et al.

and interpretation of data: Brokinkel, Stögbauer, Thomas, Senner. Online-Only Content

Drafting the article: Brokinkel, Stögbauer, Senner. Critically Supplemental material is available with the online version of the
revising the article: Brokinkel, Stögbauer, Wagner, Warneke, article.
Bunk, Grauer, Paulus, Stummer, Senner. Reviewed submitted Supplemental Figs. 1–4. https://thejns.org/doi/suppl/​10.3171/​
version of manuscript: Brokinkel, Stögbauer, Canisius, Senner. 2020.7.JNS193097.
Approved the final version of the manuscript on behalf of all
authors: Brokinkel. Statistical analysis: Brokinkel, Stögbauer. Correspondence
Administrative/technical/material support: Senner. Study supervi-
sion: Brokinkel, Senner. Benjamin Brokinkel: University Hospital Münster, Germany.
[email protected].
Supplemental Information
Videos
Video 1. https://vimeo.com/444198097.

854 J Neurosurg Volume 135 • September 2021

*42 *
 CLINICAL ARTICLE
J Neurosurg 135:855–861, 2021

Internal dose escalation associated with increased local

control for melanoma brain metastases treated with
stereotactic radiosurgery
William R. Kennedy, MD,1 Todd A. DeWees, PhD,2 Sahaja Acharya, MD,3 Mustafaa Mahmood, MD,1
Nels C. Knutson, PhD,1 S. Murty Goddu, PhD,1 James A. Kavanaugh, PhD,1
Timothy J. Mitchell, PhD,1 Keith M. Rich, MD,4 Albert H. Kim, MD, PhD,4 Eric C. Leuthardt, MD,4
Joshua L. Dowling, MD,4 Gavin P. Dunn, MD, PhD,4 Michael R. Chicoine, MD,4
Stephanie M. Perkins, MD,1 Jiayi Huang, MD,1 Christina I. Tsien, MD,1 Clifford G. Robinson, MD,1
and Christopher D. Abraham, MD1
Departments of 1Radiation Oncology and 4Neurosurgery, Washington University School of Medicine, St. Louis, Missouri;
2
Department of Biomedical Statistics and Informatics, Mayo Clinic, Scottsdale, Arizona; and 3Department of Radiation Oncology,
St. Jude Children’s Research Hospital, Memphis, Tennessee

OBJECTIVE  The internal high-dose volume varies widely for a given prescribed dose during stereotactic radiosurgery
(SRS) to treat brain metastases (BMs). This may be altered during treatment planning, and the authors have previously
shown that this improves local control (LC) for non–small cell lung cancer BMs without increasing toxicity. Here, they
seek to identify potentially actionable dosimetric predictors of LC after SRS for melanoma BM.
METHODS  The records of patients with unresected melanoma BM treated with single-fraction Gamma Knife RS be-
tween 2006 and 2017 were reviewed. LC was assessed on a per-lesion basis, defined as stability or a decrease in lesion
size. Outcome-oriented approaches were utilized to determine optimal dichotomization for dosimetric variables relative
to LC. Univariable and multivariable Cox regression analysis was implemented to evaluate the impact of collected pa-
rameters on LC.
RESULTS  Two hundred eighty-seven melanoma BMs in 79 patients were identified. The median age was 56 years
(range 31–86 years). The median follow-up was 7.6 months (range 0.5–81.6 months), and the median survival was 9.3
months (range 1.3–81.6 months). Lesions were optimally stratified by volume receiving at least 30 Gy (V30) greater than
or equal to versus less than 25%. V30 was ≥ and < 25% in 147 and 140 lesions, respectively. For all patients, 1-year
LC was 83% versus 66% for V30 ≥ and < 25%, respectively (p = 0.001). Stratifying by volume, lesions 2 cm or less (n =
215) had 1-year LC of 82% versus 70% (p = 0.013) for V30 ≥ and < 25%, respectively. Lesions > 2 to 3 cm (n = 32) had
1-year LC of 100% versus 43% (p = 0.214) for V30 ≥ and < 25%, respectively. V30 was still predictive of LC even after
controlling for the use of immunotherapy and targeted therapy. Radionecrosis occurred in 2.8% of lesions and was not
significantly associated with V30.
CONCLUSIONS  For a given prescription dose, an increased internal high-dose volume, as indicated by measures
such as V30 ≥ 25%, is associated with improved LC but not increased toxicity in single-fraction SRS for melanoma BM.
Internal dose escalation is an independent predictor of improved LC even in patients receiving immunotherapy and/or
targeted therapy. This represents a dosimetric parameter that is actionable at the time of treatment planning and war-
rants further evaluation.
https://thejns.org/doi/abs/10.3171/2020.7.JNS192210
KEYWORDS  melanoma; brain metastases; stereotactic radiosurgery; Gamma Knife; dose escalation; local control;
oncology

ABBREVIATIONS  BM = brain metastasis; DVH = dose-volume histogram; IT = immunotherapy; LC = local control; MBM = melanoma BM; MVA = multivariable analysis;
NSCLC = non–small cell lung cancer; OS = overall survival; RN = radiation necrosis; RTOG = Radiation Therapy Oncology Group; SRS = stereotactic radiosurgery; TT =
targeted therapy; UVA = univariable analysis; Vxx = metastasis volume percentages receiving at least xx Gy; WBRT = whole-brain radiation therapy.
SUBMITTED  September 11, 2019.  ACCEPTED  July 9, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS192210.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  855

*43 *
Kennedy et al.

M alignant melanoma is a leading cause of brain

metastases (BMs), accounting for 10% of all
cases and ranking behind only lung and breast
cancer as the third most common histological type.1 It is
prescription dose for metastases > 3 cm was 15 Gy; for
metastases 2.1 to 3 cm, 18 Gy; and for lesions ≤ 2 cm,
20 to 24 Gy. Dose was primarily prescribed to the 50%
isodose line. Planning parameters also included maintain-
distinctly neurotropic, with 44% of patients developing ing a gradient index (prescription isodose volume/tumor
BMs at some point during their treatment course in one volume) < 3 and an RTOG conformity index (prescription
study.2 Furthermore, the development of melanoma BMs isodose volume/tumor volume) < 2, as well as placing in-
(MBMs) portends a distinctly poor prognosis in melano- creased shots within the target to increase heterogeneity,
ma, with a 5-year overall survival (OS) of approximate- although not to any specific parameter at the time of plan-
ly 10%.3 However, much progress has been made in the ning. No additional specific optimization or plan evalua-
management of metastatic melanoma over the past several tion was performed besides these specific parameters.
years, with inhibitors of BRAF, MEK, cytotoxic T-lym- Patients underwent follow-up with brain MRI every
phocyte–associated protein 4 (CTLA4), and programmed 2–3 months. Each MRI study obtained in patients from
cell death protein 1 (PD1) all providing improvements in the time of SRS until the last follow-up or death was re-
survival compared to traditional systemic agents.4–6 viewed for local failure and RN in every treated lesion.
Local control (LC) is an important outcome in BMs. Local failure was defined as a post-SRS increase in the
Local intracranial progression can have devastating ef- size of the treated lesion except if additional imaging or
fects on patients’ quality of life. Accordingly, LC of BMs surgical pathology identified RN. Note that RN was re-
is correlated with improved neurocognition, overall quality corded if pathologically confirmed or if equivalent imag-
of life, as well as survival.7–9 There have been multiple at- ing changes were documented.
tempts to improve LC with stereotactic radiosurgery (SRS), Clinical characteristics between groups were compared
including prescription dose escalation, adding a small mar- using Fisher’s exact test and the Wilcoxon rank-sum test
gin around the target, and the use of whole-brain radiation where appropriate. For each lesion, DVH parameters from
therapy (WBRT) in conjunction with SRS. However, there the treatment planning software (Leksell GammaPlan
are potential drawbacks to each of these methods, which 10.1, Elekta) were exported and analyzed. Parameters ex-
limit their widespread use. Escalation of the prescription ported on a per-lesion basis included metastasis volume;
dose, while having a positive correlation with improved conformity indices; dose minimums to various tumor
LC, is associated with increased rates of radiation necrosis volume percentages; global minimum, maximum, me-
(RN).10 The addition of a small margin also improves LC, dian, and mean doses; and metastasis volume percentages
but this method is limited near critical structures. Finally, receiving at least xx Gy (Vxx) in 1-Gy increments from
adding WBRT prior to SRS also improves LC, but at the 20 to 40 Gy. An outcome-oriented approach based on log-
expense of increased rates of neurocognitive decline.11 rank test statistics developed by Contal and O’Quigley was
Given these limitations, identifying other factors that used to determine optimal cut points.13 The Kaplan-Meier
correlate with LC in SRS is an important goal to further method was used to estimate LC and OS with significant
improve the effectiveness of the treatment. Dose-volume differences tested through the log-rank test. Univariable
histogram (DVH) metrics representing internal high doses analysis (UVA) and multivariable analysis (MVA) were
within the treated volume may represent modifiable pa- performed using the Cox proportional hazards model. For
rameters during SRS treatment planning. Although previ- MVA, stepwise selection methods utilizing factors signifi-
ously identified as a determinant of LC in non–small cell cant on UVA were included in the MVA Cox proportional
lung cancer (NSCLC) BMs, the relationship of such mea- hazards model. To confirm appropriate selection of predic-
sures with LC in SRS for MBMs has not been reported in tive variables entered into the MVA, a backward stepwise
the literature.12 The purpose of this study was to identify (step-down) logistic regression method with selection cri-
potentially modifiable DVH parameters predictive of LC terion p < 0.10 was applied to obtain the final multivari-
in SRS for MBM. ate model. Significance was considered at p < 0.05, and
all significance levels were 2-sided. Lesions with missing
data for variables tested in these models were omitted from
Methods these analyses. Fisher’s exact test and the Wilcoxon rank-
Patients who had undergone single-fraction SRS for sum test were used to compare RN rates between patient
treatment of MBMs between 2006 and 2017 were identi- groups where appropriate. All tests were 2-tailed. Statisti-
fied from an institutional review board–approved database. cal analyses were performed using SAS (version 9.4, SAS
The Leksell Gamma Knife system (Elekta AB) was used Institute).
for single-fraction SRS in all patients. They were treated A full list of variables tested in the UVA and MVA (if
using Model C (2006 to April 2008) and Perfexion (May significant on UVA) included age; sex; race; Karnofsky
2008 to present). All target demarcation and SRS plan- Performance Status; histology; BRAF V600E mutational
ning were performed together by a radiation oncologist, status; targeted therapy (TT) use; immunotherapy (IT) use;
surgeon, and physicist. Any contrast-enhancing lesions extracranial disease status; molecular graded prognostic in-
on T1-weighted MRI defined the target volume. No fur- dex; number of shots; margin dose; metastasis volume; con-
ther margin or expansion was used. The dose prescribed formity indices; dose minimums to various tumor volume
to the margin was based on the Radiation Therapy On- percentages; global minimum, maximum, median, and
cology Group (RTOG) 90-05 protocol, with the treating mean doses; and metastasis volume percentages receiving
physicians making adjustments at their discretion.10 The at least xx Gy (Vxx) in 1-Gy increments from 20 to 40 Gy.

856 J Neurosurg Volume 135 • September 2021

 
*43 *
 Kennedy et al.

Results TABLE 1. Patient, tumor, and treatment-related characteristics

We identified a total of 287 MBMs in 79 patients who Characteristic Value (% or range)

met the aforementioned inclusion and exclusion criteria. Pa- Median age in yrs 56 (31–86)
tient and treatment characteristics are presented in Table 1.
Sex
Patients had BRAF V600E–mutated disease in 50% of the
38 patients with available testing data. The median number  M 55 (70)
of treated metastases was 2 (range 1–11 treated metastases).   F 24 (30)
IT with anti-PD1 and/or anti-CTLA4 within 3 months of KPS at time of SRS
SRS was used in 48% of patients, TT with BRAF and/or   ≤70 5 (6)
MEK inhibitors within 3 months of SRS was used in 34%  80 22 (28)
of patients, and the remaining 18% patients did not receive
  ≥90 52 (66)
IT or TT within 3 months of SRS. WBRT was given prior to
SRS in 15 patients (19%), all to 30 Gy in 10 daily fractions. Controlled extracranial disease
The median SRS dose was 20 Gy (range 15–24 Gy) to the  Yes 45 (57)
median 50% isodose line (range 45%–80%), to a median  No 34 (43)
volume of 0.377 cm3 (range 0.005–23.7 cm3). The median BRAF V600E mutation
conformality index was 1.74. The median follow-up from  Yes 19 (24)
the time of SRS was 7.6 months for all patients (range 0.5–
 No 19 (24)
81.6 months). The median OS was 9.3 months for all patients
(range 1.3–81.6 months). The 1- and 2-year OS rates were  Unknown 41 (52)
40% (95% CI 30.1–50.7) and 26% (95% CI 16–35), respec- Median melanoma molGPA 2.5
tively. There were a cumulative 53 local failures. Overall Checkpoint inhibitors w/in 3 mos of SRS
lesion-based LC at 1 year was 75% (95% CI 69–82) and at   Anti-CTLA4 alone 2 (3)
2 years it was 71% (95% CI 64–78) for all patients (Fig. 1A).   Anti-PD1 alone 19 (24)
A summary of factors associated with LC on UVA is
presented in Fig. 2. Patient and tumor characteristics such as   Anti-PD1 w/ anti-CTLA4 17 (22)
younger age, higher melanoma molecular graded prognostic  No 41 (52)
assessment score,14 and higher lesion volume and diameter BRAF/MEK inhibitors w/in 3 mos of SRS
were all associated with worse LC. Treatment characteris-  Yes 27 (34)
tics such as BRAF/MEK inhibitor use and PD1 inhibitor  No 52 (66)
use were associated with improved LC. Multiple dosimetric WBRT prior to SRS
parameters were associated with improved LC on UVA, in-
cluding increased prescription dose, V24 through V35, in-  Yes 15 (19)
creased minimum and mean dose, as well as higher RTOG  No 64 (81)
conformity index.15 Conversely, a higher (worse) Paddick’s No. of treated BMs
conformity index was associated with worse LC on UVA.16  1–3 56 (71)
Each of these variables listed above and displayed in Fig. 2   ≥4 23 (29)
were entered into the MVA. To evaluate the possible effect Tumor location
of intrasubject correlation, we computed models with and
without a sandwich variance estimator and found that there  Supratentorial 264 (92)
did not appear to be significant intrasubject correlation be-  Cerebellar 23 (8)
tween multiple metastases in the same subject (all standard Axial diameter in cm*
error estimate changes were minimal, and all covariance <   ≤2 215 (84)
0.1). MVA demonstrated that increased V30 (HR for local  2.1–3 32 (13)
recurrence = 0.46, p = 0.015) and BRAF/MEK inhibitor use   >3 8 (3)
(HR = 0.26, p = 0.006) were independently associated with
better LC outcomes, whereas an increased Paddick’s con- Median target vol in cm3 0.377 (0.00–23.7)
formity index (HR = 2.99, p = 0.001) and age < 60 years Prescription dose in Gy 20 (15–24)
(HR = 4.13, p < 0.001) were independent negative predictors V30 Gy
of LC (Fig. 3). When V30 was entered into the model, all   ≥25% 147 (51)
other dosimetric parameters were not statistically significant   <25% 140 (49)
predictors of LC other than Paddick’s conformity index.
Cut point analysis was performed for V30, which dem- KPS = Karnofsky Performance Status; molGPA = molecular graded prognostic
onstrated that the most predictive cut point was 25% for assessment.
* Data unavailable for 32 lesions.
LC (p = 0.0035), although 24% to 36% were all statistically
significant at p < 0.05. LC was then evaluated in lesions di-
chotomized into groups, with V30 of at least 25% (n = 147)
and less than 25% (n = 140), respectively. The 1-year LC axial diameter: ≤ 2 cm, > 2 to 3 cm, and > 3 cm. For lesions
was 83% (95% CI 75–90) versus 66% (95% CI 56–77) for ≤ 2 cm (n = 215), 1-year LC was 82% (95% CI 74–90) for
V30 ≥ and < 25%, respectively (p = 0.001; Fig. 1B). Lesions V30 ≥ 25% (n = 143) and 70% (95% CI 57–83) for V30 <
were further stratified into 3 groups based on maximum 25% (n = 72; p = 0.013; Fig. 1C). For lesions > 2 to 3 cm (n

J Neurosurg Volume 135 • September 2021 857

 
*43 *
Kennedy et al.

FIG. 1. Kaplan-Meier estimates of LC stratified by V30 as a function of tumor diameter. A: LC for all lesions. B: LC for all lesions,
stratified by V30 (1-year LC 83% vs 66%, p = 0.001). C: LC for lesions ≤ 2 cm (1-year LC 82% vs 70%, p = 0.013). D: LC for lesions
> 2 to 3 cm (1-year LC 100% vs 43%, p = 0.214). Figure is available in color online only.

= 32), 1-year LC was 100% for V30 ≥ 25% (n = 3) and 40% tion, such as a V30 of at least 25%, independently predict LC
(95% CI 11–76) for V30 < 25% (n = 29; p = 0.214; Fig. 1D). of MBMs after treatment with SRS, even after controlling
For lesions > 3 cm (n = 8), no patients had V30 ≥ 25%, and for prescription dose. Furthermore, V30 was still significant
1-year LC for these 8 lesions was 58% (95% CI 22–95). even after controlling for treatment with either BRAF/MEK
MVA of factors significant on UVA for OS revealed that TT or CTLA4/PD1 IT, making it a potentially valuable met-
V30 did not independently influence OS (HR = 1.055, p = ric in a broad array of patients with metastatic melanoma.
0.829). MVA did demonstrate that IT use within 3 months Although RN was uncommon in our cohort, importantly,
of SRS (HR = 0.471, p = 0.006), control of the extracranial the rate of RN was not influenced by V30. In addition to
primary disease at the time of SRS (HR = 0.401, p = 0.001), validating our prior findings in NSCLC, our work is novel in
and Karnofsky Performance Status of 90–100 (HR = 0.085, that it provides a potential actionable parameter for the treat-
p = 0.023) were all independent predictors of improved OS. ing physician at the time of treatment planning for MBM.12
The overall rate of RN on a per-lesion basis was 2.8%, As patients with metastatic melanoma continue to ex-
occurring in a total of 4 patients (5.1% of patients). Fisher’s perience gains in OS with the utilization of both BRAF/
exact test and the Wilcoxon rank-sum test revealed that MEK inhibitors and IT, LC is an increasingly important
no individual factors, including V30, prescription dose,
metric for both patients and their treating physicians. With
tumor diameter, and use of IT or TT, were significantly
associated with the rate of RN. improved survival, patients have an increased potential to
experience local failure as an event as well as the clinical
consequences of local failure. Local intracranial failure
Discussion has a high potential to be symptomatic, significantly influ-
Herein, we revealed that measures of internal dose escala- encing both patients’ quality of life and even survival.7–9,17,18

858 J Neurosurg Volume 135 • September 2021

 
*43 *
 Kennedy et al.

After SRS alone, approximately one-quarter to one-third

of patients may experience local failure.19 Although an in-
creased prescription dose has been shown to improve LC
in multiple studies, other large series have argued that in-
creasing the prescription dose beyond a standard threshold
may not provide gains in LC.20,21 Furthermore, increasing
prescription dose has been shown to increase RN in some
series, limiting the safety of this approach to maximize
LC.10,22 Therefore, other approaches to improving LC
without increasing toxicity warrant investigation.
It is important to discuss our findings within the context
of a rapidly changing paradigm of managing metastatic mel-
anoma. Recent trials of IT in patients with MBMs have all
shown reasonable intracranial objective response rates, even
in a heavily pretreated population.23 Rates appear to be im-
proved, approximately 45%–55%, with combined ipilimum-
ab/nivolumab compared to monotherapy alone, with rates
similar to those with combined dabrafenib/trametinib.24–28
Initial nonrandomized data have shown that SRS use con-
current with either IT or TT is well tolerated with favorable
local and distant intracranial responses.29–31 Expectedly,
while the BRAF V600E mutational status itself did not in-
fluence LC, BRAF/MEK inhibitor use remained an inde-
pendent predictor of LC, and PD1 and/or CTLA4 inhibitor
use was an independent predictor of OS on MVA. However,
even after accounting for the use of concurrent IT and/or TT
in our study, internal dose escalation remained a significant
independent predictor of LC, justifying its relevance in a
modern cohort in which IT and/or TT use is frequent.
Other independent predictors of LC on MVA alongside
V30 and BRAF/MEK inhibitor use included an increased
Paddick’s conformity index and a younger age, both of
which predicted worsened LC. The negative correlation of
Paddick’s conformity index with LC can be explained as
follows: an increased Paddick’s conformity index indicates
a more rapid fall-off of dose around the target, therefore
suggesting that giving an insufficient dose to the periphery FIG. 2. Significant variables (p < 0.05) on UVA for local recurrence by
of the target may compromise LC. This has been previous- forest plot representation. CI = conformity index; mol-GPA = molecular
graded prognostic assessment.
ly described by Woo and colleagues in the definitive set-
ting and is further supported by the observation that add-
ing a small margin in the postoperative setting improves
LC.32,33 Younger age is known to predict OS in metastatic differences in treatment planning and are readily modifi-
melanoma, with a subsequently increased follow-up dura- able without altering the prescription dose itself. While we
tion allowing for an increased potential for local failures, have identified a novel planning parameter representing an
possibly explaining our findings on MVA.14,34–36 escalated internal high-dose volume, the mechanisms for
Several dosimetric parameters indicating an internal why this occurs are poorly characterized. Preclinical mod-
high-dose volume were found to be predictors on UVA. els have postulated that tumor control probability can be
V30 was chosen for further analysis, as it was the most sig- improved by escalating dose to the intrinsically radioresis-
nificant parameter in our model and remained independent- tant regions of hypoxia within a tumor.37 These dose gradi-
ly predictive of LC on MVA. An important consideration ents also have important implications for normal tissue re-
is the relationship of V30 with marginal dose. Regarding pair, as selectively boosting hypoxic regions via grid ther-
V30, TT was a stronger predictor than marginal dose in our apy, in which there are alternating regions of high and low
adjusted MVA without V30 (p = 0.0672); with V30 in the radiation doses within a tumor volume, have been shown to
model, the marginal dose was not predictive (p = 0.424), have improved normal tissue repair while maintaining tu-
while V30 retained significance (p = 0.0564). However, it mor cell kill rates.38 Furthermore, utilization of PET imag-
is likely many of these parameters, V24 through V35 in our ing with 18F-fluoromisonidazole (FMISO) to identify these
current study, would have a similar impact on LC if entered hypoxic subvolumes in sites such as the lung and head and
into our MVA model. The exact dosimetric parameter is neck for treatment planning have been described, although
arguably less important than the phenomenon it represents: further clinical validation is needed as our knowledge of
an increase in the volume within the target receiving doses tumor hypoxia and dose-response, as well as identification
greater than the prescription dose. These metrics are due to and safe targeting of these regions, improves.39,40

J Neurosurg Volume 135 • September 2021 859

 
*43 *
Kennedy et al.

FIG. 3. MVA for local recurrence by forest plot representation for all lesions analyzed (n = 240).

Our study has multiple limitations, including its retro- measures is most feasible in smaller lesions, and given the
spective, single-institution design spanning an era in which relatively low frequency of metastases larger than 2 cm in
the management of metastatic melanoma has changed diameter in our series, conclusions regarding the benefit
substantially, particularly with the routine use of BRAF/ and safety of this approach in larger tumors are unclear.
MEK inhibitors and IT. Furthermore, larger lesions are be- Optimizing dosimetric measures such as the volume re-
ing more frequently treated with fractionated radiosurgery ceiving at least 30 Gy, V30 ≥ 25%, represents actionable
rather than single-fraction SRS, a likely explanation for parameters in SRS treatment planning and warrants vali-
the poor representation of larger metastases in our study. dation in a multiinstitutional or prospective manner.
The SRS technique in this series is also relatively specific
to Gamma Knife, as to achieve a V30 with a 20-Gy pre- References
scription dose would require prescribing to an isodose of
66% or less, very much on the low side for a linear ac-   1. Johnson JD, Young B. Demographics of brain metastasis.
Neurosurg Clin N Am. 1996;​7(3):​337–344.
celerator–based SRS plan. While we believe that LC is an   2. Davies MA, Liu P, McIntyre S, et al. Prognostic factors for
important outcome of interest in the management of BMs, survival in melanoma patients with brain metastases. Cancer.
some may argue that OS should be the predominant fo- 2011;​117(8):​1687–1696.
cus when reporting outcomes, especially in a population   3. Cancer Stat Facts:​Melanoma of the Skin. SEER. National
with limited survival. Intrinsic to dosimetric studies as a Cancer Institute. Accessed August 24, 2020. https:​//seer.can-
whole are concerns that multiple dosimetric parameters cer.gov/statfacts/html/melan.html
likely interact with each other, with truly isolated analyses   4. Robert C, Karaszewska B, Schachter J, et al. Improved over-
of a single measure challenging. This salient issue is why all survival in melanoma with combined dabrafenib and tra-
metinib. N Engl J Med. 2015;​372(1):​30–39.
we emphasize that V30 ≥ 25% itself is not the sole critical   5. Ribas A, Puzanov I, Dummer R, et al. Pembrolizumab versus
factor to improving LC, only that the internal high-dose investigator-choice chemotherapy for ipilimumab-refractory
volume can be optimized at the time of treatment planning melanoma (KEYNOTE-002):​a randomised, controlled,
and may improve LC without clearly increasing the risk phase 2 trial. Lancet Oncol. 2015;​16(8):​908–918.
of RN. The goal of the current study is not to construct a   6. Wolchok JD, Chiarion-Sileni V, Gonzalez R, et al. Overall
definitive dosimetric model to predict LC, but to present a survival with combined nivolumab and ipilimumab in ad-
novel hypothesis-generating model. Furthermore, it is im- vanced melanoma. N Engl J Med. 2017;​377(14):​1345–1356.
portant to highlight that given only 53 events with a variety   7. Regine WF, Scott C, Murray K, Curran W. Neurocognitive
outcome in brain metastases patients treated with accelerated-
of clinical and dosimetric variables tested without internal fractionation vs. accelerated-hyperfractionated radiotherapy:​
cross-validation or validation using external data, overfit- an analysis from Radiation Therapy Oncology Group Study
ting of our statistical models is a potential limitation of 91-04. Int J Radiat Oncol Biol Phys. 2001;​51(3):​711–717.
analysis and hence should only be interpreted as hypothesis   8. Skeie BS, Eide GE, Flatebø M, et al. Quality of life is main-
generating. As a result, our findings require independent tained using Gamma Knife radiosurgery:​a prospective study
validation with external data sets prior to more widespread of a brain metastases patient cohort. J Neurosurg. 2017;​
implementation, as they do have the potential to be practice 126(3):​708–725.
changing by providing actionable parameters at the time of   9. Ostheimer C, Bormann C, Fiedler E, et al. Malignant mela-
noma brain metastases:​Treatment results and prognostic
treatment planning that may allow for internal dose escala- factors—a single-center retrospective study. Int J Oncol.
tion without the expense of increased toxicity. 2015;​46(6):​2439–2448.
10. Shaw E, Scott C, Souhami L, et al. Single dose radiosurgical
Conclusions treatment of recurrent previously irradiated primary brain
tumors and brain metastases:​final report of RTOG protocol
For patients with MBMs treated with SRS, measures of 90-05. Int J Radiat Oncol Biol Phys. 2000;​47(2):​291–298.
dose greater than the prescription dose within the target 11. Churilla TM, Ballman KV, Brown PD, et al. Stereotactic radio-
volume are predictive of LC. Achieving high internal-dose surgery with or without whole-brain radiation therapy for lim-

860 J Neurosurg Volume 135 • September 2021

 
*43 *
 Kennedy et al.

ited brain metastases:​a secondary analysis of the North Central 30. Chen L, Douglass J, Kleinberg L, et al. Concurrent immune
Cancer Treatment Group N0574 (Alliance) randomized con- checkpoint inhibitors and stereotactic radiosurgery for brain me-
trolled trial. Int J Radiat Oncol Biol Phys. 2017;​99(5):​1173–1178. tastases in non-small cell lung cancer, melanoma, and renal cell
12. Abraham C, Garsa A, Badiyan SN, et al. Internal dose escala- carcinoma. Int J Radiat Oncol Biol Phys. 2018;​100(4):​916–925.
tion is associated with increased local control for non-small cell 31. Kotecha R, Miller JA, Venur VA, et al. Melanoma brain me-
lung cancer (NSCLC) brain metastases treated with stereotac- tastasis:​the impact of stereotactic radiosurgery, BRAF muta-
tic radiosurgery (SRS). Adv Radiat Oncol. 2017;​3(2):​146–153. tional status, and targeted and/or immune-based therapies on
13. Contal C, O’Quigley J. An application of changepoint meth- treatment outcome. J Neurosurg. 2018;​129(1):​50–59.
ods in studying the effect of age on survival in breast cancer. 32. Woo HJ, Hwang SK, Park SH, et al. Factors related to the lo-
Comput Stat Data Anal. 1999;​30(3):​253–270. cal treatment failure of gamma knife surgery for metastatic
14. Sperduto PW, Jiang W, Brown PD, et al. Estimating survival brain tumors. Acta Neurochir (Wien). 2010;​152(11):​1909–1914.
in melanoma patients with brain metastases:​an update of the 33. Soltys SG, Adler JR, Lipani JD, et al. Stereotactic radiosur-
graded prognostic assessment for melanoma using molecular gery of the postoperative resection cavity for brain metasta-
markers (Melanoma-molGPA). Int J Radiat Oncol Biol Phys. ses. Int J Radiat Oncol Biol Phys. 2008;​70(1):​187–193.
2017;​99(4):​812–816. 34. Nieder C, Hintz M, Bilger A, et al. Validation of the graded
15. Shaw E, Kline R, Gillin M, et al. Radiation Therapy Oncol- prognostic assessment for melanoma using molecular markers
ogy Group:​radiosurgery quality assurance guidelines. Int J (Melanoma-molGPA). J Clin Med Res. 2018;​10(3):​178–181.
Radiat Oncol Biol Phys. 1993;​27(5):​1231–1239. 35. Minniti G, D’Angelillo RM, Scaringi C, et al. Fractionated
16. Paddick I, Lippitz B. A simple dose gradient measurement stereotactic radiosurgery for patients with brain metastases. J
tool to complement the conformity index. J Neurosurg. 2006;​ Neurooncol. 2014;​117(2):​295–301.
105(suppl):​194–201. 36. Balermpas P, Stera S, Müller von der Grün J, et al. Repeated
17. Zindler JD, Rodrigues G, Haasbeek CJA, et al. The clinical in-field radiosurgery for locally recurrent brain metastases:​
utility of prognostic scoring systems in patients with brain feasibility, results and survival in a heavily treated patient
metastases treated with radiosurgery. Radiother Oncol. 2013;​ cohort. PLoS One. 2018;​13(6):​e0198692.
106(3):​370–374. 37. Chvetsov AV, Zeng J, Rajendran JG. Volume dependence in
18. Raizer JJ, Hwu W-J, Panageas KS, et al. Brain and leptomenin- hypoxia-targeted dose escalation. Med Phys. 2018;​45(11):​
geal metastases from cutaneous melanoma:​survival outcomes 5325–5331.
based on clinical features. Neuro Oncol. 2008;​10(2):​199–207. 38. Zhang H, Zhong H, Barth RF, et al. Impact of dose size in
19. Sahgal A, Aoyama H, Kocher M, et al. Phase 3 trials of stereo- single fraction spatially fractionated (grid) radiotherapy for
tactic radiosurgery with or without whole-brain radiation ther- melanoma. Med Phys. 2014;​41(2):​021727.
apy for 1 to 4 brain metastases:​individual patient data meta- 39. Hendrickson K, Phillips M, Smith W, et al. Hypoxia imaging
analysis. Int J Radiat Oncol Biol Phys. 2015;​91(4):​710–717. with [F-18] FMISO-PET in head and neck cancer:​potential
20. Mohammadi AM, Schroeder JL, Angelov L, et al. Impact for guiding intensity modulated radiation therapy in over-
of the radiosurgery prescription dose on the local control of coming hypoxia-induced treatment resistance. Radiother
small (2 cm or smaller) brain metastases. J Neurosurg. 2017;​ Oncol. 2011;​101(3):​369–375.
126(3):​735–743. 40. Thureau S, Dubray B, Modzelewski R, et al. FDG and
21. Shiue K, Barnett GH, Suh JH, et al. Using higher isodose FMISO PET-guided dose escalation with intensity-modulated
lines for Gamma Knife treatment of 1 to 3 brain metastases radiotherapy in lung cancer. Radiat Oncol. 2018;​13(1):​208.
is safe and effective. Neurosurgery. 2014;​74(4):​360–366.
22. Nedzi LA, Kooy H, Alexander E III, et al. Variables associat-
ed with the development of complications from radiosurgery Disclosures
of intracranial tumors. Int J Radiat Oncol Biol Phys. 1991;​ Dr. Kim has received support of non–study-related clinical or
21(3):​591–599. research effort overseen by the author from Monteris and Stryker.
23. Kluger HM, Chiang V, Mahajan A, et al. Long-term survival Dr. Leuthardt is a consultant for Monteris and has ownership in
of patients with melanoma with active brain metastases Neurolotions, Inner Cosmos, Immunovalent, Osteovantage, and
treated with pembrolizumab on a phase II trial. J Clin Oncol. Sora Imaging Solutions. Dr. Dunn has ownership of Immunova-
2019;​37(1):​52–60. lent. Dr. Tsien is a member of the speakers bureau for Varian. Dr.
24. Tawbi HA, Forsyth PA, Algazi A, et al. Combined nivolumab Robinson is a consultant for Radialogica, Varian, AstraZeneca,
and ipilimumab in melanoma metastatic to the brain. N Engl EMD Serona; has direct stock ownership in Radialogica; and has
J Med. 2018;​379(8):​722–730. received clinical or research support from Varian, Merck, Elekta,
25. Long GV, Atkinson V, Lo S, et al. Combination nivolumab and AstraZeneca for the study described.
and ipilimumab or nivolumab alone in melanoma brain
metastases:​a multicentre randomised phase 2 study. Lancet Author Contributions
Oncol. 2018;​19(5):​672–681.
26. Margolin K, Ernstoff MS, Hamid O, et al. Ipilimumab in Conception and design: Abraham, Kennedy, DeWees, Acharya.
patients with melanoma and brain metastases:​an open-label, Acquisition of data: Abraham, Kennedy, Acharya, Mahmood.
phase 2 trial. Lancet Oncol. 2012;​13(5):​459–465. Analysis and interpretation of data: Abraham, Kennedy, DeWees,
27. Goldberg SB, Gettinger SN, Mahajan A, et al. Pembrolizumab Acharya, Knutson, Goddu, Kavanaugh. Drafting the article:
for patients with melanoma or non-small-cell lung cancer and Kennedy. Critically revising the article: all authors. Reviewed
untreated brain metastases:​early analysis of a non-randomised, submitted version of manuscript: Mahmood, Knutson, Goddu,
open-label, phase 2 trial. Lancet Oncol. 2016;​17(7):​976–983. Kavanaugh, Mitchell, Rich, Kim, Leuthardt, Dowling, Dunn,
28. Davies MA, Saiag P, Robert C, et al. Dabrafenib plus trameti- Chicoine, Perkins, Huang, Tsien, Robinson. Approved the final
nib in patients with BRAFV600-mutant melanoma brain metas- version of the manuscript on behalf of all authors: Abraham.
tases (COMBI-MB):​a multicentre, multicohort, open-label, Statistical analysis: Kennedy, DeWees. Study supervision:
phase 2 trial. Lancet Oncol. 2017;​18(7):​863–873. Abraham, Kennedy, DeWees.
29. Kiess AP, Wolchok JD, Barker CA, et al. Stereotactic radio-
surgery for melanoma brain metastases in patients receiving Correspondence
ipilimumab:​safety profile and efficacy of combined treat- Christopher D. Abraham: Washington University School of Medi-
ment. Int J Radiat Oncol Biol Phys. 2015;​92(2):​368–375. cine, St. Louis, MO. [email protected].

J Neurosurg Volume 135 • September 2021 861

 
*43 *
 CLINICAL ARTICLE
J Neurosurg 135:862–870, 2021

Stereotactic radiosurgery for treatment of

radiation-induced meningiomas: a multiinstitutional study
Adomas Bunevicius, MD, PhD,1 Mohand Suleiman, MD,1 Samir Patel, MD,2
Roberto Martínez Álvarez, PhD, MD,3 Nuria E. Martinez Moreno, PhD, MD,3 Roman Liscak, MD,4
Jaromir Hanuska, MD,4 Anne-Marie Langlois, MD,5 David Mathieu, MD,5 Christine Mau, MD,6
Catherine Caldwell, BS,6 Leonard C. Tuanquin, MD,6 Brad E. Zacharia, MD,6
James McInerney, MD,6 Cheng-Chia Lee, MD, PhD,7 Huai-Che Yang, MD,7
Jennifer L. Peterson, MD,8 Daniel M. Trifiletti, MD,8 Akiyoshi Ogino, MD, PhD,9
Hideyuki Kano, MD, PhD,9 Ronald E. Warnick, MD,10 Anissa Saylany, BA,11 Love Y. Buch, BS,11
John Y. K. Lee, MD,11 Ben A. Strickland, MD,12 Gabriel Zada, MD,12 Eric L. Chang, MD,12
L. Dade Lunsford, MD,9 and Jason Sheehan, MD, PhD1
1
Department of Neurological Surgery, University of Virginia Health System, Charlottesville, Virginia; 2Division of Radiation
Oncology, Department of Oncology, University of Alberta, Edmonton, Alberta, Canada; 3Gamma Knife Radiosurgery, Hospital
Ruber Internacional, Madrid, Spain; 4Stereotactic and Radiation Neurosurgery, Na Homolce Hospital, Prague, Czech Republic;
5
Division of Neurosurgery, Université de Sherbrooke, Centre de recherche du CHUS, Sherbrooke, Québec, Canada; 6Penn State
Health, Hershey Medical Center, Hershey, Pennsylvania; 7Neurological Institute, Taipei Veterans General Hospital, and National
Yang-Ming University, Taipei, Taiwan; 8Radiation Oncology, Mayo Clinic, Jacksonville, Florida; 9Department of Neurologic
Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; 10Jewish Hospital, Mayfield Clinic, Cincinnati, Ohio;
11
Department of Neurosurgery, University of Pennsylvania, Philadelphia, Pennsylvania; and 12University of Southern California,
Los Angeles, California

OBJECTIVE  Radiation-induced meningiomas (RIMs) are associated with aggressive clinical behavior. Stereotactic
radiosurgery (SRS) is sometimes considered for selected RIMs. The authors investigated the effectiveness and safety of
SRS for the management of RIMs.
METHODS  From 12 institutions participating in the International Radiosurgery Research Foundation, the authors
pooled patients who had prior cranial irradiation and were subsequently clinically diagnosed with WHO grade I meningio-
mas that were managed with SRS.
RESULTS  Fifty-two patients underwent 60 SRS procedures for histologically confirmed or radiologically suspected
WHO grade I RIMs. The median ages at initial cranial radiation therapy and SRS for RIM were 5.5 years and 39 years,
respectively. The most common reasons for cranial radiation therapy were leukemia (21%) and medulloblastoma (17%).
There were 39 multiple RIMs (35%), the mean target volume was 8.61 ± 7.80 cm3, and the median prescription dose
was 14 Gy. The median imaging follow-up duration was 48 months (range 4–195 months). RIM progressed in 9 patients
(17%) at a median duration of 30 months (range 3–45 months) after SRS. Progression-free survival at 5 years post-SRS
was 83%. Treatment volume ≥ 5 cm3 predicted progression (HR 8.226, 95% CI 1.028–65.857, p = 0.047). Seven patients
(14%) developed new neurological symptoms or experienced SRS-related complications or T2 signal change from 1 to
72 months after SRS.
CONCLUSIONS  SRS is associated with durable local control of RIMs in the majority of patients and has an acceptable
safety profile. SRS can be considered for patients and tumors that are deemed suboptimal, poor surgical candidates,
and those whose tumor again progresses after removal.
https://thejns.org/doi/abs/10.3171/2020.7.JNS202064
KEYWORDS  stereotactic radiosurgery; radiation-induced meningioma; Gamma Knife; progression-free survival;
safety; local control; oncology

ABBREVIATIONS  CI = confidence interval; GKRS = Gamma Knife radiosurgery; HR = hazard ratio; IRRF = International Radiosurgery Research Foundation; NF1 =
neurofibromatosis type 1; NF2 = NF type 2; OS = overall survival; PFS = progression-free survival; RIM = radiation-induced meningioma; RT = radiation therapy; SRS =
stereotactic radiosurgery; WBRT = whole-brain RT.
SUBMITTED  May 29, 2020.  ACCEPTED  July 14, 2020.
INCLUDE WHEN CITING  Published online January 1, 2021; DOI: 10.3171/2020.7.JNS202064.

862 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*44 *
 Bunevicius et al.

R adiation-induced meningiomas (RIMs) are the tion. Patients who originally underwent SRS, had WHO
most common secondary intracranial neoplasms grade II or III meningiomas on histological examination,
caused by ionizing cranial irradiation. They de- or had a history of neurocutaneous disorders such as NF1
velop in patients who have undergone prior management or NF2 were excluded from the analysis. Thirty-two pa-
of a brain tumor or leukemia or treatment for tinea capi- tients (61%) did not have meningioma surgery prior to the
tis or in long-term atomic bomb survivors.1–3 The latency SRS and WHO grade was unknown.
period between cranial irradiation and RIM development In total, 60 eligible patients treated with SRS for RIM
may exceed 20 years, with a higher radiation dose being were identified in the IRRF database. The following cen-
associated with a shorter latency period to tumor develop- ters contributed the data: the University of Pittsburgh Med-
ment.2,4–6 RIMs arise within the treatment field of origi- ical Center (19 patients), Université de Sherbrooke (10 pa-
nal cranial radiation therapy (RT), are histologically and tients), Ruber International Hospital (7 patients), University
radiographically distinct from previously irradiated brain of Virginia Medical Center (6 patients), Na Homolce Hos-
tumors or lesions, and occur in patients without known pital (5 patients), University of Pennsylvania (3 patients),
genetic predisposition for brain tumors.2,7–10 RIMs possess Penn State Health–Hershey Medical Center (3 patients),
different genetic features than sporadic meningiomas;2,11,12 Taipei Veterans General Hospital (2 patients), University
compared with sporadic meningiomas, RIMs may have of Southern California (2 patients), University of Alberta
histological atypia, occur in multifocal locations, and in- Hospital (1 patient), Mayo Clinic in Florida (1 patient), and
crease in number during observation, thus making their Jewish Hospital, Mayfield Clinic (1 patient). Eight patients
gross-total resection and local tumor control more chal- were excluded from the analyses because they were di-
lenging.2,8,13–15 agnosed with WHO grade II (n = 5) or grade III (n = 1)
Resection is often considered an initial treatment for meningiomas or were originally treated with SRS (n = 2),
accessible RIMs, and it can be curative. However, com- leaving the final sample of 52 patients who were treated
plete resection of RIMs is often precluded by frequent with SRS for histologically confirmed WHO grade I RIMs
presentation with multiple RIMs, involvement of osse- (n = 20, 39%) or presumed WHO grade I RIMs (n = 32,
ous structures and blood vessels, aggressive and invasive 61%) based on MRI findings and disease course.
growth patterns, and location in surgically inaccessible
brain areas.13,16–19 RT is often considered for the manage- Patient Evaluation
ment of RIMs that are not amenable to complete resec- We obtained information regarding patient sex; age
tion.10 Experience with stereotactic radiosurgery (SRS) for at first cranial irradiation; age at SRS for RIM; indica-
the management of RIMs remains limited due to the rela- tion for first cranial irradiation; type, dose, and number
tive rarity of the disease. Prior published experience with of fractions of first cranial irradiation; history and extent
SRS for RIM is limited to single-institution series, small (gross-total resection, subtotal resection, or biopsy) of pre-
sample sizes, and limited follow-up duration.9,10,20 Further SRS resection and/or other RIM therapies preceding SRS;
research is warranted to better define the safety and ef- presenting symptoms; and RIM location. The latency of
ficacy of SRS for the management of radiation-induced RIM development was defined as an interval (in months)
intracranial neoplasms. In this large, multicenter series of from first cranial irradiation to RIM diagnosis or SRS. For
consecutive patients treated with SRS for RIMs, we inves- those patients who underwent resection of a meningioma,
tigated the effectiveness and safety of SRS for the man- we also obtained information about WHO grade of the
agement of RIMs. resected meningioma.

Methods SRS Technique

Patients were identified from institutions affiliated SRS was performed using model U, B, C, 4C, Perfex-
with the International Radiosurgery Research Foundation ion, and Icon Gamma Knife units (Elekta AB) depend-
(IRRF; protocol R-16-10) who were treated with SRS be- ing on technology availability at each of the participat-
tween 1990 and 2019. Data collection was approved by the ing centers at the time of SRS. Frame-based stereotaxy
IRBs at each of the participating centers. A database with was performed using the Leksell model G frame (Elekta
variables of interest was established by investigators at the AB), in which the patient was placed under local anes-
University of Virginia and sent to all participating centers. thesia with or without conscious sedation. Frameless SRS
Individual patient data were de-identified and pooled for using a thermoplastic mask was used for hypofractionated
the analyses. In this study, we included patients who had SRS (n = 2, 3%) or when stereotactic frame application
a past history of cranial irradiation, were diagnosed with was not technically possible. Radiosurgical planning was
meningiomas based on MRI findings and/or histological performed using high-resolution pre- and postcontrast T1-
examination (when available) with a latency period, had weighted MRI scans with 1-mm slick thickness. In rare
their RIM treated with SRS, had at least one clinical and cases in which MRI was contraindicated, stereotactic CT
radiological follow-up evaluation after SRS for RIM, and was used for SRS planning. SRS planning was performed
did not have a history of neurocutaneous disorders, such by a multidisciplinary team that included a neurosurgeon,
as neurofibromatosis type 1 (NF1) or type 2 (NF2), based radiation oncologist, and medical physicist. At each center,
on guidelines/practice at that center. MRI features of me- planning was individualized based on patient needs and
ningiomas included extraaxial location, dural involve- imaging findings. Radiosurgical parameters, including the
ment, and avid enhancement after gadolinium administra- margin and maximum dose, number of tumors treated,

J Neurosurg Volume 135 • September 2021 863

 
*44 *
Bunevicius et al.

treatment volume (cm3), number of isocenters, and number SRS for imaging-defined or confirmed WHO grade I
of fractions, were recorded for each SRS session. RIM (Table 1) based on prior resection if one had been
performed. The median patient age at the time of initial
Clinical and Radiographic Follow-Up cranial RT was 5.5 years (range 4 months to 60 years),
Imaging and clinical follow-up was performed at ap- and the median age at the time of SRS for RIM was 39
proximately 6-month intervals for the first 2 years after years (range 12–75 years). The median latency period
SRS with subsequent annual follow-up thereafter. Imaging from original cranial irradiation to RIM development was
follow-up included T1-weighted contrast-enhanced and 29 years (range 4–48 years). The most common reasons
T2-weighted MRI. All follow-up imaging was reviewed for cranial RT were leukemia (n = 11), medulloblastoma
by clinicians at the treating institution. Tumor volume on (n = 9), pituitary adenoma (n = 5), astrocytoma (n = 5),
T1-weighted contrast-enhanced MRI on post-SRS surveil- pineal mass/tumor (n = 4), and craniopharyngioma (n = 3).
lance studies was compared with that in the baseline pre- Twenty-one patients (40%) had prior meningioma resec-
SRS study. tions. RIMs were most frequently discovered incidentally
Clinical follow-up was obtained through a combination (n = 20, 39%). The most common neurological symptoms
of outpatient clinic visits, inpatient admissions, and out- at the time of RIM diagnosis were seizure (n = 10) and fo-
patient visits from referring primary care physicians. We cal neurological deficits (n = 14). The majority of patients
recorded the interval (in months) from SRS for RIM to the harboring RIMs associated with seizures or focal neuro-
last imaging follow-up, last clinical follow-up, or death. logical symptoms (n = 22) had convexity/parasagittal tu-
Patient neurological status at the last clinic follow-up visit mors (n = 16). Seven patients had multiple RIMs.
was classified by the treating team as not changed, im-
proved, or declined. We also recorded the presence, type, SRS Characteristics
and timing of any new neurological deficits or other SRS- Information from 60 SRS procedures (58 single-frac-
related complications. SRS-related adverse events were tion SRSs and 2 hypofractionated SRSs) that were per-
categorized according to the Radiation Therapy Oncology formed for 52 patients for RIMs were available for the
Group CNS toxicity criteria.21 analysis (Table 2). Sixty-five percent of SRS treatments
RIM volume at last imaging follow-up was compared were for solitary RIMs and 35% were for multiple RIMs.
with the pre-SRS tumor volume and categorized as stable, The mean target volume at each SRS procedure was 8.61
regression, or progression. A volumetric increase of RIM ± 7.80 cm3 (median 6.20 cm3, range 0.3–28.40 cm3). In
by ≥ 20% from pre-SRS baseline brain MRI was defined patients who underwent single-fraction SRS, the median
as tumor progression, while a decrease of RIM volume prescription dose was 14 Gy (range 8–20 Gy). The median
by ≤ 20% was defined as tumor regression. Tumors with maximal dose was 26 Gy (range 16–55 Gy). The median
volumetric changes within 20% of the pre-SRS tumor vol- time from initial cranial RT to SRS for RIM was 29 years
ume were noted as stable disease.22 Time to tumor volume (range 4–55 years).
change was also noted. The need for repeat SRS, radio-
therapy, resection, or chemotherapy for RIM recurrence Tumor Control
after the SRS, as well as time and cause of death, was
recorded. The median duration of imaging follow-up after SRS
for RIM was 47.6 months (range 3.8–194.6 months; Table
3, Fig. 1A). At last imaging follow-up, the vast majority of
Statistical Analysis the treated lesions were classified as stable (n = 22, 42%)
Statistical analyses were performed using IBM SPSS or regressed (n = 19, 37%). Progression was documented
Statistics for Windows (version 25.0, IBM Corp.). For all in 9 RIMs (17%). The median time to tumor progression
statistical tests, a p value < 0.05 was considered statistical- was 30 months (range 3–45 months). Thirteen patients re-
ly significant. Overall survival (OS) of the study patients ceived one or more additional treatments for index RIM
was defined as the interval (in months) from the start of after the initial SRS.
index SRS for RIM to the last follow-up or death, and pro- Actuarial PFS rates at 1, 2, and 5 years after SRS were
gression-free survival (PFS) was defined as the interval 94%, 92%, and 83%, respectively. In univariate Cox re-
(in months) from SRS for RIM to the last imaging follow- gression analyses, greater age at SRS (p = 0.031), the pres-
up or MRI-documented tumor progression, whichever oc- ence of multiple RIMs (vs a solitary RIM; p = 0.038), and
curred first. The Kaplan-Meier method was used to calcu- tumor volume ≥ 5 cm3 (p = 0.047) were associated with
late OS and PFS, and alive patients were censored at the increased risk for tumor progression. In the multivariate
last follow-up. The association of clinical and SRS treat- Cox regression model, treatment volume ≥ 5 cm3 (HR
ment factors with OS and PFS was first investigated using 8.226, 95% CI 1.028–65.857, p = 0.047) emerged as an
univariate Cox regression analyses, and significant predic- independent predictor of greater post-SRS progression
tors in univariate analyses were subsequently entered into risk of RIMs (Table 4, Fig. 2). PFS was similar in histo-
a multivariate Cox regression analysis model. The results logically confirmed versus imaging-defined WHO grade I
of Cox regression analysis are presented as hazard ratios RIMs (p = 0.83).
(HRs), 95% confidence intervals (95% CIs), and p values.
Clinical Outcomes
Results At the last follow-up visit, 39 patients were neurological-
Fifty-two patients (30 women and 22 men) underwent ly stable, 8 patients had declined, and 5 patients improved.

864 J Neurosurg Volume 135 • September 2021

 
*44 *
 Bunevicius et al.

TABLE 1. Baseline patient characteristics » CONTINUED FROM PREVIOUS COLUMN

Characteristic Value TABLE 1. Baseline patient characteristics

No. of patients 52 Characteristic Value

Median age at first SRS, yrs (range) 39 (12–75) Meningioma WHO grade, n (%)
Median age at first RT (range) 5.5 yrs (4 mos–60 yrs)   Histologically conformed WHO grade I 20 (39)
Median RIM latency, yrs (range) 29 (4–48)   Radiographically defined 32 (61)
Sex, n (%) Presentation*
 Men 22 (42)   Incidental finding 20
 Women 30 (58)  Headache 8
Indication for initial RT, n (%)  Seizure 10
 Leukemia 11 (21)   Focal neurologic deficit 14
 Medulloblastoma 9 (17)   Data not available/other 4
  Pituitary adenoma 5 (10) Median KPS score prior to SRS (range) 90 (50–100)
 Astrocytoma 5 (10) KPS = Karnofsky Performance Scale.
  Pineal mass/tumor 4 (8) * Total number > 52 because some patients had more than one symptom.
 Craniopharyngioma 3 (6)
  Tinea capitis 2 (4)
  Pilocytic astrocytoma 2 (4) Of the patients who exhibited a neurological decline, 1 pa-
 Glioma 2 (4) tient (13%) had tumor progression; of those who improved
 Meningioma 1 (2) or remained stable, 48% had stable tumors and 36% had
 Ependymoma 1 (2) decreased tumor volume. Three patients (6%) developed
 Experimental 1 (2) new neurological symptoms (grade 4 CNS toxicities) after
 Germinoma 1 (2)
SRS that included gait impairment and leg weakness. Four
patients (8%) experienced SRS-related complications that
  Langerhans cell histiocytosis 1 (2) included intratumoral hemorrhage or T2 signal change.
 Oligodendroglioma 1 (2) In 5 patients who developed a post-SRS complication and
 Retinoblastoma 1 (2) had available information about initial cranial RT, the me-
 Rhabdomyosarcoma 1 (2) dian radiation dose was 35 Gy and ranged from 10 to 64
 Subependymoma 1 (2) Gy. Indications for initial cranial irradiation in this group
Type of initial RT, n (%)
of patients included medulloblastoma (n = 2), pituitary
adenoma (n = 2), craniopharyngioma (n = 2), and leuke-
 WBRT 18 (35) mia (n = 1). The mean latency period from SRS to a new
  Other or unknown 34 (65) neurological deficit or SRS-related complication was 30
Dose of initial RT, Gy ± 29 months (range 1–72 months). Age at RT, age at SRS,
  Available data, n (%) 26 (50) interval from RT to SRS, margin dose, maximal dose, and
  Median (range) 45 (10–64) treatment volume were similar between patients who ex-
No. of fractions of initial RT
perienced a post-SRS complication or new neurological
deficit compared with those who did not (all p values ≥
  Available data, n (%) 14 (27) 0.22). Two patients died during post-SRS follow-up (Fig.
  Median (range) 25 (1–36) 1B). One death was attributed to progression, and the un-
Resection prior to SRS, n (%) 21 (40) derlying cause of death of the other patient was unknown.
Meningioma location, n (%)
 Convexity 43 (37) Discussion
 Parafalcine 24 (21) Our series exploring the safety and effectiveness of
  Cavernous sinus 7 (6) SRS for RIMs offers new insights into patterns of tumor
 Tentorial 7 (6) control and failure from a multiinstitutional cohort. SRS
  Cerebellopontine angle 5 (4) was associated with durable local control of RIMs in the
 Sphenoid 3 (3)
majority of patients and had an acceptable safety profile.
Eighty-five percent of patients had stable or improved neu-
 Sellar/parasellar 3 (3) rological status at the last follow-up, and 14% of patients
 Petroclival/petrous 3 (3) developed new neurological symptoms or SRS-related im-
 Infratemporal 2 (2) aging changes.
 Clinoid 2 (2) Our findings comprise the largest series of its type to
 Other 16 (14) date and indicate that SRS was associated with durable
local control of RIM that was comparable to the local con-
CONTINUED IN NEXT COLUMN » trol rates known from prior reports of SRS for sporadic
meningiomas. During a median imaging follow-up of 47.6

J Neurosurg Volume 135 • September 2021 865

 
*44 *
Bunevicius et al.

TABLE 2. SRS treatment of RIMs TABLE 3. Tumor control and safety

Parameter Value Variable Value
Total no. of SRS procedures, n (%) 60 Median imaging follow-up, mos (range) 47.6 (3.8–194.6)
  Single-fraction SRS 58 (97) Tumor control rate, n (%)
  Hypofractionated SRS 2 (3)  Stable 22 (42)
Total no. of RIMs treated w/ SRS 108  Regression 19 (37)
Multiplicity of RIMs, n (%)  Progression 9 (17)
 Solitary 39 (65)   Data not available 2 (4)
 Multiple 21 (35) Median clinical follow-up, mos (range) 46.91 (7.87–192.19)
Median no. of RIMs treated at each SRS 1 (1–10) Neurological status at last follow-up, n (%)
treatment (range)   No change 39 (75)
Target volume at each SRS treatment  Improved 5 (10)
  Median, cm3 (range) 6.20 (0.3–28.40)  Declined 8 (15)
  Volume ≥5 cm3, n (%) 25 (42%) New post-SRS symptom or complication, n (%) 7 (14)
Median prescription dose, Gy (range)* 14 (8–20) Median mos to new symptoms or 30 (1–72)
Median maximal dose, Gy (range) 26 (16–55) complications (range)
* Only patients who underwent single-fraction radiosurgery (n = 58). Post-SRS treatments, n (%)
  Any treatment 13 (25)
 Resection 9 (17)
months (range 3.8–194.6 months), 17% of patients expe-   Repeat SRS 4 (8)
rienced RIM progression that was documented from 3 to   Fractionated RT 2 (4)
45 months following SRS. PFS rates at 2 and 5 years after  Chemotherapy 1 (2)
the SRS were 92% and 83%, respectively. The long-term
local tumor control rate of RIMs in our series was compa- Crude mortality, n (%) 2 (4)
rable to local tumor control rates for sporadic meningio-
mas that usually exceed 85%.23,24 The local RIM control
rate in the present series was higher compared with the
RIM control rate of 75% previously reported by Kondzi- multiplicity, and invasion of osseous, vascular, or neural
olka and colleagues in a series of 19 patients with RIM structures.10 In our series, one-third of patients harbored
treated with SRS, with a median follow-up of 44 months.10 multiple RIMs and half of meningiomas were in the skull
Another single-institution series of 12 patients with RIM base. SRS should be considered for patients harboring
treated with SRS and followed for a median interval of RIMs in difficult-to-access locations, those with multiple
35 months found that local tumor control was achieved lesions, and those who are poor surgical candidates.25 De-
in all treated patients, and there were 2 cases of distant pending on the location of the RIM, the planning of repeat
tumor recurrence.20 An additional single-institution series conventional fractionated RT can be difficult in patients
of 17 patients with RIM treated with SRS also reported who have received prior wide-field RT without exceeding
a 100% 5-year local tumor control rate, with 1 treatment dose tolerances to critical normal structures, such as the
failure at 65 months after the SRS.9 The current study and optic nerves and chiasm. SRS allows spatially precise and
previously published studies strongly suggest that SRS al- conformal treatment of RIMs that helps to mitigate poten-
lows durable long-term local disease control of RIMs in tial risks associated with wider treatment fields and low-
the majority of patients. However, due to late SRS treat- dose wash (decreasing dose to surrounding structures) of
ment failures, long-term imaging surveillance should be fractionated RT approaches.
considered after SRS of RIM. Another potential limitation We excluded patients with histologically confirmed
is that it remains unknown whether RIMs that ultimately WHO grade II and III meningiomas because it is well es-
do undergo disease progression following SRS do so in a tablished that they are more resistant to SRS therapy and
more aggressive or rapid way than their sporadic counter- have more aggressive clinical courses than WHO grade
parts, or mechanistically whether additional radiation ul- I tumors, and they often require multimodality treat-
timately contributes to worsened chromosomal instability ment. However, a histological diagnosis was not avail-
or mutational burden within the RIM. able in 61% of the study patients. A higher WHO grade
Resection is often considered for initial management of meningiomas is associated with inferior local disease
of RIM.2,8 If technically possible, gross-total resection control and worse prognosis.26 For example, the 5-year lo-
can be curative and surgery allows for the assessment of cal control rate of WHO grade II non-RIMs ranges from
meningioma grade, which can be important for guidance 50% to 60%, and only 10% of WHO grade III non-RIMs
of adjuvant therapies. Another benefit of open resection are controlled with SRS.27–29 These findings indicate that
and tissue acquisition is to rule out other rare pathologies long-term control of RIMs can be inferior when compared
that may include hemangiopericytoma, lymphoma, soli- with sporadic WHO grade I meningiomas, but superior
tary fibrous tumor, and others. However, gross-total resec- to high-grade sporadic meningiomas. Indeed, the genetic
tion of RIMs is not always possible because of frequent and mutational landscape of RIMs is different from that

866 J Neurosurg Volume 135 • September 2021

 
*44 *
 Bunevicius et al.

FIG. 1. Kaplan-Meier plots of PFS (A) and OS (B) in the study patients.

of sporadic meningiomas, which might have radiobiologic should be considered for patients presenting with RIMs
implications.11,12 For example, RIMs had more common exceeding 5 cm3, and this can also have implications for
combined losses of chromosomes 1p and 22q as well as their follow-up frequency. This finding also suggests that
more complex chromosomal aberrations than sporadic earlier SRS when the RIM is smaller may portend a better
meningiomas, while targetable mutations were less com- long-term result for the patient.
mon.11 Further studies are warranted to elucidate molecu- Margin dose was previously implicated as a significant
lar characteristics and possibly identify treatment targets predictor of local control of RIMs10 and sporadic menin-
of RIM. giomas.23,35 However, prescription radiation dose was not
Treatment volume ≥ 5 cm3 was an independent predic- associated with local control of RIMs in the current study,
tor of tumor progression after SRS in multivariate regres- which can be explained by the small sample size and small
sion analyses. Greater patient age at SRS, the presence SRS treatment failure rate. The median prescription radia-
of multiple (vs solitary) RIMs, and treatment volume ≥ 5 tion dose in the present series was 14 Gy and ranged from 8
cm3 were associated with shorter PFS in univariate analy- to 22 Gy. Comparable prescription doses from RIMs were
sis. These findings correspond to the findings in a series previously reported by other groups, with median prescrip-
from the University of Pittsburgh that reported that larger tion doses of 13–14 Gy (range 8–20 Gy).9,20,28 The prescrip-
target volume was associated with shorter post-SRS PFS tion dose for RIMs corresponds to commonly used pre-
of RIM.28 A literature review of RIMs treated with SRS scription radiation doses for sporadic WHO grade I menin-
found that tumors that failed after SRS were larger when giomas,29 emphasizing that the usual radiation dose should
compared with controlled tumors (10.7 cm3 and 2.2 cm3, be used to optimize the local control of meningiomas in
respectively), while age, sex, and tumor grade were similar patients with a prior history of cranial irradiation.
between the two groups.20 Greater tumor volume is also During the median clinical follow-up of 47 months
associated with inferior local control of sporadic meningi- (range 8–192 months), 3 patients experienced new neu-
omas with SRS.28,30–34 Higher risk of progression after SRS rological symptoms after SRS and 1 patient experienced

TABLE 4. Cox regression analysis of predictors of RIM progression after SRS

Predictor Univariate Analysis Multivariate Analysis
Age at RT 1.041 (0.998–1.087), 0.063 —
Age at SRS 1.054 (1.005–1.105), 0.031 p = 0.108
Latency btwn RT & SRS 1.017 (0.989–1.067), 0.497 —
Sex 0.906 (0.468–1.754), 0.769 —
Multiple RIMs* 4.379 (1.085–17.673), 0.038 p = 0.310
Treatment volume ≥5 cm3 8.226 (1.028–65.857), 0.047 8.226 (1.028–65.857), 0.047
Prescription dose, Gy 1.096 (0.822–1.461), 0.532 —
Maximal dose, Gy 0.971 (0.860–1.097), 0.635 —
Data given as HR (95% CI), p value.
* Yes = 1 or no = 0.

J Neurosurg Volume 135 • September 2021 867

 
*44 *
Bunevicius et al.

FIG. 2. Kaplan-Meier plots of PFS stratified by SRS treatment volume (< 5 cm3 or ≥ 5 cm3).

intratumoral hemorrhage. Radiological T2 signal changes the known association between cranial irradiation and the
without associated neurological decline were noted in 3 pa- development of RIMs. The management strategy of RIMs
tients. Comparable rates (5.3%) of post-SRS morbidity and (observation vs active treatment) and timing should be
peritumoral imaging signal changes were noted in one prior selected considering RIM size, location, symptoms, prior
study of Gamma Knife radiosurgery (GKRS) for RIMs.28 therapies, ease of resection, and patient preference. It is
Jensen and colleagues described 1 case (6% of all treated expected that the threshold for initiating active treatment
patients) of increased seizure activity after SRS for RIM can be lower for RIMs versus sporadic tumors given the
that was associated with radiosurgery-related changes on higher probability of more rapid progression, as well as
imaging studies.9 Another group noted a higher incidence the patient’s prior history of a brain tumor, experience with
of SRS toxicity in the setting of RIMs (n = 2, 17%) that cranial irradiation and often other treatments such as re-
occurred within 3–4 months after the SRS and included section, and well-defined opinions regarding intervention
cranial nerve neuropathy and leg weakness/numbness.20 based on their past personal experiences.
The incidence rate of transient and permanent SRS-related
neurological complications for meningiomas is below 10%, Study Limitations
and as many as 40% of patients develop new or worsening
post-SRS peritumoral edema.36–40 These findings suggest This study has limitations typical of a retrospective,
that the safety profile of SRS for RIMs is comparable to multicenter study of a rare intracranial tumor type. Fore-
the adverse event risk after SRS for sporadic meningiomas. most, numerous tumors in this study lacked pathological
Meticulous SRS planning is of paramount importance and confirmation of being a WHO grade I meningioma and
should be exercised for complication avoidance. also lacked MIB-1 labeling data for RIMs. In the current
In our series, RIMs were discovered incidentally in study, inclusion of tumors lacking pathological confirma-
39% of patients. Imaging surveillance is often recom- tion that may have proven to be higher-grade meningio-
mended for incidentally discovered sporadic meningio- mas or ones with high MIB-1 labeling should only have
mas.41 However, a recent systematic review and meta-anal- biased the study outcomes toward less favorable PFS and
ysis reported significant variability in the management of OS following SRS. Initial cranial RT details were not avail-
incidental meningiomas that included active monitoring able for all patients; this prevented us from independently
(50.7%), surgery (27.3%), and SRS (22.0%).42 Patients with evaluating cranial irradiation fields and dose, studying the
prior histories of cranial irradiation for various brain le- association of initial cranial irradiation with RIM latency,
sions are at elevated risk for the detection of small lesions and investigating the safety and effectiveness of subsequent
such as RIMs that are asymptomatic, due to both the in- SRS. We could not confirm whether presumed RIMs devel-
creased frequency of brain imaging for an index lesion and oped within the treatment field of initial cranial irradiation
868 J Neurosurg Volume 135 • September 2021
 
*44 *
 Bunevicius et al.

for patients who did not receive whole-brain RT (WBRT).   5. Sadamori N, Shibata S, Mine M, et al. Incidence of intracra-
However, initial cranial irradiation was performed more nial meningiomas in Nagasaki atomic-bomb survivors. Int J
than 2 decades ago when less sophisticated and conformal Cancer. 1996;​67(3):​318–322.
  6. Shintani T, Hayakawa N, Kamada N. High incidence of me-
cranial irradiation methods were routinely used, raising ningioma in survivors of Hiroshima. Lancet. 1997;​349(9062):​
the possibility of at least some degree of skull irradiation 1369.
even in patients who did not receive WBRT. It is possible   7. Cahan WG, Woodard HQ, Higinbotham NL, et al. Sarcoma
that increasing the use of intensity-modulated photon RT, arising in irradiated bone:​report of eleven cases. 1948. Can-
which results in increased low-dose irradiation wash, could cer. 1998;​82(1):​8–34.
increase the risk of RIMs well outside the radiation fields.   8. Harrison MJ, Wolfe DE, Lau TS, et al. Radiation-induced
Furthermore, due to the longitudinal multicentered design meningiomas:​experience at the Mount Sinai Hospital and
of the study, both intra- and interinstitutional variation of review of the literature. J Neurosurg. 1991;​75(4):​564–574.
  9. Jensen AW, Brown PD, Pollock BE, et al. Gamma knife ra-
RIM treatment strategies is expected because SRS treat- diosurgery of radiation-induced intracranial tumors:​local
ment protocols varied between the 12 participating centers, control, outcomes, and complications. Int J Radiat Oncol
and there have been improvements in SRS techniques and Biol Phys. 2005;​62(1):​32–37.
systems over the study period.43 We also did not systemati- 10. Kondziolka D, Kano H, Kanaan H, et al. Stereotactic radio-
cally evaluate growth trajectories of RIMs before the SRS surgery for radiation-induced meningiomas. Neurosurgery.
treatment, nor did we capture race/ethnicity data. We ex- 2009;​64(3):​463–470.
pect that these variations had limited effect given that all 11. Agnihotri S, Suppiah S, Tonge PD, et al. Therapeutic radia-
tion for childhood cancer drives structural aberrations of
patients were managed at experienced GKRS centers and NF2 in meningiomas. Nat Commun. 2017;​8(1):​186.
according to the prevailing guidelines at the time of treat- 12. Shoshan Y, Chernova O, Juen SS, et al. Radiation-induced
ment. We cannot determine the effect of race/ethnicity on meningioma:​a distinct molecular genetic pattern? J Neuro-
SRS outcomes. Further longitudinal follow-up of patients pathol Exp Neurol. 2000;​59(7):​614–620.
is required to better understand the long-term outcomes of 13. Rubinstein AB, Shalit MN, Cohen ML, et al. Radiation-
SRS for RIMs. Also, while this study is the largest to date, induced cerebral meningioma:​a recognizable entity. J Neu-
additional patient accrual is needed in future studies to ad- rosurg. 1984;​61(5):​966–971.
equately power subgroup analysis to evaluate topics such as 14. Sadetzki S, Flint-Richter P, Ben-Tal T, Nass D. Radiation-
induced meningioma:​a descriptive study of 253 cases. J Neu-
SRS for asymptomatic yet progressive RIMs. In addition, rosurg. 2002;​97(5):​1078–1082.
detailed SRS treatment protocols were reported by all par- 15. Soffer D, Gomori JM, Siegal T, Shalit MN. Intracranial me-
ticipating centers. Due to possible differences in the natural ningiomas after high-dose irradiation. Cancer. 1989;​63(8):​
course of RIMs secondary to proton versus photon therapy, 1514–1519.
our results cannot be generalized to RIMs developing after 16. Borovich B, Doron Y. Recurrence of intracranial meningio-
proton beam therapy, which is commonly used for treating mas:​the role played by regional multicentricity. J Neurosurg.
pediatric patients.44 Our large sample size of RIM, which is 1986;​64(1):​58–63.
17. Borovich B, Doron Y, Braun J, et al. Recurrence of intracra-
an extremely rare disorder, is an important strength of this nial meningiomas:​the role played by regional multicentricity.
study that fortifies the study findings. Part 2:​Clinical and radiological aspects. J Neurosurg. 1986;​
65(2):​168–171.
Conclusions 18. Jääskeläinen J. Seemingly complete removal of histologically
benign intracranial meningioma:​late recurrence rate and
SRS is associated with durable local control of RIMs factors predicting recurrence in 657 patients. A multivariate
in the majority of patients and has an acceptable safety analysis. Surg Neurol. 1986;​26(5):​461–469.
profile that is comparable to that of SRS-treated sporadic 19. Wilson CB. Meningiomas:​genetics, malignancy, and the
meningiomas. SRS should be considered in the manage- role of radiation in induction and treatment. The Richard C.
ment of RIM in difficult-to-access areas and for poor re- Schneider Lecture. J Neurosurg. 1994;​81(5):​666–675.
section candidates. Future studies exploring the safety and 20. Kuhn EN, Chan MD, Tatter SB, Ellis TL. Gamma knife ste-
reotactic radiosurgery for radiation-induced meningiomas.
efficacy of surgical and SRS approaches of RIM, as well Stereotact Funct Neurosurg. 2012;​90(6):​365–369.
as the genetic landscape of RIM, are encouraged to iden- 21. Shaw E, Scott C, Souhami L, et al. Single dose radiosurgical
tify optimal treatment approaches and to improve patient treatment of recurrent previously irradiated primary brain
selection for this rare but challenging disorder. tumors and brain metastases:​final report of RTOG protocol
90-05. Int J Radiat Oncol Biol Phys. 2000;​47(2):​291–298.
22. Yang D-Y, Sheehan J, Liu Y-S, et al. Analysis of factors asso-
References ciated with volumetric data errors in gamma knife radiosur-
  1. Co JL, Swain M, Murray LJ, et al. Meningioma screening gery. Stereotact Funct Neurosurg. 2009;​87(1):​1–7.
with MRI in childhood leukemia survivors treated with 23. Sheehan JP, Starke RM, Kano H, et al. Gamma Knife radio-
cranial radiation. Int J Radiat Oncol Biol Phys. 2019;​104(3):​ surgery for sellar and parasellar meningiomas:​a multicenter
640–643. study. J Neurosurg. 2014;​120(6):​1268–1277.
  2. Umansky F, Shoshan Y, Rosenthal G, et al. Radiation-in- 24. Sheehan JP, Williams BJ, Yen CP. Stereotactic radiosurgery
duced meningioma. Neurosurg Focus. 2008;​24(5):​E7. for WHO grade I meningiomas. J Neurooncol. 2010;​99(3):​
  3. Yamanaka R, Hayano A, Kanayama T. Radiation-induced 407–416.
meningiomas:​an exhaustive review of the literature. World 25. Cohen-Inbar O, Lee CC, Schlesinger D, et al. Long-term re-
Neurosurg. 2017;​97:​635–644.e8. sults of stereotactic radiosurgery for skull base meningiomas.
  4. Preston DL, Ron E, Yonehara S, et al. Tumors of the nervous Neurosurgery. 2016;​79(1):​58–68.
system and pituitary gland associated with atomic bomb ra- 26. Pollock BE, Jacob JT, Brown PD, Nippoldt TB. Radiosurgery
diation exposure. J Natl Cancer Inst. 2002;​94(20):​1555–1563. of growth hormone-producing pituitary adenomas:​factors

J Neurosurg Volume 135 • September 2021 869

 
*44 *
Bunevicius et al.

associated with biochemical remission. J Neurosurg. 2007;​ 42. Islim AI, Mohan M, Moon RDC, et al. Incidental intracra-
106(5):​833–838. nial meningiomas:​a systematic review and meta-analysis of
27. Ding D, Starke RM, Hantzmon J, et al. The role of radiosur- prognostic factors and outcomes. J Neurooncol. 2019;​142(2):​
gery in the management of WHO Grade II and III intracra- 211–221.
nial meningiomas. Neurosurg Focus. 2013;​35(6):​E16. 43. Bunevicius A, Sheehan D, Vance ML, et al. Outcomes of
28. Kondziolka D, Mathieu D, Lunsford LD, et al. Radiosurgery Cushing’s disease following Gamma Knife radiosurgery:​
as definitive management of intracranial meningiomas. Neu- effect of a center’s growing experience and era of treatment. J
rosurgery. 2008;​62(1):​53–60. Neurosurg. 2021;134(2):547–554.
29. Pollock BE, Stafford SL, Link MJ. Stereotactic radiosurgery 44. Scaringi C, Minniti G, Bozzao A, et al. Radiation-induced
of intracranial meningiomas. Neurosurg Clin N Am. 2013;​ malignant meningioma following proton beam therapy for a
24(4):​499–507. choroidal melanoma. J Clin Neurosci. 2015;​22(6):​1036–1037.
30. Ge Y, Liu D, Zhang Z, et al. Gamma Knife radiosurgery for
intracranial benign meningiomas:​follow-up outcome in 130
patients. Neurosurg Focus. 2019;​46(6):​E7. Disclosures
31. Starke RM, Przybylowski CJ, Sugoto M, et al. Gamma Knife Dr. Zacharia reports being a consultant for Medtronic, Inc.,
radiosurgery of large skull base meningiomas. J Neurosurg. and being on the speakers bureau for NICO Corp. Dr. Lunsford
2015;​122(2):​363–372. reports direct stock ownership in Elekta and being a consultant
32. Park K-J, Kano H, Iyer A, et al. Gamma Knife stereotactic to Insightec and DSMB. Dr. McInerney reports receiving funding
radiosurgery for cavernous sinus meningioma:​long-term for non–study-related research from Elekta.
follow-up in 200 patients. J Neurosurg. 2019;​130(6):​1799–
1808. Author Contributions
33. Sheehan JP, Starke RM, Kano H, et al. Gamma Knife radio-
surgery for posterior fossa meningiomas:​a multicenter study. Conception and design: Sheehan. Acquisition of data: Sheehan,
J Neurosurg. 2015;​122(6):​1479–1489. Bunevicius, Suleiman, Patel, Martínez Álvarez, Martinez
34. Starke R, Kano H, Ding D, et al. Stereotactic radiosurgery of Moreno, Liscak, Hanuska, Mathieu, Mau, Caldwell, Tuanquin,
petroclival meningiomas:​a multicenter study. J Neurooncol. Zacharia, McInerney, CC Lee, Yang, Peterson, Trifiletti, Ogino,
2014;​119(1):​169–176. Kano, Warnick, Saylany, Buch, JYK Lee, Strickland, Zada,
35. Mehta GU, Zenonos G, Patibandla MR, et al. Outcomes of Chang, Lunsford. Analysis and interpretation of data: Sheehan,
stereotactic radiosurgery for foramen magnum meningiomas:​ Bunevicius, Patel, Langlois. Drafting the article: Bunevicius,
an international multicenter study. J Neurosurg. 2018;​129(2):​ Trifiletti. Critically revising the article: Sheehan, Bunevicius,
383–389. Patel, Martínez Álvarez, Martinez Moreno, Liscak, Hanuska,
36. Cohen-Inbar O, Lee C-C, Sheehan JP. The contemporary role Langlois, Mathieu, Mau, Caldwell, Tuanquin, Zacharia,
of stereotactic radiosurgery in the treatment of meningiomas. McInerney, CC Lee, Yang, Peterson, Trifiletti, Ogino, Kano,
Neurosurg Clin N Am. 2016;​27(2):​215–228. Warnick, Saylany, Buch, JYK Lee, Strickland, Zada, Chang,
37. Kollová A, Liscák R, Novotný J Jr, et al. Gamma Knife Lunsford. Reviewed submitted version of manuscript: Sheehan,
surgery for benign meningioma. J Neurosurg. 2007;​107(2):​ Bunevicius, Suleiman, Patel, Martínez Álvarez, Martinez Moreno,
325–336. Liscak, Hanuska, Langlois, Mathieu, Mau, Caldwell, Tuanquin,
38. Cai R, Barnett GH, Novak E, et al. Principal risk of peritu- Zacharia, McInerney, CC Lee, Yang, Peterson, Ogino, Kano,
moral edema after stereotactic radiosurgery for intracranial Warnick, Saylany, Buch, JYK Lee, Zada, Chang, Lunsford.
meningioma is tumor-brain contact interface area. Neurosur- Approved the final version of the manuscript on behalf of all
gery. 2010;​66(3):​513–522. authors: Sheehan. Statistical analysis: Bunevicius. Administrative/
39. Unger KR, Lominska CE, Chanyasulkit J, et al. Risk factors technical/material support: Sheehan. Study supervision: Sheehan.
for posttreatment edema in patients treated with stereotactic
radiosurgery for meningiomas. Neurosurgery. 2012;​70(3):​ Correspondence
639–645. Jason Sheehan: University of Virginia Health System, Charlottes-
40. Sheehan JP, Lee C-C, Xu Z, et al. Edema following Gamma ville, VA. [email protected].
Knife radiosurgery for parasagittal and parafalcine menin-
giomas. J Neurosurg. 2015;​123(5):​1287–1293.
41. Goldbrunner R, Minniti G, Preusser M, et al. EANO guide-
lines for the diagnosis and treatment of meningiomas. Lancet
Oncol. 2016;​17(9):​e383–e391.

870 J Neurosurg Volume 135 • September 2021

 
*44 *
 CLINICAL ARTICLE
J Neurosurg 135:871–880, 2021

Neuropsychological assessments before and after awake

surgery for incidental low-grade gliomas
Sam Ng, MD,1 Guillaume Herbet, PhD,1–3 Anne-Laure Lemaitre, PhD,1,2
Jérôme Cochereau, MD, MSc,3,4 Sylvie Moritz-Gasser, PhD,1–3 and Hugues Duffau, MD, PhD1–3
1
Department of Neurosurgery, Gui de Chauliac Hospital, Montpellier University Medical Center, Montpellier; 2Department of
Speech–Language Pathology, University of Montpellier; 3INSERM U1191, Team “Plasticity of Central Nervous System, Human
Stem Cells, and Glial Tumors,” Institute of Functional Genomics, Montpellier; and 4Department of Neurosurgery, Poitiers
University Hospital, Poitiers, France

OBJECTIVE  Early surgery in presumed asymptomatic patients with incidental low-grade glioma (ILGG) has been sug-
gested to improve maximal resection rates and overall survival. However, no study has reported on the impact of such
preventive treatment on cognitive functioning. The aim of this study was to investigate neuropsychological outcomes in
patients with ILGG who underwent preventive surgery.
METHODS  This was a retrospective analysis of a consecutive series of patients with ILGG who underwent awake sur-
gery and who had presurgical and 3-month postsurgical neuropsychological assessments. Data were normalized into z-
scores and regrouped by cognitive domains. Clinicoradiological data, histomolecular profile, and differences in z-scores
(Δz-scores) were analyzed.
RESULTS  Forty-seven patients were included (mean age 39.2 ± 11.3 years). Twenty-eight patients (59.6%) underwent
supratotal or total resections. All patients were still alive after a mean follow-up of 33.0 ± 30.8 months. Forty-one patients
(87.2%) had stable (n = 34, 72.3%) or improved (Δz-score > 1; n = 7, 14.9%) neurocognitive outcomes after surgery. Six
patients (12.8%) presented a slight impairment (Δz-score < −1) in at least one cognitive domain. The mean presurgical
and postsurgical z-scores were comparable except in the psychomotor speed and attention domain. A significant cor-
relation between presurgical executive functioning and tumor volume was reported, whereas the extent of resection and
histomolecular profile did not impact neuropsychological outcomes.
CONCLUSIONS  Early surgical treatment in presumed asymptomatic patients with ILGG was associated with stable or
improved neuropsychological outcomes in 87.2% of patients at 3 months, with only mild cognitive decline observed in 6
patients. In return, supratotal or total resections were achieved in most patients, and all patients were still alive at the end
of the follow-up.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201507
KEYWORDS  awake surgery; brain mapping; diffuse low-grade glioma; incidental glioma; neuropsychological
assessment; quality of life; oncology; surgical technique

D iffuse low-grade gliomas (LGGs—i.e., WHO

grade II gliomas) are frequently discovered inci-
dentally due to increasing access to neuroimaging.
Incidental LGGs (ILGGs) precede the symptomatic stage
servative approach.7 Nonetheless, early surgical treatment
for ILGG remains a controversial issue, because patients
are presumed asymptomatic and may be enjoying active
social and professional lives, even though they may suffer
in the natural history of glioma, which is later inevitably from low to moderate neuropsychological disturbances.8
characterized by a migration of the tumor and an unpre- Thanks to awake intraoperative mapping, several surgical
dictable risk of malignant transformation.1–3 A consider- series have suggested better oncological outcomes due to a
able improvement in long-term survival has been reported greater extent of resection (EOR) in patients with ILGG,9
after early and maximal resection in LGG,4–6 thereby without treatment-related neurological deficits,10 with a
doubling the overall survival in comparison with a con- low rate of postoperative seizures,11 and with a high rate

ABBREVIATIONS  AED = antiepileptic drug; EOR = extent of resection; ILGG = incidental LGG; LGG = low-grade glioma; RL-RI16 = Rappel libre et Rappel indicé à 16
items; ROCF = Rey-Osterrieth complex figure; TCF = Taylor complex figure; TMT = Trail-Making Test; WAIS-IV = Wechsler Adult Intelligence Scale–Fourth Edition.
SUBMITTED  April 28, 2020.  ACCEPTED  July 6, 2020.
INCLUDE WHEN CITING  Published online December 4, 2020; DOI: 10.3171/2020.7.JNS201507.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  871

*44 *
Ng et al.

of return to professional activities.12 Although expecta- neuropsychologist (G.H. or A.-L.L.). Language abilities
tions must be set very high in terms of functional results were assessed by a dedicated speech therapist (S.M.-G.) in
in this subpopulation of LGG, no study to date has in- patients with left-sided tumors and/or left-handed patients.
vestigated neuropsychological outcomes after “prophy- The neurocognitive assessment battery was selected to
lactic” surgery. Indeed, neuropsychological functioning cover all cognitive domains: 1) language and semantics
is an increasingly important outcome measure in neuro- were assessed with the DO80 naming task,18 the Pyra-
oncology,13,14 because patients with LGG may suffer from mids and Palm Trees Test,21 and phonological and seman-
deficits in various cognitive domains resulting either from tic verbal fluencies; 2) psychomotor speed and attention
the tumor itself or from current treatment modalities.15,16 were assessed using the subtest “code” from the Wechsler
Such disorders may impact health-related quality of life.17 Adult Intelligence Scale–Fourth Edition (WAIS-IV),22 the
In this study we aimed to assess neuropsychological out- subtests “color naming” and “reading” from the Stroop
comes by comparing pre- and postoperative cognitive as- task,23 and part A of the Trail-Making Test (TMT);24 (3)
sessments in a consecutive series of patients with ILGG executive functioning and working memory were assessed
who underwent prophylactic awake surgery. using part B of the TMT, phonological and semantic ver-
bal fluencies, Stroop interference minus naming tasks,
and forward and backward digit span from the WAIS-IV;
Methods 4) episodic verbal memory was assessed with the Rappel
Study Design and Participants libre et Rappel indicé à 16 items (RL-RI16), a French ad-
We conducted a single-center retrospective analysis of aptation of the Free and Cued Selective Reminding Test;25
a consecutive series of patients with ILGG, who under- and 5) visuospatial functioning was assessed with the
went surgery between December 2011 and May 2019. All Rey-Osterrieth complex figure (ROCF), the Taylor com-
included patients were asymptomatic (i.e., without clinical plex figure (TCF),26 the Bells test,27 and the line bisection
manifestations as assessed by a standard neurological ex- task.28 All participants’ raw scores were appropriately
amination and without a history of seizures) from initial di- aligned to published French normative data (adjusted ac-
agnosis to surgery. Exclusion criteria were as follows: pre- cording to educational level, age, and sex) and subsequent-
vious chemotherapy or radiotherapy, previous neurological ly converted into z-scores, except for results from the line
disease with neurocognitive consequences (e.g., multiple bisection tasks, which were converted to a simple cutoff
sclerosis), non–French-speaking foreigners, and patients score. The difference between postsurgical z-scores and
lacking perioperative neuropsychological assessments. presurgical z-scores (i.e., Δz-score) was calculated.
Note that some additional neuropsychological tasks
Interventions were tailored to the location of the tumor and to the perfor-
Patients were assigned to surgery if 2 or more MRI ses- mances of the patients in a specific domain (e.g., reading
sions, spaced at least 3 months apart, demonstrated an ob- of regular words, irregular words, pseudowords, and text
jective expansion of FLAIR signal abnormalities. All pa- reading in right-handed patients with left-sided tumors)
tients underwent surgery with awake mapping, performed and may thus not have been conducted for all patients.
by the senior author (H.D.).
Clinical and Radiological Measures
Intraoperative cortical and subcortical mapping was
performed using direct electrostimulation, through an The following parameters were obtained: sex, age at
asleep-awake-asleep protocol. Intraoperative motor, sen- the time of surgery, handedness, reason for initial radio-
sory, and cognitive mapping included counting and DO80 logical examination, the interval between diagnosis (i.e.,
picture naming18 in concert with contralateral movement the date of the first MRI) and surgery, the follow-up time,
(dual tasks). Complementary tasks were also used, de- and the overall survival.
pending on the tumor location, the patient’s presurgical The tumor volumes (volume at diagnosis, presurgical
language and neuropsychological assessment, and the volume, and postsurgical volume) were calculated using
patient’s expectations regarding his/her work and hob- dedicated software (Myrian; Intrasense). The mean veloc-
bies (e.g., word reading, semantic association, visuospatial ity of diameter expansion was obtained from the first MRI
tasks, mentalizing). The technical procedures involved and the presurgical MRI. The mean diameter (Dmean) was
with this awake mapping have been extensively described deduced from the volume (V) by using the previously de-
in previous reports.19,20 Postsurgical MRI was performed scribed formula:29,30 Dmean = (2 × V)1/3.
within 24 hours following the surgery and then at 3 months. The EOR and residual tumor volume (if any) were
Antiepileptic drugs (AEDs) were systematically adminis- measured on the MRI sequences performed 3 months af-
tered during at least the first 3 months following surgery. ter surgery. In accordance with previous studies,31–33 the
All patients received intensive cognitive rehabilitation at EOR classification was defined as supratotal when there
home after surgery for a period of at least 3 months. were no residual FLAIR-weighted signal abnormalities
and when the surgical cavity was larger than the presur-
gical FLAIR tumor volume; as total when no residual
Measures FLAIR-weighted signal abnormalities were present; as
Neuropsychological Assessment subtotal when the residual volume was < 10 cm3; and as
All participants underwent a neuropsychological ex- partial when residual volume was ≥ 10 cm3. In addition,
amination the day prior to surgery and 3 months after the designation gross-total resection combined both total
surgery; the examinations were performed by the same and supratotal resections.

872 J Neurosurg Volume 135 • September 2021

 
*44 *
 Ng et al.

Primary and Secondary Outcomes coxon signed-rank test, according to the normality of the
Primary Outcome distribution, which was assessed with the Shapiro-Wilk
In order to assess neuropsychological changes before test. The effect of the side of the tumor on Δz-scores was
and 3 months after surgery, the following outcomes were assessed using the Mann-Whitney test. The different loca-
analyzed: 1) mean z-scores regrouped by cognitive do- tions of the tumor, histologies, and types of resection were
main (i.e., language, psychomotor speed and attention, ex- compared using a 1-way ANOVA or the Kruskal-Wallis
ecutive functioning, episodic verbal memory, and visuo- test. Spearman’s rho correlation coefficient was used to as-
spatial functioning); 2) number of patients who showed sess the relationship between presurgical z-scores or Δz-
stability (Δz-score = [−1;1]), decline (Δz-score < −1), or scores and the following: tumor volume, patients’ age, and
improvement (Δz-score > 1) by cognitive domain; and 3) tumor velocity of diameter expansion. For all statistical
number of patients who showed stability in all cognitive analyses, p < 0.05 was considered statistically significant.
domains (Δz-score = [−1;1]), decline in at least one cogni- Statistical tests were performed with GraphPad Prism 7.0
tive domain (Δz-score < −1), or improvement in at least (GraphPad Software, Inc.).
one cognitive domain (Δz-score > 1).
Patients with poor cognitive function were considered Results
to have a z-score < −1 (indicating that cognitive perfor- Participants
mance was lower than that of 85% of the reference popu-
lation), patients with a deficit were considered to have a z- Patient demographics and radiological features are pro-
score < −1.65 (indicating that cognitive performance was vided in Table 1. The reasons for initial imaging, leading
lower than that of 95% of the reference population), and to ILGG diagnosis in most patients, were headache or ear,
patients reporting a variation of > 1 SD (Δz-score < −1 or nose, and throat symptoms (n = 28, 59.6%). All reasons
> 1) were considered to have a significant change in their for initial imaging are detailed in Table 2. Forty-seven pa-
performance. A similar methodology has been described tients were selected, according to the inclusion criteria (31
in previous studies on this topic.8,16,34,35 women and 16 men), with a mean age of 39.2 ± 11.3 years
Note that patients were used as their own control by (range 21–65 years) at the time of surgery. The mean delay
means of paired presurgical and postsurgical evaluations. between image results and surgery was 3.0 ± 3.62 years
This main outcome was designed to specifically assess the (range 0.17–14.2 years). The mean presurgical tumor vol-
effect of surgery on neuropsychological changes, taking ume was 23.2 ± 23.9 cm3 (range 1.4–110 cm3).
into account the possibility that later neuropsychological
assessment may have been biased by the progression of Surgical Results
the disease itself, or by the administration of adjuvant on- Type of resection, EOR, and histomolecular features
cological treatments. are provided in Table 1. Twelve patients (25.5%) under-
went a supratotal resection, and 16 (34.0%) underwent a
Secondary Outcomes total resection. Nineteen patients (40.4%) had a subtotal
Comparisons of Δz-scores (regrouped by cognitive do- resection, with a mean tumor residual of 1.4 ± 2.6 cm3
main) were performed according to the hemispheric loca- (range 0.4–9 cm3). A microfocus with endothelial prolif-
tion of the tumor, the lobal location of the tumor, the type eration was diagnosed in 3 patients (6.4%).
of surgical resection, and histomolecular results. Correla-
tion analyses for each cognitive domain were conducted Early Postsurgical Outcomes
between mean presurgical z-scores/Δz-scores and tumor Twenty-six patients (55.3%) showed a transient imme-
characteristics (e.g., age, velocity of diameter expansion, diate postsurgical worsening (including 17 mild language
tumor volume). disorders, 3 visuospatial deficits, 3 with hemiparesis, and
3 supplementary motor area syndromes). The mean length
Standard Protocol Approvals and Patient Consent of hospital stay was 4.1 days (median 4 days, range 2–7
All patients gave informed consent for the retrospective days). All patients underwent intensive motor and cogni-
collection of their clinical data. Patients were not subject- tive rehabilitation at home. There were no postsurgical
ed to procedures and were not required to follow rules of general complications (including hematomas or venous
behavior outside routine clinical care. Data and imaging thrombotic events) except for 1 wound infection that re-
were analyzed after anonymization in accordance with the quired surgical scar revision and antibiotics.
Personal Data Protection Act and the Code of Conduct for
Responsible Use of Human Tissue and Medical Research. Clinical and Oncological Follow-Up
This work was conducted in compliance with the standard Three months after surgery, all patients had totally re-
ethics of our institution for retrospective study. covered from transient neurological worsening, assessed
by a standard clinical examination. Eight patients (17.0%)
Statistical Analysis experienced a transient partial seizure in the postsurgical
Continuous variables were expressed as the mean ± course, and 3 among them (6.4% of the cohort) experi-
SD, and categorical variables were expressed as numbers enced repeated seizures beyond 12 months following sur-
and proportions. Categorical variables were compared gery. All patients were still receiving AEDs at 3 months
using Fisher’s exact test, and continuous variables were after surgery. Thirty patients (63.8%) had stopped AEDs
compared with either the paired Student t-test or the Wil- at 12 months after surgery. Among 41 patients who were

J Neurosurg Volume 135 • September 2021 873

 
*44 *
Ng et al.

TABLE 1. Clinical, radiological, and histopathological features in TABLE 2. Reasons for initial imaging
47 patients with ILGG
Reason for Imaging Total Population, n = 47
Total Population,
Headache 19 (40.4)
Feature n = 47
Head trauma 3 (6.4)
Demographics ENT symptoms 9 (19.1)
  Mean age, yrs (SD) 39.2 (11.3) Follow-up for another disease 2 (4.3)
  Sex, female, no. (%) 31 (66.0) Research protocol 1 (2.1)
  Right-handed, no. (%) 34 (72.3) Screening test for familial history of glioma 1 (2.1)
  Ambidextrous, no. (%) 2 (4.3) Retinal migraine 3 (6.4)
  Mean education, yrs (SD) 14.3 (3.1) Anosmia 2 (4.3)
  Time from MRI finding to surgery, yrs (SD) 3.0 (3.62) Facial dysesthesia 5 (10.6)
Tumor location, no. (%) Algodystrophy 1 (2.1)
 Left 27 (57.4) Meningitis 1 (2.1)
 Right 20 (42.6)
ENT = ear, nose, throat.
 Frontal 18 (38.3)
 Temporal 8 (17.0)
 Parietal 4 (8.5)
 Insular 17 (36.2) are detailed in Table 3. Episodic memory was the cognitive
Tumor radiological characteristics domain that was most impaired (mean z-score: −0.48 ±
0.85; percentage of patients with z-score < −1: 30.3%). Pho-
  Mean vol at diagnosis, cm3 (SD) 16.0 (20.4)
nological and categorical fluency task scores were slightly
  Mean preop vol, cm3 (SD) 23.2 (23.9) decreased (mean z-scores: −0.32 ± 1.20 and −0.18 ± 1.26,
  Mean preop velocity of diameter expansion, 3.5 (3.4) respectively), affecting both language (mean z-score: −0.17
  mm/yr (SD) ± 0.80; percentage of patients with z-score < −1: 10.6%)
  Mean postop vol, cm3 (SD) 1.4 (2.6) and executive functioning (mean combined z-score: −0.11
  Supratotal resection, no. (%) 12 (25.5) ± 0.67; percentage of patients with z-score < −1: 10.6%)
  Total resection, no. (%) 16 (34.0) domains. Overall, 16 patients (34.0%) presented with poor
functioning (z-score < −1) in at least one cognitive domain.
  Subtotal resection, no. (%) 19 (40.4)
Among these, 5 patients (10.6%) presented a deficit (z-score
  Partial resection, no. (%) 0 (0.0) < −1.65) in at least one cognitive domain.
  GTR, no. (%) 28 (59.6)
Histopathology, no. (%) Postsurgical Neuropsychological Assessment at 3
  Microfocus of endothelial proliferation 3 (6.4) Months and Comparisons With Baseline
  Diffuse astrocytoma IDH wild-type* 10 (21.3) The results of the 3-month postsurgical neuropsycho-
  Diffuse astrocytoma IDH mutant* 13 (27.7) logical tests are detailed in Table 3. Additional results in-
  Oligodendroglioma IDH mutant/1p19q 23 (48.9)
cluding the proportion of patients with z-scores < −1 and
 codeletion*
> 1 are provided in Supplemental Table 1.
No statistically significant difference was found be-
  Oligodendroglioma NOS* 1 (2.1) tween presurgical and postsurgical z-scores except in
GTR = gross-total resection (calculated as supratotal plus total resections); psychomotor speed and attention abilities (mean z-scores:
IDH = isocitrate dehydrogenase; NOS = not otherwise specified. 0.12 ± 0.72 vs −0.22 ± 1.05, respectively; W = −313; p =
* According to the 2016 WHO classification. 0.035; Fig. 1). The proportion of patients with cognitive
disturbances (z-score < −1) was comparable before and
after surgery within all cognitive domains. No patient pre-
employed in normal professional activity before surgery, sented with significant deviations from normal, either pre-
38 (92.7%) returned to work on average 5.8 ± 3.9 months operatively or postoperatively, on the line bisection task.
after surgery (range 1–18 months). Overall, 87.2% of patients showed no decline in their
No patient received oncological adjuvant treatment cognitive performances (Δz-score ≥ −1) in any cognitive
within the first 6 months following surgery. Eleven patients domain at 3 months postsurgery (Fig. 2A). The distribu-
(23.4%) received chemotherapy on average 30.2 ± 20.9 tion of patients’ Δz-scores by cognitive domain is detailed
months after surgery (range 8–78 months). Five patients in Fig. 2B. Among 6 patients (12.8%) who showed a de-
(10.6%) received radiotherapy on average 32.0 ± 19.7 months cline in their cognitive performance (Δz-score < −1) in at
after surgery (range 10–48 months). The mean follow-up least one cognitive domain, 4 had an impairment of their
was 33.0 ± 30.8 months (range 3–96 months) after surgery. psychomotor speed and attention abilities. All 6 patients
All patients were still alive at the end of the follow-up. resumed their professional activities after surgery.

Presurgical Neuropsychological Assessment Subgroup Comparisons for Δz-Scores

The results of the baseline neuropsychological testing No impact of lateralization of the tumor on Δz-scores

874 J Neurosurg Volume 135 • September 2021

 
*44 *
 Ng et al.

TABLE 3. Preoperative and 3 months postsurgical was found (Fig. 3A). There was, however, a significant ef-
neuropsychological test results fect of tumor location on executive functioning (H(2) =
Mean z-Score (SD) 6.21, p = 0.0448; Fig. 3B). There was no significant effect
Presurgical, Postsurgical,
of the type of resection (supratotal vs total vs subtotal) on
Test n = 47 n = 47 p Value*
Δz-scores in any cognitive domain (Fig. 3C). Similarly,
there was no significant effect of histomolecular profile on
Language Δz-scores in any cognitive domain (Fig. 3D).
 DO80† 0.13 (1.31) 0.19 (0.94)
 PPTT† −0.05 (1.40) 0.04 (1.30) Correlations
  Phonological fluency −0.32 (1.20) −0.47 (1.09) Spearman’s rho correlation coefficient in each cogni-
  Categorical fluency −0.18 (1.26) −0.37 (1.29) tive domain was used to assess the relationship between
presurgical z-scores and tumor volume, patient age, and
  Reading text† −0.71 (0.69) −0.78 (1.06)
tumor velocity of diameter expansion. The same method
  Regular words† −0.07 (0.82) −0.05 (0.65) was used to assess the relationship between Δz-scores and
  Irregular words† 0.56 (0.59) 0.63 (0.57) tumor volume, patient age, and tumor velocity of diameter
 Pseudowords† −0.23 (1.15) −0.26 (1.05) expansion. Two-tailed tests were conducted. A signifi-
  Combined z-scores −0.17 (0.80) −0.35 (0.98) 0.070 cant correlation was found between presurgical z-scores
Psychomotor speed and in the executive functioning domain and tumor volume (r
attention = −0.356 [95% CI −0.59, −0.068], p = 0.014). The pre-
surgical z-score in the language domain and tumor vol-
  WAIS-IV code 0.29 (0.98) 0.14 (1.02)
ume showed a tendency to be statistically correlated (r =
  TMT A 0.55 (0.55) 0.43 (1.05) −0.281 [95% CI −0.53, 0.015], p = 0.055). The Δz-score in
  Stroop naming −0.06 (1.02) −0.30 (1.41) the episodic memory domain and patient age also showed
  Stroop reading −0.39 (1.41) −0.94 (1.57) a tendency to be statistically correlated (r = −0.38 [95%
  Combined z-scores 0.12 (0.72) −0.22 (1.05) 0.035 CI −0.68, 0.020], p = 0.055). Other correlations were non-
Executive functioning significant.
  TMT B 0.17 (1.01) 0.19 (1.46)
  TMT B-A −0.21 (1.13) 0.10 (0.74) Discussion
  Stroop interference −0.13 (0.84) −0.03 (0.96) At 3 months after prophylactic surgery for ILGG, stan-
  Stroop I-D −0.17 (0.89) −0.00 (0.94) dard clinical examinations revealed that all 47 patients
had completely recovered from their transient deficits,
  Phonological fluency −0.32 (1.20) −0.47 (1.10)
and additionally, multimodal neurocognitive assessments
  Categorical fluency −0.18 (1.26) −0.37 (1.29) indicated that 87.2% of patients presented no significant
  Forward span 0.05 (1.15) −0.23 (1.56) decline in their higher cognitive functions. Interestingly,
  Backward span −0.01 (0.83) −0.03 (1.08) nearly 15% of patients showed an improvement in their
  Combined z-scores −0.11(0.67) −0.11 (0.74) 0.967 cognitive performance, whereas 12.8% of patients did
Episodic memory not show complete recovery of their neuropsychological
  RL-RI16: 1st free recall −0.62 (1.06) −0.56 (1.48)
abilities in the first 3 months following surgery. Overall,
psychomotor speed and attention was found to be the cog-
  RL-RI16: 2nd free recall −0.37 (0.98) −0.32 (0.94) nitive domain that was most impacted by surgery. The
  RL-RI16: 3rd free recall −0.59 (1.03) −0.59 (0.98) strengths of this study include the following: 1) we provide
  RL-RI16: delayed recall −0.35 (0.96) −0.80 (1.12) the largest series of patients with LGG who underwent
  Combined z-scores −0.48 (0.85) −0.70 (1.19) 0.471 systematic presurgical and postsurgical neuropsychologi-
Visuospatial functioning cal assessment after awake surgery at a prespecified fol-
  ROCF/TCF: copy 0.70 (0.49) 0.65 (0.50)
low-up time; 2) the patients presented in this series were
used as their own controls by means of paired presurgical
  ROCF/TCF: immediate recall 0.43 (0.85) 0.22 (1.07) and postsurgical evaluations; 3) neuropsychological out-
  ROCF/TCF: delayed recall 0.39 (0.95) 0.10 (1.14) comes have not previously been investigated in a selected
  Bells test: omissions 0.33 (0.98) −0.07 (0.89) population of incidentally discovered LGG; and 4) none
  Bells test: differences 0.06 (0.82) 0.14 (0.68) of the selected patients experienced potential confounding
  Combined z-scores 0.39 (0.53) 0.17 (0.54) 0.147 factors (such as seizures) prior to baseline neuropsycho-
logical assessment, nor did they receive adjuvant oncologi-
PPTT = Pyramids and Palm Trees Test. cal treatment that may have biased postsurgical neuropsy-
* Student t-tests were used to compare combined z-scores for language and chological results.36
executive functioning domains, because variables were normally distributed.
Wilcoxon signed-rank tests were used for other cognitive domains, because
variables were non-normally distributed. Preoperative Neuropsychological Assessment
† Some language tasks were assessed only for patients with left-sided tumors Many controversies have arisen in the neuro-oncolog-
and/or left-handed patients. ical community on how surgery, chemotherapy, and ra-
diotherapy may contribute to neurocognitive impairments
that can have an impact on social life, professional abili-

J Neurosurg Volume 135 • September 2021 875

 
*44 *
Ng et al.

FIG. 1. Box plot showing mean z-scores by cognitive domain: presurgical versus 3-month postsurgical results. Data are presented
as medians, quartiles, and extreme values. Student t-tests were used to compare z-scores for language and executive function-
ing domains, because variables were normally distributed. Wilcoxon signed-rank tests were used for other cognitive domains,
because variables were non-normally distributed. ns = nonsignificant. *p < 0.05.

ties, and consequently health-related quality of life.15 As a In a recent study investigating cognitive outcomes in a
result of these debates, neuropsychological assessment has population of 49 adult patients with symptomatic LGG,
emerged as an increasingly important objective measure surgery was even associated with an improvement in
for evaluation of the impact of current treatment modali- memory and executive functions, with no evidence for
ties. Pretherapeutic neuropsychological data in ILGG have deterioration in other cognitive domains,34 although it is
been rarely reported, but recent studies suggest that most worth mentioning that only 23 patients underwent awake
patients who have undergone a cognitive screening, such surgery in that series, and that postsurgical neuropsycho-
as the Montreal cognitive assessment,8,16 usually suffer logical assessments were not performed at a prespecified
from short-term memory loss, reduced attention, and re- follow-up time. Nevertheless, no study has investigated
duced verbal fluency,16 with impairments of their psycho- neuropsychological outcomes in ILGG, although func-
motor speed and executive functioning.35 Our series sup- tional outcome measures are categorically required in this
ports the finding that more than one-third of patients with context of an incidental discovery. Our results suggest that
ILGG present with poor functioning in at least one cogni- few patients show an incomplete recovery at 3 months, and
tive domain, and more than 10% suffered from an objec- that most of these had concerns with psychomotor speed
tive cognitive deficit. These results not only confirm that and attention. The variation in performance in psychomo-
patients with ILGG are not, strictly speaking, asymptom- tor speed and attention processes must be interpreted with
atic,8,16 but also suggest that neuropsychological outcomes caution, because the mean postsurgical z-score in this do-
are related to the natural course of the disease prior to any main was −0.22 ± 1.05, which remains close to the score
therapeutic choice. Indeed, presurgical executive function- of the general population and may not significantly impact
ing deficits were correlated with the volume of the tumor, health-related quality of life, especially given that 97.1%
emphasizing that infiltration of the glioma already impacts of patients with ILGG were previously reported to have
cognitive functioning in many so-called presymptomatic resumed their active professional life after surgery.12 Inter-
patients,37 and that symptoms (e.g., seizures) usually occur estingly, correlation analyses indicated that the variation
at an overcompensated stage of brain plasticity.38 in z-scores induced by surgery is not correlated with the
volume of the tumor, suggesting that EOR and neuropsy-
Postoperative Neuropsychological Assessment chological outcomes are not related. In addition, subgroup
Recent series reporting functional results after awake comparisons based on the type of resection (supratotal
surgery for ILGG suggest that complete neurological re- vs total vs subtotal) did not show significant differences
covery from transient postoperative deficits, as assessed by in neuropsychological outcomes, supporting the need to
a simple clinical examination, is standard at 3 months.9,10 achieve a maximal resection of the tumor for oncological
876 J Neurosurg Volume 135 • September 2021
 
*44 *
 Ng et al.

FIG. 2. Graphs showing Δz-scores at 3-month postsurgical assessment from baseline. A: Percentage of patients with impairment
(Δz-score < −1), stability (Δz-score = [−1;1]), and improvement (Δz-score > 1). B: Distribution of Δz-scores by cognitive domain.
a
For overall score, patients presenting both Δz-scores < −1 and > 1 in at least one domain were considered as impaired. Reported
percentages may not add up to 100% due to rounding. Figure is available in color online only.

purposes. Interestingly, patient age tended to be correlated strategy is still open to debate in ILGG, because many
with the differences in z-scores after surgery (especially neuro-oncologists remain reluctant to propose a prophylac-
in the episodic memory domain), suggesting that cognitive tic surgical treatment in so-called asymptomatic patients.
recovery and neuroplasticity decrease with age. The long delay observed in the present study between tu-
mor discovery and surgery (the mean delay was 3 years) is
Rationale for Performing Early Prophylactic Surgery for a reflection of this conservative “watch-and-wait” attitude;
ILGG most patients from this series were diagnosed and followed
Although early maximal resection significantly improves in other institutions before being referred to our center.
overall survival in symptomatic LGG by delaying the risk First, ILGG refers to a clinically silent stage belonging
of malignant transformation,39,40 the optimal therapeutic to a continuum in the natural history of diffuse glioma (the
J Neurosurg Volume 135 • September 2021 877
 
*44 *
Ng et al.

FIG. 3. Box plots showing z-scores and Δz-scores at 3-month postsurgical assessment from baseline by cognitive domain, based
on hemispheric side of the tumor (A), lobe location of the tumor (B), type of resection (C), and histopathological results (D). *p <
0.05. The Mann-Whitney test was conducted to compare hemispheric sides of the tumors. The Kruskal-Wallis test was conducted
for other comparisons. Figure is available in color online only.

median delay before reaching the symptomatic stage has The estimation of the onco-functional balance in in-
been reported to be approximately 48 months)41 and shares cidentally discovered LGG must take into account the
the same characteristic of permanent migration along the oncological benefits reported in the present series (no pa-
white matter pathways.1,2,29 Malignant transformation is tients died during a mean follow-up of 33 months) and
unavoidable and unpredictable, with a risk of acute trans- from other published series.9,16 In addition, our results
formation to high-grade glioma at any stage of the disease3 support other studies’ findings that awake-guided surgery
(it is worth pointing out that we found 3 patients with a allows not only the avoidance of permanent neurological
microfocus of endothelial proliferation in this series). deficits but also a high degree of preservation of cognitive
Furthermore, recent studies have reported that EOR was performance, with an acceptable rate of mild cognitive
an independent prognostic factor for longer survival in pa- decline postsurgery, keeping in mind that such declines
tients with LGG,4,6,42 thus advocating for the principle of may still be compatible with a return to normal daily liv-
early resection. In the present series, we report a high rate ing and professional activities.12 Interestingly, it should
of supratotal (25.5%) and total (34.0%) resections, with lim- also be mentioned that 14.9% of patients experienced an
ited tumor residuals (mean postoperative volume 1.4 cm3; 0 improvement of their neurocognitive status after surgery.
partial resections). Indeed, the smaller preoperative tumor These results will assist medical and surgical neuro-on-
volume observed in ILGG allows achievement of higher cologists to better advise patients with newly diagnosed
EOR through preventive surgery and better oncological ILGG. In this setting, developing an optimal treatment
outcomes,9 with the possibility to postpone postoperative strategy at the individual level in so-called asymptomat-
adjuvant oncological treatment43 (in this series only 11/47 ic patients remains challenging, and it is crucial to con-
patients received chemotherapy, with a mean delay of 32.0 sider the immediate and delayed consequences of taking
± 20.9 months after surgery; and 5/47 received radiothera- a conservative versus a surgical approach. Thus, patients
py, with a mean delay of 32.0 ± 19.7 months after surgery). with ILGG must be referred to tertiary centers where they
878 J Neurosurg Volume 135 • September 2021
 
*44 *
 Ng et al.

can benefit from a personalized therapeutic strategy44 tai-   3. Cochereau J, Herbet G, Rigau V, Duffau H. Acute progres-
lored to the characteristics of the tumor (volume, location, sion of untreated incidental WHO Grade II glioma to glio-
velocity of expansion, and potential to achieve a total or blastoma in an asymptomatic patient. J Neurosurg. 2016;​
124(1):​141–145.
supratotal resection that would impact the natural course   4. McGirt MJ, Chaichana KL, Attenello FJ, et al. Extent of
of the disease); to patient demands (their understanding surgical resection is independently associated with survival
of the concept of a controlled chronic disease and mo- in patients with hemispheric infiltrating low-grade gliomas.
tivation for a prophylactic treatment); and to the socio- Neurosurgery. 2008;​63(4):​700–708.
environmental aspects.12   5. Schomas DA, Laack NNI, Rao RD, et al. Intracranial low-
Based on the oncological benefits, excellent neurologi- grade gliomas in adults:​30-year experience with long-term
cal outcomes, and acceptable cognitive risks observed af- follow-up at Mayo Clinic. Neuro Oncol. 2009;​11(4):​437–445.
ter so-called prophylactic awake surgery, our results sup-   6. Capelle L, Fontaine D, Mandonnet E, et al. Spontaneous and
port the implementation of an MRI screening policy in the therapeutic prognostic factors in adult hemispheric World
Health Organization Grade II gliomas:​a series of 1097 cases:​
healthy population, as has already been suggested by some clinical article. J Neurosurg. 2013;​118(6):​1157–1168.
authors.2,45 Therefore, a cost-effectiveness analysis and   7. Jakola AS, Myrmel KS, Kloster R, et al. Comparison of a
identification of the optimal subpopulation to be screened strategy favoring early surgical resection vs a strategy favor-
are now urgently needed. ing watchful waiting in low-grade gliomas. JAMA. 2012;​
Our study did have some limitations. First, areas of 308(18):​1881–1888.
cognitive performance that were slightly decreased after   8. Cochereau J, Herbet G, Duffau H. Patients with incidental
surgery (mostly psychomotor speed and attention) may WHO grade II glioma frequently suffer from neuropsychologi-
be related to an incomplete recovery during the first 3 cal disturbances. Acta Neurochir (Wien). 2016;​158(2):​305–312.
  9. Ius T, Cesselli D, Isola M, et al. Incidental low-grade glio-
months following surgery. Later neuropsychological as- mas:​single-institution management based on clinical, surgi-
sessments were not systematically performed at a spe- cal, and molecular data. Neurosurgery. 2020;​86(3):​391–399.
cific endpoint. Second, all patients received AEDs during 10. Duffau H. Awake surgery for incidental WHO grade II glio-
the first 3 months following surgery, which may have sig- mas involving eloquent areas. Acta Neurochir (Wien). 2012;​
nificantly influenced mood and cognition performance. 154(4):​575–584.
Another limitation is that some neuropsychological tasks 11. Lima GL, Duffau H. Is there a risk of seizures in “preven-
were tailored to the location of the tumor. Consequent- tive” awake surgery for incidental diffuse low-grade gliomas?
ly, patients who were expected to suffer from a specific J Neurosurg. 2015;​122(6):​1397–1405.
disturbance (e.g., language abilities in right-handed pa- 12. Ng S, Herbet G, Moritz-Gasser S, Duffau H. Return to
work following surgery for incidental diffuse low-grade
tients with left-sided tumors) had no missing data for the glioma:​a prospective series with 74 patients. Neurosurgery.
related neuropsychological tasks, whereas others more 2020;87(4):720–729.
frequently had missing data. The introduction of this se- 13. Taphoorn MJ, Klein M. Cognitive deficits in adult patients
lection bias may have led to postsurgical results being with brain tumours. Lancet Neurol. 2004;​3(3):​159–168.
underrated. 14. Klein M. Health-related quality of life aspects in patients
with low-grade glioma. Adv Tech Stand Neurosurg. 2010;​35:​
213–235.
Conclusions 15. Klein M. Neurocognitive functioning in adult WHO grade II
In this consecutive series of 47 patients with inciden- gliomas:​impact of old and new treatment modalities. Neuro
tally discovered LGG who underwent early surgery under Oncol. 2012;​14(suppl 4):​iv17–iv24.
the awake condition, 87.2% had stable or improved neu- 16. Racine CA, Li J, Molinaro AM, et al. Neurocognitive func-
tion in newly diagnosed low-grade glioma patients undergo-
rocognitive outcomes at 3 months after surgery. Among ing surgical resection with awake mapping techniques. Neu-
them, 7 patients (14.9%) showed an early improvement of rosurgery. 2015;​77(3):​371–379.
their cognitive abilities in comparison to their presurgi- 17. Jalali R, Dutta D. Factors influencing quality of life in adult
cal status. Six patients (12.8%) presented a mild cogni- patients with primary brain tumors. Neuro Oncol. 2012;​
tive decline, without consequences for the resumption of 14(suppl 4):​iv8–iv16.
their professional activities. In addition, supratotal or total 18. Metz-Lutz MN, Kremin H, Deloche G, et al. Standardisation
resections were achieved in 59.6% of patients and all pa- d’un test de dénomination orale:​contrôle des effets de l’âge,
tients were still alive at the end of the follow-up. In light du sexe et du niveau de scolarité chez les sujets adultes nor-
maux. Rev Neuropsychol. 1991;​1(1):​73–95.
of these favorable oncological outcomes resulting from 19. Duffau H, Gatignol P, Mandonnet E, et al. New insights into
an early prophylactic surgical management, these novel the anatomo-functional connectivity of the semantic system:​
neurocognitive results may help neuro-oncologists to bet- a study using cortico-subcortical electrostimulations. Brain.
ter estimate the onco-functional balance in patients with 2005;​128(Pt 4):​797–810.
newly diagnosed ILGG, in order to devise an optimal and 20. Duffau H. Stimulation mapping of white matter tracts to
personalized therapeutic strategy. study brain functional connectivity. Nat Rev Neurol. 2015;​
11(5):​255–265.
21. Howard D, Patterson K. The Pyramids and Palm Trees Test:​
References A Test of Semantic Access From Words and Pictures. Pear-
  1. Pallud J, Fontaine D, Duffau H, et al. Natural history of in- son Assessment;​1992.
cidental World Health Organization grade II gliomas. Ann 22. Doppelt JE, Wallace WL. Standardization of the Wechsler
Neurol. 2010;​68(5):​727–733. Adult Intelligence Scale for older persons. J Abnorm Psychol.
  2. Mandonnet E, de Witt Hamer P, Pallud J, et al. Silent diffuse 1955;​51(2):​312–330.
low-grade glioma:​toward screening and preventive treat- 23. Stroop JR. Studies of interference in serial verbal reactions. J
ment? Cancer. 2014;​120(12):​1758–1762. Exp Psychol. 1935;​18(6):​643–662.

J Neurosurg Volume 135 • September 2021 879

 
*44 *
Ng et al.

24. Tombaugh TN. Trail Making Test A and B:​normative data 41. Pallud J, Mandonnet E, Duffau H, et al. Prognostic value of
stratified by age and education. Arch Clin Neuropsychol. initial magnetic resonance imaging growth rates for World
2004;​19(2):​203–214. Health Organization grade II gliomas. Ann Neurol. 2006;​
25. Buschke H. Cued recall in amnesia. J Clin Neuropsychol. 60(3):​380–383.
1984;​6(4):​433–440. 42. Cesselli D, Ius T, Isola M, et al. Application of an artificial
26. Tremblay M-P, Potvin O, Callahan BL, et al. Normative data intelligence algorithm to prognostically stratify grade II glio-
for the Rey-Osterrieth and the Taylor complex figure tests in mas. Cancers (Basel). 2019;​12(1):​50.
Quebec-French people. Arch Clin Neuropsychol. 2015;​30(1):​ 43. Duffau H, Taillandier L. New concepts in the management of
78–87. diffuse low-grade glioma:​proposal of a multistage and in-
27. Gauthier L, Dehaut F, Joanette Y. The Bells test:​a quantita- dividualized therapeutic approach. Neuro Oncol. 2015;​17(3):​
tive and qualitative test for visual neglect. Int J Clin Neuro- 332–342.
psychol. 1989;​11(2):​49–54. 44. Duffau H. Paradoxes of evidence-based medicine in lower-
28. Jewell G, McCourt ME. Pseudoneglect:​a review and meta- grade glioma:​to treat the tumor or the patient? Neurology.
analysis of performance factors in line bisection tasks. Neu- 2018;​91(14):​657–662.
ropsychologia. 2000;​38(1):​93–110. 45. Kelly PJ. Gliomas:​Survival, origin and early detection. Surg
29. Mandonnet E, Delattre J-Y, Tanguy M-L, et al. Continuous Neurol Int. 2010;​1(1):​96.
growth of mean tumor diameter in a subset of grade II glio-
mas. Ann Neurol. 2003;​53(4):​524–528.
30. Mandonnet E, Pallud J, Clatz O, et al. Computational model- Disclosures
ing of the WHO grade II glioma dynamics:​principles and The authors report no conflict of interest concerning the materi-
applications to management paradigm. Neurosurg Rev. 2008;​ als or methods used in this study or the findings specified in this
31(3):​263–269. paper.
31. Berger MS, Deliganis AV, Dobbins J, Keles GE. The effect of
extent of resection on recurrence in patients with low grade Author Contributions
cerebral hemisphere gliomas. Cancer. 1994;​74(6):​1784–1791.
32. Yordanova YN, Moritz-Gasser S, Duffau H. Awake surgery Conception and design: Duffau. Acquisition of data: Herbet,
for WHO Grade II gliomas within “noneloquent” areas in the Lemaitre, Moritz-Gasser, Duffau. Analysis and interpretation
left dominant hemisphere:​toward a “supratotal” resection. of data: all authors. Drafting the article: Ng. Critically revising
Clinical article. J Neurosurg. 2011;​115(2):​232–239. the article: Herbet, Lemaitre, Cochereau, Moritz-Gasser, Duffau.
33. Rossi M, Ambrogi F, Gay L, et al. Is supratotal resection Reviewed submitted version of manuscript: Herbet, Lemaitre,
achievable in low-grade gliomas? Feasibility, putative factors, Cochereau, Moritz-Gasser, Duffau. Approved the final version of
safety, and functional outcome. J Neurosurg. 2020;​132(6):​ the manuscript on behalf of all authors: Ng. Statistical analysis:
1692–1705. Ng, Herbet. Study supervision: Duffau.
34. Barzilai O, Ben Moshe S, Sitt R, et al. Improvement in cogni-
tive function after surgery for low-grade glioma. J Neuro- Supplemental Information
surg. 2019;​130(2):​426–434. Online-Only Content
35. van Kessel E, Snijders TJ, Baumfalk AE, et al. Neurocogni- Supplemental material is available with the online version of the
tive changes after awake surgery in glioma patients:​a retro- article.
spective cohort study. J Neurooncol. 2020;​146(1):​97–109. Supplemental Table 1. https://thejns.org/doi/suppl/​10.3171/​
36. McAleer MF, Brown PD. Neurocognitive function following 2020.7.JNS201507.
therapy for low-grade gliomas. Semin Radiat Oncol. 2015;​
25(3):​210–218. Correspondence
37. Herbet G, Maheu M, Costi E, et al. Mapping neuroplastic
Sam Ng: Gui de Chauliac Hospital, Montpellier University
potential in brain-damaged patients. Brain. 2016;​139(Pt 3):​
Medical Center, Montpellier, France. [email protected].
829–844.
38. Szalisznyo K, Silverstein DN, Duffau H, Smits A. Pathologi-
cal neural attractor dynamics in slowly growing gliomas sup-
ports an optimal time frame for white matter plasticity. PLoS
One. 2013;​8(7):​e69798.
39. Duffau H. Long-term outcomes after supratotal resection of
diffuse low-grade gliomas:​a consecutive series with 11-year
follow-up. Acta Neurochir (Wien). 2016;​158(1):​51–58.
40. Jakola AS, Skjulsvik AJ, Myrmel KS, et al. Surgical resection
versus watchful waiting in low-grade gliomas. Ann Oncol.
2017;​28(8):​1942–1948.

880 J Neurosurg Volume 135 • September 2021

 
*44 *
 LITERATURE REVIEW
J Neurosurg 135:881–889, 2021

Resection of vestibular schwannomas after stereotactic

radiosurgery: a systematic review
Max Whitmeyer, BS,1 Bledi C. Brahimaj, MD,2 André Beer-Furlan, MD, PhD,2 Sameer Alvi, MD,3
Madeline J. Epsten, BA,4 Fred Crawford, BS,4 Richard W. Byrne, MD,2 and R. Mark Wiet, MD2,5
1
Ohio State University College of Medicine, Columbus, Ohio; 2Department of Neurological Surgery, Rush University Medical
Center, Chicago; 3Northwestern Medicine, Geneva; 4Rush Medical College, Chicago; and 5Department of Otolaryngology, Rush
University Medical Center, Chicago, Illinois

OBJECTIVE  Multiple short series have evaluated the efficacy of salvage microsurgery (MS) after stereotactic radio-
surgery (SRS) for treatment of vestibular schwannomas (VSs); however, there is a lack of a large volume of patient data
available for interpretation and clinical adaptation. The goal of this study was to provide a comprehensive review of tumor
characteristics, management, and surgical outcomes of salvage of MS after SRS for VS.
METHODS  The Medline/PubMed, Scopus, CINAHL, Cochrane Library, and Google Scholar databases were queried
according to PRISMA guidelines. All English-language and translated publications were included. Studies lacking ad-
equate study characteristics and outcomes were excluded. Cases involving neurofibromatosis type 2, previous MS, or
malignant transformation were excluded when possible.
RESULTS  Twenty studies containing 297 cases met inclusion criteria. Three additional cases from Rush University
Medical Center were added for 300 total cases. Tumor growth with or without symptoms was the primary indication for
salvage surgery (92.3% of cases), followed by worsening of symptoms without growth (4.6%) and cystic enlargement
(3.1%). The average time to MS after SRS was 39.4 months. The average size and volume of tumor at surgery were 2.44
cm and 5.92 cm3, respectively. The surgical approach was retrosigmoid (42.8%) and translabyrinthine (57.2%); 59.5%
of patients had a House-Brackmann (HB) grade of I or II. The facial nerve was preserved in 91.5% of cases. Facial
nerve preservation and HB grades were lower for the translabyrinthine versus retrosigmoid approach (p = 0.31 and p =
0.18, respectively); however, fewer complications were noted in the translabyrinthine approach (p = 0.29). Gross-total
resection (GTR) was completed in 55.7% of surgeries. Studies that predominantly used subtotal resection (STR) were
associated with a lower rate of facial nerve injury (5.3% vs 11.3%, p = 0.07) and higher rate of HB grade I or II (72.9% vs
48.0%, p = 0.00003) versus those using predominantly GTR. However, majority STR was associated with a recurrence
rate of 3.6% as compared to 1.4% for majority GTR (p = 0.29).
CONCLUSIONS  This study showed that the leading cause of MS after SRS was tumor growth at an average of 39.4
months after radiation. There were no significant differences in outcomes of facial nerve preservation, postoperative HB
grade, or complication rate based on surgical approach. Patients who underwent STR showed statistically significant
better HB outcomes compared with GTR. MS after SRS was considered by most authors to be more difficult than prima-
ry MS. These data support the notion that the surgical goals of salvage surgery are debulking of tumor mass, decreasing
compression of the brainstem, and not necessarily pursuing GTR.
https://thejns.org/doi/abs/10.3171/2020.7.JNS2044
KEYWORDS  acoustic neuroma; microsurgery; salvage surgery; vestibular schwannoma; oncology; stereotactic

V
radiosurgery

estibular schwannomas (VSs) are typically be- radic; however, bilateral tumors are seen in association
nign tumors arising most commonly from the with neurofibromatosis type 2 (NF2).7
nerve sheath of the superior division of cranial Treatment options for VS may be divided into three
nerve VIII. The incidence, as determined by modern pop- main groups: conservative management, microsurgery
ulation-based studies, is between 1.1 and 4.2 per 100,000 (MS), and stereotactic radiosurgery (SRS). Historically,
people.1–6 The majority of tumors are unilateral and spo- MS was the predominant treatment of choice. Neverthe-

ABBREVIATIONS  CK = CyberKnife; fSRT = fractionated stereotactic radiotherapy; GKS = Gamma Knife surgery; GTR = gross-total resection; HB = House-Brackmann;
LINAC = linear accelerator; MS = microsurgery; NF2 = neurofibromatosis type 2; SRS = stereotactic radiosurgery; STR = subtotal resection; VS = vestibular schwannoma.
SUBMITTED  January 5, 2020.  ACCEPTED  July 7, 2020.
INCLUDE WHEN CITING  Published online November 27, 2020; DOI: 10.3171/2020.7.JNS2044.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  881

*45 *
Whitmeyer et al.

less, the incremental experience and long-term follow-up were included in the review if the sum of patients that fell
with fractionated stereotactic radiotherapy (fSRT) and into these categories did not make up more than 50% of
SRS led to their increased use as an adjunct to surgery the patient population.
or as the primary treatment modality for VS in the past
decades. There is a growing body of literature on SRS as Data Extraction
the primary treatment of VS. Tumor control rates are gen- Data was extracted by the first author and verified by
erally good for fSRT and SRS, ranging from 81% to 100% the second or third authors. Variables included number of
at 5 years.8–10 Along with the increased use of SRS as a participants, average age, sex, laterality, previous surgery,
primary treatment for VS, broader indications, imperfect NF2, malignancy, follow-up duration, SRS method, SRS
control, and an aging patient population have led to an in- dose, indications for surgery, time between SRS and MS,
crease in the number of patients requiring MS after failed tumor size/volume, MS approach, extent of resection, tu-
primary SRS.11 mor control, facial nerve preservation, preoperative HB
Despite the robust literature on primary SRS for VS, grade, postoperative HB grade, postoperative hearing,
there is a relative paucity of studies addressing patients difficulty with dissection, complications, and additional
who experience treatment failure. The goal of this sys- notes.
tematic review is to synthesize the outcomes of previously
published literature and add our experience on patients
with VS requiring salvage surgery after failed SRS. Data Processing
Potential bias in the studies was not assessed when
compiling and processing variables. All variables col-
Methods lected were summed and averaged based on the number of
The review protocol was conducted in accordance patients in each study. In the comparison between surgi-
with PRISMA guidelines. Five online databases (Med- cal approaches, a 2-sided t-test was used for significance.
line/PubMed, Scopus, CINAHL, Cochrane Library and In the comparison between extent of resection, a 1-sided
Google Scholar) were systematically searched by two in- t-test was used. Analysis was overseen by a statistician.
dependent reviewers using filters for English-language ar-
ticles, publication between 1990 and 2019, and the follow-
ing search string: (((“Neuroma, Acoustic”[Mesh] or (ves- Results
tibular and (schwannoma or schwannomas)) or “acoustic The initial query of databases yielded 1738 articles (Fig.
neuroma”)) AND (“Radiosurgery”[Mesh] or radiosurgery 1). The title and abstract of the 1131 results were screened
OR radiosurgical or SRS or “cyberknife” or “gamma by two reviewers for any type of surgery that occurred fol-
knife” or radiotherapy or X-knife or Novalis or “stereotac- lowing SRS treatment of VS. After screening, 236 articles
tic radiation treatment”)) AND (“Microsurgery”[Mesh] or were reviewed in full to determine eligibility based on the
microsurgery or surgery OR surgical or resection). inclusion and exclusion criteria. One hundred eighty-six
Titles and abstracts were screened by the first author for of these studies reported the tumor control rate following
surgery that occurred following irradiation of VSs. The SRS but provided no further details on the subsequent sal-
remaining articles were then reviewed by the first author vage surgery in the full text. A further 19 papers related
based on the criteria outlined below and reviewed by the to malignant transformations of VS following SRS were
second or third authors. Cases were also added from the excluded. Three studies that evaluated only MS following
authors’ institution and categorized as “Rush data” in the primary treatment of VS with a combination of surgery
review. and SRS were excluded. Five studies that met all other
criteria were excluded because all their patients were used
Inclusion Criteria in a later paper, which was included in the review. Three
Full articles in which at least 1 patient underwent pri- studies that passed the title and abstract review were not
mary fSRT or SRS for a VS that required subsequent sal- included because an English translation of the full study
vage MS were included. Studies needed to have at least 3 could not be obtained.
of the following variables in regard to tumor characteris- Ultimately, 20 studies met all inclusion and exclusion
tics: type of radiation treatment, radiation treatment dose, criteria for this review. Data compiled from the Rush Uni-
surgery indication, time interval between SRS and MS, versity Vestibular Schwannoma database were added to
and tumor size; and at least 3 of the following variables in the total results to bring the total number of studies to 21.
regard to surgical outcome: approach, resection success,
tumor control, facial nerve preservation, preoperative Patient and Radiation Characteristics
House-Brackmann (HB) grade, postoperative HB grade, The 21 studies reviewed included 300 patients who re-
and complications. quired MS following failed SRS for treatment of a VS.
The average age at the time of surgical salvage was 53.2
Exclusion Criteria years. Females made up 63.9% of the total population and
To aid in having a homogenous patient population, 60.7% of the tumors were right-sided (Table 1).
cases of malignant transformation of VS following SRS, Individual cases of prior MS, NF2, or malignant trans-
patients with NF2, or those with previous MS were ex- formation are included in the pooled data, as these cases
cluded from our analysis. In instances in which the data could not be parsed from the aggregate data of their re-
reported were pooled, the study and patient populations spective studies. Three of the 21 studies included at least 1

882 J Neurosurg Volume 135 • September 2021

 
*45 *
 Whitmeyer et al.

FIG. 1. Flow diagram of our systematic review according to PRISMA guidelines. Figure is available in color online only.

case from one of these categories. Patients falling into the surgeries, respectively. According to the authors, 68.5% of
categories of prior MS, NF2, or malignant transformation salvage surgeries were described as more difficult than a
represent just 5.0%, 6.9%, and 0.7%, respectively, of the primary MS for VS (Table 2).
total population analyzed.
The average tumor size at the time of surgery was 2.44 Surgical Outcomes and Complications
cm and the average volume was 5.92 cm3 (Table 1). Gam- Facial nerve anatomical preservation was 91.5% in
ma Knife surgery (GKS; Elekta) was the most frequent studies that reported this outcome. The average postoper-
type of SRS (n = 218) that required surgical salvage in the ative HB grade among all studies was 2.58, with 59.5% of
studies listed, followed by linear accelerator (LINAC; n = patients having an HB grade of I or II (Table 2).
30), fSRT (n = 23), and CyberKnife (CK; Accuray; n = 10). The facial nerve preservation rates and postoperative
The weighted average doses of the SRS treatments were HB grades were broken down by surgical approach—ret-
12.7 Gy (GK), 27.1 Gy (LINAC), 31.5 Gy (fSRT), and 21.5 rosigmoid or translabyrinthine—in studies that provided
Gy (CK). the data separately. For retrosigmoid approach surgeries,
the facial nerve preservation rate was 89.7% and 61.0% of
Surgical Indication and Management patients had an HB grade of I or II. For translabyrinthine
Tumor growth, with or without symptoms, comprised approaches, the facial nerve preservation rate was 87.8%
92.3% of the indications for surgery following failed and 51.0% of patients had an HB grade of I or II. Nine
SRS; 4.6% of surgeries occurred following a worsening complications were noted in the 68 retrosigmoid cases,
of symptoms (i.e., neurological deterioration, unbearable while there were 8 complications in the 98 translabyrin-
facial pain, or trigeminal neuralgia) after SRS despite no thine cases.
tumor growth. Additionally, 3.1% of surgeries occurred There were 10 reported CSF leaks as a result of salvage
due to cystic enlargement of the tumor (Table 1). MS, for a total rate of 3.3%. Other surgical complications
The surgical approaches used for salvage MS of VS included 3 patients with meningitis, 3 pseudomeningo-
after failed SRS were retrosigmoid approaches and trans- celes, 2 cerebral strokes, 2 incidences of cerebellar edema,
labyrinthine approaches, representing 42.8% and 57.2% of 2 cases of syndrome of inappropriate antidiuretic hormone

J Neurosurg Volume 135 • September 2021 883

 
*45 *
Whitmeyer et al.

TABLE 1. Included studies, demographics, tumor characteristics, and indications for surgery
Mean Mean Mean
Tumor Tumor Duration Btwn No. w/ No. w/ No. w/
No. of Age Size Volume SRS and MS Tumor Growth & Symptoms
Authors & Year Pts (yrs) Females (%) Rt-Sided (cm) (cm3) (mos) Growth Symptoms Alone
Aboukaïs et al., 2018 12
11 50.2 7 (63.6) NA NA 1.84 49.7 11 0 0
Battista & Wiet, 200013 5 NA NA NA 1.9 NA NA NA NA NA
Breshears et al., 201714 10 54 7 (70) NA 2.2 NA 36 3 7 0
Couldwell & Mohan, 200215 1 48 1 (100) 1 3.4 NA 3 0 1 0
Friedman et al., 201116 73 48.4 46 (63) NA 2.5 NA 43.2 “Most” 0 “Some”
Gerganov et al., 201217 15 51 9 (60) NA 3.2 NA 30.7 NA NA NA
Hong et al., 201418 4 NA NA NA NA 16.25 10.5 4 0 0
Husseini et al., 201319 13 55.4 8 (61.5) 7 3.24 NA 34.8 0 13 0
Iwai et al., 201620 18 54 11 (61.1) NA 2.12 NA 26 16 0 0
Lee et al., 201421 5 44 10 (200) NA NA 5.56 40.4 2 0 2
Lee et al., 200322 4 58 4 (100) 2 1.9 NA 18 0 4 0
Lee et al., 201723 6 40.5 6 (100) NA NA 12.1 52.7 6 0 0
Limb et al., 200524 9 57 8 (88.9) NA 2.6 NA NA NA NA NA
Nonaka et al., 201611 39 51.8 24 (61.5) NA 2.5 NA 50.4 36 0 3
Pollock et al., 199825 13 65 NA NA NA 5.6 27 10 0 3
Roche et al., 200826 23 NA NA NA NA 4.67 39 10 13 0
Schulder et al., 199927 1 33 1 (100) 1 1.7 NA 26 0 1 0
Shuto & Matsunaga, 201628 2 63.5 1 (50) 0 NA 2.57 96 0 0 0
Shuto et al., 200829 8 57.3 4 (50) NA NA 6.9 48.2 5 0 0
Wise et al., 201630 37 61 19 (51.4) NA 2 NA 36 37 0 1
Rush data 3 58.2 2 (66.7) 2 2 NA 17.7 0 2 0
Total/average 300 53.2 168 (63.9) 37 (60.7%) 2.44 5.92 39.4 140 (71.4%) 41 (20.9%) 9 (4.6%)
NA = not available; Pts = patients.

secretions, 2 transient vocal cord paralyses, 1 brainstem least 4.67 cm3, with the notable exception of Aboukaïs et
stroke, 1 hemiparesis, 1 hematoma, 1 venous thrombosis, al., who reported an average tumor volume of just 1.84 cm3
1 dehiscence, 1 wound infection, 1 persistent headache, 1 at the time of surgery.12 Because tumor growth was the ma-
pneumocephalus, and 1 epileptic seizure. jor indicator for surgical salvage, it is unsurprising that the
Gross-total resection (GTR) was achieved in 55.7% average tumor in most studies was moderately large.
of patients, compared with 44.3% of subtotal resection Regarding SRS treatment, GKS was the most frequent
(STR; n = 291 total GTR or STR cases). Studies in which SRS modality that required surgical salvage in the stud-
a majority of patients underwent STR had a lower rate of ies listed. This is likely due to GKS historically being the
facial nerve severance than those where the majority of most frequently used modality for SRS for primary VS.9
patients underwent GTR: 5.3% versus 11.3%, respectively No study mentioned any differences in surgical techniques
(p = 0.07). Postoperative HB grade I/II was 72.9% for or outcomes based on SRS treatment type. Radiation doses
STR versus 48.0% for GTR (p = 0.00003). While more for treatment of VS have been continually decreasing in re-
studies are needed directly comparing STR and GTR, cent years; the impact this will have on salvage surgery fol-
preferably with institutional and patient controls, these lowing radiation is unknown. Adverse radiation effects may
data show that among published studies, using primarily decrease, but tumor control rates from primary SRS may
STR led to better facial nerve outcomes. fall with decreasing doses, leading to a greater incidence
of recurrent VS. Schumacher et al. report that despite SRS
Discussion doses decreasing from a median of 16 Gy in older studies to
Patient and Radiation Characteristics 12.5 Gy in newer studies, there are similar reported rates of
The average tumor size at the time of surgery was 2.44 tumor control.31 Within our study, we were unable to inves-
cm and the average volume was 5.92 cm3. This average tigate the superiority of tumor control for individual types
tumor size would classify as a grade 3 House and Koos of SRS, consistent with the published literature.
score, indicating a moderately large tumor. All studies with
a moderate or large sample size (> 5 patients) had an aver- Surgical Indication and Management
age tumor size of at least 2 cm or average volume of at The International Stereotactic Radiosurgery Society pub-

884 J Neurosurg Volume 135 • September 2021

 
*45 *
 Whitmeyer et al.

TABLE 2. Surgical approaches, resection success, and outcomes

Authors No. CN VII HB Grade Difficulty in
& Year of Pts RS TL GTR STR Severed (%) I/II (%) Dissection
Aboukaïs et al., 201812 11 0 11 8 3 2 (18.2) 1 (9) 6
Battista & Wiet, 200013 5 NA NA NA NA NA NA NA
Breshears et al., 201714 10 4 6 7 3 0 (0) 8 (74.1) NA
Couldwell & Mohan, 200215 1 NA NA 1 0 0 (0) 0 (0) NA
Friedman et al., 201116 73 0 73 58 15 10 (13.7) 36 (57.5) NA
Gerganov et al., 201217 15 15 0 15 0 2 (13.3) 8 (54.9) NA
Hong et al., 201418 4 4 0 3 1 0 (0) 3 (63.2) NA
Husseini et al., 201319 13 0 12 11 1 1 (7.7) NA 12
Iwai et al., 201620 18 18 0 0 18 2 (11.1) 15 (79.6) NA
Lee et al., 201421 5 5 0 0 4 0 (0) 4 (69) NA
Lee et al., 200322 4 4 0 2 2 0 (0) 3 (63.2) 4
Lee et al., 201723 6 0 5 0 6 0 (0) 2 (31.6) 6
Limb et al., 200524 9 8 1 3 6 NA 4 (42.4) 9
Nonaka et al., 201611 39 NA NA 13 26 0 (0) 32 (84.5) 27
Pollock et al., 199825 13 11 1 7 6 NA 3 (22.7) 8
Roche et al., 200826 23 5 18 15 8 0 (0) 12 (51) 10
Schulder et al., 199927 1 0 1 1 0 0 (0) 0 (0) 1
Shuto & Matsunaga, 201628 2 NA NA NA NA 0 (0) NA NA
Shuto et al., 200829 8 8 0 0 8 2 (25) 5 (51.9) 4
Wise et al., 201630 37 23 14 18 19 2 (5.4) 27 (71.6) 0
Rush data 3 2 1 0 3 1 (33.3) 0 (0) 2
Total/average 300 107 (42.8%) 143 (57.2%) 162 (55.7%) 129 (44.3%) 25 (8.5) 166 (59.5) 89 (68.5%)
CN = cranial nerve; RS = retrosigmoid approach; TL = translabyrinthine approach.

lished consensus practice guidelines on the management of sion from tumor expansion.15 Barring exceptions as previ-
VS with respect to SRS. While this guides physicians in pri- ously mentioned, our institutional preference is to avoid
mary diagnosis of VS, there is less literature regarding the surgical intervention during the period of transient expan-
management of VS after failed SRS. Continued growth of sion, with judicious use of steroids as necessary.
the VS is the leading indication for the requirement of MS Most recurrent VSs grow at a steady rate following SRS;
(> 90% of salvage surgeries) in our review. The other indica- however, Nonaka et al. reported that 10% of the tumors they
tions for salvage surgery listed by authors were worsening treated with salvage MS were quiescent for many years pri-
of symptoms (i.e., neurological deterioration, unbearable fa- or to rapid growth.11 Iwai et al. speculate that this may be
cial pain, and trigeminal neuralgia) and cystic enlargement.8 due to malignant transformation from radiation injury.20 An
Transient expansion occurs in the majority of VSs treat- additional complication that can arise is cystic expansion.
ed with SRS. Nagano et al. found that 74% of VSs treated Shuto and Matsunaga present 2 cases of symptomatic cyst
with GKS had a volume increase of at least 10%, with formation occurring 4 and 12 years following GKS treat-
the average tumor expanding 47% at its peak. The mean ment for noncystic VS.28 They speculate that this may be
time to shrink back to the original size prior to SRS was due to repeated minor hemorrhages or the extravasation of
12 months.32 Due to these transient expansions, adverse fluid from the lesion. Thus, many authors call for SRS stud-
radiation effects occur most commonly between 6 and 18 ies with longer follow-up times to better determine the rate
months and self-resolve in more than half of cases within of delayed radiation injury or tumor control failure.
3–6 months. Furthermore, the cranial nerves and brainstem It must be stated that publications have advocated for the
are most susceptible to surgical injury during this time.21 use of repeat SRS in instances of persistent growth of VSs
Many authors of the included studies advocate for avoiding after initial SRS. Kano et al. published a study on a small
surgery during these periods. Nonaka et al. advocate for a series of 6 patients who received repeat SRS to the same tu-
3-year minimum waiting period between SRS and surgical mor for progressive growth at a median of 63 months after
salvage, barring the development of severe symptoms.11 initial treatment. In all of these patients, tumor control was
Most studies in our review avoided surgery during the obtained at a median 29 months of follow-up.33 Thus, while
transient period as the average duration between SRS and our review does not address this method of treatment, it is
MS was 39.4 months among all studies. However, there a viable option to be considered for patients who may be
are quite a few exceptions to this general recommendation considered at high risk for surgical intervention. Our insti-
in the literature, often due to severe symptomatic progres- tutional preference in patients with persistent growth after

J Neurosurg Volume 135 • September 2021 885

 
*45 *
Whitmeyer et al.

SRS would be surgical intervention, as it provides critical to SRS treatment may carry even greater growth potential
debulking of mass effect and, in our experience, superior than a residual tumor left from primary MS, and therefore
tumor control and clinical outcome. recurrence rates after primary STR MS may underesti-
Retrosigmoid and translabyrinthine approaches were mate the rate for these tumors. Husseini et al. similarly
utilized nearly equally across all studies (42.8% vs 57.2%, recommend attempting GTR in all cases, except cases of
respectively; n = 250). The translabyrinthine approach is elderly patients or those with cystic tumor when the cyst
often cited as providing the best exposure of the cerebello- wall is in contact with vital structures.19
pontine angle for VS removal. The surgery does not spare Wise et al. evaluated the potential of GTR on a case-
hearing, but the vast majority of patients requiring salvage by-case basis. They aimed to achieve GTR in all cases, but
surgery following failed SRS have little to no hearing. were quick to change to STR based on both intraoperative
Limb et al. preferred using a retrosigmoid approach as it impression and electroprognostic testing. The GTR rate
allowed them to have early exposure to the cranial nerves from their study was 49%.30
within the internal auditory canal.24 Our analysis included a breakdown of studies by GTR
In our analysis comparing studies that used only a retro- versus STR, when reported (n = 291). Studies in which
sigmoid or translabyrinthine approach (n = 166), the retro- the results between GTR and STR could not be separat-
sigmoid approach led to higher rates of facial nerve anatom- ed were placed into whichever category a majority of the
ical preservation (89.7% vs 87.8% for translabyrinthine, p = cases were. The biggest difference between the GTR and
0.31) and lower postoperative HB scores (61.0% vs 50.6% STR populations was the difference in the proportion of
HB grade I/II for translabyrinthine, p = 0.18). However, there patients who had postoperative HB grade I/II: 72.9% for
were lower complication rates listed for surgeries performed STR versus 48.0% for GTR (p < 0.05). While more studies
via the translabyrinthine approach: 8.2% for translabyrin- are needed directly comparing STR to GTR, preferably
thine and 13.2% for retrosigmoid (p = 0.29). Overall, there with institutional and patient controls, these data show that
is not a strong indication for using one approach over the among published studies, using primarily STR led to bet-
other in salvage MS for VSs treated with primary SRS. We ter facial nerve outcomes.
advocate for the proper surgery for the proper patient in this Additional intervention following salvage surgery, be it
scenario, prioritizing surgeon comfort with an approach. radiation or further surgery, was low in both groups. The
Difficulty in dissection is a subjective variable, and there tumor control rate was 98.6% for the GTR group versus
appear to be differences in how different authors classify 96.3% for STR. The overall occurrence of tumor regrowth
dissections as difficult. For the sake of comparison, tumors was low in each population (5 total recurrences), and this
were counted as being difficult dissections if the authors was not statistically significant (p = 0.29). Additionally,
mentioned the surgery or dissection as being more difficult there was a wide variety in reported follow-up times. At
than average. By these criteria, 68.5% of classified tumors this point, it appears that STR does lead to higher recur-
were rated as more difficult than normal. Tumors could rence rates, but the true rate and effect of this is unknown
be more difficult to resect for several reasons. The most because of the low rate of occurrence and variability in
frequently cited reasons were adhesion to cranial nerves follow-up. A summary of results based on surgical ap-
and lack of a facial plane.17,26 As a consequence of greater proach and extent of tumor resection is shown in Fig. 2.
adhesion between the tumor and the cranial nerves fol-
lowing SRS, STR is often performed out of necessity or Surgical Outcomes and Complications
by surgeon choice. GTR was completed on 55.7% of the The facial nerve anatomical preservation rate for MS
surgeries that reported extent of resection (n = 291), with following primary SRS was 91.5%. This is lower than the
the rest of the surgeries leaving at least some residual tu- preservation rate of 95% for primary MS.34,35 Just 59.5%
mor. For primary MS for VS, the GTR rate is much higher of patients in this review had an HB grade of I or II in the
at 81.6%.6 The decision of whether to aim for GTR or plan postoperative period. Across 30 studies of primary MS
for STR is a significant debate in the literature reviewed. for VS, the rate of postoperative HB grade I or II was be-
Friedman et al. achieved GTR in 78.9% of MSs fol- tween 55.8% and 98.6%.34 The lower rates of facial nerve
lowing SRS prior to 2005.16 After 2005, GTR in their se- preservation and HB grade I or II bolster the opinion of
ries decreased to 56.5% of cases following a switch to a many authors that surgery is more difficult in tumors that
more conservative approach in the hope of improving fa- have previously been irradiated.
cial nerve outcomes. The results of the change have been Among studies with at least 10 participants, an HB
promising early on, as 86% of patients since 2005 have grade of I or II was achieved in more than 70% of cases by
had HB grade I or II 1 year postoperatively. Furthermore, Breshears et al.,14 Iwai et al.,20 Nonaka et al.,11 and Wise et
among all patients who had a preoperative HB grade of I al.30 Breshears et al. reported their HB grades at 14 months
or II, the rate of postoperative HB grade I/II was 42.9% postoperatively, whereas most other studies recorded post-
for GTR (n = 42) and 84.6% for STR (n = 13). Given these operative HB grades in the first week following surgery. It
results, they advocate for planned STR in these patients, is also worth noting that Iwai, Nonaka, and Wise all used
although they caution that there are still limited data on STR on more than half of their MSs, which may have aid-
recurrence rates for residual tumors. ed in better than average postoperative HB grades.
In contrast, Gerganov et al. continue to recommend A total of 34 complications were noted among the 300
GTR, as the follow-up periods of authors who advocate cases (11.3%). The most common complication was CSF
for STR is too short (up to 36 months) to make conclusions leak, which occurred in 3.3% of cases. The total complica-
about its efficacy.17 Additionally, a tumor that is resistant tion rate as well as the CSF leak rate are higher than would be
886 J Neurosurg Volume 135 • September 2021
 
*45 *
 Whitmeyer et al.

FIG. 2. Comparison of surgical approach and extent of resection in salvage surgery.

expected from primary VS MS. A large study across multi- Limitations

ple institutions of primary MS patients without NF2 showed There are several limitations to this study. The retro-
a total complication rate of 5.3%, with a CSF leak rate of spective nature of the work made it difficult to fully ana-
2.03%. The next leading complications noted in this primary lyze data in the studies included. Due to the pooled data re-
MS study were dysphagia, CSF leak, urinary tract infection, port of many studies, comparing cause and effect between
intracerebral hemorrhage, vocal cord paralysis, and subdural variables such as resection amount and tumor control rate
hematoma.36 The translabyrinthine approach was found to was challenging. With raw data from each of these studies,
be a contributing factor to an increased risk of CSF leak in stronger conclusions could be made about the efficacy of
primary VS MS, along with longer operative time.37 approach types and extents of resection.
It is our practice to pursue the maximal safe resection.
This typically translates to a near-total or subtotal resec-
tion, with the goal of anatomical preservation of the facial Conclusions
nerve, and close observation in the postoperative period Microsurgery following SRS for VSs has proved to
with continual assessment for further SRS for life. A rep- be more difficult than primary surgery. Our study shows
resentative case is shown in Fig. 3. lower rates of GTR, facial nerve preservation, and good

FIG. 3. Axial MR images obtained in a 60-year-old man with a history of a 2.2-cm right VS who underwent GKS therapy 4 years
prior to presentation (A). Two years following GKS, a decrease in tumor size (to 1.9 cm) was noted (B). Four years following GKS,
serial imaging showed the tumor size began to increase (2.6 cm; C) and the patient developed a severe sensorineural hearing loss
of the right ear. The patient underwent a right translabyrinthine approach for STR of the lesion with full anatomical preservation of
cranial nerve VII. Postoperatively, his facial nerve function was normal. MRI was obtained 6 months following surgery (D).

J Neurosurg Volume 135 • September 2021 887

 
*45 *
Whitmeyer et al.

outcomes (defined as HB grade I or II) in this population 16. Friedman RA, Berliner KI, Bassim M, et al. A paradigm shift
of patients. In the case of salvage surgery, we conclude in salvage surgery for radiated vestibular schwannoma. Otol
that a conservative approach to resection with an empha- ​
Neurotol. 2011;32(8):1322–1328. ​
17. Gerganov VM, Giordano M, Samii A, Samii M. Surgical
sis on brainstem decompression and cranial nerve pres- treatment of patients with vestibular schwannomas after failed
ervation over a GTR is preferred for an optimal clinical previous radiosurgery. J Neurosurg. 2012;116(4):713–720. ​ ​
outcome. 18. Hong B, Krauss JK, Bremer M, et al. Vestibular schwannoma
microsurgery for recurrent tumors after radiation therapy or
previous surgical resection. Otol Neurotol. 2014;35(1):171–181. ​ ​
Acknowledgments 19. Husseini ST, Piccirillo E, Taibah A, et al. Salvage surgery of
We thank Jennifer Westrick, Prasanna Vaduvathiriyan, and the vestibular schwannoma after failed radiotherapy:​the Gruppo
Library of Rush University Medical Center for helping to generate Otologico experience and review of the literature. Am J Oto-
and refine the search criteria. ​
laryngol. 2013;34(2):107–114. ​
20. Iwai Y, Ishibashi K, Nakanishi Y, et al. Functional outcomes of
salvage surgery for vestibular schwannomas after failed gam-
References ma knife radiosurgery. World Neurosurg. 2016;90:385–390. ​ ​
  1. Kleijwegt M, Ho V, Visser O, et al. Real incidence of vestibu- 21. Lee C-C, Wu H-M, Chung W-Y, et al. Microsurgery for
lar schwannoma? Estimations from a national registry. Otol vestibular schwannoma after Gamma Knife surgery:​chal-
Neurotol. 2016;​37(9):​1411–1417. lenges and treatment strategies. J Neurosurg. 2014;121(suppl): ​ ​
  2. Kshettry VR, Hsieh JK, Ostrom QT, et al. Incidence of ves- 150–159.
tibular schwannomas in the United States. J Neurooncol. 22. Lee DJ, Westra WH, Staecker H, et al. Clinical and his-
2015;​124(2):​223–228. topathologic features of recurrent vestibular schwannoma
  3. Babu R, Sharma R, Bagley JH, et al. Vestibular schwanno- (acoustic neuroma) after stereotactic radiosurgery. Otol Neu-
mas in the modern era:​epidemiology, treatment trends, and ​ ​
rotol. 2003;24(4):650–660.
disparities in management. J Neurosurg. 2013;​119(1):​121–130. 23. Lee H-J, Kim MJ, Koh SH, et al. Comparing outcomes fol-
  4. Stangerup S-E, Caye-Thomasen P. Epidemiology and natural lowing salvage microsurgery in vestibular schwannoma
history of vestibular schwannomas. Otolaryngol Clin North patients failing gamma-knife radiosurgery or microsurgery.
Am. 2012;​45(2):​257–268, vii. ​
Otol Neurotol. 2017;38(9):1339–1344. ​
  5. Marinelli JP, Lohse CM, Carlson ML. Incidence of vestibular 24. Limb CJ, Long DM, Niparko JK. Acoustic neuromas after
schwannoma over the past half-century:​a population-based failed radiation therapy:​challenges of surgical salvage. La-
study of Olmsted County, Minnesota. Otolaryngol Head ​
ryngoscope. 2005;115(1):93–98. ​
Neck Surg. 2018;​159(4):​717–723. 25. Pollock BE, Lunsford LD, Kondziolka D, et al. Vestibular
  6. Carlson ML, Habermann EB, Wagie AE, et al. The chang- schwannoma management. Part II. Failed radiosurgery and
ing landscape of vestibular schwannoma management in the the role of delayed microsurgery. J Neurosurg. 1998;89(6): ​ ​
United States—a shift toward conservatism. Otolaryngol 949–955.
Head Neck Surg. 2015;​153(3):​440–446. 26. Roche P-H, Khalil M, Thomassin J-M, et al. Surgical re-
  7. Mathieu D, Kondziolka D, Flickinger JC, et al. Stereotactic moval of vestibular schwannoma after failed Gamma Knife
radiosurgery for vestibular schwannomas in patients with radiosurgery. Prog Neurol Surg. 2008;21:152–157. ​ ​
neurofibromatosis type 2:​an analysis of tumor control, com- 27. Schulder M, Sreepada GS, Kwartler JA, Cho ES. Microsurgi-
plications, and hearing preservation rates. Neurosurgery. cal removal of a vestibular schwannoma after stereotactic
2007;​60(3):​460–470. radiosurgery:​surgical and pathologic findings. Am J Otol.
  8. Tsao MN, Sahgal A, Xu W, et al. Stereotactic radiosurgery ​ ​
1999;20(3):364–368.
for vestibular schwannoma:​International Stereotactic Ra- 28. Shuto T, Matsunaga S. Two cases of cystic enlargement of
diosurgery Society (ISRS) Practice Guideline. J Radiosurg vestibular schwannoma as a late complication following
SBRT. 2017;​5(1):​5 –24. gamma knife surgery. J Clin Neurosci. 2016;33:239–241. ​ ​
  9. Persson O, Bartek J Jr, Shalom NB, et al. Stereotactic radio- 29. Shuto T, Inomori S, Matsunaga S, Fujino H. Microsurgery for
surgery vs. fractionated radiotherapy for tumor control in vestibular schwannoma after gamma knife radiosurgery. Acta
vestibular schwannoma patients:​a systematic review. Acta ​
Neurochir (Wien). 2008;150(3):229–234. ​
Neurochir (Wien). 2017;​159(6):​1013–1021. 30. Wise SC, Carlson ML, Tveiten OV, et al. Surgical salvage of
10. Kondziolka D, Mousavi SH, Kano H, et al. The newly diag- recurrent vestibular schwannoma following prior stereotactic
nosed vestibular schwannoma:​radiosurgery, resection, or radiosurgery. Laryngoscope. 2016;126(11):2580–2586. ​ ​
observation? Neurosurg Focus. 2012;​33(3):​E8. 31. Schumacher AJ, Lall RR, Lall RR, et al. Low-dose Gamma
11. Nonaka Y, Fukushima T, Watanabe K, et al. Surgical man- Knife radiosurgery for vestibular schwannomas:​tumor con-
agement of vestibular schwannomas after failed radiation trol and cranial nerve function preservation after 11 Gy. J
treatment. Neurosurg Rev. 2016;​39(2):​303–312. Neurol Surg B Skull Base. 2017;78(1):2–10. ​ ​
12. Aboukaïs R, Bonne N-X, Touzet G, et al. Progression of ves- 32. Nagano O, Higuchi Y, Serizawa T, et al. Transient expansion
tibular schawnnoma after GammaKnife radiosurgery:​a chal- of vestibular schwannoma following stereotactic radiosur-
lenge for microsurgical resection. Clin Neurol Neurosurg. ​
gery. J Neurosurg. 2008;109(5):811–816. ​
2018;​168:​77–82. 33. Kano H, Kondziolka D, Niranjan A, et al. Repeat stereotactic
13. Battista RA, Wiet RJ. Stereotactic radiosurgery for acoustic radiosurgery for acoustic neuromas. Int J Radiat Oncol Biol
neuromas:​a survey of the American Neurotology Society. ​ ​
Phys. 2010;76(2):520–527.
Am J Otol. 2000;​21(3):​371–381. 34. Taha I, Hyvärinen A, Ranta A, et al. Facial nerve function
14. Breshears JD, Osorio JA, Cheung SW, et al. Surgery after and hearing after microsurgical removal of sporadic vestibu-
primary radiation treatment for sporadic vestibular schwan- lar schwannomas in a population-based cohort. Acta Neuro-
nomas:​case series. Oper Neurosurg (Hagerstown). 2017;​ ​
chir (Wien). 2020;162(1):43–54. ​
13(4):​441–447. 35. Hong W, Cheng H, Wang X, Feng C. Influencing factors
15. Couldwell WT, Mohan AL. Enlargement of a vestibular analysis of facial nerve function after the microsurgical re-
schwannoma after stereotactic radiotherapy. Acta Neurochir section of acoustic neuroma. J Korean Neurosurg Soc. 2017;​
(Wien). 2002;​144(12):​1319–1322. ​
60(2):165–173.

888 J Neurosurg Volume 135 • September 2021

 
*45 *
 Whitmeyer et al.

36. Hatch JL, Bauschard MJ, Nguyen SA, et al. National trends Reviewed submitted version of manuscript: Brahimaj, Beer-
in vestibular schwannoma surgery:​influence of patient char- Furlan, Epsten. Approved the final version of the manuscript on
acteristics on outcomes. Otolaryngol Head Neck Surg. 2018;​ behalf of all authors: Wiet. Statistical analysis: Whitmeyer. Study
159(1):​102–109. supervision: Wiet, Byrne.
37. Copeland WR, Mallory GW, Neff BA, et al. Are there modi-
fiable risk factors to prevent a cerebrospinal fluid leak fol- Supplemental Information
lowing vestibular schwannoma surgery? J Neurosurg. 2015;​ Previous Presentations
122(2):​312–316.
Parts of this study were presented as an abstract at the North
American Skull Base Society annual meeting, February 9, 2020,
Disclosures as “Surgical resection after radiosurgery for the management of
vestibular schwannomas: a systematic review.” The abstract was
The authors report no conflict of interest concerning the materi- published by the Journal of Neurological Surgery: Skull Base in
als or methods used in this study or the findings specified in this February 2020.
paper.
Correspondence
Author Contributions R. Mark Wiet: Rush University Medical Center, Chicago, IL.
Conception and design: Wiet, Whitmeyer, Brahimaj, Beer- [email protected].
Furlan, Alvi, Byrne. Acquisition of data: Whitmeyer, Brahimaj,
Alvi, Crawford. Analysis and interpretation of data: Whitmeyer,
Beer-Furlan. Drafting the article: Whitmeyer. Critically revis-
ing the article: Wiet, Whitmeyer, Brahimaj, Beer-Furlan, Epsten.

J Neurosurg Volume 135 • September 2021 889

 
*45 *
 J Neurosurg 135:890–892, 2021

EDITORIAL
The quintessential nerve injury model
Daniel Umansky, MD, and Rajiv Midha, MSc, MD
Department of Clinical Neurosciences, Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary, Alberta,
Canada

N erve surgeons and scientists have been attempting

to create the ideal nerve injury model for decades,
with a surge occurring after World War II.1–4 Sem-
inal works of Kline and colleagues attempting to emulate
netic engineering to decipher molecular and mechanistic
pathways that may underlie the development of new thera-
pies.
Mahan and colleagues have gathered significant ex-
both war and civilian injuries have investigated several perience in developing a stretch injury model in rodents
animal models, namely dogs, rhesus monkeys, baboons, that provides us with a better understanding of the bio-
and chimpanzees,3,4 with detailed comparative results of mechanical, histological, and possible outcomes. Their
different regeneration patterns and variability between excellent work18–20 uses a rapid-stretch technique in an at-
different animal species when investigating gross and his- tempt to more closely emulate the mechanism of clinical
tological characteristics, neuroma formation, and nerve nerve stretch injuries, thus providing a potentially better
conductivity. These findings supported very early astute rodent model for closed traction injuries in humans. In
observations by Ramon y Cajal.5 Liu and colleagues6 dem- their biomechanical results,18 sciatic rat nerves subjected
onstrated structural failure of freshly obtained cadaver to both slow- and rapid-stretch injuries showed thresh-
nerves when stretched beyond 10%–20%, and provided old- and directional loading-susceptible ruptures, while
one of the first stress–strain curves and histopathological nerve branch anatomy seemed the most important factor
consequences of stress-induced failure, creating a basis for contributing to tensile strength, thus having clinical im-
the appreciation of injury pathophysiology. This work was portance. Moreover, injuries appeared as discontinuous
later revisited by Sunderland and Bradley in human spinal processes that actually fall into different elastic, inelastic,
nerve roots.7 Subsequent work by Lundborg and Rydevik8 and rupture biomechanical patterns, a finding that further
demonstrated intraneural circulation and blood-nerve supports seminal observations by Sunderland and Brad-
barrier disruption at even an 8% stretch, stressing the im- ley.7 In their subsequent histological results paper,19 in
portance of microarchitecture’s contribution in nerve in- which they evaluate previous biomechanical damage pat-
jury models. The above classic papers used slow, graded terns, in-depth graded injury was studied, demonstrating
stretch, and indeed, Highet and Sanders9 speculated that undulation straightening of axons and graded endoneurial
the degree of nerve damage is dependent on the total ten- injury concurrent with stretch severity, whereas injury to
sion applied rather than on the rate at which the injury is both epineurium and perineurium was minimal and did
occurring. not predict transition from one grade to another. This find-
Although a significant understanding of both the mo- ing supported previous conclusions made by Sunderland
lecular processes of peripheral nerve regeneration as well that the epineurium provides little resistance to stretch in-
as injury mechanism and histopathological processes has jury even under rapid stretching conditions. In their transi-
occurred,10–14 a single gold standard, verified, and repeat- tion to a mouse model,20 inelastic and elastic injuries were
able injury model is still lacking. Furthermore, models correlated with both behavioral and histological assays,
have repeatedly used rodents for their primary investiga- further demonstrating neuroma formation (simultaneous
tive purposes for multiple reasons,15,16 including relatively regrowth of neural fibers and excessive fibrous scar tissue
low costs, even though they entail several limitations, such proliferation) in inelastic injuries. Finally, in their current
as small size, relatively short regeneration distance, spe- publication,21 Mahan and colleagues report their results
cies-specific neurobiological regenerative profile, faster of traction injuries to the sciatic nerve in mice, evaluat-
regenerating capacity, and bifascicular nerves in contrast ing their model in comparison with crush and transec-
to polyfascicular human nerves. Still, attempting to prove tion injuries (with and without repair), and perform motor
translation into larger nonhuman primates17 as well as and sensory behavioral studies in addition to histological
clinical treatment heavily relies on these smaller animal analysis. Their data demonstrate neuroma formation in the
models. Moreover, a well-developed and accepted rodent stretch injury group in comparison to other injury models.
nerve injury model would harness the powerful tool of ge- Because the current paper is a continuation of the au-

ACCOMPANYING ARTICLE  See pp 893–903. DOI: 10.3171/2020.5.JNS193448.

890 J Neurosurg Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law
 
*46 *
 Editorial

TABLE 1. Nerve injury model: species comparison

Animal Supported Lit & Translational Genetic & Mechanistic Critical Nerve No. of Targeted Nerve Ethical
Species Available Publications Factor* Studies Gap (cm)† Sites Investigated Approval Price
Mouse Abundant Low Very high ~0.5–0.7 High Easy Low
Rat Abundant Low High ~1.3–1.5 High Easy Low
Rabbit Abundant Low Moderate ~3 Moderate Easy Moderate
Cat Moderate Intermediate Low ~3–4 Low Moderate High
Dog Moderate Intermediate Low ~3–4 Low Moderate High
Pig Moderate Intermediate Low to moderate‡ ~4 Low Moderate High
Sheep Sparse Intermediate Low ~4 Low Difficult High
NHP Sparse High Low ~4 Extremely low Very difficult Extremely high
Lit = literature; NHP = nonhuman primates.
* Clinical resemblance through verified studies and assessment methodology.
† Critical nerve gap is defined as a nerve gap over which no recovery will occur without some form of nerve grafting or bridging when repair investigations are con-
ducted.
‡ “Humanized” pigs.

thors’ line of investigation, more emphasis should have References

been given to their previous results, to provide the reader   1. Weiss P. The technology of nerve regeneration:​a review.
with some continuity of the work attempted, especially Sutureless tubulation and related methods of nerve repair. J
given that the initial work was done in rat models and Neurosurg. 1944;​1(6):​400–450.
the later work was done in transgenic mice. Although the   2. Woodhall B. Modern history of peripheral nerve surgery;​
authors have briefly noted that they have not encountered World War II and the postwar study of peripheral nerve
regeneration. J Am Med Assoc. 1949;​139(9):​564–566.
significant differences between the models used in terms   3. Kline DG, Hayes GJ, Morse AS. A comparative study of
of elasticity or rupture threshold,20 except inelastic maxi- response of species to peripheral-nerve injury. J Neurosurg.
mum strain injury—believed to represent injury technique 1964;​21(11):​968–979.
differences and not intermodel differences—a more de-   4. Kline DG, Hayes GJ, Morse AS. A comparative study of
tailed account for using the later models would have been response of species to peripheral-nerve injury. II. Crush
appreciated. Furthermore, the authors could have under- and severance with primary suture. J Neurosurg. 1964;​21:​
lined the key differences in the two models, and why they 980–988.
have chosen to extend the work to mice, including pros   5. Ramon y Cajal S. Degeneration and Regeneration of the
Nervous System. May RM, trans. Oxford University Press;​
and cons of the two rodent models in comparison to other 1928.
larger-animal models (Table 1). In our experience, obtain-   6. Liu CT, Benda CE, Lewey FH. Tensile strength of human
ing behavioral results in mouse models, especially evalua- nerves;​an experimental physical and histologic study. Arch
tion of sensation using the von Frey monofilament testing Neurol Psychiatry. 1948;​59(3):​322–336.
technique, is difficult and potentially unreliable, a notion   7. Sunderland S, Bradley KC. Stress-strain phenomena in hu-
that is further supported by published literature.22,23 This man spinal nerve roots. Brain. 1961;​84(1):​120–124.
is especially so, because a combination of pain and deaf-   8. Lundborg G, Rydevik B. Effects of stretching the tibial nerve
ferentation can mask results, as the authors have rightfully of the rabbit. A preliminary study of the intraneural circula-
noted in their discussion. That being said, the authors’ tion and the barrier function of the perineurium. J Bone Joint
Surg Br. 1973;​55(2):​390–401.
histological section (as in their previous publications) has   9. Highet WB, Sanders FK. The effects of stretching nerves
the strongest and most convincing results presented, de- after suture. BMJ. 1943;​30(120):​355–369.
serving more emphasis than their behavioral analysis. Fi- 10. Shakhbazau A, Kawasoe J, Hoyng SA, et al. Early regenera-
nally, use of this method, especially in the different stretch tive effects of NGF-transduced Schwann cells in peripheral
injuries, might be technically difficult to reproduce and nerve repair. Mol Cell Neurosci. 2012;​50(1):​103–112.
warrants further validation by other research laboratories. 11. Martini R, Fischer S, López-Vales R, David S. Interactions
As we enter an era of fast translation from the bench between Schwann cells and macrophages in injury and in-
to the clinic, we believe that Mahan and colleagues’ col- herited demyelinating disease. Glia. 2008;​56(14):​1566–1577.
12. Midha R. Mechanisms and pathology of injury. In:​KimDH,
lective work can provide a fundamental basis from which Kline DG, eds. Kline and Hudson’s Nerve Injuries:​Opera-
future investigations are carried forward for studies on tive Results for Major Nerve Injuries, Entrapments and
nerve injury, regeneration, and therapeutic modalities. The Tumors. 2nd ed. Saunders Elsevier;​2008:​23–42.
work in mice is especially important, because it permits 13. Lundborg G. Nerve injury and repair – a challenge to the
the study of genetically altered animals harboring a poten- plastic brain. J Peripher Nerv Syst. 2003;​8(4):​209–226.
tial transgenic mutation to allow a better appreciation of 14. Fu SY, Gordon T. The cellular and molecular basis of periph-
mechanistic interactions between Schwann cell and other eral nerve regeneration. Mol Neurobiol. 1997;​14(1-2):​67–116.
cellular and molecular participants in the nerve injury and 15. Kaplan HM, Mishra P, Kohn J. The overwhelming use of
rat models in nerve regeneration research may compromise
regeneration response.24 designs of nerve guidance conduits for humans. J Mater Sci
https://thejns.org/doi/abs/10.3171/2020.6.JNS201799 Mater Med. 2015;​26(8):​226.

J Neurosurg Volume 135 • September 2021 891

 
*46 *
Editorial

16. Umansky D, Midha R. Commentary:​Rapid-stretch injury clinical injuries. Drs. Umansky and Midha suggested that
to peripheral nerves:​histologic results. Neurosurgery. 2020;​ a more detailed comparison between rodent models—rats
86(3):​E333–E334. in the initial work and mice in the current work—would
17. Fadia NB, Bliley JM, DiBernardo GA, et al. Long-gap pe-
ripheral nerve repair through sustained release of a neuro-
be beneficial; as they point out, the greater transgenic flex-
trophic factor in nonhuman primates. Sci Transl Med. 2020;​ ibility of the mouse model was a clear benefit. Although it
12(527):​eaav7753. would be both enlightening and worthwhile to understand
18. Mahan MA, Yeoh S, Monson K, Light A. Rapid stretch variation between rodent species in nerve biomechanics,
injury to peripheral nerves:​biomechanical results. Neurosur- nerve histology, and neurological recovery, we considered
gery. 2019;​85(1):​E137–E144. a detailed study that extended beyond our published sta-
19. Warner WS, Yeoh S, Light A, et al. Rapid-stretch injury to tistical comparison to be a less pressing concern for the
peripheral nerves:​histologic results. Neurosurgery. 2020;​ transition to transgenic animals that would more closely
86(3):​437–445.
20. Mahan MA, Warner WS, Yeoh S, Light A. Rapid-stretch reflect humans.
injury to peripheral nerves:​implications from an animal We also agree that behavioral work such as von Frey
model. J Neurosurg. Published online October 4, 2019. doi:​ testing can be challenging in rodent species, because simi-
10.3171/2019.6.JNS19511 lar monofilament testing has been interpreted as a test for
21. Yeoh S, Warner WS, Eli I, Mahan MA. Rapid-stretch both hypesthesia1 and allodynia.2 Our attempts to over-
injury to peripheral nerves:​comparison of injury models. come that challenge, including Hargreaves testing,3 proved
J Neurosurg. Published online November 6, 2020. doi:​ inconclusive. These results indicate that the contributions
10.3171/2020.5.JNS193448
22. Chaplan SR, Bach FW, Pogrel JW, et al. Quantitative assess-
from both neuropathic pain and loss of afferent fibers are
ment of tactile allodynia in the rat paw. J Neurosci Methods. complex phenomena that require careful experimental
1994;​53(1):​55–63. design and data assessment, as ceaselessly expressed in
23. Bonin RP, Bories C, De Koninck Y. A simplified up-down the dynamic spectrum of pain and its perception that our
method (SUDO) for measuring mechanical nociception in clinical patients describe.
rodents using von Frey filaments. Mol Pain. 2014;​10(1):​26. We wholeheartedly concur that further validation and
24. Stratton JA, Holmes A, Rosin NL, et al. Macrophages regu- advancement to larger animals is necessary and will pro-
late Schwann cell maturation after nerve injury. Cell Rep. vide new insights into the biomechanics and cellular re-
2018;​24(10):​2561–2572.e6.
sponse of polyfascicular nerves. Our rapid-stretch injury
Disclosures model is flexible and can be applied in both rodent and
The authors report no conflict of interest.
larger-animal models, much as controlled cortical impact,
contusion weight drop, and fluid percussion have all be-
Correspondence come relatively standardized models for injury to the brain
Rajiv Midha: Department of Clinical Neurosciences, University and spinal cord. To take advantage of the various benefits
of Calgary, Alberta, Canada. [email protected]. listed in the editorial’s Table 1, we and our partners are
currently using our stretch-injury model in both rats and
INCLUDE WHEN CITING  larger animals for experiments in imaging, electrophysiol-
Published online November 6, 2020; DOI: 10.3171/2020.6.JNS201799. ogy, and behavioral assessment to more fully answer bio-
mechanical and physiological questions regarding nerve
Response stretch injury, regeneration, and neuroma formation. We
intend to continue our work in mice to take advantage of
Mark A. Mahan, MD,1 Wesley S. Warner, BA,1 tools such as genetic sequencing and transgenic modifi-
Stewart Yeoh, PhD,1 and Alan Light, PhD2 cation to probe cellular and molecular contributions to
failed regeneration, with the goal of using this knowledge
Department of Neurosurgery, Clinical Neurosciences Center, University to develop therapeutic solutions for improving nerve in-
1

of Utah; and 2Department of Anesthesiology, University of Utah, Salt jury outcomes.

Lake City, Utah
We thank Drs. Umansky and Midha for their insightful References
and considered editorial. As they point out, nerve stretch   1. Bradman MJG, Ferrini F, Salio C, Merighi A. Practical
injury has long been known to cause severe dysfunction mechanical threshold estimation in rodents using von Frey
hairs/Semmes–Weinstein monofilaments:​towards a rational
and pathophysiology; however, the study and repair of pe- method. J Neurosci Methods. 2015;​255:​92–103.
ripheral nerve injury is a field that has lacked standardiza-   2. Chaplan SR, Bach FW, Pogrel JW, et al. Quantitative assess-
tion and application of a clinically relevant injury model. ment of tactile allodynia in the rat paw. J Neurosci Methods.
Most of the current animal research investigates the cues 1994;​53(1):​55–63.
of successful regeneration. We would also benefit from   3. Mahan MA, Warner WS, Yeoh S, Light A. Rapid-stretch
understanding the mechanisms that lead to failed regen- injury to peripheral nerves:​implications from an animal
eration, which is a common clinical outcome after trau- model. J Neurosurg. Published online October 4, 2019. doi:​
10.3171/2019.6.JNS19511.
matic injury.
Drs. Umansky and Midha’s discussion of the benefits INCLUDE WHEN CITING 
and drawbacks of different animal models highlights a key Published online November 6, 2020; DOI: 10.3171/2020.7.JNS202241.
issue that all preclinical research faces: translating animal
findings into clinical relevance. Experimental animals are
not humans and experimental injuries often fail to mimic

892 J Neurosurg Volume 135 • September 2021

 
*46 *
 LABORATORY INVESTIGATION
J Neurosurg 135:893–903, 2021

Rapid-stretch injury to peripheral nerves: comparison of

injury models
Stewart Yeoh, PhD, Wesley S. Warner, BA, Ilyas Eli, MD, and Mark A. Mahan, MD
Department of Neurosurgery, Clinical Neurosciences Center, University of Utah, Salt Lake City, Utah

OBJECTIVE  Traditional animal models of nerve injury use controlled crush or transection injuries to investigate nerve
regeneration; however, a more common and challenging clinical problem involves closed traction nerve injuries. The
authors have produced a precise traction injury model and sought to examine how the pathophysiology of stretch injuries
compares with that of crush and transection injuries.
METHODS  Ninety-five late-adolescent (8-week-old) male mice underwent 1 of 7 injury grades or a sham injury (n > 10 per
group): elastic stretch, inelastic stretch, stretch rupture, crush, primary coaptation, secondary coaptation, and critical gap.
Animals underwent serial neurological assessment with sciatic function index, tapered beam, and von Frey monofilament
testing for 48 days after injury, followed by trichrome and immunofluorescent nerve histology and muscle weight evaluation.
RESULTS  The in-continuity injuries, crush and elastic stretch, demonstrated different recovery profiles, with more
severe functional deficits after crush injury than after elastic stretch immediately following injury (p < 0.05). However,
animals with either injury type returned to baseline performance in all neurological assessments, accompanied by mini-
mal change in nerve histology. Inelastic stretch, a partial discontinuity injury, produced more severe neurological deficits,
incomplete return of function, 47% ± 9.1% (mean ± SD) reduction of axon counts (p < 0.001), and partial neuroma forma-
tion within the nerve. Discontinuity injuries, including immediate and delayed nerve repair, stretch rupture, and critical
gap, manifested severe, long-term neurological deficits and profound axonal loss, coupled with intraneural scar forma-
tion. Although repaired nerves demonstrated axon regeneration across the gap, rupture and critical gap injuries demon-
strated negligible axon crossing, despite rupture injuries having healed into continuity.
CONCLUSIONS  Stretch-injured nerves present unique pathology and functional deficits compared with traditional
nerve injury models. Because of the profound neuroma formation, stretch injuries represent an opportunity to study the
pathophysiology associated with clinical injury mechanisms. Further validation for comparison with human injuries will
require evaluation in a large-animal model.
https://thejns.org/doi/abs/10.3171/2020.5.JNS193448
KEYWORDS  peripheral nerve; stretch injury; neuroma; neuroma-in-continuity; nerve repair; nerve coaptation

T raumaticnerve injury is a significant cause of pain models used to elucidate the mechanisms of nerve regen-
and disability, with an estimated 18,000 new cases eration. Crush injuries generally create a temporary dys-
annually in the United States.1,2 Although periph- function that recovers fully,6,7 whereas transection models
eral nerves possess a remarkable facility for recovery and show reinnervation3 and functional recovery.8 Although
regrowth, the regenerative program may be impaired by helpful in understanding successful reinnervation and neu-
the severity and mechanism of injury, resulting in limited rological recovery, these models do not mimic the com-
or failed reinnervation in a majority of cases. Histopatho- plexity and concomitant severity of clinical injuries.9 For
logical evaluation of injured nerves often reveals the for- example, one of the most frequent mechanisms of severe
mation of an intraneural fibrous scar, which has been asso- nerve injury results from rapid-stretch trauma sustained
ciated with poor outcomes.3–5 Understanding the specific during falls or high-speed injuries, such as motor vehicle
pathophysiology of failed regeneration is fundamental to accidents.10,11 Because of the different injury mechanisms,
advancing treatment options. animal models of crush or transection injuries do not de-
Nerve crush or transection injuries are common animal velop the characteristic neuroma-in-continuity of clinical

ABBREVIATIONS  DAPI = 4′,6-diamidino-2-phenylindole; GLUT1 = glucose transporter–1; MBP = myelin basic protein; NF200 = neurofilament protein 200; SFI = sciatic
function index.
ACCOMPANYING EDITORIAL  See pp 890–892. DOI: 10.3171/2020.6.JNS201799.
SUBMITTED  February 19, 2020.  ACCEPTED  May 13, 2020.
INCLUDE WHEN CITING  Published online November 6, 2020; DOI: 10.3171/2020.5.JNS193448.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  893

*46 *
Yeoh et al.

injuries,12,13 nor do they provide insight into the pathophys- Primary Repair
iology of closed stretch injuries.14 The midsciatic nerve was cut with sharp scissors and
To address the disparity between clinical injury mecha- immediately coapted with two interrupted stitches through
nism and basic science models, we have developed a re- the epineurium with a 9-0 suture.
producible, rapid-stretch nerve injury model for use in ro-
dents that stratifies injury severity based on biomechanical Secondary Repair
characteristics. We have previously demonstrated that our Transection was performed in the same manner as
model generates reproducible injury grades with histologi- outlined for primary repair, except that after transection,
cal features and functional outcomes similar to those of proximal and distal nerve stumps were separated by 15
human nerve injuries.15 These animal models of stretch mm and lightly sutured to muscle on opposite sides of
injuries exhibit long zones of fiber disruption that generate the incision. The wound was then closed, and the animals
a disorganized matrix that shares many characteristics of were returned to their home cages for 4 weeks to repli-
clinical neuroma histology.12,16 cate delayed clinical repair.19,20 At 4 weeks the nerve was
To further validate this novel model, we compared rap- reexposed, and the proximal and distal stumps were care-
id-stretch injury against established crush and transection fully dissected from scar tissue. The nerve stumps were
protocols to highlight the histological and functional dif- freshened to normal-appearing nerve and coapted with a
ferences between injury mechanisms. 9-0 suture as discussed above. The 4-week delay in repair
approximates 2–3 years in humans.21
Methods
Animals Critical Gap
Ninety-five (n > 10 per group) adolescent (8-week- To assess nerve discontinuity greater than critical gap
old) male mice were used to simulate the primary clini- distance,22,23 the nerves were exposed and a 15-mm-long
cal demographic for peripheral nerve injury.17 The trans- nerve segment was removed.
genic B6.Cg-Tg(Thy1-YFP)HJrs/J; Jax mice (no. 003782;
The Jackson Laboratory) that were used express a mosaic Sham Injury
distribution of yellow fluorescent protein in the axons of Sham injury consisted of surgical exposure and place-
motor and sensory fibers. Animals had ad libitum access ment of the nerve on the weight drop system hook for 30
to food and water while housed on a 12-hour light–dark seconds without rapid stretch.
cycle. Experiments were upheld to the standards of the
University of Utah Institutional Animal Care and Use Behavioral Evaluation
Committee. Animals were tested at a consistent time and acclimat-
ed to the testing room for 30 minutes before testing for
Nerve Injury Models all experimental time points. Functional performance was
Mice were anesthetized with 2% isoflurane and se- assessed by sciatic function index (SFI) for fine locomo-
cured to a water-recirculating heating pad. The sciatic tor function according to the methods of Bain et al.,24 in
nerve was exposed from the knee to the sciatic notch with which mouse paw prints are analyzed for factors such as
blunt dissection. One of 7 injury grades or sham injury, as print length, step distance, and toe spread to determine a
described below, was applied to the nerve before closing final SFI score. The tapered beam walking task was used
the incision with nylon suture. to assess gross locomotor function, the plantar von Frey
monofilament test for mechanical sensibility, and the Har-
Rapid-Stretch Injury greaves method for temperature sensibility, as previously
Three grades of rapid-stretch injury were performed detailed.18 Before nerve injury, animals underwent 2 days
by using a weight drop on a cable and pulley/hook sys- of training in which 2 trials of SFI and 3 trials with the
tem, with severity graduated by maximum hook displace- tapered beam were performed. Baseline evaluation was
ment: 1) elastic stretch injury (7.5-mm stretch excursion); obtained after 2 days of training (day −1), and all animals
2) inelastic stretch injury (12.5 mm); and 3) stretch rup- were trained to acceptable standards. A priori time points
ture (> 15 mm). Nerve stretch injury was confirmed by were determined for testing, as follows: after recovery
calculation of maximum strain and persistent deforma- from injury (day 0) and on days 1, 2, 6, 9, 15, 23, 30, and
tion, which were determined using high-speed (5000 fps) 48. For secondary repair, these time points were obtained
video of fiducial ink markers applied to the nerve before after the delay in repair, except for baseline, which was
injury as previously described.18 For rupture injury, nerve obtained and recorded prior to initial surgery.
ends were reapproximated without suture or other coapta-
tion. Histological Assessment
Animals were euthanized 48 days after injury. The in-
Crush Injury jured sciatic nerve was extracted from anterior to the ham-
Crush injury was applied to the midsciatic nerve by string to posterior of the distal trifurcation. Proximal and
using a titanium mini-aneurysm clip (Aesculap FT680T) distal segments (1–2 mm) were reserved for axial section-
with a jaw width of 1.0 mm and a crush force of 1.08 N. ing for axon counts, and the zone of injury was sectioned
Crush was applied for 30 seconds, after which the clip was longitudinally for evaluation of the injury architecture.
removed and the incision closed. Nerve histomorphometry, trichrome staining, and immu-

894 J Neurosurg Volume 135 • September 2021

 
*46 *
 Yeoh et al.

FIG. 1. Serial neurological assessment. A: SFI score. Animals with sham and elastic stretch injuries showed no significant differ-
ence in SFI scores across the testing period. Inelastic stretch generated significant (p < 0.01) acute impairment, with moderate
recovery. Stretch rupture showed severe impairment (p < 0.001), with minimally recovering SFI over the testing period. In contrast,
nerve crush generated only transient SFI deficit (p < 0.05 on days 0, 1, 2, and 6), recovering by day 9. All transection models
(primary repair, secondary repair, nerve gap) showed immediate and complete SFI impairment (p < 0.001). B: Tapered beam test.
The performance of animals with sham injuries was consistent across testing days, and no difference from baseline was observed.
Although an increase in foot faults was observed acutely after injury for elastic stretch, this behavior was not significant. Crush
injury demonstrated significantly increased foot faults until day 6 (p < 0.01), and inelastic deficits persisted through day 9 (p <
0.01), yet both injury grades recovered to baseline levels by day 15. Primary repair, nerve gap, and rupture injury grades demon-
strated markedly similar performances with significant deficits acutely after injury and progressive recovery until day 9 (p < 0.01,
p < 0.001, p < 0.0001, respectively). Although deficits remained statistically significant at day 15, recovery proved to be transient
and functional deficits worsened through the end of the testing paradigm (no repair, p < 0.05; primary and stretch rupture, p <
0.0001). Immediately after repair, secondary repair demonstrated a similar performance to nerve gap when compared at the same
days after primary injury. Secondary repair followed with progressive deficits as other discontinuity injuries through the end of the
experiment (p < 0.01). C: von Frey test. No difference in response was observed between sham and elastic injuries (p = 0.18).
Crush injury demonstrated only sensation-related deficits the day of surgery (p < 0.05), whereas inelastic, primary, secondary, and
nerve gap injuries demonstrated transient deficits (p < 0.05). A significant decrease in sensitivity was observed in stretch rupture
(p < 0.01) until the final day of testing. D: Wet muscle weight. Crush and elastic injuries did not demonstrate a statistical decrease
in gastrocnemius wet muscle weight in comparison with sham injuries. Inelastic injury demonstrated slight atrophy (p < 0.01),
whereas the stretch rupture, primary, secondary, and nerve gap subgroups demonstrated statistically decreased muscle weight on
harvest (p < 0.0001). Figure is available in color online only.

nofluorescent imaging were performed as previously de- and Hargreaves performances; error bars are reported
scribed.18 Specifically, Lille’s trichrome imaging was used with standard error of the mean. SFI, wet muscle weight,
for evaluation of collagen deposition, cellularity, and com- and nerve fiber histomorphometry were assessed with
parison to clinical staining. Immunofluorescent antibodies one-way ANOVA (Tukey’s post hoc test); error bars for
were used to identify changes to structural components of these 3 values are reported with standard deviation. Sig-
the nerve after injury: perineurium (glucose transporter–1 nificance was defined as p < 0.05.
[GLUT1], 1:200, Abcam); myelin (myelin basic protein
[MBP], 1:500, Abcam); axons (neurofilament protein 200
[NF200], 1:500, Sigma); endoneurial extracellular matrix
Results
(laminin, 1:800, Sigma); and nuclei (4′,6-diamidino-2-phe- Behavioral Evaluation
nylindole [DAPI]). The gastrocnemius and tibialis ante- Fine Locomotor Assessment (SFI)
rior muscles of the injured and contralateral sides were Sham and elastic injuries demonstrated no statistically
harvested and weighed immediately for an indirect mea- significant changes from baseline (Fig. 1A), whereas crush
surement of reinnervation. injuries exhibited a profound decrease in SFI immediately
after injury (−71 ± 16 [SD], p < 0.001), with recovery to
Statistical Evaluation sham or baseline condition by day 9. Severe rapid-stretch
GraphPad Prism 6 was used for statistical analysis. The injury (inelastic stretch and stretch rupture) shared poor
Kruskal-Wallis test with Dunn’s multiple comparison post recovery with surgical discontinuity injuries (primary
hoc test was used to assess the tapered beam, von Frey, repair, secondary repair, and critical gap injuries), dem-

J Neurosurg Volume 135 • September 2021 895

 
*46 *
Yeoh et al.

FIG. 2. Lillie’s trichrome staining of longitudinally sectioned (15-μm slice thickness) sciatic nerve extracted 48 days after injury for
assessment of fiber arrangement. Sham-injured nerves (A) demonstrated no change compared with normal control nerves (not
shown). Elastic (B) and crush (E) injuries had intact fibers with no visible changes to the epineurium and perineurium. Inelastic
injuries (C), primary repair (F), and secondary repair (G) demonstrated partial fiber integrity; a number of fibers traverse the fringe
of the injury and maintain uninterrupted connection. Other fibers are intensely snarled at the injury and suture site, suggestive of
frustrated regeneration. Rupture (D) and critical gap (H) demonstrated the most progressive pathology, with robust engorgement,
collagen deposition, and fibrous entanglement characteristic of neuroma. Bar = 50 μm. Figure is available in color online only.

onstrating highly impaired function (inelastic p < 0.01; critical gap injuries was statistically significant on day 0
rupture, primary repair, secondary repair, nerve gap; p < (p < 0.05), yet deficits were not significant on other testing
0.001) immediately after injury. Final outcomes ranged days. Stretch rupture was the only injury to demonstrate
from moderate but highly variable impairment for animals significant and persistent sensory deficits through day 23
in the inelastic injury group (SFI −47 ± 19, p = 0.07) to se- after injury (p < 0.01).
vere impairment for all discontinuity models (rupture −71
± 23, primary repair −69 ± 4, secondary repair −87 ± 18, Temperature Sensitivity (Hargreaves)
nerve gap −90 ± 14; all p < 0.001). There was no statistical All animals retained sensitivity to heat stimulus after
difference between severe rapid-stretch injuries and surgi- injury. Although instances of significant differences from
cal discontinuity injuries at the experimental endpoint. sham injury were observed acutely after injury, no persis-
tent or meaningful assumptions can be made with respect
Gross Locomotor Assessment (Tapered Beam) to injury grade.
All injuries demonstrated a transient increase in foot
faults compared with sham injury (Fig. 1B). Elastic injury Histological Assessment
demonstrated only mild changes, whereas crush and in- Wet Muscle Weight
elastic injuries demonstrated significantly increased defi- Rupture, primary and secondary repair, and critical
cits until day 15 (p < 0.01), but no statistical difference gap injury were significantly atrophied in comparison
from sham injury at the experiment’s endpoint. In con- with sham injury (p < 0.0001) (Fig. 1D). Inelastic injury
trast, discontinuity injuries (rupture, primary repair, sec- also demonstrated a significant loss in wet muscle weight
ondary repair, and critical gap) demonstrated persistent (p < 0.01), whereas crush and elastic injuries were not sta-
deficits throughout the testing paradigm (p < 0.01, two- tistically different compared with sham injury (p = 0.07
way ANOVA). and p = 0.81, respectively).
Light Touch (von Frey) Trichrome Staining
Sham and elastic injuries were not statistically differ- A positive association between fiber pathology and
ent from one another, although both demonstrated a non- injury severity was observed in longitudinal sections of
significant slight loss of sensation after surgery (Fig. 1C). specimens (Fig. 2). Compact and oscillating fibers of
Decreased sensation was observed in crush, inelastic, and sham-injured nerves recapitulated control nerve histol-
secondary repair acutely after injury or repair; however, ogy (not shown). Elastic and crush injuries demonstrated
differences were not statistically significant (p > 0.05 for a similar pattern of loose and straightened endoneurial
all comparisons). Loss of sensation for primary repair and fibers, with intact external layers of perineurium and epi-

896 J Neurosurg Volume 135 • September 2021

 
*46 *
 Yeoh et al.

FIG. 3. Immunofluorescent longitudinal images (15-μm slice thickness) of the area of maximal injury 48 days after injury for elastic
and crush injuries. In comparison with the sham injury model (A–E), there were minimal architectural changes in elastic (F–J)
and crush (K–O) injuries. Antilaminin depicted minimal change in endoneurial tube thickness and continuity in sham, elastic, and
crush injuries (A, F, and K). MBP (C, H, and M) consistently encapsulated NF200 (B, G, and L) across injury grades; thus, MBP
similarly attenuated as NF200 decreased with increasing injury severity. GLUT1 delineated the exterior perineurial layer of sham,
elastic, and crush injuries, respectively (D, I, and N). DAPI revealed hypercellularity in all injuries except elastic ones (E, J, and O).
Original magnification ×200 (bar = 50 μm). Figure is available in color online only.

neurium. Inelastic, primary repair, and secondary repair decrease of MBP at the coaptation site of primary and sec-
injuries revealed extensive collagen deposition and disor- ondary repairs, as well as the intact portions of inelastic
ganized fibers. For repair injuries, this phenomenon ex- injuries. In contrast, dense MBP immunofluorescence and
ists most prevalently around the suture site; however, for a distinct tube structure—similar to histological findings
inelastic injuries, a dichotomy in pathology presented— in sham injury—were observed both distal and proximal
with one fascicle apparently uninjured, whereas the other to coaptation. Rupture injuries demonstrated the most sub-
fascicle manifested neuroma histology. Rupture and criti- stantial reduction in MBP, whereas scattered MBP bodies
cal gap injuries were hallmarked by intense collagen de- were seen at the stump of the critical gap injury. GLUT1,
position, dense and persistent hypercellularity, and fiber a component of the perineurium and blood-nerve barrier,
anarchy. was restricted to the exterior of the nerve bundle in sham,
elastic, and crush injuries. In contrast, GLUT1 infiltrated
Immunofluorescence Imaging the endoneurium of severe stretch, primary, secondary,
Robust remodeling of nerve architecture increased with and critical gap injuries, crisply defining the endoneurial
progressive injury severity (Figs. 3 and 4). Antilaminin tubes throughout inelastic stretch–injured nerves. GLUT1
depicted minimal change in endoneurial tube thick- was sequestered primarily at the suture site in primary
ness and continuity in sham, elastic, and crush injuries. and secondary injuries and intensified around the suture
In contrast, inelastic injuries exhibited chaotic laminin material, which is a feature of a foreign body reaction (Fig.
architecture. In primary and secondary repair, laminin 5).25 Interestingly, although the ruptured nerves healed in
tubules proximal to the coaptation site were roughly the continuity, there was no convincingly linear or tubular
same diameter and resembled those of sham injury, with GLUT1 staining in the neuroma between the distal and
consistent linearity and continuity. However, around the proximal ends of ruptured nerves, as occurred after suture
site of coaptation, endoneurial tube integrity was compro- repair. DAPI staining showed minimal increase in cel-
mised with a honeycomb-like matrix of laminin. Tubular lularity in crush and elastic injuries in comparison with
morphology was not evident in rupture and critical gap in- sham injuries, except at the site of the crush (Fig. 4). Pri-
juries. NF200 identified continuous fibers in sham, elastic, mary, secondary, and critical gap injuries display hyper-
and crush injuries. Axons were not linear and were more cellularity similar to severe stretch injuries.
widely spaced in inelastic, primary, and secondary repair
injuries, whereas they were seemingly nonexistent in rup- Nerve Histomorphometry
ture and critical gap injuries. Dense and tubular MBP en- Sham-injured and elastic stretch–injured nerves dem-
veloped NF200-positive axons in sham, elastic, and crush onstrated minimal change and did not differ significantly
injuries. Attenuated NF200 was accompanied by a similar in number of axons (Fig. 6), average axon size, or myelin

J Neurosurg Volume 135 • September 2021 897

 
*46 *
Yeoh et al.

FIG. 4. Immunofluorescent longitudinal images (15-μm slice thickness) of the area of maximal injury 48 days after injury for severe
stretch, repair, and critical gap. In comparison with the sham injury model (see Fig. 3A–E), severe stretch (inelastic [A–E] and
rupture [F–J]), primary (K–O) and secondary (P–T) repair, and critical gap (U–Y) injuries demonstrated robust pathology. Laminin
(A, F, K, P, and U) and NF200 (B, G, L, Q, and V) immunofluorescence revealed aberrant remodeling of endoneurial tubules and
limited axonal regeneration of primary and secondary repair and critical gap injuries, mirroring severe stretch injury. MBP (C, H,
M, R, and W) consistently encapsulated NF200 across injury grades; thus, MBP similarly attenuated as NF200 decreased with
increasing injury severity. Both MBP and NF200 were depleted in rupture injury, whereas some MBP persisted in critical gap injury
despite the dearth of NF200. Profound GLUT1 infiltration hallmarks severe stretch injuries (inelastic and rupture), transection inju-
ries (primary and secondary repair), and critical gap (D, I, N, S, and X). Interestingly, this infiltration demarcates the architecture
of endoneurial tubules in inelastic, primary, and secondary repair, whereas stretch rupture and critical gap are typified by diffuse
staining. Finally, DAPI revealed hypercellularity in all injuries except elastic ones (E, J, O, T, and Y). Original magnification ×200
(bar = 50 μm). Figure is available in color online only.

thickness (data not included). All injury models except nerve subjected to rapid-stretch injuries, which fit into three
elastic had significantly (p < 0.01) fewer axons distal to thresholds of severity: elastic stretch, inelastic stretch, and
the injury zone. Additionally, inelastic, rupture, and criti- stretch rupture.15 The biomechanical severities reflected
cal gap injuries showed fewer axons proximal to the injury consistent patterns of acute histological changes26 that also
(p < 0.01). There appears to be a loosely proportional in- correlated with longer-term neurological recovery and neu-
crease in fibrous stroma of the distal nerve, with complete roma formation. Specifically, stretch within the elastic lim-
fibrous replacement of the distal nerve as a consequence of it resembles a neurapraxic injury, whereas inelastic injury
critical gap and rupture injuries. demonstrates cardinal features of axonotmesis with partial
recovery. Stretch rupture of a nerve appears to obliterate
successful regeneration.18 Thus, the recovery potential of
Discussion pure stretch injury follows the biomechanical and tissue
We previously defined the biomechanical properties of structural properties of the nerve.
898 J Neurosurg Volume 135 • September 2021
 
*46 *
 Yeoh et al.

FIG. 5. Wide-field longitudinal immunofluorescent images (15-μm slice thickness) proximal and distal to the zone of maximal injury.
Rich GLUT1 staining is consistent with the perineurium and distinctly outlined the nerve bundle in sham (A), elastic (B), and crush
(E) injuries. In contrast, all other injuries demonstrated aberrant perineurial remodeling, with either GLUT1 outlining endoneurial
tubes or chaotic GLUT1 infiltration. The endoneurial tubes are clearly defined in inelastic (C), primary repair (F), and secondary
repair (G) injuries. In nerve coaptations, this GLUT1 remodeling begins slightly proximal to the suture site, circumnavigates the
suture, and then continues to infiltrate distally. In stretch rupture (D) and critical gap (H) injuries, we visualized chaotic GLUT1
infiltration at the neuroma site with progression to the distal aspect of the specimen. Original magnification ×100 (bar = 1000 μm).
Figure is available in color online only.

In this study, we sought to examine how the gradu- Crush is distinct from elastic stretch in the severity of
ated stretch injury severities compare with the traditional deficits immediately after injury, with significantly worse
models used to study nerve regeneration: crush, transec- performance in SFI, tapered beam, and von Frey testing. A
tion, and critical gap.3,27 Our panel of behavioral and his- focal crush injures axons and creates wallerian degenera-
tological evaluations demonstrated a clear spectrum of tion and regeneration, as observed by others.13,34,35 How-
injury, which can be summarized into three groups: nerve ever, recovery after crush injury, as demonstrated by wet
injury–in-continuity (elastic and crush), which demon- muscle weight, histology, and function returning to base-
strated neurological recovery to baseline; partial disconti- line, has been attributed to the preservation of endoneu-
nuity (inelastic), which demonstrated mixed recovery and rial architecture, which facilitates successful guidance of
histological outcomes; and discontinuity models (stretch regenerating axons to their prior distal targets.36 In support
rupture, primary and secondary repair, and critical gap), of this, we found that there were no histological changes
which shared similarly poor functional outcomes but di- to crushed nerves, except a small zone of persistently in-
vergent histopathological findings. creased cellularity, which was suggestive of focal scar and
potentially related to inflammation or the process of wal-
lerian degeneration.37
Nerve-in-Continuity Injuries In contrast, elastic stretch injuries demonstrated mod-
Crush injury is a common experimental nerve injury est to insignificant behavioral alterations, presumably due
model because of its reproducibility and ease of surgi- to the distribution of stretching energy along the nerve—
cal manipulation.28,29 Many studies demonstrate complete based on continuum mechanics of stretch and our prior
neurological recovery and little visible histopathological evaluation of histology immediately after stretch inju-
change.30–33 This provides an excellent comparison with ries.26 Axonal injury is presumably shielded by the inter-
elastic stretch injury, which similarly has shown scant nal redundancy of endoneurial undulations, the so-called
neurological deficits and histopathological change, despite bands of Fontana.38,39 Thus, the ephemeral functional defi-
experiencing an average strain of nearly 50%.18 cits after elastic stretch injury recover too quickly to be at-
J Neurosurg Volume 135 • September 2021 899
 
*46 *
Yeoh et al.

FIG. 6. A: Cross-sectional histomorphometry of sections far proximal and far distal to the zone of injury. Sham, elastic, and crush
injuries retained large axons in proximal and distal sections. There were fewer large fibers in inelastic and primary and second-
ary repair injuries, and large fibers were effectively nonexistent in rupture or critical gap. In the distal segment, a greater degree
of fibrous stroma accompanied the absence of large fibers. Osmium tetroxide staining, 1-μm slice thickness, poststained in 1%
toluidine blue before imaging. Original magnification ×400 (bar = 50 µm). B: Histomorphometric analysis. Distal axon counts were
significantly reduced (p < 0.01) in all discontinuity (stretch rupture, primary and secondary repair, critical gap) and nerve crush
models. Axon size decreased in repair, crush, and gap models, and myelin thickness decreased both proximally and distally in all
models except elastic, with significant difference (p < 0.01) between proximal and distal in repair models (data not shown). Figure
is available in color online only.

900 J Neurosurg Volume 135 • September 2021

 
*46 *
 Yeoh et al.

tributed to substantial axon degeneration and regeneration, axonal regeneration (primary and secondary repair) and
and are substantiated by minimal histological changes, no axonal regeneration (critical gap and rupture). The per-
similar to crush injury. The deficits after elastic stretch are sistent motor deficits after effective repair are consistent
more likely to be due to a conduction block from minute with findings from other groups29,31–33,40 and may indicate
and reversible architectural changes, which may be further difficulty in reestablishing functional proprioception and
elucidated through electrophysiological investigation. motor control. Conversely, the apparent deficits may sim-
ply be the result of the animal’s habituation or adaptation
Partial Discontinuity Injury to the tests or the limitation of the tests as the mouse learns
Stretch beyond the limits of elasticity demonstrates in- to walk with significant inversion or supination of the paw.
ternal fragmentation,26 as hallmarked by persistent elonga- Alternatively, poor performance could be explained by
tion of the nerve.15 After longer survival, excised nerves persistent pain as opposed to compromised function.
display neuromatous areas as well as partially abnor- Plantar von Frey fiber testing revealed differences be-
mal, linear axonal fibers, similar in appearance to elastic tween the discontinuity injuries, yet created uncertainty in
stretched nerves.18 interpretation. Both primary and secondary repair dem-
This pattern of injury is unique with respect to all oth- onstrated return of the withdrawal reflex to sham injury
er injury mechanisms, in both neurological recovery and levels, which may reflect successful regeneration seen his-
histology. Inelastic injury demonstrated more functional tologically. In particular, small-fiber regeneration appears
impairment and axonal loss than crush or elastic stretch, to be more robust after transection.41 However, the recov-
but less than discontinuity injuries. Muscle atrophy was ery of withdrawal reflex in the critical gap injury suggests
similar to an in-continuity or successful repair of discon- that alternative pathways may play a role, such as from
tinuity injury. The inelastic injury was uniquely defined the uninjured saphenous nerve. At the same time, the im-
by deviation between functional results on the SFI and ta- poverished sensory response of stretch-ruptured nerves
pered beam test. Whereas all other injury grades either re- is evidence against this same conclusion. It may be that
covered (in-continuity injuries) or demonstrated persistent a combination of pain and deafferentation underlies these
deficits (discontinuity injuries) for both functional tests, results; monofilament testing has been used to test both
inelastic injury did not recover completely on the SFI test pain and sensation in rodent models.42 The experimental
but did for the tapered beam. This dichotomized recovery design could have benefited from additional sensation as-
suggests two considerations. First, the partial lesion of in- sessments to provide further detail, particularly concern-
elastic injury may provide for improved gross locomotor ing cautious interpretation of von Frey testing in nerve
recovery. Second, SFI is a more sensitive test for differ- literature.43
ences in injury and recovery, despite the preservation of Interestingly, although stretch-ruptured nerves healed
tactile sensibility after inelastic stretch. in continuity, they showed no evidence of regeneration,
The histological findings of inelastic injury shared which confirms that stretch rupture is a unique patho-
features of both in-continuity and discontinuity injuries. physiology. Stretch rupture displayed impressive pathol-
First, one fascicle of the inelastic injury demonstrated ogy characterized by complete absence of axons (NF200)
continuity but apparent whorl-like neuroma in the other and myelination (MBP), extensive perineurial (GLUT1)
fascicle. The finding of intact but pathologically remod- infiltration, riotous and occasionally swirled extracellu-
eled fascicles is a pattern we have previously observed lar matrix (laminin), and a massive increase in cellular-
in human neuroma-in-continuity specimens.12 Second, ity (DAPI). This dysfunctional landscape shares histology
inelastic nerves demonstrated massive increases in cellu- with only the stump of the critical gap injury and the neu-
larity (DAPI), perineurial cell (GLUT1) infiltration, and romatous portion of the inelastic injury. The haphazard
loss of axons (NF200)—as also seen in the discontinuity matrix may represent a particular challenge for neurite
injuries. Myelination (MBP) is also reduced, characteristic outgrowth during regeneration, as suggested by others as
of wallerian degeneration. This injury pattern may be most the cause for poor recovery in humans.44
faithful to human neuroma-in-continuity lesions, in which
nerve regeneration and consequent muscle and functional Limitations
recovery are frequently partial and variable.9 We selected 48 days as the endpoint for functional and
histological remodeling, based on a correlated duration of
Discontinuity Injuries approximately 2 years in humans.21 Additional changes
Primary (immediate) repair and secondary (delayed) may manifest past the 48-day experimental window, as
repair represent models of clinical surgical repair and evidenced by the worsening tapered beam scores for all
demonstrate histological evidence of axonal regeneration. nerve discontinuity models. This abridged examination of
In contrast, a critical gap injury is used as a negative con- recovery curves does not encompass the long-duration im-
trol for regeneration studies. With a substantial part of the provement and may miss additional changes, particularly
nerve removed, the likelihood of spontaneous regeneration in the nerve repair injury models. Eight-week-old animals
is limited by the necessity of overcoming the critical gap. were used to reflect the clinically predominant population
These extremes of discontinuity injuries were used for of adolescents and young adults, because injury response
comparison with stretch-ruptured nerves. and regenerative capacity appear to diminish with advanc-
Transection injuries demonstrated consistent and per- ing age.45,46
sistent deficits in function; however, motor testing did not Variation in injury severity may account for differences
distinguish between injuries with histological evidence of in outcome. For example, more severe crush or less suc-
J Neurosurg Volume 135 • September 2021 901
 
*46 *
Yeoh et al.

cessful surgical repair can create differential outcomes; References

however, our results are comparable to those of previously   1. Karsy M, Watkins R, Jensen MR, et al. Trends and cost
published rodent models. Stretch injuries were validated analysis of upper extremity nerve injury using the National
with postinjury video review; however, despite identical (Nationwide) Inpatient Sample. World Neurosurg. 2019;123: ​ ​
injury thresholds, the actual strain each nerve suffers is e488–e500.
within a range and not identical to the others. Our initial   2. Foster CH, Karsy M, Jensen MR, et al. Trends and cost-
characterization of stretch injury used cadaveric rat nerve analysis of lower extremity nerve injury using the National
Inpatient Sample. Neurosurgery. 2019;85(2):250–256.
​ ​
tissue,15 and while some differences in elastic/inelastic   3. Cattin AL, Burden JJ, Van Emmenis L, et al. Macrophage-
strain thresholds and ultimate rupture limits were ob- induced blood vessels guide Schwann cell-mediated regen-
served, rat and mice nerves were broadly similar in their eration of peripheral nerves. Cell. 2015;162(5):1127–1139.
​ ​
response to stretch injury.18 Groups investigating nerve   4. Parrinello S, Napoli I, Ribeiro S, et al. EphB signaling di-
stretch injury should be aware that stretch magnitude, rects peripheral nerve regeneration through Sox2-dependent
vector, and strain rate all will influence injury character- Schwann cell sorting. Cell. 2010;143(1):145–155.
​ ​
istics, and experiments should be designed accordingly.   5. Lin JC, Schwentker-Colizza A, Curtis CG, Clarke HM. Final
results of grafting versus neurolysis in obstetrical brachial
Electrophysiological testing may be helpful for evaluating plexus palsy. Plast Reconstr Surg. 2009;123(3):939–948.
​ ​
the mechanisms underlying behavioral changes, and as-   6. Brushart TM. Nerve Repair. Oxford University Press;​2011.
sessment of the dorsal root ganglion and spinal cord may   7. Dulin JN, Antunes-Martins A, Chandran V, et al. Transcrip-
reveal neuronal viability after the different mechanisms of tomic approaches to neural repair. J Neurosci. 2015;35(41):​ ​
nerve injuries. Both of these questions are under current 13860–13867.
investigation.   8. Koka R, Hadlock TA. Quantification of functional recovery
Last, direct application of these results to human clini- following rat sciatic nerve transection. Exp Neurol. 2001;​
168(1):192–195.
​
cal injuries is not appropriate. Human nerves have more   9. Kim DH, Midha R, Murovic JA, et al, eds. Kline & Hud-
connective tissue associated with the polyfascicular struc- son’s Nerve Injuries:​Operative Results for Major Nerve
ture. Thus, stretch injuries in human nerves may have a Injuries, Entrapments, and Tumors. 2nd ed. W.B. Saunders;​
different injury pattern. Further understanding of rapid- 2007.
stretch injuries in a large-animal model remains necessary 10. Noble J, Munro CA, Prasad VS, Midha R. Analysis of upper
for application to clinical injuries. and lower extremity peripheral nerve injuries in a popula-
tion of patients with multiple injuries. J Trauma. 1998;45(1):
​ ​
116–122.
Conclusions 11. Birch R. Surgical Disorders of the Peripheral Nerves.
Springer;​2011.
We compared the regenerative outcomes between bio- 12. Mahan MA, Abou-Al-Shaar H, Karsy M, et al. Pathologic
mechanically graded rapid-stretch nerve injuries and clas- remodeling in human neuromas:​insights from clinical speci-
sic experimental models of nerve injury. Stretch-injured mens. Acta Neurochir (Wien). 2019;161(12):2453–2466.
​ ​
nerves differed substantially from existing models and 13. Kline DG, Hayes GJ, Morse AS. A comparative study of
demonstrate a distinct pathophysiology. For in-continuity response of species to peripheral-nerve injury. II. Crush
injuries, crush injuries exhibited more substantial deficits, and severance with primary suture. J Neurosurg. 1964;21: ​ ​
consistent with wallerian degeneration and regeneration, 980–988.
14. Kaplan HM, Mishra P, Kohn J. The overwhelming use of
whereas elastic stretch produced a more subtle neura- rat models in nerve regeneration research may compromise
praxia. When a nerve is stretched beyond its elastic limit, designs of nerve guidance conduits for humans. J Mater Sci
substantial fragmentation occurs. This partial disconti- Mater Med. 2015;26(8):226.
​ ​
nuity injury leads to diminished recovery, particularly 15. Mahan MA, Yeoh S, Monson K, Light A. Rapid stretch inju-
fine-motor control, and histology consistent with partial ry to peripheral nerves:​biomechanical results. Neurosurgery.
neuroma-in-continuity. Full discontinuity injuries demon- 2019;85(1):E137–E144.
​ ​
strated a bifurcation in recovery. Transection with repair 16. van Vliet AC, Tannemaat MR, van Duinen SG, et al. Human
neuroma-in-continuity contains focal deficits in myelination.
displayed axons crossing the gap and less significant mus- J Neuropathol Exp Neurol. 2015;74(9):901–911.
​ ​
cle mass loss. In contrast, stretch rupture injury, although 17. Midha R. Epidemiology of brachial plexus injuries in a multi-
healed into continuity, demonstrated similar results to the trauma population. Neurosurgery. 1997;40(6):1182–1189.
​ ​
negative control, which was critical gap injury. Stretch- 18. Mahan MA, Warner WS, Yeoh S, Light A. Rapid-stretch
injured nerves present a unique pathology and represent, injury to peripheral nerves:​implications from an animal
in contrast to crush or transection, an opportunity to study model. J Neurosurg. Published online October 4, 2019. doi:​
the pathophysiology associated with clinical injury mech- 10.3171/2019.6.JNS19511
19. Woodhall B. Modern history of peripheral nerve surgery;​
anisms. Further work is necessary to define the cellular World War II and the postwar study of peripheral nerve re-
proliferation and molecular cues inhibiting axonal regen- generation. J Am Med Assoc. 1949;139(9):564–566.
​ ​
eration. 20. Kline DG. Timing for exploration of nerve lesions and evalu-
ation of the neuroma-in-continuity. Clin Orthop Relat Res.
1982;(163):42–49.
​ ​
Acknowledgments 21. Fox JG, Newcomer C, Smith A, et al, eds. The Mouse in Bio-
We thank Kristin Kraus, MSc, for editorial assistance with this medical Research. 2nd ed. Elsevier;​2007.
paper and Nancy Chandler for osmium sample preparation at the 22. Labroo P, Hilgart D, Davis B, et al. Drug-delivering nerve
University of Utah Electron Microscopy Core Laboratory. This conduit improves regeneration in a critical-sized gap. Bio-
study was partially funded by the AANS/CNS Joint Section on technol Bioeng. 2019;116(1):143–154.
​ ​
Disorders of the Spine and Peripheral Nerves. 23. Madduri S, Gander B. Growth factor delivery systems and

902 J Neurosurg Volume 135 • September 2021

 
*46 *
 Yeoh et al.

repair strategies for damaged peripheral nerves. J Control 39. Pourmand R, Ochs S, Jersild RA Jr. The relation of the bead-
Release. 2012;​161(2):​274–282. ing of myelinated nerve fibers to the bands of Fontana. Neu-
24. Bain JR, Mackinnon SE, Hunter DA. Functional evaluation roscience. 1994;​61(2):​373–380.
of complete sciatic, peroneal, and posterior tibial nerve le- 40. Kemp SW, Phua PD, Stanoulis KN, et al. Functional recovery
sions in the rat. Plast Reconstr Surg. 1989;​83(1):​129–138. following peripheral nerve injury in the transgenic Thy1-GFP
25. Menovsky T, Beek JF. Carbon dioxide laser-assisted nerve rat. J Peripher Nerv Syst. 2013;​18(3):​220–231.
repair:​effect of solder and suture material on nerve regenera- 41. Lindenlaub T, Sommer C. Partial sciatic nerve transection as
tion in rat sciatic nerve. Microsurgery. 2003;​23(2):​109–116. a model of neuropathic pain:​a qualitative and quantitative
26. Warner W, Yeoh S, Light A, et al. Rapid-stretch injury to pe- neuropathological study. Pain. 2000;​89(1):​97–106.
ripheral nerves:​histologic results. Neurosurgery. 2020;​86(3):​ 42. Li Y, Dorsi MJ, Meyer RA, Belzberg AJ. Mechanical hyper-
437–445. algesia after an L5 spinal nerve lesion in the rat is not de-
27. Lewis GM, Kucenas S. Perineurial glia are essential for mo- pendent on input from injured nerve fibers. Pain. 2000;​85(3):​
tor axon regrowth following nerve injury. J Neurosci. 2014;​ 493–502.
34(38):​12762–12777. 43. Deuis JR, Dvorakova LS, Vetter I. Methods used to evaluate
28. Geuna S. The sciatic nerve injury model in pre-clinical re- pain behaviors in rodents. Front Mol Neurosci. 2017;​10:​284.
search. J Neurosci Methods. 2015;​243:​39–46. 44. Tos P, Battiston B, Ciclamini D, et al. Primary repair of crush
29. Wolthers M, Moldovan M, Binderup T, et al. Comparative nerve injuries by means of biological tubulization with mus-
electrophysiological, functional, and histological studies of cle-vein-combined grafts. Microsurgery. 2012;​32(5):​358–363.
nerve lesions in rats. Microsurgery. 2005;​25(6):​508–519. 45. Vaughan DW. Effects of advancing age on peripheral nerve
30. Gladman SJ, Huang W, Lim SN, et al. Improved outcome regeneration. J Comp Neurol. 1992;​323(2):​219–237.
after peripheral nerve injury in mice with increased levels 46. Verdú E, Ceballos D, Vilches JJ, Navarro X. Influence of ag-
of endogenous ω-3 polyunsaturated fatty acids. J Neurosci. ing on peripheral nerve function and regeneration. J Peripher
2012;​32(2):​563–571. Nerv Syst. 2000;​5(4):​191–208.
31. Inserra MM, Yao M, Murray R, Terris DJ. Peripheral nerve
regeneration in interleukin 6-deficient mice. Arch Otolaryn-
gol Head Neck Surg. 2000;​126(9):​1112–1116. Disclosures
32. Yao M, Moir MS, Wang MZ, et al. Peripheral nerve regenera- Dr. Mahan has served as a consultant for AxoGen and Joimax.
tion in CNTF knockout mice. Laryngoscope. 1999;​109(8):​
1263–1268. Author Contributions
33. Yao M, Inserra MM, Duh MJ, Terris DJ. A longitudinal,
functional study of peripheral nerve recovery in the mouse. Conception and design: Mahan, Yeoh, Warner. Acquisition of
Laryngoscope. 1998;​108(8 Pt 1):​1141–1145. data: Yeoh, Warner, Eli. Analysis and interpretation of data:
34. Dailey AT, Avellino AM, Benthem L, et al. Complement de- Mahan, Yeoh, Warner. Drafting the article: Mahan, Yeoh,
pletion reduces macrophage infiltration and activation during Warner. Critically revising the article: Mahan, Yeoh, Warner.
Wallerian degeneration and axonal regeneration. J Neurosci. Reviewed submitted version of manuscript: Mahan, Yeoh, Warner.
1998;​18(17):​6713–6722. Approved the final version of the manuscript on behalf of all
35. Zochodne DW, Ho LT. Endoneurial microenvironment and authors: Mahan. Figure production: Yeoh, Warner.
acute nerve crush injury in the rat sciatic nerve. Brain Res.
1990;​535(1):​43–48. Correspondence
36. Nguyen QT, Sanes JR, Lichtman JW. Pre-existing pathways Mark A. Mahan: Clinical Neurosciences Center, The University
promote precise projection patterns. Nat Neurosci. 2002;​5(9):​ of Utah, Salt Lake City, UT. [email protected].
861–867.
37. Moalem-Taylor G, Li M, Allbutt HN, et al. A preconditioning
nerve lesion inhibits mechanical pain hypersensitivity follow-
ing subsequent neuropathic injury. Mol Pain. 2011;​7:​1.
38. Alvey LM, Jones JFX, Tobin-O’Brien C, Pickering M. Bands
of Fontana are caused exclusively by the sinusoidal path of
axons in peripheral nerves and predict axon path;​evidence
from rodent nerves and physical models. J Anat. 2019;​234(2):​
165–178.

J Neurosurg Volume 135 • September 2021 903

 
*46 *
 LABORATORY INVESTIGATION
J Neurosurg 135:904–911, 2021

Femoral nerve decompression and sartorius-to-quadriceps

nerve transfers for partial femoral nerve injury:
a cadaveric study and early case series
Colin W. McInnes, MD,1 Austin Y. Ha, MD,2 Hollie A. Power, MD,3 Thomas H. Tung, MD,2 and
Amy M. Moore, MD4
1
Department of Surgery, Division of Plastic Surgery, Fraser Health Authority/University of British Columbia, New Westminster,
British Columbia, Canada; 2Department of Surgery, Division of Plastic and Reconstructive Surgery, Washington University
School of Medicine in St. Louis, Missouri; 3Department of Surgery, Division of Plastic Surgery, University of Alberta, Edmonton,
Alberta, Canada; and 4Department of Plastic and Reconstructive Surgery, The Ohio State University Wexner Medical Center,
Columbus, Ohio

OBJECTIVE  Partial femoral nerve injuries cause significant disability with ambulation. Due to their more proximal and
superficial location, sartorius branches are often spared in femoral nerve injuries. In this article, the authors report the
benefits of femoral nerve decompression, demonstrate the feasibility of sartorius-to-quadriceps nerve transfers in a ca-
daveric study, describe the surgical technique, and report clinical results.
METHODS  Four fresh-frozen cadaveric lower limbs were dissected for anatomical analysis of the sartorius nerve. In
addition, a retrospective review of patients with partial femoral nerve injuries treated with femoral nerve decompression
and sartorius-to-quadriceps nerve transfers was conducted. Pre- and postoperative knee extension Medical Research
Council (MRC) grades and pain scores (visual analog scale) were collected.
RESULTS  Up to 6 superficial femoral branches innervate the sartorius muscle just distal to the inguinal ligament. Each
branch yielded an average of 672 nerve fibers (range 99–1850). Six patients underwent femoral nerve decompression
and sartorius-to-quadriceps nerve transfers. Four patients also had concomitant obturator-to-quadriceps nerve transfers.
At final follow-up (average 13.4 months), all patients achieved MRC grade 4−/5 or greater knee extension. The average
preoperative pain score was 5.2, which decreased to 2.2 postoperatively (p = 0.03).
CONCLUSIONS  Femoral nerve decompression and nerve transfer using sartorius branches are a viable tool for restor-
ing function in partial femoral nerve injuries. Sartorius branches serve as ideal donors in quadriceps nerve transfers
because they are expendable, are close to their recipients, and have an adequate supply of nerve fibers.
https://thejns.org/doi/abs/10.3171/2020.6.JNS20251
KEYWORDS  femoral nerve injury; nerve to sartorius; nerve decompression; nerve transfer; cadaveric study;
peripheral nerve

uadriceps muscle function is critical for normal, and access to the proximal stump is difficult. The ben-
unassisted ambulation without the use of braces or efits of nerve transfers are well established: they provide
splints. Patients with femoral nerve injury have an a source of viable axons closer to the target muscle than
impeded gait and are often unable to run or climb stairs. nerve grafts, and, compared with tendon transfers, they
Possible etiologies include trauma, malignancy, virus-me- reinnervate the native muscles, leading to an optimized
diated phenomena, or, most commonly, iatrogenic injury.1 length/tension ratio, more natural function, and poten-
Methods to restore quadriceps muscle function in- tially greater strength. Nerve transfers also avoid the
clude direct nerve repair, nerve grafting, tendon transfers, need for extended postoperative immobilization. Obtu-
functional muscle transfers, and prosthetics. Choices for rator branches have been previously described as suit-
intervention and outcomes are dependent on the time able nerve transfer donors for reconstruction of complete
from injury, etiology, and prognosis for recovery.1–7 Nerve femoral nerve palsy.8,9 From a technical standpoint, there
transfers are a viable option when the injury is proximal are two main types of nerve transfers utilized to restore

ABBREVIATIONS  EMG = electromyography; ETE = end-to-end; MRC = Medical Research Council; MUAP = motor unit action potential; ROM = range of motion; SETS =
supercharged end-to-side.
SUBMITTED  January 24, 2020.  ACCEPTED  June 30, 2020.
INCLUDE WHEN CITING  Published online November 6, 2020; DOI: 10.3171/2020.6.JNS20251.

904 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*47 *
 McInnes et al.

motor function. An end-to-end (ETE) nerve transfer is

intuitive; the distal end of the nerve from the donor mus-
cle is coapted to the proximal end of the recipient nerve.
The downside to utilizing an ETE transfer for a partial
nerve injury is that it requires the partially injured nerve
to be transected, eliminating the possibility of any native
recovery within that nerve.
The supercharged end-to-side (SETS) transfer, also
referred to as a reverse end-to-side transfer, is a surgical
technique to augment function of a regenerating nerve
while allowing native regeneration to occur. To date, the
SETS transfer has not yet been described for femoral
nerve injuries.10 SETS transfers involve creating an epi-
neural window and sewing the transferred nerve to the
side of an otherwise intact nerve. Studies have demon-
strated that the transferred nerve will regenerate down the
injured nerve to the new target muscle.11,12 SETS transfers
have also been studied clinically for both acute traumatic
nerve injury and compression neuropathy and are an in-
creasingly utilized option.10–15
In previous femoral nerve decompression and nerve
transfer procedures for femoral nerve injuries, we ob-
served that the sartorius branches were often spared. We
noted this pattern by stimulating branches of the femoral FIG. 1. Incision for femoral nerve decompression and sartorius-to-femo-
nerve with a handheld nerve stimulator. Given the relative ral nerve transfers. Figure is available in color online only.
expendability of the sartorius nerve branches and their
proximity to other femoral nerve branches, we began ex-
ploring their use as donors for restoring quadriceps muscle
function. The objective of this article was to demonstrate senior authors (T.H.T. and A.M.M.) was conducted. Pa-
the importance of femoral nerve decompression and the tients with virus/immune-mediated femoral nerve palsy
feasibility of sartorius-to-quadriceps nerve transfers in a (e.g., acute flaccid myelitis) and patients with less than 3
cadaveric study, describe the surgical technique in patients months of postoperative follow-up were excluded. Preop-
with partial femoral nerve injuries, and report early clini- erative workup included electromyography (EMG) testing
cal results. of the quadriceps muscles to determine the extent of na-
tive recovery, if any. This was completed by neurologists
Methods with fellowship training in electrodiagnostic testing. We
do not perform EMG of the sartorius muscle and instead
Part A: Cadaveric Study rely on intraoperative assessment with a nerve stimulator.
Four fresh-frozen cadaveric lower limbs were dissected Pre- and postoperative Medical Research Council (MRC)
for anatomical analysis of the femoral nerve and its branch- strength and pain scores were evaluated. Descriptive sta-
es to the sartorius. Surface anatomy landmarks, branching tistics were calculated. Normality of continuous variables
patterns, and nerve fiber counts of terminal branches great- was assessed using the Shapiro-Wilk test. A two-tailed,
er than 1 mm in diameter were recorded. The methodol- paired Student t-test was used to compare pre- and post-
ogy of nerve sample processing and histological analysis operative pain scores (version 24.0, IBM SPSS Statistics,
used in this study has been previously reported.16,17 Briefly, IBM Corp.).
the nerves were fixed in 3% glutaraldehyde solution in 0.1
M phosphate-buffered saline (pH 7.2) at 4°C prior to sec- Surgical Technique
tioning. The tissues were then postfixed with 1% osmium The patient is positioned supine on the operating table
tetroxide and serially dehydrated in ethanol. Specimens with the hip in partial external rotation. An 8-cm longitu-
were embedded in Araldite 502 (Polysciences Inc.), sec- dinal incision is marked just below the inguinal crease and
tioned into 1-µm-thick cross-sections, and stained with 1% lateral to the palpable femoral artery (Fig. 1). This corre-
toluidine blue and evaluated via light microscopy. Micro- sponds to a point about two-thirds of the distance from the
scopic images were analyzed with an automated digital im- pubic tubercle to the anterior superior iliac spine. Decom-
age analysis system linked to histomorphometry software pression of the femoral nerve requires release of 3 layers
to quantify the total nerve fiber count. of investing fascia: the superficial fascia, the fascia of the
sartorius muscle, and the femoral sheath. After incising
Part B: Retrospective Case Series through the superficial fascia within the subcutaneous fat,
After obtaining institutional review board approval, dissection continues to the muscular fascia of the sartorius
a retrospective review of all patients with quadriceps muscle, which is incised over the medial border (Fig. 2
muscle weakness treated with femoral nerve decompres- left). The sartorius muscle is then retracted laterally, and
sion and sartorius-to-quadriceps nerve transfers by the its motor nerves can be seen entering the muscle on the

J Neurosurg Volume 135 • September 2021 905

 
*47 *
McInnes et al.

FIG. 2. Left: Fascia overlying the sartorius muscle. Right: Femoral

sheath. Both are incised to expose the femoral neurovascular bundle
and nerve branches to the sartorius muscle. Figure is available in color FIG. 4. Patient 4. Intraoperative photograph showing a right sartorius-
online only. to–vastus medialis SETS nerve transfer. Figure is available in color
online only.

deep surface of its medial border. These nerve branches

are superficial relative to the quadriceps nerve branches.
The femoral sheath is then incised overlying the femo- with a handheld nerve stimulator (Checkpoint Surgical,
ral neurovascular bundle, distal to the inguinal ligament Inc.) to assess function. The sartorius branches must ex-
and lateral to the palpable vessels (Fig. 2 right). The sarto- hibit strong response to stimulation at 0.5 mA and 100 Hz
rius branches are traced proximally to the femoral nerve to be deemed appropriate donors. An ETE nerve trans-
trunk. Exposure and dissection of the femoral vessels fer is completed in cases in which the recipient nerve has
should be avoided, as this may disrupt lymphatic drainage. no response (i.e., no visible contraction) and preoperative
Once the femoral nerve trunk is identified, the fascia over- EMG shows no motor unit action potentials (MUAPs). An
lying the nerve branches proximally and distally along the SETS nerve transfer is completed if there is at least partial
nerve, freeing it from any scar tissue. The inguinal liga- preservation of the recipient function and if there is evi-
ment is identified and retracted. Complete decompression dence of available motor endplates on preoperative EMG
is achieved when it is possible to guide a finger over the in the form of fibrillations and positive sharp waves (Fig.
femoral nerve into the retroperitoneal space (Fig. 3). Re- 4). When possible, we preserve 1–2 sartorius branches to
lease of the inguinal ligament is not required for a com- maintain sartorius function.
plete decompression. To perform SETS transfers, the donor sartorius nerve
The femoral nerve branches are then neurolyzed dis- branches are transected distally and brought to the re-
tally to their insertions. They are individually stimulated cipient nerve branches. The lower extremity is then taken
through a full range of motion (ROM) to ensure tension-
free repair. Epineurial and perineurial windows are then
created in the recipient nerve at these locations. The win-
dow size should match the caliber of the donor nerve. Four
simple interrupted sutures are placed under microscope
magnification using 9-0 nylon suture, and fibrin glue is
used to reinforce the coaptation. A closed-suction drain
is placed and removed at 1 week postoperatively. A knee
immobilizer is used for 3 weeks postoperatively. Weight
bearing is allowed as tolerated. Physical therapy is initi-
ated at the time of initial evaluation of patients with lower-
extremity nerve injuries to ensure supple joints with maxi-
mum passive ROM and continued in the postoperative pe-
riod for donor activation–focused rehabilitation.18

Results
Part A: Cadaveric Study
The femoral nerve was located 63% (range 59%–67%)
FIG. 3. The femoral nerve trunk after complete decompression. The of the distance from the pubic tubercle to the anterior su-
sartorius muscle is being retracted laterally. Figure is available in color perior iliac spine. Five to six nerve branches greater than
online only. 1 mm in diameter innervated the sartorius muscle in its

906 J Neurosurg Volume 135 • September 2021

 
*47 *
 McInnes et al.

Postop Knee Postop

(MRC grade) Score

Pain

3
6

1
Extension

4+ (bilat)
4−
4+

4+

4
RF, VM, VI, VL: weak;

RF, VL: weak; VM, VI:

RF, VM, VI, VL: weak
Intraop Stimulation

strong; VM: weak

RF: strong; VM, VI,

VL/RF: twitch only;

obturator: none
Response

VM, VI: none

RF, RF, VI, VL:

twitch only
VL: weak
reduced MUAPs; VM/VL: 3+ fibs

& PSWs, mildly reduced MUAPs

Bilat VM: 3+ fibs & PSWs, severely
RF, VM, VL: 2+ fibs & PSWs, mod-

duced MUAPs; VM, VL: 2+ fibs

Fibs = fibrillations; FU = follow-up; PSW = positive sharp wave; pt = patient; RF = rectus femoris; VI = vastus intermedius; VL = vastus lateralis; VM = vastus medialis.
RF: 2+ fibs & PSWs, moderately

RF, VM, VL: 2+ fibs & PSWs, no

RF: no fibs & PSWs, mildly re-
RF, VM, VL: 2+ fibs & PSWs,
severely reduced MUAPs

erately reduced MUAPs

All patients underwent femoral nerve decompression and had femoral nerve branches to the sartorius muscle with strong response to electrical stimulation.
Preop EMG

& PSWs, no MUAPs

reduced MUAPs

MUAPs
FIG. 5. Cadaveric dissection of the left thigh showing a typical branch-
ing pattern of the sartorius nerve branches. Figure is available in color
online only.
Preop Knee Preop

(MRC grade) Score

Pain

3
8

3
0 VM, 2 VL,
Extension

proximal third, all branching from the femoral nerve at

3 (bilat)
1 RF

2
2

or just proximal to the inferior aspect of the inguinal liga-

ment (Fig. 5). The first branch innervated the muscle 5
cm (range 4–7 cm) below the inguinal ligament and the
Rt sartorius to VL (×2), RF SETS;

Lt sartorius to VM, VL, RF SETS

most distal branch entered 12 cm (range 11–14 cm) be-

obturator to VM, VI, RF SETS

Lt sartorius to VI, RF ETE; obtu-

Bilat sartorius to VI SETS; obtu-
low. Histological analysis of 25 sartorius nerve branches
Lt sartorius to VL (×2) SETS;
obturator to VM, VI SETS

showed an average of 672 nerve fibers (range 99–1850) in

Rt sartorius to VM SETS
TABLE 1. Summary of clinical cases, workup, intervention, and outcomes

each branch. The sartorius branches were always the most

superficial of the femoral nerve branches. The most dis- rator to VM SETS

rator to VM ETE
Op

tal sartorius branch continues through the sartorius to the

overlying skin as a sensory branch. All branches could be
transposed onto the quadriceps femoral branches without
tension.

Part B: Retrospective Case Series

Op Interval Duration
(mos)
10.5

Between August 2017 and February 2019, 6 patients

FU

12

24
4

24

with partial femoral nerve palsy underwent femoral nerve

decompression and sartorius nerve transfers for knee ex-
Injury to

tension. Patient 5 underwent an identical procedure in the

(mos)
6.5

15

10

8
6

bilateral lower extremities. The data are summarized in

Table 1.
Three of the 6 (50%) patients were male, and the aver-
arthroplasty
Iliopsoas com-
Radiofrequen-

hematoma
cy ablation

age age at time of surgery was 46.8 years (range 14–69

syndrome
partment
Etiology

wound
wound

years). The average interval from injury to surgery was

Iliopsoas
Gunshot
Gunshot

Total hip

8.6 months (range 6–15 months), and the average follow-

up was 13.4 months (range 4–24 months). Etiology of the
injury included gunshot wounds (2/6, 33.3%), iliopsoas
No. (yrs), Sex
Age at

hematoma and compartment syndrome (2/6, 33.3%), and

48, M

26, M

69, M
Injury

65, F

59, F
14, F

iatrogenic (2/6, 33.3%).

The preoperative knee extension MRC grade was 2/5 in
4 patients (66.7%) and 3/5 in 2 patients (33.3%). All had full
Pt

6
1

J Neurosurg Volume 135 • September 2021 907

 
*47 *
McInnes et al.

FIG. 6. Patient 3. A and B: Photographs obtained at 6 months after femoral nerve injury from iliopsoas compartment syndrome,
showing atrophy and weak contraction of his rectus femoris and no contraction of the vastus medialis and lateralis. C: The same
patient at 12 months postoperatively with full extension of the knee with firing of the rectus femoris, vastus medialis, and vastus
lateralis muscles as well as improved muscle bulk in all quadriceps muscles. Figure is available in color online only.

passive ROM but required bracing, and 4 required assist its function in the setting of a femoral nerve palsy. Nerve
devices (crutches and/or wheeled walker) for ambulation transfers offer several advantages over tendon transfers,
preoperatively. All patients achieved MRC grade 4−/5 or including the potential to reinnervate the native quadri-
greater postoperatively and subjectively noted improved ceps for ultimately stronger and more natural function,
strength, increased muscle bulk, and more natural gait reduced donor site morbidity, and shorter immobiliza-
over time (Fig. 6). None were using braces or assist de- tion postoperatively.6 Our group has previously reported
vices at the final follow-up. The average preoperative pain early results of obturator-to-femoral nerve transfers.9 In
score was 5.2 (range 3–8), which decreased to 2.2 (range this article, we demonstrate the feasibility of sartorius-to-
0–6) postoperatively (p = 0.03). Preoperative EMG was quadriceps nerve transfers in a cadaveric study, describe
completed in all patients and demonstrated fibrillations the surgical technique for femoral nerve decompression
and positive sharp waves in one or more of the quadriceps and nerve transfers in patients with partial femoral nerve
muscle groups. The presence of fibrillations and positive injuries, and report clinical results.
sharp waves indicates the availability of unoccupied mo-
tor endplates that can be reinnervated.10–12,14 In all patients, Anatomy of the Sartorius Nerve Branches
EMG testing also showed reduced, but present, MUAPs in Cadaver dissection revealed 5–6 branches of the femo-
one or more quadriceps muscles, indicating a partial nerve ral nerve to the sartorius muscle with a diameter of 1 mm
injury. or greater, which is consistent with the findings of Tanaka
The results of intraoperative stimulation can be found et al.19 Nerve branches innervated the muscle on the deep
in Table 1. A strong response of the sartorius muscle is a surface of its medial border and were accompanied by
requirement for its use as a donor, and this was present in small vascular branches. A previous study demonstrated
all 6 patients. In cases in which there was at least a weak that the sartorius muscle was innervated by 1–2 main
response to electrical stimulation, an SETS nerve transfer branches arising from the femoral nerve with an average
was completed to preserve and augment the native mo- length of 10 cm (range 5–14 cm) and an average diameter
tor nerve function. In patient 6, whose quadriceps muscle of 1.7 mm (range 0.7–2.5 mm).20 More recently, Buckland
group had no response to electrical stimulation, an ETE et al. reported there were 1 (40%) or 2 (60%) main nerve
nerve transfer was completed. In 4 patients, obturator-to- branches that arborized closer to the muscle,21 findings
quadriceps nerve transfers were also performed. Patient 2 similar to our study and others.19,22
was not a candidate for this transfer, as her obturator nerve
was also injured and did not respond to electrical stimula-
tion. Patient 4 had a strong response in all quadriceps mus- Suitability of the Sartorius as a Donor Nerve
cles except for the vastus medialis, and the sartorius trans- The sartorius muscle is a secondary hip flexor, thigh
fer alone was deemed adequate to augment its function. adductor, and knee flexor, and is widely considered ex-
pendable.21 It is commonly used in reconstructive surgery
as a muscle flap for small groin wounds, often following
Discussion vascular bypass surgery. Although the sartorius muscle
Intact quadriceps muscle function is essential for unas- contributes to knee flexion through its insertion at the
sisted, independent ambulation without the use of splints, pes anserinus, it has been used successfully for enhanc-
braces, or other devices. As such, it is critical to restore ing knee extension as a tendon transfer.6 It was found to

908 J Neurosurg Volume 135 • September 2021

 
*47 *
 McInnes et al.

adapt and function in phase with the remaining quadri- in function and especially muscle bulk are more likely a
ceps muscles because of its common origin of innervation result of nerve transfers, as the regeneration of nerve fibers
from the femoral nerve.23 This finding supports its use as reinnervates previously unoccupied motor endplates, lead-
a donor for quadriceps reinnervation and in many ways ing to muscle contraction and hypertrophy.
facilitates the donor activation–focused rehabilitation after
the nerve transfer.18 Treatment Algorithm
Potentially due to its more proximal and superficial Our treatment algorithm for all nonsharp injuries to
branching location, sartorius branches are often spared the femoral nerve, such as blunt trauma, thermal inju-
in partial femoral nerve injuries. They are also close to ries, and gunshot wounds, typically involves a period of
the quadriceps branches, which allows for a graft-free and monitoring for spontaneous recovery with serial physi-
tension-free coaptation through all ranges of motion of the cal examination and electrodiagnostic testing starting at
lower extremity. Lastly, our histological analysis suggests 3 months postinjury. If there is delayed return of function
that the sartorius branches supply enough nerve fibers to or no evidence of spontaneous recovery (e.g., no MUAPs)
power the larger quadriceps muscles. In this sample of 25 on electrodiagnostic testing, we plan for operative inter-
sartorius nerve branches, an average of 672 nerve fibers vention by 6 months. At minimum, all patients undergo
(range 99–1850) were present in each branch. This is com- decompression of the femoral nerve. Whether they are
parable to the 1019 fibers present in the branch to the vas- candidates for nerve transfers, and if so, which type of
tus medialis muscle and 455 fibers present in the branch transfer, depends on findings on their preoperative EMG
to the vastus lateralis muscle, as reported by Moore et al.16 study and intraoperative nerve stimulation, as stated
This suggests that there are abundant axons available to re- above.
store partial femoral nerve injuries. If none of the quadri- Based on our understanding of nerve transfers in the
ceps muscles stimulate after decompression and sartorius upper extremity, we interpret the presence of fibrillations
branches are preserved, they can be utilized in conjunction and positive sharp waves on EMG as evidence that the
with obturator nerve branches to add additional axons to muscle has unoccupied motor endplates that can be rein-
the muscles dedicated to knee extension. If there is a par- nervated with an increased supply of regenerating nerve
tial femoral nerve injury (e.g., clinical weakness, signs of fibers; therefore, their presence is a requirement for pur-
limited recovery to specific branches), sartorius branches suing nerve transfers.10–12,14 In contrast, the presence of
may be utilized alone in an end-to-side fashion with no MUAPs is evidence that there is spontaneous recovery
morbidity. of the injured nerve, and response to intraoperative nerve
stimulation confirms the presence of intact nerve fibers
Clinical Results (Video 2).
The patients in this series all had substantial quadri- VIDEO 2. Patient 3. Intraoperative stimulation 3 following femoral
ceps muscle weakness preoperatively that significantly nerve decompression showing a strong response in the rectus
improved following femoral nerve decompression and femoris muscle but a weak response in the vastus medialis and
sartorius-to-quadriceps nerve transfers. Four patients also lateralis. Copyright Amy M. Moore. Published with permission. Click
underwent concomitant obturator-to-quadriceps nerve here to view.
transfers. The average interval from injury to surgery was In such cases, an SETS transfer is performed to pre-
9 months, suggesting that these patients had meaningful serve and augment the regenerating native motor nerve.
knee extension that failed to recover spontaneously, and In cases in which MUAPs and response to intraoperative
the improvement seen postoperatively is due to the sur- stimulation are absent, an ETE transfer is performed. Our
gical intervention. Postoperatively, all patients achieved a treatment algorithm is summarized in Fig. 7.
knee extension MRC grade of 4−/5 or greater and noted
improved strength, increased muscle bulk, and more natu- Limitations
ral gait over time (Video 1). The main limitations of this study are its small sample
VIDEO 1. Patient 3. Part 1: The patient has an abnormal preopera- size and retrospective design. The patients in this series
tive gait and weak contraction of the rectus femoris only. Part 2: also underwent a combination of interventions, namely,
The same patient jogging at 12 months postoperatively with restora- femoral nerve decompression in conjunction with nerve
tion of more normal gait and improved strength. Copyright Amy M. transfer(s), making it difficult to isolate the effect of any
Moore. Published with permission. Click here to view. one of these interventions. Many patients also received
The decrease in postoperative pain scores was also sta- multiple nerve transfers including both the sartorius and
tistically significant. These improvements were seen de- obturator nerves. Therefore, the findings of this case series
spite the variability in the etiology of femoral nerve injury should not be construed as conclusive evidence for their
and patient age and sex. effectiveness. However, this is a technique based on prin-
We attribute improvement of function in our patients ciples established at the basic science level and utilized in
to both the femoral nerve decompression and the nerve the upper extremity, and we are encouraged by the signifi-
transfers; however, it is difficult to ascertain the exact con- cant improvement in function and pain experienced by our
tribution of each component. A subgroup of patients, more patients. We are in the process of building a prospective
specifically, patient 4 in this series, experienced an imme- database that includes more objective outcome measures
diate difference postoperatively. This is likely due to the at defined time intervals, including PROMIS (Patient-
resolution of ischemic conduction block within the femo- Reported Outcomes Measurement Information System)
ral nerve. In contrast, sustained long-term improvement scores.
J Neurosurg Volume 135 • September 2021 909
 
*47 *
McInnes et al.

FIG. 7. Our treatment algorithm for partial femoral nerve injuries. EDX = electrodiagnostic studies; fibs = fibrillations; PSW =
positive sharp wave. *Electrodiagnostic studies have limitations with regard to the subjectivity and skill of the electrodiagnostician.
Intraoperative confirmation with neurophysiology can be used to confirm the recovery and activity of specific muscles.

Conclusions   5. MacArthur IR, McInnes CW, Dalke KR, et al. Patient report-
ed outcomes following lower extremity soft tissue sarcoma
Femoral nerve injuries cause significant morbidity, resection with microsurgical preservation of ambulation. J
and every effort should be made to restore its function. Reconstr Microsurg. 2019;​35(3):​168–175.
Branches to the sartorius are often spared in femoral nerve   6. Pritsch T, Malawer MM, Wu C-C, et al. Functional recon-
palsies and serve as ideal donors due to their expendability, struction of the extensor mechanism following massive tumor
proximity to the quadriceps branches, and adequate supply resections from the anterior compartment of the thigh. Plast
Reconstr Surg. 2007;​120(4):​960–969.
of nerve fibers. Furthermore, postoperative outcomes can   7. Willcox TM, Smith AA, Beauchamp C, Meland NB. Func-
be maximized by the relatively intuitive donor-activated tional free latissimus dorsi muscle flap to the proximal lower
rehabilitation. extremity. Clin Orthop Relat Res. 2003;​410(410):​285–288.
  8. Campbell AA, Eckhauser FE, Belzberg A, Campbell JN.
Obturator nerve transfer as an option for femoral nerve
Acknowledgments repair:​case report. Neurosurgery. 2010;​66(6)(Suppl
We acknowledge Daniel Hunter for his assistance with histo- Operative):​375.
logical analysis.   9. Tung TH, Chao A, Moore AM. Obturator nerve transfer for
femoral nerve reconstruction:​anatomic study and clinical
application. Plast Reconstr Surg. 2012;​130(5):​1066–1074.
References 10. Davidge KM, Yee A, Moore AM, Mackinnon SE. The super-
  1. Kim DH, Murovic JA, Tiel RL, Kline DG. Intrapelvic and charge end-to-side anterior interosseous-to-ulnar motor nerve
thigh-level femoral nerve lesions:​management and outcomes transfer for restoring intrinsic function:​clinical experience.
in 119 surgically treated cases. J Neurosurg. 2004;​100(6):​ Plast Reconstr Surg. 2015;​136(3):​344e–352e.
989–996. 11. Kale SS, Glaus SW, Yee A, et al. Reverse end-to-side nerve
  2. Hallock GG. Restoration of quadriceps femoris function with transfer:​from animal model to clinical use. J Hand Surg Am.
a dynamic microsurgical free latissimus dorsi muscle trans- 2011;​36(10):​1631–1639.e2.
fer. Ann Plast Surg. 2004;​52(1):​89–92. 12. Farber SJ, Glaus SW, Moore AM, et al. Supercharge nerve
  3. Ihara K, Shigetomi M, Kawai S, et al. Functioning muscle transfer to enhance motor recovery:​a laboratory study. J
transplantation after wide excision of sarcomas in the ex- Hand Surg Am. 2013;​38(3):​466–477.
tremity. Clin Orthop Relat Res. 1999;​(358):​140–148. 13. Power HA, Kahn LC, Patterson MM, et al. Refining indica-
  4. Kuhlmann RF, Bell JF. A clinical evaluation of tendon trans- tions for the supercharge end-to-side anterior interosseous to
plantations for poliomyelitis affecting the lower extremities. J ulnar motor nerve transfer in cubital tunnel syndrome. Plast
Bone Joint Surg Am. 1952;​34 A(4):​915–926. Reconstr Surg. 2020;​145(1):​106e–116e.

910 J Neurosurg Volume 135 • September 2021

 
*47 *
 McInnes et al.

14. Baltzer H, Woo A, Oh C, Moran SL. Comparison of ulnar 23. Malawer MM, Kellar-Graney K. Soft tissue reconstruction
intrinsic function following supercharge end-to-side anterior after limb-sparing surgery for tumors of the upper and lower
interosseous-to-ulnar motor nerve transfer:​a matched cohort extremities. Oper Tech Orthop. 2004;​14(4):​276–287.
study of proximal ulnar nerve injury patients. Plast Reconstr
Surg. 2016;​138(6):​1264–1272.
15. Koriem E, El-Mahy MM, Atiyya AN, Diab RA. Compari- Disclosures
son between supercharged ulnar nerve repair by anterior Dr. Moore: Consultant for Checkpoint Surgical, Inc.
interosseous nerve transfer and isolated ulnar nerve repair in
proximal ulnar nerve injuries. J Hand Surg Am. 2020;​45(2):​ Author Contributions
104–110.
16. Moore AM, Krauss EM, Parikh RP, et al. Femoral nerve Conception and design: Moore, McInnes, Power, Tung. Acquisi-
transfers for restoring tibial nerve function:​an anatomical tion of data: all authors. Analysis and interpretation of data: all
study and clinical correlation:​a report of 2 cases. J Neuro- authors. Drafting the article: McInnes, Ha. Critically revising the
surg. 2018;​129(4):​1024–1033. article: all authors. Reviewed submitted version of manuscript: all
17. Hunter DA, Moradzadeh A, Whitlock EL, et al. Binary imag- authors. Approved the final version of the manuscript on behalf
ing analysis for comprehensive quantitative histomorphom- of all authors: Moore. Statistical analysis: Ha, Power. Administra-
etry of peripheral nerve. J Neurosci Methods. 2007;​166(1):​ tive/technical/material support: Moore, Power. Study supervision:
116–124. Moore, Tung.
18. Kahn LC, Moore AM. Donor activation focused rehabilita-
tion approach:​maximizing outcomes after nerve transfers. Supplemental Information
Hand Clin. 2016;​32(2):​263–277. Videos
19. Tanaka C, Ide MR, Junior AJR. Anatomical contribution to Video 1. https://vimeo.com/440969013.
the surgical construction of the sartorius muscle flap. Surg Video 2. https://vimeo.com/440968782.
Radiol Anat. 2006;​28(3):​277–283.
20. Yang D, Morris SF, Sigurdson L. The sartorius muscle:​ana- Previous Presentations
tomic considerations for reconstructive surgeons. Surg Ra-
diol Anat. 1998;​20(5):​307–310. This study was previously presented as a podium presentation
21. Buckland A, Pan WR, Dhar S, et al. Neurovascular anatomy using Microsoft PowerPoint slides at the Canadian Society of
of sartorius muscle flaps:​implications for local transposition Plastic Surgeons 72nd Annual Meeting, Jasper, Alberta, Canada,
and facial reanimation. Plast Reconstr Surg. 2009;​123(1):​ June 19–23, 2018.
44–54.
22. Yu D-Z, Liu A-T, Dang R-S, et al. Intramuscular innervations Correspondence
of muscle flaps that are commonly used in clinical settings. Amy M. Moore: The Ohio State University, Columbus, OH.
Surg Radiol Anat. 2010;​32(7):​637–646. [email protected].

J Neurosurg Volume 135 • September 2021 911

 
*47 *
 LITERATURE REVIEW
J Neurosurg 135:912–922, 2021

Meralgia paresthetica treated by injection, decompression,

and neurectomy: a systematic review and meta-analysis of
pain and operative outcomes
Victor M. Lu, MD, PhD,1 S. Shelby Burks, MD,1 Rainya N. Heath, MS,1 Tizeta Wolde, MPH,1
Robert J. Spinner, MD,2 and Allan D. Levi, MD, PhD1
1
Department of Neurological Surgery, University of Miami Miller School of Medicine, Miami, Florida; and 2Department of
Neurologic Surgery, Mayo Clinic, Rochester, Minnesota

OBJECTIVE  Meralgia paresthetica is caused by entrapment of the lateral femoral cutaneous nerve (LFCN) and often
presents with pain. Multiple treatment options targeting the LFCN can be pursued to treat the pain should conservative
measures fail, with the most common options being injection, neurolysis, and neurectomy. However, their efficacy in
causing pain relief and their clinical outcomes have yet to be directly compared. The aim of this study was to interrogate
the contemporary literature and quantitatively define how these options compare.
METHODS  The electronic databases Ovid Embase, PubMed, SCOPUS, and the Cochrane Library were interrogated
from inception to May 2020 following the PRISMA guidelines. Candidate articles were screened against prespecified
criteria. Outcome data were abstracted and pooled by random-effects meta-analysis of proportions.
RESULTS  There were 25 articles that satisfied all criteria, reporting outcomes for a total of 670 meralgia paresthetica
patients, with 78 (12%) treated by injection, 496 (74%) by neurolysis, and 96 (14%) by neurectomy. The incidence of com-
plete pain relief was 85% (95% CI 71%–96%) after neurectomy, 63% (95% CI 56%–71%) after neurolysis, and 22% (95%
CI 13%–33%) after injection, which were all statistically different (p < 0.01). The incidence of revision procedures was
12% (95% CI 4%–22%) after neurolysis and 0% (95% CI 0%–2%) after neurectomy, which were significantly lower than
81% (95% CI 64%–94%) after injection (p < 0.01). The incidences of treatment complications were statistically compa-
rable across all three treatments, ranging from 0% to 5% (p = 0.34).
CONCLUSIONS  There are multiple treatment options to target pain in meralgia paresthetica. The incidence of complete
pain relief appears to be the greatest among the 3 interventions after neurectomy, accompanied by the lowest incidence
of revision procedures. These findings should help inform patient preference and expectations. Greater exploration of the
anatomical rationale for incomplete pain relief after surgical intervention will assist in optimizing further surgical treatment
for meralgia paresthetica.
https://thejns.org/doi/abs/10.3171/2020.7.JNS202191
KEYWORDS  meralgia paresthetica; lateral femoral cutaneous; injection; neurolysis; decompression; neurectomy;
surgery; peripheral nerve; pain

M eralgiaparesthetica (Greek, with µηροs mean- the lateral femoral cutaneous nerve (LFCN), which most
ing thigh, αλγοs meaning pain) is a relatively commonly travels beneath the lateral portion of the ingui-
rare, spontaneous entrapment syndrome charac- nal ligament near the anterior superior iliac spine (ASIS).3
terized by pain, paresthesia, and abnormal skin sensation Pain in this syndrome can be disabling to a patient, and
on the anterolateral surface of the thigh.1 This syndrome often initial efforts to address the pain are conservative
more frequently occurs in middle-aged adults and has no in nature, which include pharmacological and physiother-
particular sex predilection, and estimates suggest an in- apeutic means with highly variable degrees of success.1
cidence of 0.4 per 10,000 population.2 It has been estab- Should these conservative managements fail, however,
lished that this presentation results from compression of other interventional treatment options are available to tar-

ABBREVIATIONS  ASIS = anterior superior iliac spine; GRADE = Grading of Recommendations, Assessment, Development and Evaluation; LFCN = lateral femoral cutane-
ous nerve; MOOSE = Meta-analysis of Observational Studies in Epidemiology; PICOS = Population, Intervention, Comparison, Outcome, Study type; PRISMA = Preferred
Reporting Items for Systematic Reviews and Meta-Analyses.
SUBMITTED  June 7, 2020.  ACCEPTED  July 31, 2020.
INCLUDE WHEN CITING  Published online January 15, 2021; DOI: 10.3171/2020.7.JNS202191.

912 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*48 *
 Lu et al.

get the LFCN, including steroid injection, decompression following injection were considered only after single in-
by means of surgical neurolysis, and nerve transection by jection, rather than after a series or number of injections
means of surgical neurectomy. over an extended period of time. The purpose of injection,
The risk of treatment complications for meralgia pares- temporizing or curative, did not impact inclusion. Multiple
thetica is low, and indeed with respect to serious morbidity, injections in a study were considered revision procedures.
rare.4 As such, outcomes that may guide intervention se- When there were duplicate studies with overlapping co-
lection include patient quality of life, which would incor- horts from the same institutions, only the most complete
porate pain relief, given that pain is one of the most com- study was included. Studies were limited to English-lan-
mon reasons for presentation.1 Consensus trends among guage publications; database studies, review articles, con-
the contemporary literature have yet to be established for ference abstracts or presentations, and editorials or expert
meralgia paresthetica patients treated by these interven- opinions were not included.
tions. To date, there has only been one attempt5 at agglom-
erating the metadata in the literature, and, although the Data Handling
authors’ interpretation of the limited literature at the time Data could be abstracted from text, tables, and figures
was that neurectomy could confer superior pain results of all studies, with the incidence of events being the pri-
to neurolysis, they were not able to attempt to quantify mary summary statistics of this study. Incidence rates
specific pain outcomes, nor did they assess the efficacy of were calculated first with initial variance using Fisher’s
injection therapy in the overall treatment paradigm. Con- exact test for binomial data and followed by Freeman-
sequently, the aim of this study was to interrogate the con- Tukey transformation to stabilize any variances.7 Statis-
temporary literature and analyze its metadata describing tics were pooled by means of a meta-analysis of propor-
pain outcomes of meralgia paresthetica patients treated by tions using the random-effects model described by Der-
injection, neurolysis, and neurectomy, to assist in the treat- Simonian and Laird.8 As these pooled estimates relied on
ment selection and formation of patient expectations. study weighting, sums of different pain outcomes by the
same treatment were not expected to sum to 100% exactly.
Methods Heterogeneity was assessed using I2 values.9 Linear meta-
regression was used to evaluate trends across incidence
Search Strategy
rates against study-level characteristics. All p values were
According to the Population, Intervention, Compari- 2-sided, with significance defined using an alpha threshold
son, Outcome, Study type (PICOS) question format, our of 0.05. All statistical analyses were conducted using Stata
search strategy design was as follows: “In patients with 14.1 (StataCorp).
meralgia paresthetica (Population) treated with 1) injec-
tion, 2) surgical decompression, or 3) neurectomy (Inter- Quality, Bias, and Certainty Assessments
ventions & patient-reported outcomes (Study type)?” This
review was conducted in accordance with the Preferred Each article was appraised using the Meta-analysis of
Reporting Items for Systematic Reviews and Meta-Analy- Observational Studies in Epidemiology (MOOSE) assess-
ses (PRISMA) guidelines.6 Electronic searches were per- ment10 to determine the quality of the original study de-
formed using Ovid Embase, PubMed, SCOPUS, Cochrane sign. Publication bias was assessed via the generation of a
Central Register of Controlled Trials, Cochrane Database funnel plot, and small study biases were assessed by Eg-
of Systematic Reviews, American College of Physicians ger’s linear regression test and Begg’s correlation tests.11,12
Journal Club, and Database of Abstracts of Review of Ef- This was only performed in the cases of groups with ≥ 10
fectiveness, from their dates of inception to May 2020. contributing studies. A trim-and-fill method was used to
Database searches were completed using the following recalculate the pooled summary statistic if any bias was
search string: (“meralgia paresthetica” OR “lateral femo- suspected.13 The certainty of the pooled outcomes was
ral cutaneous nerve”) AND (“decompression” OR “neu- then evaluated using the Grading of Recommendations,
rolysis” OR “neurectomy” OR “injection”). Assessment, Development and Evaluation (GRADE) cri-
teria.14
Selection Criteria
All articles were preliminarily screened against prede- Results
termined selection criteria independently by two investi- Search Strategy
gators (R.N.H. and T.W.) to identify all candidate articles. Our initial search identified 821 articles to evaluate.
Any discrepancy was resolved by discussion. Inclusion There were 295 duplicate citations, which were removed.
criteria were patients 1) with clinically diagnosed meralgia Titles and abstracts of the remaining articles were then
paresthetica; 2) managed by injection, neurolysis, or neu- screened against the selection criteria (Fig. 1). There were
rectomy; 3) for whom pain outcomes were categorizable 38 articles that appeared to potentially satisfy all criteria
into a three-tier response (complete, partial, or no relief); based on title and abstract, and then their full texts were
and 4) with age ≥ 18 years. When pain response was cat- individually evaluated. Ultimately, a total of 25 articles
egorized, complete relief and no relief corresponded to the describing 33 distinct treatment cohorts published be-
highest and lowest categories used, with partial response tween 1988 and 2019 satisfied all criteria for inclusion in
deemed for all intermediate categories in between. When our study (Table 1). For treatment by injection, there were
pain response was quantified, these tiers corresponded to 6 retrospective cohorts;4,15–19 for treatment by neurolysis,
< 20%, 20%–80%, and > 80% pain response. Outcomes there were 16 retrospective cohorts,4,16,20–33 2 prospective

J Neurosurg Volume 135 • September 2021 913

 
*48 *
Lu et al.

cohorts,34,35 and 1 randomized cohort;36 and for treatment

by neurectomy, there were 5 retrospective cohorts,4,23,24,37,38
2 prospective cohorts,34,35 and 1 randomized cohort.36

Demographics and Clinical Features

There was a total of 670 meralgia paresthetica patients
across all 33 cohorts, with 352 (53%) males, a median co-
hort age of 49 years (range 11–70 years), and a median
follow-up of 25 months (range 1.3–200 years) (Table 1).
When reported, the median symptom duration before in-
tervention was 2.6 years (range 0.4–6.4 years), with the
paresthetica being bilateral in 8% of cases and idiopathic
in 64% of cases. In terms of treatment, 78 (12%) were
treated by injection, 496 (74%) by neurolysis, and 96 (14%)
by neurectomy, with statistically comparable demograph-
ics and clinical features between them. Within the injec-
tion group, all treatments were ultrasound guided, and the
steroid agents used were betamethasone, dexamethasone
methylprednisolone, and triamcinolone (Table 2).
FIG. 1. Search strategy results as per PRIMSA guidelines. Figure is
Pain Relief available in color online only.
Following injection, the incidence of complete pain re-
lief was estimated to be 22% (95% CI 13%–33%; I2 = 0%;
p = 0.55); partial pain relief, 37% (95% CI 12%–67%; I2 =
84%; p < 0.01); and no pain relief, 31% (95% CI 4%–67%; rectomy, 0% (95% CI 0%–2%; I2 = 0%; p > 0.99) from 6
I2 = 89%; p < 0.01) from 6 cohorts. Following neurolysis, cohorts (Table 2). All revision procedures following single
the incidence of complete pain relief was estimated to be injection involved either repeat injection or neurolysis.
63% (95% CI 56%–71%; I2 = 54%; p < 0.01); partial pain Overall, the incidence of revision procedures was statisti-
relief, 22% (95% CI 18%–26%; I2 = 4%; p = 0.41); and no cally significantly greater after injection, followed by neu-
pain relief, 8% (95% CI 3%–15%; I2 = 74%; p < 0.01) from rolysis and then neurectomy (p < 0.01) (Fig. 3).
19 cohorts. Following neurectomy, the incidence of com-
plete pain relief was estimated to be 85% (95% CI 71%– Meta-Regression for Complete Pain Relief
96%; I2 = 53%; p = 0.04); partial pain relief, 11% (95% CI The potential of design parameters to impact the
4%–20%; I2 = 11%; p = 0.35); and no pain relief, 2% (95% trends of complete pain relief was investigated by means
CI 0%–8%; I2 = 25%; p = 0.23) based on 8 cohorts. of meta-regression of individual rates against component
Overall, complete pain relief was statistically signifi- study characteristics. For each modality, the study size,
cantly more common after neurectomy, followed by neu- age, symptom duration, and follow-up duration, as well as
rolysis and then injection (p < 0.01) (Fig. 2). There were no proportions of males, bilateral presentation, and idiopathic
statistically significant differences between modalities in meralgia paresthetica, did not significantly influence the
terms of partial pain relief (p = 0.06) (Supplementary Fig.
rates of complete pain relief (Supplementary Table 1).
1). No pain relief was statistically significantly least com-
mon after neurectomy, followed by neurolysis and injec-
tion (p = 0.04) (Supplementary Fig. 2). Quality Assessment
With respect to study design, answering our PICOS
Complications questions, 4 of 6 (67%) injection cohorts, 19 of 19 (100%)
The incidence of complications following injection was neurolysis cohorts, and 7 of 8 (88%) neurectomy cohorts
estimated to be 0% (95% CI 0%–6%; I2 = NA) from 2 co- were deemed moderate and high quality per the MOOSE
horts; neurolysis, 5% (95% CI 2%–8%; I2 = 9%; p = 0.36) criteria (Supplementary Table 2).
from 10 cohorts; and neurectomy, 0% (95% CI 0%–32%;
I2 = NA) from 1 cohort. The most common complications Bias Assessment
were hematoma and wound infection (Table 2). Overall, in Risks of bias were assessed for neurolysis, as it was the
terms of complications, there were no statistically signifi- only modality with ≥ 10 individual cohorts. Asymmetry
cant differences between modalities (p = 0.34) (Supple- of the funnel plot was suspected in the case of revision
mentary Fig. 1). procedures only. A trim-and-fill approach to this outcome
showed a higher incidence of 18% (95% CI 9%–37%; p
Revision Procedures < 0.01), indicating that this outcome may have been un-
The incidence of revision procedures following single derestimated by possible publication bias. This suspicion
injection was estimated to be 81% (95% CI 64%–94%; I2 was validated by positive Egger’s (p = 0.04) and Begg’s (p
= 48%; p = 0.11) from 5 cohorts; neurolysis, 12% (95% CI = 0.04) tests, which were then incorporated into the cer-
4%–22%; I2 = 79%; p < 0.01) from 16 cohorts; and neu- tainty assessment of this outcome.

914 J Neurosurg Volume 135 • September 2021

*48 *
 TABLE 1. Characteristics and demographics of all included studies
Authors Years of Design (no. No. of No. of Mean Age in Mean Sx Duration, No. of No. of Mean Follow-Up
& Year Country Study of institutions) Pts Males (%) Yrs (range) Yrs (range) Bilat (%) Idiopathic (%) in Mos (range)
Injection
  Ahmed et al., 201615 India 2014–2015 ROCS (1) 6 2 (33) 46 ± 10 0.4 (0.3–0.5) 6 (100) 3
  Edelson & Stevens, 199416 US 1985–1994 ROCS (30) 14 8 (57) 11 (3–17) 3.2 (2.1–5.0) 35 (4–60)
  Elavarasi et al., 201917 India 2013–2017 ROCS (1) 8 4 (50) 51 1 (13) 25 (4–41)
  Ivins, 20004 US 1992–1996 ROCS (1) 15 6 (40) 48 1 (7) 44 (28–60)
  Jiang & Xu, 198818 China NR ROCS (1) 15 15 (100) 44 18 (10–32)
  Klauser et al., 201619 Austria 2008–2013 ROCS (1) 20 9 (45) 61 (46–75) 12
  Total (%) or median 78 44 (56) 47 1.8 2/23 (9) 6/6 (100) 22
Neurolysis
  Alberti et al., 200920 Germany 1989–2004 ROCS (1) 55 33 (60) 50 ± 13 2.7 ± 2.9 7 (13) 38 (6–156)
  Ataizi et al., 201921 Turkey 2012–2017 ROCS (1) 13 4 (31) 59 (33–73) 0 7 (54) 38 (26–55)
  Benezis et al., 200722 France 1987–2003 ROCS (1) 167 107 (64) 52 (17–80) 3.8 (0–25) 7 (4) 105 (62) 98 (20–212)
  de Ruiter et al., 201223 The Netherlands 1999–2010 ROCS (1) 10 8 (80) 51 ± 11 2.6 1 (10) 10 (100) 1.3
  de Ruiter et al., 201534 The Netherlands 2012–2014 POCS (1) 8 5 (63) 50 2.9 0 2
  Ducic et al., 200624 US 2002–2005 ROCS (1) 29 12 (41) 49 1.5 (0.5–2.5) 4 (14) 10 (33) 16
  Edelson & Stevens, 199416 US 1985–1994 ROCS (30) 13 8 (62) 11 (3–17) 3.2 (2.1–5.0) 6 (46) 35 (4–60)
  Emamhadi, 201235 Iran 2000–2008 POCS (1) 5 4 (80) 43.8 1.7 (0.5–4.0) 0 5 (100) 18
  Hanna, 201925 US 2011–2016 ROCS (1) 19 7 (37) 49 (25–72) 0 19 (100) 12
  Ivins, 20004 US 1992–1996 ROCS (1) 4 0 43 (36–58) 0 60 (36–72)
  Macnicol & Thompson, 199026 UK 1971–1985 ROCS (1) 25 14 (56) 46 (22–77) 2.1 (0.3–5.0) 25 (100) 66 (24–180)
  Malessy et al., 201927 The Netherlands 2013–2017 ROCS (1) 17 5 (29) 50 (27–61) 5.1 (0.8–19) 2 (12) 30 (22–39)

*48 *
  Morimoto et al., 201828 Japan 2015–2017 ROCS (1) 12 3 (25) 70 (62–75) 1.8 (0.3–7.0) 12 (100) 19 (12–27)
  Nahabedian & Dellon, 199529 US NR ROCS (1) 23 7 (30) 37 (18–54) 2.7 (0.5–9.0) 3 (13) 1 (4) 9 (2–33)
  Nouraei et al., 200730 UK 2000–2005 ROCS (1) 20 10 (50) 47 ± 12 2.2 ± 1.8 2 (10) 25 (7–63)
  Schwaiger et al., 201831 Australia 2015–2016 ROCS (1) 13 7 (54) 51 (17–70) 4 (31) 12 (4–19)
  Siu & Chandran, 200532 Australia 1996–2000 ROCS (1) 42 21 (50) 53 ± 13 2.6 (0.3–18) 48
 
  Son et al., 201233 Korea 2003–2010 ROCS (1) 11 5 (45) 58 ± 8 0.7 ± 0.3 33 (12–60)
  van Eerten et al., 199536 The Netherlands 1974–1992 RCT (1) 10 4 (40) 40 (21–71) 4.1 (0.5–30) 46 (10–155)
  Total (%) or median 496 264 (53) 50 2.6 32/395 (8) 198/316 (63) 30
Neurectomy
  Berini et al., 201437 US NR ROCS (1) 7 4 (57) 44 (27–78) 6.4 (2–15) 1 (14) 5 (0.5–31)
  de Ruiter et al., 201223 The Netherlands 1999–2010 ROCS (1) 8 4 (50) 51 ± 11 1.7 1 (13) 8 (100) 8
  de Ruiter et al., 201534 The Netherlands 2012–2014 POCS (1) 14 9 (64) 59 4.6 1 (7) 2
  Ducic et al., 200624 US 2002–2005 ROCS (1) 19 7 (37) 49 1.5 (0.5–2.5) 3 (16) 6 (32) 16

J Neurosurg Volume 135 • September 2021

  Emamhadi, 201235 Iran 2000–2008 POCS (1) 9 3 (33) 48 1.2 (0.8–2.0) 0 9 (100) 18
CONTINUED ON PAGE 916 »

915
Lu et al.
Lu et al.

Certainty Assessment

Mean Follow-Up
Idiopathic (%) in Mos (range)
The certainty of each outcome was assessed using the

200 (48–300)
116 (24–192)
60 (36–72)
GRADE criteria14 (Table 3). Certainties in pain outcomes

17
following neurectomy were broadly more certain than
those after neurolysis and injection, due to their greater
degree of consistency and narrower effect sizes. Similarly,
certainties of complication outcomes were higher follow-

23/36 (64)
ing neurectomy than neurolysis and injection. With re-
No. of

spect to revision procedures, these outcomes were deemed

to confer very low certainty following neurolysis and in-
jection due to their less common and more heterogeneous

NR = not reported; POCS = prospective observational cohort study; pts = patients; RCT = randomized controlled trial; ROCS = retrospective observational cohort study; Sx = symptoms.
reporting.
6/61 (10)
Bilat (%)
No. of

Discussion
Our study sought to evaluate the heterogeneity and
Mean Sx Duration,

certainty of pain and surgical outcomes in meralgia par-

1.0 (0.2–6.0)
4.1 (0.5–30)
Yrs (range)

esthetica patients treated by injection, decompression,

1.8

and neurectomy as reported in the literature. Our results

estimate that the incidence of complete pain relief is sta-
tistically highest after neurectomy (approximately 4 in 5
patients), followed by neurolysis (approximately 3 in 5 pa-
tients) and then injection (approximately 1 in 5 patients),
Mean Age in

52 (45–66)
Yrs (range)

48 (19–68)
40 (21–71)

with the inverse trend true for no pain relief and com-
49

parable performance with respect to partial pain relief.

These trends likely correlate with the statistically signifi-
cantly greater incidence of revision procedures follow-
ing injection compared with neurolysis and neurectomy.
Yet, given that the incidences of complications associated
Males (%)

3 (75)
4 (36)

44 (46)
10 (42)
No. of

with all procedures were relatively low and comparable,

the data presented here support the position that there are
multiple treatment options for meralgia paresthetica that
can be considered at the initial presentation to achieve
Design (no. No. of

4
11
24
96
of institutions) Pts

pain relief.
There are few studies in the current literature that in-
volve multiple treatments for meralgia paresthetica in the
ROCS (1)

ROCS (1)
RCT (1)

same cohort, but in the studies that have, the trends of

superior complete pain relief following neurectomy com-
pared with neurolysis4,23,24,36,39 and injection4 are consistent
TABLE 1. Characteristics and demographics of all included studies

with our pooled trends. Interestingly, the partial and no

1992–1996
1974–1992

pain relief outcomes were more heterogeneous and likely

Years of
Study

attributable to the more subjective nature of interpreting

NR

incomplete versus complete pain relief. Nevertheless, the

increase in revision surgery for meralgia paresthetica pa-
tients treated by injection versus surgery suggests there is
The Netherlands

likely an order of increasing patient satisfaction with pain

relief for these treatments.40 This is most apparent when
Country

US

US

considering that revision procedures following injection

were either repeat injection or neurolysis and revision pro-
cedures following neurolysis were either repeat neurolysis
or neurectomy, and that no revision procedures were re-
ported following neurectomy.
These trends in pain relief would suggest that neurec-
  van Eerten et al., 199536

tomy is the most effective procedure to consider in the set-

  Williams & Trzil, 199138
» CONTINUED FROM PAGE 915

Neurectomy (continued)

ting of meralgia paresthetica. However, the recommenda-

  Total (%) or median
Authors

tion of this approach versus neurolysis or injection cannot

& Year

necessarily be made universal. With respect to injection,

  Ivins, 20004

although complete pain relief occurs less than in other

treatments, there exists a subset of patients who report
complete pain relief following a single steroid injection
for multiple years.18 However, this may be the exception

916 J Neurosurg Volume 135 • September 2021

*48 *
 Lu et al.

TABLE 2. Clinical characteristics and outcomes of all included studies

Pain Relief (%) Complications Revision Procedures
Authors & Year Agent/Technique Complete Partial None No. (%) Details No. (%) Details
Injection
  Ahmed et al., 201615 Methylprednisolone 1 (17) 5 (83) 0 0 6 (100) Injection
  Edelson & Stevens, 199416 Betamethasone 1 (7) 0 13 (93) 13 (93) Neurolysis
  Elavarasi et al., 201917 Triamcinolone 2 (25) 1 (13) 5 (62) 5 (63) Injection
  Ivins, 20004 Methylprednisolone 5 (33) 10 (66) 0 9 (60) Injection
  Jiang & Xu, 198818 Dexamethasone 5 (33) 5 (33) 5 (33)
  Klauser et al., 201619 Triamcinolone 4 (20) 11 (55) 5 (25) 0 16 (80) Injection
Neurolysis
  Alberti et al., 200920 Decompression 27 (49) 21 (38) 7 (13) 5 (9) 2 seroma, 2 infection, 1 2 (4) Neurolysis
hematoma
  Ataizi et al., 201921 Decompression 8 (62) 5 (38) 0 0
  Benezis et al., 200722 Decompression 102 (61) 28 (17) 37 (22) 12 (7) 8 hematoma, 4 wound 16 (10) Neurolysis
rupture
  de Ruiter et al., 201223 Decompression 6 (60) 4 (40) 0 1 (10) 1 hematoma 4 (40) Neurectomy
  de Ruiter et al., 201534 Decompression 2 (25) 1 (13) 5 (63) 3 (38) Neurolysis
  Ducic et al., 200624 Decompression 16 (55) 8 (28) 5 (17)
  Edelson & Stevens, 199416 Decompression 8 (62) 3 (23) 2 (15) 1 (8) Neurolysis
  Emamhadi, 201235 Decompression 4 (80) 1 (20) 0 5 (100) Neurectomy
  Hanna, 201925 Decompression 9 (47) 5 (26) 5 (26) 3 (16) 2 wound infection, 1 2 (11) Neurolysis
hematoma
  Ivins, 20004 Decompression 2 (50) 2 (50) 0 4 (100) Neurectomy
  Macnicol & Thompson, Decompression 11 (44) 6 (24) 8 (32) 4 (16) Neurolysis
 199026
  Malessy et al., 201927 Decompression 17 (89) 2 (11) 0 1 (6) 1 seroma
  Morimoto et al., 201828 Decompression 9 (75) 3 (25) 0 0 0
  Nahabedian & Dellon, Decompression 18 (78) 4 (17) 1 (4) 0 1 (4) Neurolysis
 199529
  Nouraei et al., 200730 Decompression 17 (85) 3 (15) 0 0 3 (15) Neurolysis
  Schwaiger et al., 201831 Decompression 9 (64) 5 (36) 0 1 (8) Neurolysis
  Siu & Chandran, 200532 Decompression 33 (73) 9 (20) 0 4 (10) 3 hematoma, 1 infection 0
  Son et al., 201233 Decompression 9 (81) 2 (19) 0 0 0
  van Eerten et al., 199536 Decompression 3 (30) 3 (30) 4 (40)
Neurectomy
  Berini et al., 201437 Transection 6 (86) 1 (14) 0 0
  de Ruiter et al., 201223 Transection 6 (75) 1 (13) 1 (13) 0 0
  de Ruiter et al., 201534 Transection 10 (71) 4 (29) 0 0
  Ducic et al., 200624 Transection 11 (58) 4 (21) 4 (21)
  Emamhadi, 201235 Transection 9 (100) 0 0 0
  Ivins, 20004 Transection 4 (100) 0 0 0
  van Eerten et al., 199536 Transection 9 (82) 2 (18) 0
  Williams & Trzil, 199138 Transection 23 (96) 1 (4) 0 0

in most cases, since the remaining injection studies all re- a temporizing measure rather than a curative one, and its
ported the administration of further injections or neuroly- role in that regard should not be discounted.
sis treatment to relieve pain. Nevertheless, given the mini- With respect to neurolysis, the appeal of this surgi-
mally invasive nature of an injection, this may appeal to cal procedure versus neurectomy lies within keeping the
particular patients who are adverse to surgical procedures LFCN intact, meaning that sensation is preserved in the
or those wanting to consider a less-invasive option first. In LFCN distribution. However, in a study of 24 meralgia
other circumstances, injection may prove useful more as paresthetica patients treated by neurectomy, none reported
J Neurosurg Volume 135 • September 2021 917
 
*48 *
Lu et al.

FIG. 2. Incidence of complete pain relief after treatment. The effect size (ES) of incidence, its 95% CI, and the relative weightings
are represented by the middle of the square, the horizontal line, and the relative size of the square, respectively. Figure is available
in color online only.

dissatisfaction or impact on quality of life by the loss of It is worth noting, however, that, even following neurec-
that sensation.38 The only other series to survey patients’ tomy, approximately 1 in 5 meralgia patients do not report
opinions on numbness was by de Ruiter et al.,23 who re- complete pain relief.24 This would appear counterintuitive
ported that 1 of 8 patients was frequently bothered by this. if all pain transmission pathways are severed via LFCN
Greater evaluation in future quality-of-life evaluations for transection. Two possible reasons arise for this. The first
different treatments of meralgia paresthetica are needed reason is that the LFCN may be vulnerable to stump neu-
and encouraged to better guide management.27 Ultimately, roma formation and associated pain syndromes after tran-
the loss of sensation needs to be discussed as a conse- section.41 Justification then of neurectomy over neurolysis
quence of neurectomy, and patient preference must remain (without neuroma risk) in these circumstances may require
a key contributor to the selection algorithm of neurectomy a comparison of preoperative meralgia paresthetica pain
versus the other options. with postoperative neuroma pain in the future. The second
918 J Neurosurg Volume 135 • September 2021

*48 *
 Lu et al.

FIG. 3. Incidence of revision procedures. The effect size (ES) of incidence, its 95% CI, and the relative weightings are represented
by the middle of the square, the horizontal line, and the relative size of the square, respectively. Figure is available in color online
only.

reason for incomplete pain relief following neurectomy is from adjacent genitofemoral or femoral nerves; 2) their lo-
anatomical variation of the LFCN. Although the LFCN cation with respect to the inferior inguinal ligament inser-
largely exits the pelvis at the ASIS intact as a single nerve, tion point and surrounding fascia and inserting muscles;
it has been suggested that its course varies in up to 30% and 3) the anatomical course of the LFCN over, through,
of patients.3,42 In their anatomical study, Keegan and Holy- or below the sartorius muscle. Furthermore, it has been
oke3 found that these variations specifically related to a reported that meralgia paresthetica patients can present
number of factors, including 1) the nervous contributions with as many as 4 branches of the LFCN with multiple
J Neurosurg Volume 135 • September 2021 919
 
*48 *
Lu et al.

TABLE 3. GRADE assessment for reported outcomes

Incidence No. of Certainty Assessment Overall

Outcome (95% CI) Cohorts Type of Evidence Quality Consistency Directness Effect Size Quality Certainty
Injection
  Pain relief
  Complete 22% (13−33%) 6 +2 −1 +1 0 +1 +2 Low
  Partial 37% (12−67%) 6 +2 −1 −1 0 0 0 Very low
  None 31% (4−67%) 6 +2 −1 −1 0 0 0 Very low
 Complications 0% (0−6%) 2 +2 −2 −1 −1 +2 0 Very low
  Revision procedure 81% (64−94%) 5 +2 −1 0 0 0 +1 Very low
Neurolysis
  Pain relief
  Complete 63% (56−71%) 19 +2 −1 0 0 +1 +2 Low
  Partial 22% (18−26%) 19 +2 −1 +1 0 +2 +4 High
  None 8% (3−15%) 19 +2 −1 0 0 +1 +2 Low
 Complications 5% (2−8%) 10 +2 −1 +1 0 +2 +4 High
  Revision procedure 12% (4−22%) 16 +2 −1 −1 0 +1 +1 Very low
Neurectomy
  Pain relief
  Complete 85% (71−96%) 8 +2 −1 +1 0 0 +2 Low
  Partial 11% (4−20%) 8 +2 −1 +1 0 +1 +3 Moderate
  None 2% (0−8%) 8 +2 −1 0 0 +2 +3 Moderate
 Complications 0% (0−32%) 1 +2 −2 −1 −1 0 −2 Moderate
  Revision procedure 0% (0−2%) 6 +2 −1 +1 0 +2 +4 High
The overall quality score is determined based on the sum of the included domains. Type of evidence is based on design of the included studies (range +2 to +4). The
study quality reflects the blinding and allocation, follow-up and withdrawals, sparsity of data, and methodological concerns (range −3 to 0). Consistency is graded
based on heterogeneity of the included population and study end points with respect to one another (range −1 to +1). Directness is graded based on generalizability of
included results (range −2 to 0). Effect size is graded based on the number of percent deciles the pooled 95% CI overlap (range 0 to +2). The overall quality of results
for each outcome can be considered high (≥ 4 points), moderate (3 points), low (2 points), or very low (≤ 1 point).

neuromas, making single nerve transection less effective tion can alleviate the pain syndrome without the need for
if multiple components and compressions are involved.38 LFCN transection.
Although physical compression of the LFCN at the Attempts have been made to identify predictors of sur-
ASIS is largely alleviated by both neurolysis and neurec- gical treatment success in meralgia paresthetica, largely
tomy, our results indicate that neither surgical approach without success. Multiple studies utilizing neurolysis eval-
provides universal pain relief.38 Enhancing decompres- uated age, sex, and duration of pretreatment symptoms as
sion with nerve transposition following neurolysis has predictors of pain relief and revision surgery, but none re-
proven anecdotally to be even more effective than neurol- ported any statistical significance.30,32,33 Obesity has been
ysis alone, supporting the idea that the entire anatomical often proposed as a contributor to meralgia paresthetica
course of the LFCN in these presentations may be as im- onset,38 but its role in impacting posttreatment pain relief
portant as the course at the ASIS.25 The recent report by is less consistent,27 with one study32 reporting it to be an
Malessy et al.27 of 17 of 19 (89%) patients with complete independent significant predictor of relief and another
pain relief after neurolysis with dynamic decompression study33 reporting it to be nonsignificant. It remains to be
(defined as the additional transection of fibrous structures seen if treatment outcomes can be successfully modeled
found to cause LFCN compression during thigh move- based on demographic and clinical parameters.
ment) also suggests that the entrapment elements of me- Outside of the interventions discussed in this study,
ralgia paresthetica may in fact extend beyond the ASIS. there are emerging interventions that one may consider if
This could explain why some pain persists in some neu- the pain of meralgia paresthetica is not relieved by neu-
rolysis patients despite complete nerve decompression at rectomy. The most reported of these is radiofrequency
the ASIS only, as well as the discrepancy with the pooled ablation. The principle of this treatment is to use a high
incidence rate for complete pain relief for the neurolysis temperature to induce wallerian degeneration leading to
group overall, which included many older studies that did a deafferentation effect within the LFCN to block pain
not account for this possibility. Ultimately, continued sur- transmission.43 Despite the specific evidence in the me-
gical and anatomical investigation will reveal whether or ralgia paresthetica setting being limited, small case series
not meralgia paresthetica is caused by single-site entrap- have demonstrated immediate pain relief in the majority of
ment only, and if multisite decompression or transposi- patients following ablation, although the long-term effect
920 J Neurosurg Volume 135 • September 2021

*48 *
 Lu et al.

remains unclear, with series reporting complete pain relief patients before pain relief is fully appreciable, making ear-
at 1 year postprocedure ranging from 50% to 100%.44–46 ly comparisons may be premature.22,37 Yet, whether or not
Another technique to consider is spinal cord stimulation, this will prove to be statistically and clinically significant
which can inhibit peripheral noxious stimuli by concomi- requires more time to understand, as the follow-up time
tant stimulation of the dorsal columns and disinhibition did not impact trends in pain relief per our meta-regression
of descending analgesia pathways.47 This stimulation has analysis of the current literature.
been reported to confer satisfactory pain control in the
short-term at least.48 Finally, although not targeting the Conclusions
LFCN, other techniques such as cryoablation of the pos-
terior femoral cutaneous nerve49 and shockwave therapy Meralgia paresthetica is a spontaneous entrapment
to the femoral nerve50 have also been described for pain syndrome of the LFCN, and treatments by injection, neu-
treatment, and may too have a role in meralgia paresthetica rolysis, or neurectomy can all afford pain relief. Our study
treatment refractory to the surgical techniques discussed of the contemporary data indicates that the incidence of
in this study. complete pain relief is greatest following neurectomy, ac-
There are limitations to this study. First, sampling bias companied by the lowest incidence of revision procedures.
is a primary concern in the pooled outcomes, given the Nevertheless, there exist other aspects of patient prefer-
relatively small sizes of most included cohorts, particu- ence that should be accommodated when selecting the
optimal treatment for meralgia paresthetica, as well as the
larly for neurectomy, and given the vast anatomical varia-
understanding that no treatment to date has proven uni-
tions inherent to one surgeon’s approach versus another.
versally definite in achieving complete pain relief for all
The concern for small-study bias would be somewhat al- patients. Furthermore, greater anatomical understanding
layed if more studies were available for post hoc analy- of compression along the entire LFCN will assist further
ses, and such studies should be incorporated into future in optimizing surgical techniques.
neurectomy evaluations over time. Although we utilized
a random-effects model with size-based weightings, it is
possible that overrepresentation of favorable outcomes and References
selection biases in favor of one technique over the others   1. Harney D, Patijn J. Meralgia paresthetica:​diagnosis and
in the literature may overestimate the efficacy of the treat- management strategies. Pain Med. 2007;​8(8):​669–677.
ments we report. However, GRADE assessment showed   2. van Slobbe AM, Bohnen AM, Bernsen RM, et al. Incidence
that, at least with respect to surgical treatments, there ap- rates and determinants in meralgia paresthetica in general
practice. J Neurol. 2004;​251(3):​294–297.
pears to be a higher degree of consistency between studies,   3. Keegan JJ, Holyoke EA. Meralgia paresthetica. An anatomi-
advocating the validity of their results. cal and surgical study. J Neurosurg. 1962;​19:​341–345.
It is difficult in retrospect to clarify the number of sur-   4. Ivins GK. Meralgia paresthetica, the elusive diagnosis:​clini-
gical cases preceded by failed pain control after single in- cal experience with 14 adult patients. Ann Surg. 2000;​232(2):​
jection, or indeed other treatment. More data are needed 281–286.
to determine if the injection agent or its use indeed can   5. Payne R, Seaman S, Sieg E, et al. Evaluating the evidence:​is
impact the effectiveness of surgical intervention. Ideally, neurolysis or neurectomy a better treatment for meralgia par-
esthetica? Acta Neurochir (Wien). 2017;​159(5):​931–936.
future analyses will answer the question of whether pain   6. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred re-
control can be achieved by multiple injections without the porting items for systematic reviews and meta-analyses:​the
need for surgical intervention, as the included studies did PRISMA statement. PLoS Med. 2009;​6(7):​e1000097.
not separate outcomes based on treatment intent, another   7. Freeman MF, Tukey JW. Transformations related to the angu-
weakness of retrospective studies. Prospective cohort data lar and the square root. Ann Math Stat. 1950;​21(4):​607–611.
from subspecialized surgeon experiences will prove more   8. DerSimonian R, Laird N. Meta-analysis in clinical trials.
insightful in optimizing approach algorithms in the future. Control Clin Trials. 1986;​7(3):​177–188.
  9. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring
Next, the subjective nature of patient-reported pain out- inconsistency in meta-analyses. BMJ. 2003;​327(7414):​557–560.
comes may have heterogenized our pooling, but for the 10. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
majority of outcomes, did not appear to impact the risk observational studies in epidemiology:​a proposal for report-
of publication and small-study biases. Given that different ing. Meta-analysis Of Observational Studies in Epidemiology
studies reported pain outcomes differently, some qualita- (MOOSE) group. JAMA. 2000;​283(15):​2008–2012.
tively and some quantitatively, the partial response tier is 11. Begg CB, Mazumdar M. Operating characteristics of a rank
the outcome most subject to inconsistency versus more re- correlation test for publication bias. Biometrics. 1994;​50(4):​
1088–1101.
producible definitions of pain freedom and persistent pain. 12. Egger M, Davey Smith G, Schneider M, Minder C. Bias in
Therefore, translation of our trends is more likely valid meta-analysis detected by a simple, graphical test. BMJ.
when discussing complete pain relief rather than partial 1997;​315(7109):​629–634.
pain relief. Then, retrospective reports are vulnerable to 13. Duval S, Tweedie R. Trim and fill:​a simple funnel-plot-based
patient selection bias, as they may have different clinical, method of testing and adjusting for publication bias in meta-
block, and conduction criteria to classify meralgia par- analysis. Biometrics. 2000;​56(2):​455–463.
esthetica. Finally, more consideration should be given to 14. Atkins D, Best D, Briss PA, et al. Grading quality of evidence
and strength of recommendations. BMJ. 2004;​328(7454):​1490.
longer-term follow-up in the future. Given that the overall 15. Ahmed A, Arora D, Kochhar AK. Ultrasound-guided alcohol
benefit of injection is most likely best experienced in the neurolysis of lateral femoral cutaneous nerve for intractable
early postprocedural period, whereas a refractory period meralgia paresthetica:​a case series. Br J Pain. 2016;​10(4):​
of postoperative pain may persist for months in surgical 232–237.

J Neurosurg Volume 135 • September 2021 921

 
*48 *
Lu et al.

16. Edelson R, Stevens P. Meralgia paresthetica in children. J 39. de Ruiter GC, Kloet A. Comparison of effectiveness of dif-
Bone Joint Surg Am. 1994;​76(7):​993–999. ferent surgical treatments for meralgia paresthetica:​Results
17. Elavarasi A, Goyal V, Singh MB, Padma Srivastava MV. Is of a prospective observational study and protocol for a ran-
triamcinolone an easy and efficient way to treat meralgia domized controlled trial. Clin Neurol Neurosurg. 2015;​134:​
paresthetica? A cohort study. Ann Indian Acad Neurol. 2019;​ 7–11.
22(3):​308–310. 40. Khalil N, Nicotra A, Rakowicz W. Treatment for meralgia
18. Jiang GX, Xu WD. Meralgia paraesthetica of spinal origin:​ paraesthetica. Cochrane Database Syst Rev. 2012;​12(12):​
brief report. J Bone Joint Surg Br. 1988;​70(5):​843–844. CD004159.
19. Klauser AS, Abd Ellah MM, Halpern EJ, et al. Meralgia par- 41. Zimmermann M. Pathobiology of neuropathic pain. Eur J
aesthetica:​ultrasound-guided injection at multiple levels with Pharmacol. 2001;​429(1-3):​23–37.
12-month follow-up. Eur Radiol. 2016;​26(3):​764–770. 42. Carai A, Fenu G, Sechi E, et al. Anatomical variability of
20. Alberti O, Wickboldt J, Becker R. Suprainguinal retroperito- the lateral femoral cutaneous nerve:​findings from a surgical
neal approach for the successful surgical treatment of meral- series. Clin Anat. 2009;​22(3):​365–370.
gia paresthetica. J Neurosurg. 2009;​110(4):​768–774. 43. Choi EJ, Choi YM, Jang EJ, et al. Neural ablation and regen-
21. Ataizi ZS, Ertilav K, Ercan S. Surgical options for meralgia eration in pain practice. Korean J Pain. 2016;​29(1):​3 –11.
paresthetica:​long-term outcomes in 13 cases. Br J Neuro- 44. Abd-Elsayed A, Gyorfi MJ, Ha SP. Lateral femoral cutaneous
surg. 2019;​33(2):​188–191. nerve radiofrequency ablation for long-term control of refrac-
22. Benezis I, Boutaud B, Leclerc J, et al. Lateral femoral cuta- tory meralgia paresthetica. Pain Med. 2020;​21(7):​1433–1436.
neous neuropathy and its surgical treatment:​a report of 167 45. Ghai B, Dhiman D, Loganathan S. Extended duration pulsed
cases. Muscle Nerve. 2007;​36(5):​659–663. radiofrequency for the management of refractory meralgia
23. de Ruiter GC, Wurzer JA, Kloet A. Decision making in the paresthetica:​a series of five cases. Korean J Pain. 2018;​31(3):​
surgical treatment of meralgia paresthetica:​neurolysis versus 215–220.
neurectomy. Acta Neurochir (Wien). 2012;​154(10):​1765–1772. 46. Lee JJ, Sohn JH, Choi HJ, et al. Clinical efficacy of pulsed
24. Ducic I, Dellon AL, Taylor NS. Decompression of the lateral radiofrequency neuromodulation for intractable meralgia
femoral cutaneous nerve in the treatment of meralgia pares- paresthetica. Pain Physician. 2016;​19(3):​173–179.
thetica. J Reconstr Microsurg. 2006;​22(2):​113–118. 47. Dubuisson D. Effect of dorsal-column stimulation on gelati-
25. Hanna A. Transposition of the lateral femoral cutaneous nosa and marginal neurons of cat spinal cord. J Neurosurg.
nerve. J Neurosurg. 2019;​130(2):​496–501. 1989;​70(2):​257–265.
26. Macnicol MF, Thompson WJ. Idiopathic meralgia parestheti- 48. Barna SA, Hu MM, Buxo C, et al. Spinal cord stimulation
ca. Clin Orthop Relat Res. 1990;​(254):​270–274. for treatment of meralgia paresthetica. Pain Physician. 2005;​
27. Malessy MJA, Eekhof J, Pondaag W. Dynamic decompres- 8(3):​315–318.
sion of the lateral femoral cutaneous nerve to treat meralgia 49. Joshi DH, Thawait GK, Del Grande F, Fritz J. MRI-guided
paresthetica:​technique and results. J Neurosurg. 2019;​131(5):​ cryoablation of the posterior femoral cutaneous nerve for the
1552–1560. treatment of neuropathy-mediated sitting pain. Skeletal Ra-
28. Morimoto D, Kim K, Kokubo R, et al. Deep decompression diol. 2017;​46(7):​983–987.
of the lateral femoral cutaneous nerve under local anesthesia. 50. Hausdorf J, Lemmens MA, Heck KD, et al. Selective loss of
World Neurosurg. 2018;​118:​e659–e665. unmyelinated nerve fibers after extracorporeal shockwave
29. Nahabedian MY, Dellon AL. Meralgia paresthetica:​etiology, application to the musculoskeletal system. Neuroscience.
diagnosis, and outcome of surgical decompression. Ann Plast 2008;​155(1):​138–144.
Surg. 1995;​35(6):​590–594.
30. Nouraei SA, Anand B, Spink G, O’Neill KS. A novel ap-
proach to the diagnosis and management of meralgia pares- Disclosures
thetica. Neurosurgery. 2007;​60(4):​696–700. Dr. Levi: grant support from the Department of Defense.
31. Schwaiger K, Panzenbeck P, Purschke M, et al. Surgical de-
compression of the lateral femoral cutaneous nerve (LFCN) Author Contributions
for Meralgia paresthetica treatment:​Experimental or state of
the art? A single-center outcome analysis. Medicine (Balti- Conception and design: Lu, Burks, Spinner, Levi. Acquisition of
more). 2018;​97(33):​e11914. data: Lu, Heath, Wolde. Analysis and interpretation of data: Lu,
32. Siu TL, Chandran KN. Neurolysis for meralgia paresthetica:​ Heath, Wolde. Drafting the article: all authors. Critically revising
an operative series of 45 cases. Surg Neurol. 2005;​63(1):​ the article: Lu, Burks, Spinner, Levi. Reviewed submitted ver-
19–23. sion of manuscript: all authors. Approved the final version of the
33. Son BC, Kim DR, Kim IS, et al. Neurolysis for meralgia par- manuscript on behalf of all authors: Lu. Statistical analysis: Lu.
esthetica. J Korean Neurosurg Soc. 2012;​51(6):​363–366. Study supervision: Spinner, Levi.
34. de Ruiter GC, Wurzer JA, Kloet A. Recurrence of meralgia
paresthetica years after a neurexeresis procedure:​a case re- Supplemental Information
port. Br J Neurosurg. 2015;​29(6):​885–887. Online-Only Content
35. Emamhadi M. Surgery for meralgia paresthetica:​neurolysis Supplemental material is available with the online version of the
versus nerve resection. Turk Neurosurg. 2012;​22(6):​758–762. article.
36. van Eerten PV, Polder TW, Broere CA. Operative treatment Supplementary Tables and Figures. https://thejns.org/doi/
of meralgia paresthetica:​transection versus neurolysis. Neu- suppl/10.3171/2020.7.JNS202191.
rosurgery. 1995;​37(1):​63–65.
37. Berini SE, Spinner RJ, Jentoft ME, et al. Chronic meralgia Correspondence
paresthetica and neurectomy:​a clinical pathologic study. Victor M. Lu: University of Miami Miller School of Medicine,
Neurology. 2014;​82(17):​1551–1555. Miami, FL. [email protected].
38. Williams PH, Trzil KP. Management of meralgia parestheti-
ca. J Neurosurg. 1991;​74(1):​76–80.

922 J Neurosurg Volume 135 • September 2021

*48 *
 CLINICAL ARTICLE
J Neurosurg 135:923–933, 2021

The role of indocyanine green fluorescence in endoscopic

endonasal skull base surgery and its imaging correlations
Mostafa Shahein, MD,1,3 Daniel M. Prevedello, MD,1,2 Thomas L. Beaumont, MD, PhD,1
Khalid Ismail, MD,3 Radwan Nouby, MD,3 Marilly Palettas, MPH,4 Luciano M. Prevedello, MD, MPH,5
Bradley A. Otto, MD,1,2 and Ricardo L. Carrau, MD1,2

Departments of 1Neurological Surgery, 2Otolaryngology–Head and Neck Surgery, 4Biomedical Informatics—Center for
Biostatistics, and 5Radiology, The Ohio State University Wexner Medical Center, Columbus, Ohio; and 3Department of
Neurosurgery, Aswan University Hospitals, Aswan University, Aswan, Egypt

OBJECTIVE  The use of endoscope-integrated indocyanine green (E-ICG) has recently been introduced in skull base
surgery. The quantitative correlation between E-ICG and T1-weighted gadolinium-enhanced (T1WGd) images for skull
base tumors has not been previously assessed, to the authors’ knowledge. In this study, the authors investigated the
indications for use and the limitations of E-ICG and sought to correlate the endoscopic fluorescence pattern with MRI
contrast enhancement.
METHODS  Following IRB approval, 20 patients undergoing endoscopic endonasal skull base surgery between June
2017 and August 2018 were enrolled in the study. Tumor fluorescence was measured using a blue color value and blood
fluorescence as a control. Signal intensities (SIs) of tumor T1WGd images were measured and the internal carotid artery
(ICA) SI was used as a control. For pituitary adenoma, the pituitary gland fluorescence was also measured. The relation-
ships between ICG fluorescence and MRI enhancement measurements were analyzed.
RESULTS  Data showed that in pituitary adenoma there was a strong correlation between the ratios of gland/blood fluo-
rescence to gland/ICA SI (n = 8; r = 0.92; p = 0.001) and tumor/blood fluorescence to tumor/ICA SI (n = 9; r = 0.82; p =
0.006). In other pathologies there was a strong correlation between the ratios of tumor/blood fluorescence and tumor/
ICA SI (n = 9; r = 0.74; p = 0.022). The ICG fluorescence allowed perfusion assessment of the pituitary gland as well as
of the nasoseptal flaps. Visualization of the surrounding vasculature was also feasible.
CONCLUSIONS  Defining the indications and understanding the limitations are critical for the effective use of E-ICG.
Tumor fluorescence seems to correlate with preoperative MRI contrast enhancement.
https://thejns.org/doi/abs/10.3171/2020.6.JNS192775
KEYWORDS  neuroendoscopy; skull base surgery; pituitary; indocyanine green; ICG

I ndocyanine green (ICG) has been used for decades in ies measured the fluorescence of pituitary adenomas based
vascular neurosurgery since its introduction by Raabe on a blue color value,17,18 with results suggesting that better
et al.1 Several studies have described the benefits of visualization is associated with higher fluorescence ratios
using microscope-integrated ICG in various spinal and of gland to tumor.18 In addition, E-ICG has been suggested
cranial tumors.2–6 Recent studies have suggested the ad- as a tool to enhance visualization of vasculature, assess-
ministration of high doses of ICG the day before surgery ment of blood supply to nasoseptal flaps, and tumor iden-
to take advantage of the fluorescence induced in menin- tification.17,19–22
giomas, gliomas, and brain metastases.7–9 Some investigators have concluded that there is a
Endoscope-integrated ICG (E-ICG) has been used in correlation between tumor fluorescence and radiologi-
visualizing vascular structures10–13 and performing biopsy cal enhancement on T1-weighted gadolinium-enhanced
of intraventricular tumors.14 In skull base surgery, E-ICG (T1WGd) MR images.2,9 To our knowledge, however, there
was first introduced to assess its feasibility in pituitary have been no described criteria for the selection of cases in
tumors.15 Adenomas were found to be hypofluorescent in which E-ICG may be useful, and there has been no quanti-
comparison with the pituitary gland.15,16 Subsequent stud- tative correlation of fluorescence to preoperative imaging.

ABBREVIATIONS  E-ICG = endoscope-integrated ICG; ICA = internal carotid artery; ICG = indocyanine green; SI = signal intensity; T1WGd = T1-weighted gadolinium-
enhanced.
SUBMITTED  October 12, 2019.  ACCEPTED  June 18, 2020.
INCLUDE WHEN CITING  Published online November 13, 2020; DOI: 10.3171/2020.6.JNS192775.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  923

*48 *
Shahein et al.

Methods cence of the tumors, pituitary gland, and stalk. 2) At the

Design and Demographic Data end of surgery, a second dose was used for the assessment
of tumor residual or pituitary gland fluorescence and of
A prospective cohort study of intraoperative E-ICG the integrity of vascular structures in the tumor bed.
was conducted in patients undergoing endoscopic skull The latency of fluorescence signal onset was recorded
base surgery at our center between June 2017 and August by timing the start of the internal carotid artery (ICA)
2018. The study was approved by the Institutional Review fluorescence, after ICG injection, which is the first struc-
Board in accordance with the Health Insurance Portabil- ture to be fluorescent. A blue color value ranging from 0
ity and Accountability Act (HIPAA). In total, 20 patients to 255, using ImageJ23 software, was used to measure the
(age range 20–70 years, median 47.5 years; 7 males and 13 fluorescence of the gland, tumor, and blood in pituitary
females) were included after providing informed consent. adenoma cases and of the tumor and blood in other pathol-
We divided the enrolled patients into two subgroups based ogies. Single or two divided doses were used in all cases.
on pathology: patients with pituitary adenoma (n = 10) and All the fluorescence measurements were taken only after
those with other skull base pathologies (n = 10). A total of the first ICG dose to avoid any possible additive effect.
20 vials of ICG (25 mg each) were allowed for the study, Snapshots were generated from ICG surgical videos
10 for each subgroup. In both subgroups, patients were divided into 1-second intervals. The snapshots selected
enrolled consecutively. As the design of the study was to for blood fluorescence measurement were those obtained
include 10 pituitary adenomas and 10 other tumors, the when maximum blue color value was reached in the
pituitary adenoma subgroup was finalized earlier during blood. For pituitary adenoma cases, the snapshots selected
the period of enrollment as this condition occurred more for the gland and tumor fluorescence measurements were
frequently. Patient exclusion criteria included pregnancy taken when the maximum blue color value was reached
and allergy to penicillin, sulfa, iodide, or any other dye. in the pituitary gland. For other tumor cases, the snap-
Tumor pathologies included the following: pituitary ad- shots selected for the tumor fluorescence measurement
enoma, including 1 carcinoma (n = 10); chordoma (n = 4); were when the maximum blue color value was reached
tuberculum sellae meningioma (n = 3); chondrosarcoma in the tumor. The distance of the endoscope from areas
(n = 1); esthesioneuroblastoma (n = 1); and Rathke’s cleft measured ranged from 1 cm to 2 cm. The mean of a color
cyst (n = 1). range in a 40-pixel-diameter circular area (300 DPI) that
was selected for each structure mentioned was used for
Equipment and Technique the final statistical analysis. The circular area was selected
A standard endoscopic approach was achieved depend- from parts that were not contaminated with blood to avoid
ing on the type and site of pathology, varying between overlapping fluorescence.
transsellar, transtubercular-transplanum, and transclival On T1WGd images, the signal intensities (SIs) of the
approaches. One vial of ICG (25 mg) was allowed for each gland, tumor, and ICA in pituitary adenoma cases and of
patient. Each vial was diluted in 10 ml of saline to reach a the tumor and ICA in other pathologies were measured
final concentration of 2.5 mg/ml administered in either a using a Lexmark NilRead image viewer. Circular areas
single bolus dose or 2 divided doses of 12.5 mg each. ICG of 1-mm and 2.5-mm diameter for the ICA and tumor SI,
was administered once the surgical corridor and the expo- respectively, were used for the measurement. The mean
sure were completed and at the end of surgery if a second of the SI range in each circular area was used for the final
dose was needed. Standard 0° and 45°, 4-mm-diameter, statistical analysis.
18-cm endoscopes (Karl Storz Endoscopy America, Inc.)
were used for surgery. For visualization of ICG fluores- Statistical Methods
cence, a separate endoscope consisting of a Cold Light
Fountain D-LIGHT P and an IMAGE1S H3-Z FI Three- Patient demographic characteristics were summarized
Chip FULL HD camera attached to a 5.8-mm-diameter, using descriptive statistics. ICG fluorescence was ex-
19-cm-length endoscope (all Karl Storz Endoscopy Amer- pressed as the ratios of pituitary gland/blood and tumor/
ica, Inc.) was used. Shifting between the ICG and the nor- blood blue color values in the pituitary adenoma subgroup
mal mode was effected using a foot pedal. and the ratio of tumor/blood blue color value in the other
pathologies subgroup. MRI enhancement was expressed
as the ratios of pituitary gland/ICA and tumor/ICA SI in
Video and Imaging Analysis the pituitary subgroup and the ratio of tumor/ICA SI in the
A sample of how the measurements were achieved is other pathologies subgroup. Pearson’s correlation coeffi-
presented in Fig. 1. cients were calculated to determine correlations between
Fluorescent scenes were interpreted during surgery; in ICG fluorescence and MRI enhancement measures, re-
addition, all surgical videos were archived in full length spective to each subgroup. Analyses were performed in
and analyzed postoperatively. E-ICG was used at dif- SAS 9.4 (SAS Institute) and in R version 3.5.2 (R Founda-
ferent stages during surgery depending on the aim of its tion for Statistical Computing).
utilization: 1) At the beginning of the tumor dissection, a
continuous ICG mode was used to identify the parasellar
or paraclival carotid artery, intercavernous sinus, basilar Results
plexus, condition of nasoseptal flap, abnormal dural vas- No adverse effects were noted related to the drug in a
culature, and tumor or pituitary gland fluorescence; and follow-up period of 6 months. The findings were catego-
in the middle of surgery, it was used to assess the fluores- rized as reported in the following sections.

924 J Neurosurg Volume 135 • September 2021

 
*48 *
 Shahein et al.

FIG. 1. Measurement of the MRI SI and ICG fluorescence. A: Screenshot of axial T1WGd showing the measurement of the SI of
both the pituitary gland and the pituitary adenoma. B: Screenshot of the image representing when the maximum blue color value
was reached in the pituitary gland, with attached screenshots from the measurements generated using ImageJ software for both
the pituitary gland and the tumor. All similar measurements in addition to the control measurements were grouped and ratios were
calculated and then statistically analyzed. G = gland; T = tumor. Figure is available in color online only.

Intraoperative Tumor Fluorescence (based on pathology) that required careful consideration. Clear distinction be-
Pituitary Adenoma Subgroup tween the gland and the tumor preoperatively on T1WGd
In 8 of the 10 cases, E-ICG was useful and identifica- images predicted helpful ICG scenes. However, in recur-
tion of the pituitary gland and the adenoma was success- rent pituitary adenoma and pituitary carcinoma cases,
ful. In 2 cases, E-ICG was not helpful (Video 1). E-ICG was not helpful as clear distinctions could not be
VIDEO 1. Sample of pituitary adenoma cases in patients enrolled in found preoperatively on radiological imaging. The pres-
the study. CS = cavernous sinus; G = gland; T = tumor. Copyright ence of an intact pseudocapsule, which represents a com-
Daniel M. Prevedello. Used with permission. Click here to view. pressed pituitary gland and helps in extracapsular resec-
The pituitary gland was found to be more fluorescent tion,24 and the absence of marked intratumoral bleeding
than the adenoma (Fig. 2). Careful interpretation of the made E-ICG helpful. In relatively large macroadenomas,
ICG scenes in these cases was essential due to the other the pseudocapsule was disrupted and intratumoral bleed-
surrounding fluorescent structures. ing was more prominent. As the blood was fluorescent
We categorized our findings based on certain criteria during the early phase of ICG injection, the fluorescent

J Neurosurg Volume 135 • September 2021 925

 
*48 *
Shahein et al.

the tumor. In the other 2 cases, where there was weak or

no enhancement of the tumor, weak or no fluorescence, re-
spectively, was noted. The 1 chondrosarcoma case showed
mixed enhancement radiologically, with the central part
being neither enhancing nor fluorescent; however, the
bilateral posterior clinoidal parts were radiologically en-
hancing and fluorescent with E-ICG.
All meningiomas in the study showed enhancement
on T1WGd images. This finding correlated well with the
tumor fluorescence seen on E-ICG. The site of the dural
attachment of the meningiomas could be localized based
on the abnormal dural vasculature viewable with E-ICG.
After tumor resection, enhancing dural residuals could be
detected for removal or cauterization.
ICG Fluorescence/MRI Enhancement Correlation
In the pituitary adenoma subgroup (Table 1), in each
case the ratio of the blue color value (representing the ICG
fluorescence) of the pituitary gland compared with that of
the blood (mean ± SD 0.722 ± 0.13) and of the tumor com-
pared with the blood (0.538 ± 0.16) were compared, re-
spectively, with the ratios of pituitary gland/ICA (0.675 ±
0.11) and tumor/ICA (0.494 ± 0.087) SIs (representing the
FIG. 2. Fluorescence of the tumors (pituitary adenoma)—T1WGd MR
images. Axial (A) and coronal (B) cuts show pituitary microadenoma MRI enhancement) in T1WGd images. The degree of the
where localization might be difficult. The gland covering the whole sel- differential ICG fluorescence between the pituitary gland
lar surface (C) and the fluorescent gland covering the tumor and initial and the tumor was correlated to the degree of differential
incision based on the preoperative radiological images at the suction SI between the gland and tumor in the T1WGd images.
tip (D) are shown. Effects of minimal dural cautery reflected on the The mean ratios of gland/tumor ICG fluorescence (1.476
fluorescence are indicated (↓). The gland and tumor under an ordinary ± 0.249) and gland/tumor MRI SI (1.405 ± 0.100) were
endoscope (E) are well differentiated based on the pattern of the fluo- calculated. In the other pathologies subgroup (Table 2), the
rescence obtained using E-ICG (F). G = gland; T = tumor. Figure is
available in color online only.
tumor/blood blue color value ratio was compared with the
tumor/ICA SI ratio for each case.
Scatterplots were used to visually display the ICG fluo-
rescence and MRI SI correlations (Fig. 5). In the pituitary
pituitary gland and nonfluorescent adenoma could not be adenoma subgroup, there was a strong correlation between
easily distinguished. In these cases, the fluorescent pitu- the pituitary gland/blood and gland/ICA ratios (n = 8; r =
itary gland was only visualized after clearance of ICG 0.92; p = 0.001) and tumor/blood and tumor/ICA ratios (n
from the blood. Nonfluorescent residual tumor was en- = 9; r = 0.82; p = 0.006). However, negligible associations
countered against surrounding fluorescent structures, such were detected for the differences between the pituitary
as the pituitary gland or the cavernous sinus. Finally, pre- gland/blood and tumor/blood ratios in images with ICG
sentation at the sellar surface helped in early identification in comparison with the differences between the pituitary
of the tumor/gland interface. The presence of a thinned gland/ICA and tumor/ICA ratios in MR images (n = 8; r
normal pituitary gland covering the area exposed neces- = 0.09; p = 0.83). In the subgroup of patients with other
sitated an initial incision based on the predicted location. pathologies, there was a strong correlation between the tu-
E-ICG was helpful afterward and not initially during sur- mor/blood and tumor/ICA ratios (n = 9; r = 0.74; p = 0.22).
gery. An integrated summary of our subjective remarks In the pituitary adenoma subgroup, comparison of the
and proposed flowchart is shown in Fig. 3. gland/blood and tumor/blood ICG fluorescence ratios re-
vealed a statistically significant difference (t = 4.39, p =
Other Pathologies Subgroup 0.001). In addition, comparison of the gland/ICA MRI SI
Different findings were present and depended mainly and tumor/ICA MRI SI ratios revealed a statistically sig-
on the preoperative imaging enhancement (Video 2). nificant difference (t = 11.33, p < 0.001).
VIDEO 2. Samples of tumors other than pituitary adenomas (chor- Two cases were excluded from the correlation mea-
doma and tuberculum sellae meningioma) in patients enrolled in the surements. The first case was in a patient with a recurrent
study. Copyright Daniel M. Prevedello. Used with permission. Click pituitary adenoma, which was excluded because there
here to view. was no clear distinction between the ICG fluorescence
Four cases of chordomas and 1 chondrosarcoma with and MRI SI measurements for the gland and tumor. The
different enhancement patterns were studied (Fig. 4). In second was the Rathke’s cleft cyst case, in which thin-
the 4 chordoma cases, 2 demonstrated contrast enhance- wall cyst fluorescence could not be accurately measured.
ment on T1WGd images and were fluorescent using E- The chondrosarcoma case had two values for each of the
ICG. The tumor-specific fluorescence helped delineate the ICG scene and MR image measurements representing
tumor and suspected small foci of tumor seedings around the two different fluorescence and enhancement patterns,

926 J Neurosurg Volume 135 • September 2021

 
*48 *
 Shahein et al.

FIG. 3. A proposed flowchart with integrated data for the use of E-ICG in pituitary adenoma cases for which the perspective of the
fluorescence of the tumor would be beneficial. Figure is available in color online only.

respectively; however, one value representing the center the tumor, and assessment of the flap condition using ICG
of the tumor was included for statistical analysis. In the demonstrated no fluorescent signal. The flap was reuti-
pituitary carcinoma case, the aim of surgery was biopsy lized and no CSF leak was encountered intra- or postop-
of the lesion as the patient had liver metastasis, so mea- eratively. The third was an esthesioneuroblastoma case, in
surement of the pituitary gland fluorescence and SI was which there was a decreased flow in the harvested naso-
not achieved. septal flap noted by the surgeons and confirmed with E-
ICG in comparison with the surrounding normal mucosa.
Diffusion to Structures Multilayered reconstruction was needed, and no CSF leak
Visualization of the pituitary gland was helpful at dif- was encountered postoperatively.
ferent stages of surgery, especially in pathologies close to
the gland. The gland was usually fluorescent for hours and Visualization of the Vasculature
fluorescence often lasted until the end of the surgery. In Before tumor dissection, different skull base vasculature
the Rathke’s cleft cyst case that required pituitary gland could be viewed with ICG depending on the location of the
splitting to reach the retrochiasmatic portion of the tumor, pathology (Fig. 6). The ICA was the first structure to be
the fluorescence showed slower diffusion to the right side, fluorescent. Time from ICG injection to ICA fluorescence
possibly indicating decreased blood supply, in comparison ranged from 3 to 48.85 seconds (average 18.125 seconds).
with the left side. This matched the postoperative T1WGd The most rapid ICA fluorescence was in a chordoma case
images. The pituitary stalk and posterior lobe of the pitu- with a patent foramen ovale. E-ICG helped mainly in iden-
itary gland were also viewable in some cases. tification of the parasellar lateral limits of the approach in
The mucosal fluorescence helped in assessment of the tuberculum sellae meningioma and pituitary adenoma cas-
condition of the nasoseptal flaps of 3 cases during the cor- es. However, failed clear ICA detection was noted in 2 of 3
ridor stage. The first was a recurrent chordoma case, and cases of acromegalic somatotrophs. Failed detection of the
the flap was elevated from its position, which was used in paraclival ICA was noted in 3 out of 5 clival lesions. Thick-
previous surgery. The flap demonstrated good flow based ness of bone might be a reasonable explanation for this de-
on the degree of the fluorescence signal with E-ICG and tection failure. In the pituitary carcinoma case, the tumor
was reutilized. The second was a chordoma case in which was completely surrounding the ICA and avidly fluores-
there was unintentional partial violation of the pedicle of cent, which caused failure of ICA localization using E-ICG.
the nasoseptal flap, due to the nature of the extension of Other fluorescent vessels were the anterior and pos-

J Neurosurg Volume 135 • September 2021 927

 
*48 *
Shahein et al.

FIG. 4. Fluorescence of the tumors—chordoma and chondrosarcoma T1WGd MR images. Case 1. Axial (A) and coronal (B) cuts
show left-sided recurrent chordoma, with the enhancing tumor on ICG mode (C), with small seedings (▼) that could be missed
*
during resection in normal mode (D). Comparable parts on normal and ICG modes are indicated ( ). Case 2. Axial (E) and coronal
(F) cuts show a clival chordoma; the tumor after completion of the approach on ICG mode with the nonfluorescent tumor and the
slightly fluorescent gland covered by the dura (G). Later close inspection of very faint fluorescence of the tumor shown in panel G
(H). Case 3. Axial (I) and coronal (J) cuts show the bilateral enhancing posterior clinoidal parts of the low-grade chondrosarcoma
and the central hypoenhancing part, reflected in the ICG mode (K) showing the fluorescent part in comparison with the normal
mode (L) (►). (By that time, the lack of a 45° ICG endoscope made the visualization of clinoidal parts difficult.) Gl = gland; T =
tumor. Figure is available in color online only.

terior ethmoidal vessels in the transcribriform approach. and assessed in some cases, including the cavernous sinus,
In this case, fluorescence helped in decreasing bleeding intercavernous sinus, and basilar plexus.
by accurate localization of such vessels. Abnormal dural
blood vessels in meningiomas were also viewable with E- Discussion
ICG, which helped in the localization and cauterization of
feeding blood vessels before resection. McConnell’s artery ICG with a special microscope has been used in the
was noted to be hypertrophied and fluorescent in one case neurosurgical field for decades.1 Since the introduction of
and was cauterized. Visualization of these vessels allowed E-ICG, many articles have described the usefulness of this
the surgeons to achieve proper tumor devascularization. technology in various applications.10–21,25
After tumor resection, vascular fluorescence helped in as- The variations among studies of E-ICG in the skull base
sessing the condition of the vessels in the tumor bed. In field, though relatively few, are notable. The dose of ICG
suprasellar lesions, the condition of the anterior cerebral varied between 5,18 6.25,14 12.5,17 and 10 or 2515 mg/dose.
arteries and the superior hypophyseal arteries could be The timing for detection of the fluorescence in the pituitary
detected. In clival lesions, E-ICG allowed assessment of gland ranged from 17 seconds17 to 10 minutes.14 Previous
the integrity of small and large brainstem perforators and studies also showed variations in fluorescence timing of dif-
also helped in assessing the flow in larger vessels, such as ferent structures.17 Quantification of the fluorescence was by
the basilar artery. Venous sinuses could also be viewable either subjective15 or objective17,18 measurements. The tim-
928 J Neurosurg Volume 135 • September 2021
 
*48 *
 Shahein et al.

TABLE 1. Demographic and pathological data of the pituitary adenoma cases and postoperative measurements of the ratios of
fluorescence and the intensity of both the gland and tumor in relation to the blood and ICA, respectively
Differential Differential
Case Age (yrs)/ Pituitary Cell Size (max A: Gland/Blood B: Tumor/Blood C: Gland/ICA D: Tumor/ICA Fluorescence Intensity
No.* Sex Lineage diameter, cm) Fluorescence Fluorescence Intensity Intensity (A & B)† (C & D)†
1 38/F Corticotroph 1.2 0.767 0.448 0.674 0.471 0.319 0.203
2 70/F Gonadotroph 2.1 — — — — — —
3 36/M Somatotroph 1.6 0.493 0.26 0.530 0.395 0.232 0.134
4 70/F Null cell 2.1 0.640 0.418 0.634 0.46 0.222 0.173
5 54/F Corticotroph 3.36 — 0.8 — 0.584 — —
6 25/F Somatotroph 2.9 0.569 0.469 0.534 0.36 0.1 0.174
7 46/F Corticotroph 0.7 0.778 0.638 0.772 0.522 0.14 0.249
8 25/F Corticotroph 0.4 0.861 0.658 0.828 0.592 0.203 0.236
9 68/M Somatotroph 1.4 0.847 0.639 0.729 0.603 0.207 0.126
10 57/M Gonadotroph 1.6 0.814 0.507 0.699 0.485 0.307 0.24
Mean ± SD 0.722 ± 0.13 0.538 ± 0.16 0.675 ± 0.11 0.494 ± 0.087
Data based on measurements taken from surgical videos: A, ratios of the pituitary gland to blood ICG fluorescence; B, ratios of the pituitary adenoma to blood ICG
fluorescence measurements. Data based on measurements taken from MRI: C, ratios of the pituitary gland to ICA signal intensity; D, ratios of the pituitary adenoma to
ICA signal intensity.
* Case 2 was a recurrent pituitary adenoma that had no clear distinction between the pituitary gland and tumor. Case 5 proved to be a pituitary carcinoma with liver
metastasis; the aim of surgery was biopsy and debulking, so pituitary gland visualization was not achieved.
† Statistically significant differences between A and B (t = 4.39, p = 0.001) and C and D (t = 11.33, p < 0.001).

ing of ICA fluorescence was variable, which might be at- ICA fluorescence in a case with a patent foramen ovale in
tributable to the differences among patient characteristics, our series. For this reason, we found that setting a specific
including cardiac output and rate of clearance, as well as the timing for visualization of each structure was not practical.
rate of ICG injection. This justifies the fastest occurrence of The gadolinium-based contrast agents are small par-

TABLE 2. Demographic and pathological data of the other tumor pathologies and the postoperative measurements of
the ratios of fluorescence and the SI of the tumor in relation to the blood and ICA, respectively
Tumor Type & Case No. Age (yrs)/Sex Tumor Size, cm A: Tumor/Blood Fluorescence B: Tumor/ICA Intensity
Chordoma
 1 34/F 1.2 × 1 × 1.2 0.996 0.708
 2 20/M 3.4 × 3.2 × 4 0.204 0.283
 3 70/M 5.7 × 2.7 × 7 0.69 0.684
 4 38/F 1.1 × 1 × 1.3 0.331 0.425
Meningioma
 1 49/M 0.9 × 0.9 × 0.6 0.581 0.818
 2 67/F 0.5 × 0.9 × 0.8 0.782 0.563
 3 57/F 2.1 × 2 × 2 0.644 0.56
Chondrosarcoma
 1 57/M 3.1 × 1.4 × 4.7 0.948,* 0.407† 0.73,* 0.37†
Esthesioneuroblastoma
 1 62/M 1.1 × 0.6 × 0.8 0.486 0.564
Rathke’s cleft cyst
 1 30/F 1.1 × 1 × 1.2,‡ 1 × 0.9 × 0.9§ — —
Mean ± SD 0.569 ± 0.242 0.553 ± 0.171
A, ratio of tumor to blood ICG fluorescence measurements taken from surgical videos; B, ratio of tumor to ICA signal intensity measurements
taken from MRI.
* Posterior clinoidal part of the tumor, not included in final statistical analysis.
† Central clival part of the tumor, included in final statistical analysis.
‡ Sellar component of the cyst.
§ Retrochiasmatic part of the cyst.

J Neurosurg Volume 135 • September 2021 929

 
*48 *
Shahein et al.

FIG. 5. Scatterplots showing the relationship between the ICG fluorescence and the MRI T1WGd SI measured postoperatively in
the pituitary adenoma and other pathologies subgroups. A: The relationship between the MRI tumor/ICA SI ratio on the vertical
axis and the tumor/blood ICG fluorescence ratio on the horizontal axis in both subgroups.  FIG. 5. (continued) →

930 J Neurosurg Volume 135 • September 2021

 
*48 *
 Shahein et al.

FIG. 5. B: The relationship between the pituitary gland/ICA MRI SI ratio on the vertical axis and the pituitary gland/blood ICG fluo-
rescence ratio on the horizontal axis in the pituitary adenoma subgroup. C: The relationship of the difference between the ratios
of MRI gland/ICA and tumor/ICA SI on the vertical axis to the difference between the ratios of ICG gland/blood and tumor/blood
fluorescence on the horizontal axis in the pituitary adenoma subgroup. Figure is available in color online only.

ticles that bind to the plasma protein and depend on de- pattern has been either rarely or unclearly mentioned in
fects in the blood-brain barrier to enhance the lesions on previous articles.2,9,17–19,27 To our knowledge, objective
T1WGd images.26 These agents and ICG are hypothesized quantification of endoscopic ICG fluorescence was never
to share a similar mechanism of action.7 The correlation correlated with the SI of MRI enhancement in the litera-
between the MRI enhancement and the ICG fluorescence ture before the present study.

FIG. 6. Visualization of the vasculature, before (left panels, A–F) and after (right panels, G–L) tumor resection. A: Transcribiform
approach to a case of esthesioneuroblastoma in normal mode. B: ICG mode of the same case. The left anterior (▼) and both
posterior ethmoidal (▲) arteries are marked in A and B. The left anterior ethmoidal artery was cauterized earlier. C: Transtuber-
cular approach to a meningioma with a hypertrophied McConnell artery (►) that needed cautery before dural opening and helped
in decreasing the bleeding (D). E: Transtubercular approach after complete cautery of the dural attachment of a tuberculum sellae
meningioma bordered by the fluorescent ICA on both sides. F: Transclival approach for a small chordoma with fluorescence of
the surrounding vascular structures. G: A case of post–tuberculum sellae meningioma resection showing both anterior cerebral
arteries (↑) using a 45° endoscope. H: The view using a 0° endoscope in ICG mode with fluorescent (↑) anterior cerebral arteries.
Note that the blood clot (●) masked visualization of underlying structures. The stalk and superior hypophyseal arteries (→) are
visualized on normal (I) and ICG (J) modes. K: A transclival approach for resection of a chordoma. L: Patency of the brainstem
perforators and branches of the basilar artery in the tumor bed after resection. The small hypofluorescent portion (►►) repre-
sents a small part that was invaded by the tumor. BP = basilar plexus; G = gland; I = internal carotid artery; ICS = intercavernous
sinus; PS = pituitary stalk; T = tuberculum sellae meningioma. Figure is available in color online only.

J Neurosurg Volume 135 • September 2021 931

 
*48 *
Shahein et al.

We correlated the ICG fluorescence using the blue it might lead to the risk of sampling normal tissue.33,34 This
color value with the MRI T1WGd images using the SI. method also carries a higher risk of bias during sampling
Each case was used as its own control, and ratios, instead as the surgeon may be influenced by the concern of caus-
of absolute values, were used for statistical analysis. Using ing any harm to the patient. The current study stands as a
this method helped in diminishing the possible variations preliminary step to expand and test the usefulness of ICG
related to each case, including pathological type and ICG in specific pathologies separately. Subsequent separate
characteristics. The degrees of the difference between the specific studies including a larger number of patients for
gland/tumor ICG fluorescence ratio and the gland/tumor each pathology are needed.
MRI enhancement ratio are not correlated. This finding
might be attributable to the different pathological subtypes Conclusions
and needs to be further investigated in subsequent studies.
In pituitary adenoma cases, the various patterns of en- E-ICG is a useful tool in skull base surgery. Careful
hancement seen in MR images are due to the different consideration of the indications and identification of the
vasculature between the gland and the tumor.28 We always limitations and imaging correlations help in selecting the
found a differential fluorescence, though variable, with the patients who will benefit from the use of this method. The
gland being more fluorescent. Based on our experience, we main purpose of this technique is to assess the vasculature
postulated that presentation to the sellar surface, presence around the tumor and its integrity and perfusion to normal
of an intact pseudocapsule, and intratumoral bleeding were structures or to stain the tumor itself.
the main factors that should be considered in selection of
the cases and the way of interpretation. In other types of Acknowledgments
tumors, E-ICG was helpful in fluorescent tumors that were This project has been supported in part by the Neuroscience
enhancing radiologically. In nonfluorescent tumors, the Research Institute at The Ohio State University.
lack of fluorescent background deemed the tumor detec-
tion unreliable. E-ICG might help in the future to achieve References
more precise resection, though the specificity and sensitiv-
ity of the exact tumor margins need further investigation   1. Raabe A, Beck J, Gerlach R, et al. Near-infrared indocyanine
green video angiography:​a new method for intraoperative as-
with pathological confirmation. sessment of vascular flow. Neurosurgery. 2003;52(1):132–139.
​ ​
In the perspective of visualization of the vascular struc-   2. Acerbi F, Vetrano IG, Sattin T, et al. The role of indocyanine
tures, failure of ICA detection in some cases was due to green videoangiography with FLOW 800 analysis for the
limited penetration of the fluorescent ICG through tissues, surgical management of central nervous system tumors:​an
which ranged from 2 to 10 mm.7,15 Bone thickness, espe- update. Neurosurg Focus. 2018;44(6):E6.
​ ​
cially in patients with acromegaly or during assessment of   3. Takami T, Naito K, Yamagata T, et al. Benefits and limita-
the paraclival regions, and encasement by the tumor were tions of indocyanine green fluorescent image-guided surgery
for spinal intramedullary tumors. Oper Neurosurg (Hagers-
the main obstacles. We believe that Doppler and naviga- town). 2017;13(6):746–754.
​ ​
tion systems are still needed in addition to E-ICG for lo-   4. Asayama B, Sato K, Fukui T, et al. Skull bone tumor resec-
calization purposes. tion with intraoperative indocyanine green fluorescence im-
In previous studies, authors concluded that E-ICG is aging:​a series of four surgical cases. Interdiscip Neurosurg.
better in visualization during aneurysm surgery29 or in pi- 2017;9:8–13.
​ ​
tuitary surgery,30 attributing this to the closer distance of   5. Hojo M, Arakawa Y, Funaki T, et al. Usefulness of tumor
observation and better illumination. Such factors affected blood flow imaging by intraoperative indocyanine green
the degree of the blue color value, and this was ignored videoangiography in hemangioblastoma surgery. World Neu-
rosurg. 2014;82(3-4):e495–e501.
​ ​
in many previous studies. However, using the ratios with   6. Benedetto N, Aquila F, Vannozzi R. Use of near-infrared
similar distances for analysis made the effect of the dis- indocyanine videoangiography and Flow 800 in the resec-
tance on the blue color value not statistically significant.31 tioning of a spinal cord haemangioblastoma. Br J Neurosurg.
More technical advancement to help detect minor dif- 2013;27(6):847–849.
​ ​
ferences in the fluorescence will help in the differentia-   7. Lee JYK, Pierce JT, Thawani JP, et al. Near-infrared fluo-
tion of similarly fluorescent structures, especially in the rescent image-guided surgery for intracranial meningioma. J
cases of the pituitary adenomas. Different systems32 and Neurosurg. 2018;128(2):380–390.
​ ​
techniques27 can affect the ability to visualize ICG micro-   8. Lee JYK, Pierce JT, Zeh R, et al. Intraoperative near-infrared
optical contrast can localize brain metastases. World Neuro-
scopically and consequently endoscopically. Real-time in- surg. 2017;106:120–130.
​ ​
traoperative correlation between the fluorescence and the   9. Lee JYK, Thawani JP, Pierce J, et al. Intraoperative near-
enhancement may help in improving interpretation of the infrared optical imaging can localize gadolinium-enhancing
ICG scenes. gliomas during surgery. Neurosurgery. 2016;79(6):856–871.
​ ​
This study carries some limitations; there was no path- 10. Wachter D, Behm T, von Eckardstein K, Rohde V. Indocya-
ological confirmation of the fluorescent or nonfluorescent nine green angiography in endoscopic third ventriculostomy.
portions of the tumor to better identify the specificity, Neurosurgery. 2013;73(1 ​ Suppl Operative):ons67–ons73.
​
11. Bruneau M, Appelboom G, Rynkowski M, et al. Endoscope-
sensitivity, positive predictive value, and negative predic- integrated ICG technology:​first application during intracra-
tive value of the tumors’ margins. Though these data are ​
nial aneurysm surgery. Neurosurg Rev. 2013;36(1):77–85. ​
needed for idealization of the results in all the fluorescent 12. Mielke D, Malinova V, Rohde V. Comparison of intraopera-
materials, including ICG, 5-aminolevulinic acid, and so- tive microscopic and endoscopic ICG angiography in aneu-
dium fluorescein, this method carries some controversy as ​
rysm surgery. Neurosurgery. 2014;10(suppl ​
3):418–425.

932 J Neurosurg Volume 135 • September 2021

 
*48 *
 Shahein et al.

13. Yoshioka H, Kinouchi H. The roles of endoscope in aneurys- 31. Lee JYK, Cho SS, Zeh R, et al. Folate receptor overexpres-
mal surgery. Neurol Med Chir (Tokyo). 2015;​55(6):​469–478. sion can be visualized in real time during pituitary adenoma
14. Tsuzuki S, Aihara Y, Eguchi S, et al. Application of indocya- endoscopic transsphenoidal surgery with near-infrared imag-
nine green (ICG) fluorescence for endoscopic biopsy of intra- ing. J Neurosurg. 2018;​129(2):​390–403.
ventricular tumors. Childs Nerv Syst. 2014;​30(4):​723–726. 32. Cho SS, Zeh R, Pierce JT, et al. Comparison of near-infrared
15. Litvack ZN, Zada G, Laws ER Jr. Indocyanine green fluo- imaging camera systems for intracranial tumor detection.
rescence endoscopy for visual differentiation of pituitary Mol Imaging Biol. 2018;​20(2):​213–220.
tumor from surrounding structures. J Neurosurg. 2012;​116(5):​ 33. Stummer W, Pichlmeier U, Meinel T, et al. Fluorescence-
935–941. guided surgery with 5-aminolevulinic acid for resection of
16. Catapano G, Sgulò F, Laleva L, et al. Multimodal use of in- malignant glioma:​a randomised controlled multicentre phase
docyanine green endoscopy in neurosurgery:​a single-center III trial. Lancet Oncol. 2006;​7(5):​392–401.
experience and review of the literature. Neurosurg Rev. 2018;​ 34. Cho SS, Salinas R, Lee JYK. Indocyanine-green for fluores-
41(4):​985–998. cence-guided surgery of brain tumors:​evidence, techniques,
17. Hide T, Yano S, Shinojima N, Kuratsu J. Usefulness of the in- and practical experience. Front Surg. 2019;​6:​11.
docyanine green fluorescence endoscope in endonasal trans-
sphenoidal surgery. J Neurosurg. 2015;​122(5):​1185–1192.
18. Verstegen MJT, Tummers QRJG, Schutte PJ, et al. Intra- Disclosures
operative identification of a normal pituitary gland and an Storz provided the endoscopic equipment for ICG and the ICG
adenoma using near-infrared fluorescence imaging and vials. D. M. Prevedello reports being a consultant for Stryker,
low-dose indocyanine green. Oper Neurosurg (Hagerstown). Medtronic, and Integra; being a patent holder in Mizuho, KLS-
2016;​12(3):​260–268. Martin, and ACE-Medical; receiving royalties from KLS-Martin;
19. Geltzeiler M, Nakassa ACI, Turner M, et al. Evaluation of and receiving honoraria from Mizuho and Storz.
intranasal flap perfusion by intraoperative indocyanine green
fluorescence angiography. Oper Neurosurg (Hagerstown). Author Contributions
2018;​15(6):​672–676.
20. Inoue A, Ohnishi T, Kohno S, et al. Usefulness of an image Conception and design: DM Prevedello, Shahein, Nouby, LM
fusion model using three-dimensional CT and MRI with Prevedello, Carrau. Acquisition of data: DM Prevedello, Shahein,
indocyanine green fluorescence endoscopy as a multimodal Beaumont, LM Prevedello, Otto. Analysis and interpretation
assistant system in endoscopic transsphenoidal surgery. Int J of data: DM Prevedello, Shahein, Ismail. Drafting the article:
Endocrinol. 2015;​2015:​694273. Shahein. Critically revising the article: DM Prevedello, Ismail,
21. Kerr EE, Jamshidi A, Carrau RL, et al. Indocyanine green Nouby, LM Prevedello, Otto, Carrau. Reviewed submitted version
fluorescence to evaluate nasoseptal flap viability in endo- of manuscript: all authors. Approved the final version of the
scopic endonasal cranial base surgery. J Neurol Surg B Skull manuscript on behalf of all authors: DM Prevedello. Statistical
Base. 2017;​78(5):​408–412. analysis: Shahein, Palettas. Administrative/technical/material
22. Simal Julián JA, Sanromán Álvarez P, Miranda Lloret P, support: Shahein, Beaumont, LM Prevedello, Otto, Carrau. Study
Botella Asunción C. Endo ICG videoangiography:​localizing supervision: DM Prevedello, Carrau.
the carotid artery in skull-base endonasal approaches. Acta
Neurochir (Wien). 2016;​158(7):​1351–1353. Supplemental Information
23. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to Videos
ImageJ:​25 years of image analysis. Nat Methods. 2012;​9(7):​ Video 1. https://vimeo.com/436753965.
671–675. Video 2. https://vimeo.com/436753551.
24. Oldfield EH, Vortmeyer AO. Development of a histological
pseudocapsule and its use as a surgical capsule in the exci- Previous Presentations
sion of pituitary tumors. J Neurosurg. 2006;​104(1):​7–19.
25. Yokoyama J, Ishibashi K, Shiramizu H, Ohba S. Impact of Portions of the manuscript have been presented previously as
endoscopic indocyanine green fluorescence imaging on su- a podium presentation at the 28th and 29th Annual Meetings
perselective intra-arterial chemotherapy for recurrent cancer of the North American Skull Base Society in San Diego, CA,
of the skull base. Anticancer Res. 2016;​36(7):​3419–3424. February 16–18, 2018, and Orlando, FL, February 15–17, 2019.
26. Provenzale JM, Mukundan S, Dewhirst M. The role of The abstracts have been published in J Neurol Surg B. 2018;79(S
blood-brain barrier permeability in brain tumor imaging and 01):S1–S188 and J Neurol Surg B. 2019;80(S 01):S1–S244.
therapeutics. AJR Am J Roentgenol. 2005;​185(3):​763–767.
27. Simal-Julián JA, Miranda-Lloret P, Evangelista-Zamora R, Correspondence
et al. Indocyanine green videoangiography methodological Daniel M. Prevedello: The Ohio State University, Columbus, OH.
variations:​ review. Neurosurg Rev. 2015;​38(1):​49–57. [email protected].
28. Di Ieva A, Weckman A, Di Michele J, et al. Microvascular
morphometrics of the hypophysis and pituitary tumors:​from
bench to operating theatre. Microvasc Res. 2013;​89:​7–14.
29. Fischer G, Rediker J, Oertel J. Endoscope- versus micro-
scope-integrated near-infrared indocyanine green videoan-
giography in aneurysm surgery. J Neurosurg. 2018;​131(5):​
1413–1422.
30. Sandow N, Klene W, Elbelt U, et al. Intraoperative indocya-
nine green videoangiography for identification of pituitary
adenomas using a microscopic transsphenoidal approach.
Pituitary. 2015;​18(5):​613–620.

J Neurosurg Volume 135 • September 2021 933

 
*48 *
 CLINICAL ARTICLE
J Neurosurg 135:934–942, 2021

Postoperative 30-day emergency department utilization

after 7294 cranial neurosurgery procedures at a tertiary
neuroscience center
Douglas A. Hardesty, MD,1,2 Michael A. Mooney, MD,1 Benjamin K. Hendricks, MD,1
Joshua S. Catapano, MD,1 Scott T. Brigeman, MD,1 Michael A. Bohl, MD,1 John P. Sheehy, MD,1 and
Andrew S. Little, MD1
1
Department of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona;
and 2Department of Neurological Surgery, The Ohio State University Wexner Medical Center, Columbus, Ohio

OBJECTIVE  Hospital readmission and the reduction thereof has become a major quality improvement initiative in or-
ganized medicine and neurosurgery. However, little research has been performed on why neurosurgical patients utilize
hospital emergency rooms (ERs) with or without subsequent admission in the postoperative setting.
METHODS  This study was a retrospective, single-center review of data for all surgical cranial procedures performed
from July 2013 to July 2016 in patients who survived to discharge. The study was approved by the institutional review
board of the participating medical center.
RESULTS  The authors identified 7294 cranial procedures performed during 6596 hospital encounters in 5385 patients.
The rate of postoperative ER utilization within 30 days after surgical hospitalization across all procedure types was 13.1
per 100 surgeries performed. The two most common chief complaints were pain (30.7%) and medical complication
(18.2%). After identification of relevant surgical and patient factors with univariable analysis, a multivariable backward
elimination logistic regression model was constructed in which Ommaya reservoir placement (OR 2.65, p = 0.0008) and
cranial CSF shunt placement (OR 1.40, p = 0.0001) were associated with increased ER utilization. Deep brain stimula-
tion electrode placement (OR 0.488, p = 0.0004), increasing hospital length of stay (OR 0.935, p < 0.0001), and increas-
ing patient age (OR 0.988, p < 0.0001) were associated with lower rates of postoperative ER utilization. One-half (50%)
of ER visit patients were readmitted to the hospital. New/worsening neurological deficit chief complaint (OR 1.99, p =
0.0088), fever chief complaint (OR 2.41, p = 0.0205), altered mentation chief complaint (OR 2.71, p = 0.0002), patient
chronic kidney disease (OR 3.31, p = 0.0037), brain biopsy procedure type (OR 3.50, p = 0.0398), and wound infection
chief complaint (OR 31.4, p = 0.0008) were associated with increased rates of readmission to the hospital from the ER in
multivariable analysis.
CONCLUSIONS  The authors report the rates of and reasons for ER utilization in a large cohort of postoperative cranial
neurosurgical patients. Factors identified were associated with both increased and decreased use of the ER after cranial
surgery, as well as variables associated with readmission to the hospital after postoperative ER visitation. These findings
may direct future quality improvement via prospective implementation of care pathways for high-risk procedures.
https://thejns.org/doi/abs/10.3171/2020.8.JNS202404
KEYWORDS  emergency department; postoperative readmission; cranial surgery

P hospital readmission analyses and mission in the postoperative setting. Emergency depart-
ostprocedural
reductions have become major quality improvement ment use after surgery represents a potentially significant
initiatives in American medicine.1–3 The literature fiscal burden to patients and hospital systems. Furthermore,
on unexpected neurosurgical hospital readmissions contin- presentation to an ER often represents the first step in sub-
ues to grow rapidly.4–10 However, little research has been sequent postprocedural hospital readmission. Lastly, ER
performed on why neurosurgical patients utilize hospital visits may negatively affect a patient’s quality of life and/
emergency rooms (ERs) with or without subsequent ad- or negatively affect their subjective view of the operative

ABBREVIATIONS  ASA = American Society of Anesthesiologists; DBS = deep brain stimulation; ER = emergency room; LOS = length of stay.
SUBMITTED  June 21, 2020.  ACCEPTED  August 3, 2020.
INCLUDE WHEN CITING  Published online January 29, 2021; DOI: 10.3171/2020.8.JNS202404.

934 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*49 *
 Hardesty et al.

TABLE 1. Rate of postoperative ER utilization and relative risk by procedure type

No. of Pts
ER Use Pt ER
Procedure Yes No Total Use, % RR 95% CI p Value
Ommaya reservoir placement 17 50 67 25.4% 1.95 1.29–2.95 0.0016
ICP monitor 9 36 45 20.0% 1.53 0.85–2.76 0.156
CSF leak repair 7 33 40 17.5% 1.34 0.68–2.63 0.4
CSF cranial shunt placement/revision 229 1104 1333 17.2% 1.41 1.23–1.61 <0.0001
Chiari decompression 20 97 117 17.1% 1.31 0.87–1.96 0.19
Epilepsy 17 88 105 16.2% 1.24 0.798–1.92 0.34
Wound revision 33 179 212 15.6% 1.19 0.867–1.64 0.278
Vascular: intracranial bypass 11 63 74 14.9% 1.13 0.655–1.96 0.65
Vascular: aneurysm clipping 51 336 387 13.2% 1.005 0.772–1.31 0.972
Transsphenoidal (endoscopic or microscopic, all pathologies) 71 489 560 12.7% 0.964 0.769–1.21 0.748
Cyst resection/fenestration 12 83 95 12.6% 0.962 0.565–1.64 0.887
Brain tumor resection 280 1960 2240 12.5% 0.933 0.819–1.06 0.297
Brain biopsy (open or stereotactic) 18 130 148 12.2% 0.926 0.598–1.43 0.729
Microvascular decompression 21 152 173 12.1% 0.926 0.619–1.39 0.7
Vascular: cavernoma resection 22 184 206 10.7% 0.81 0.543–1.21 0.3
Cranioplasty 13 109 122 10.7% 0.81 0.483–1.36 0.423
Laser interstitial thermal therapy 10 85 95 10.5% 0.8 0.444–1.44 0.459
Trauma (EDH/SDH/abscess/empyema/hemicraniectomy) 71 618 689 10.3% 0.768 0.611–0.966 0.024
Vascular: AVF/AVM resection 12 113 125 9.6% 0.728 0.424–1.25 0.251
Endoscopic 3rd ventriculostomy 5 54 59 8.5% 0.644 0.278–1.49 0.305
DBS (electrode) 28 374 402 7.0% 0.517 0.360–0.742 0.0004
AVF = arteriovenous fistula; AVM = arteriovenous malformation; EDH = epidural hematoma; ICP = intracranial pressure; pt = patient; SDH = subdural hematoma.
Rate of postoperative ER utilization and RR with 95% CIs for each studied cranial procedure type. Also see Fig. 1 for graphical representation. Boldface type indicates
univariate analysis statistically significant at the level of p < 0.05.

care provided. To address this knowledge gap, we sought placement) in a single hospital admission were counted in
to describe the rates and factors associated with postoper- each procedure category. Charts of patients who presented
ative emergency department utilization via a retrospective to our hospital ER within 30 days of surgery were manu-
review of our tertiary neuroscience center experience. ally reviewed to extract the chief complaint, the results of
laboratory and radiographic studies, and disposition.
Methods
Study Design Statistical Considerations
The study was approved by our local institutional re- Univariable and multivariable statistical analyses using
view board at St. Joseph’s Hospital and Medical Center, chi-square tests (or Fisher exact tests for subgroups of <
Phoenix, Arizona. We identified all sequential cranial 5 individuals), Student t-tests, and backward elimination
operations in an institutional database from June 2013 to logistic regression analysis with categorical and continu-
June 2016 in patients who survived to discharge and who ous variables were performed where appropriate using
were not discharged to hospice. These dates represent the MedCalc software version 15.0. Backward elimination
beginning of our hospital electronic medical record to the multivariable models were constructed from all relevant
date of database creation. Stereotactic radiosurgery and variables deemed statistically significant (p < 0.05) in uni-
endovascular interventions were excluded. Patient records variable analysis.
were then semiautonomously electronically queried for
procedure type, length of postoperative admission (i.e., Results
days admitted after surgery performed), patient age at time
of surgery, sex, and categorical ASA (American Society of We identified 7294 cranial procedures performed dur-
Anesthesiologists) functional status classification (grades ing 6596 hospital encounters in 5385 patients over the
1–5). We dichotomized ASA classification (grades 1–2 as study period. Postoperative ER use within 30 days was
“low” and grades > 2 as “high”) to reduce patient dilution identified after 957 procedures (rate 13.1 per 100 proce-
in subset analysis. Patients who underwent more than one dures). The rate of ER use varied significantly between
surgical procedure (e.g., tumor resection and CSF shunt procedure types (Table 1, Fig. 1). The average day of re-

J Neurosurg Volume 135 • September 2021 935

 
*49 *
Hardesty et al.

FIG. 1. Relative risk of ER utilization within 30 days of cranial neurosurgery sorted by procedure type, compared to all other
procedures. Bars are 95% confidence intervals. AVF = arteriovenous fistula; AVM = arteriovenous malformation; ICP = intracranial
pressure.

turn for a postoperative ER visit was postoperative day operative ER utilization included younger age (median
12.7 (range 0–29 days). Most patients (74.6%) received 51 vs 57 years, p < 0.0001; Table 3) and shorter hospital
neurosurgical consultation during their ER encounter. length of stay (LOS) during surgical admission (median
Patients were allowed more than one chief complaint 2.5 vs 2.6 days postoperatively, p < 0.001). Patient sex and
for their ER visitation. The two most common chief com- preoperative ASA classification (grades dichotomized
plaints were pain (30.7%) and medical complications such into low, 1–2; and high, > 2) had no significant correlation
as urinary tract infection, constipation, nausea/vomiting, with ER utilization (p > 0.05).
venous thromboembolism, or pneumonia (18.2%). Other
chief complaints (categorized as altered mentation; wound Multivariable Analysis Factors Associated With ER
concern without infection; neurological deficit; seizure; Utilization
fever without infection; wound infection; ground-level
fall; concern for CSF leak from wound, nose, or ear; med- We performed a multivariable backward elimina-
ication-related issues; and other) were less common and tion logistic regression utilizing the univariable signifi-
constituted a myriad of visit reasons distributed across the cant findings of procedure type, patient age, and length
different procedural category types (Table 2). of index hospitalization (Table 4) to calculate odds ra-
tios (ORs). In this model, DBS electrode placement (OR
Univariate Analysis of Factors Associated With 0.488, p = 0.0004), increasing hospital LOS (OR 0.935, p <
Postoperative ER Use 0.0001), and increasing patient age (OR 0.988, p < 0.0001)
were associated with less postoperative ER utilization.
Surgery procedure type was associated with postopera- Ommaya reservoir placement (OR 2.65, p = 0.0008) and
tive ER utilization on univariate analysis, with the highest cranial CSF shunt placement (OR 1.40, p = 0.0001) were
relative risk (RR) among patients undergoing indwelling conversely associated with heightened postoperative ER
CSF reservoir (Ommaya) placement (RR 1.95, p = 0.0016),
or new or revision ventricular CSF shunt placement (RR utilization. Trauma as a procedure type was rendered non-
1.41, p < 0.0001). Deep brain stimulation (DBS) electrode significant in this multivariable model.
placement had a significant protective effect when com-
pared to all other procedure types (RR 0.517, p = 0.0004). 30-Day Readmissions From the ER
In univariable analysis only, trauma (subdural hematoma, Of the 957 cases of ER readmission within 30 days
epidural hematoma, empyema or cerebral abscess, and postoperatively, about one-half (50.1%, n = 480) led to pa-
hemicraniectomy) as a procedure type also was associ- tient readmission. Patients readmitted from the ER were
ated with reduced postoperative ER utilization (RR 0.768, on average older (readmitted median age 55 years vs dis-
p = 0.024). charged median 49 years, p = 0.0012) and more likely to
Patient factors associated with increased 30-day post- be male (men, 56% readmission rate, vs women, 45%;

936 J Neurosurg Volume 135 • September 2021

 
*49 *
 TABLE 2. Reasons for ER utilization, sorted by index surgical procedure type
Pain Medical Altered Wound Neurological Wound CSF Medication Total
Procedure Control Complication Mentation Concern Deficit Seizure Fever Infection Fall Leak Issue Other CCs
Brain biopsy (open or stereotactic) 4 2 3 0 2 2 1 1 1 1 0 0 17
Brain tumor resection 59 54 33 31 31 40 10 15 5 12 5 5 300
Chiari decompression 11 4 0 6 1 0 0 0 2 1 0 0 25
Cranioplasty 0 5 1 3 1 1 1 2 0 0 0 0 14
CSF leak repair 1 0 0 3 0 0 1 1 0 1 0 0 7
CSF shunt placement/revision 120 39 27 20 16 7 9 7 8 1 0 0 254
Cyst resection/fenestration 6 2 0 1 1 1 0 0 0 1 0 0 12
DBS (electrode) 8 8 4 1 1 1 2 1 4 0 0 1 31
Endoscopic ventriculostomy 2 0 1 0 1 0 1 0 0 0 0 0 5
Epilepsy 7 2 2 0 1 6 0 0 0 0 0 0 18
ICP monitor/external ventriculostomy 7 1 0 0 0 0 0 0 0 1 0 0 9
Laser interstitial thermal therapy 2 1 1 0 4 2 0 0 0 0 1 0 11
Microvascular decompression 14 1 0 3 2 0 1 0 0 1 0 0 22
Ommaya reservoir placement 3 3 8 1 0 2 1 0 0 0 0 0 18
Transsphenoidal 18 33 2 9 0 1 2 0 0 1 2 1 69

*49 *
Trauma (EDH/SDH/stroke/abscess) 9 14 16 5 13 11 5 1 5 1 0 1 81
Vascular: aneurysm clipping 26 9 2 2 4 3 3 0 1 2 1 0 53
Vascular: AVF/AVM 4 0 0 4 4 0 1 0 0 0 0 0 13
Vascular: cavernoma resection 10 3 3 1 2 2 1 0 0 0 0 0 22
Vascular: intracranial bypass 2 1 0 3 3 1 0 0 0 0 2 0 12
 
Wound revision 3 5 4 9 0 3 3 7 1 0 0 0 35
Total CCs across all procedure types, no. 316 187 107 102 87 83 42 35 27 23 11 8
Total CCs across all procedure types, % 30.8% 18.2% 10.4% 9.9% 8.5% 8.1% 4.1% 3.4% 2.6% 2.2% 1.1% 0.8%
CC = chief complaint.
Total numbers may add up to more than 957 postoperative encounters because patients may have had more than one presenting complaint.

J Neurosurg Volume 135 • September 2021

937
Hardesty et al.
Hardesty et al.

TABLE 3. Univariable analysis of postoperative ER utilization by patient factors

Variable 30-Day ER Use No ER Use p Value
Total procedures 957 (13.1%) 6337 (86.9%)
Patient age, yrs 51 ± 18 57 ± 18 <0.0001
Sex (M:F), no. 449:508 3143:3194 NS
Index admission LOS, days 2.5 ± 4 2.6 ± 10 <0.0001
ASA grade (% of group)
  Low: 1–2 213 (22.3%) 1336 (21.1%) NS
  High: >2 744 (77.7%) 5001 (78.9%) NS
NS = not significant.
Median values are presented ± SD. Also see Table 1 for analysis by procedure type.

p < 0.001) (Table 5). Hospital LOS during the index surgi- sion rate from the ER. New or worsening neurological
cal admission did not differ significantly between those deficit chief complaint (OR 1.99, p = 0.0088), fever chief
patients readmitted from the ER versus those who were complaint (OR 2.41, p = 0.0205), altered mentation chief
discharged (median 2.6 vs 2.4 days, p > 0.05). Preexisting complaint (OR 2.71, p = 0.0002), patient chronic kidney
chronic kidney disease (stages 2–5) was associated with disease (OR 3.31, p = 0.0037), brain biopsy procedure type
readmission (RR 1.51, p = 0.002), while the other ana- (OR 3.50, p = 0.0398), and wound infection chief com-
lyzed medical comorbidities of atrial fibrillation, diabetes plaint (OR 31.4, p = 0.0008) remained independently sta-
mellitus, and chronic obstructive pulmonary disease were tistically associated with increased rates of readmission
not (all p > 0.05). Index procedure types with statistically from the ER in multivariable analysis. Ground-level fall
significant heightened rates of postoperative readmission chief complaint (OR 0.204, p = 0.0016) and pain control
from the ER when compared to all procedure types were chief complaint (OR 0.594, p = 0.0027) remained inde-
brain biopsy (RR 1.55, p = 0.03) and cranial wound revi- pendently statistically associated with decreased rates of
sion (RR 1.45, p = 0.013; Table 6). Chief complaints asso- readmission from the ER in multivariable analysis.
ciated with increased rates of readmission to the hospital
from the ER were CSF leak (from the wound, nose, or ear; Discussion
RR 1.92, p < 0.0001; Table 6), wound infection (RR 1.87, p
< 0.0001), altered mentation without focal deficit (RR 1.47, We report for the first time the rate of and reasons for ER
p < 0.0001), fever without surgical wound infection (RR utilization in postoperative cranial neurosurgical patients
1.42, p = 0.004), and new or worsening focal neurologi- at a major neuroscience center. We identified the most
cal deficit (RR 1.3, p = 002). Conversely, chief complaints common chief complaints, as well as procedure types, and
associated with significantly decreased rates of readmis- patient factors that were associated with increased or de-
sion from the ER compared to the entire cohort were creased postoperative ER utilization. Overall, patients re-
ground-level fall (RR 0.43, p = 0.002), pain control (RR ceived ER care within 30 days postoperatively after 13.1%
0.69, p < 0.0001), and medical complication (RR 0.86, p of cranial neurosurgical procedures.
= 0.035). We again constructed a backward elimination
logistic regression model (Table 7) to assess the relative Reasons for ER Utilization Within 30 Days of Cranial
contributions of these statistically significant univariable Neurosurgery
findings on hospital readmission from the ER. Within this The most common presenting chief complaint for ER
multivariable model, the factors of patient age, patient sex, visitation was pain control (i.e., headache). In a retrospec-
wound revision procedure type, CSF leak chief complaint, tive review of records, it was difficult at times to ascertain
and medical complication chief complaint were rendered if this was incisional pain or exacerbation of presurgical
nonsignificant (p > 0.05) with regard to hospital readmis- migraine symptoms. An appropriate pain management

TABLE 4. Multivariable analysis of postoperative ER utilization

Variable OR (95% CI) p Value
DBS electrode placement (procedure type) 0.488 (0.327–0.727) 0.0004
Hospital LOS, days (patient characteristic) 0.934 (0.919–0.950) <0.0001
Patient age at op (patient characteristic) 0.988 (0.985–0.992) <0.0001
CSF shunt (procedure type) 1.40 (1.19–1.66) 0.0001
Ommaya reservoir placement (procedure type) 2.65 (1.50–4.70) 0.0008
Results of multivariable backward elimination regression model utilizing factors found statistically significant in univari-
able analysis. ORs < 1 are associated with less postoperative ER utilization, and ORs > 1 are associated with higher
postoperative ER utilization. Hospital LOS and patient age are continuous variables.

938 J Neurosurg Volume 135 • September 2021

 
*49 *
 Hardesty et al.

TABLE 5. Univariable analysis of hospital readmissions after ER utilization

Variable Readmitted From ER Discharged From ER p Value
Postop ER visits, no. 480 477 —
Patient age, yrs 55 ± 18 49 ± 18 0.0012
Sex (M:F), no. 253:227 196:281 <0.001
Index admission LOS, days 2.6 ± 4.0 2.4 ± 3.2 NS
ASA grade (% of group)
  Low: 1–2 96 (20%) 117 (25%) NS
  High: >2 384 (80%) 360 (75%) NS
Median values are presented ± SD. Univariable analysis of subsequent readmissions to the hospital from the ER (n =
480) among the entire study cohort utilizing the ER within 30 days of surgery (n = 957).

TABLE 6. Readmission rates after ER utilization by procedure type and chief complaint
Readmitted RR of
Variable After ER Use, % Readmission p Value
Procedure type
  Brain biopsy (open or stereotactic) 77.80% 1.6 0.03
  Wound revision 72.70% 1.5 0.01
  Ommaya reservoir placement 70.60% 1.4 NS
  Endoscopic ventriculostomy 60.00% 1.2 NS
  Laser interstitial thermal therapy 60.00% 1.2 NS
  CSF leak repair 57.10% 1.1 NS
 Transsphenoidal 56.30% 1.1 NS
  Vascular: aneurysm clipping 54.90% 1.1 NS
  Brain tumor resection 52.90% 1.1 NS
  Trauma (EDH/SDH/stroke/abscess) 46.50% 0.9 NS
 Cranioplasty 46.20% 0.9 NS
  Vascular: cavernoma resection 45.50% 0.9 NS
  CSF shunt placement/revision 45.00% 0.9 NS
  ICP monitor 44.40% 0.9 NS
  Cyst resection/fenestration 41.70% 0.8 NS
  Vascular: AVF/AVM 41.70% 0.8 NS
 Epilepsy 41.20% 0.8 NS
  Chiari decompression 40.00% 0.8 NS
  DBS (electrode) 35.70% 0.7 NS
  Microvascular decompression 33.30% 0.7 NS
  Vascular: intracranial bypass 27.30% 0.5 NS
Chief complaint
  CSF leak 100.00% 1.92 <0.0001
  Wound infection (requiring antibiotics or revision) 97.10% 1.87 <0.0001
  Altered mentation 76.60% 1.47 <0.0001
  Fever w/o wound infection 73.80% 1.42 0.004
  Neurological deficit 67.80% 1.3 0.002
 Seizure 57.80% 1.11 NS
  Medication issue (allergy/side effect, need for refill, etc.) 54.50% 1.05 NS
  Wound concern (not infected) 47.10% 0.9 NS
  Total medical complications (UTI, DVT, A-fib, etc.) 44.90% 0.86 0.035
  Pain control 36.10% 0.69 <0.0001
  Ground-level fall 22.00% 0.43 0.002
  Other (motor vehicle accident, animal bite) 0.00% 0 NS
A-fib = atrial fibrillation; DVT = deep venous thrombosis; UTI = urinary tract infection.

J Neurosurg Volume 135 • September 2021 939

 
*49 *
Hardesty et al.

TABLE 7. Multivariable analysis of hospital readmission after ER utilization

Variable OR (95% CI) p Value
Ground-level fall (chief complaint) 0.204 (0.0764–0.5472) 0.0016
Pain control (chief complaint) 0.594 (0.422–0.835) 0.0027
Neurological deficit (chief complaint) 1.99 (1.19–3.33) 0.0088
Fever (chief complaint) 2.41 (1.14–5.06) 0.0205
Altered mentation (chief complaint) 2.71 (1.62–4.56) 0.0002
Chronic kidney disease (patient characteristic) 3.31 (1.47–7.41) 0.0037
Brain biopsy (procedure type) 3.50 (1.06–11.6) 0.0398
Wound infection (chief complaint) 31.4 (4.21–233.6) 0.0008
Results from a multivariable backward elimination regression model utilizing the factors found statistically significant
in univariable analysis. ORs < 1 are associated with higher rates of readmission after postoperative ER utilization, and
ORs > 1 are associated with lower rates of readmission after postoperative ER utilization.

strategy discussed with the patient at the time of discharge, stantial pre- and postoperative care by a multidisciplinary
sufficient quantity of pain medications, and special atten- team. This paradigm cannot be replicated for all of cranial
tion to those patients undergoing craniotomy with comor- neurosurgery, but further study of DBS patients may eluci-
bid migraines may reduce this rate. Aggressive outpatient date improvements that can be generalized.
management protocols such as the ability to obtain same- Although medical complication was a relatively com-
day outpatient cranial imaging in nontoxic patients may mon chief complaint, as discussed previously, increasing
also reduce this ER use rate; many of the ER visits for pain patient medical comorbidities as measured by the ASA
control may be driven by concern for hydrocephalus, hem- physical condition scale commonly used by anesthesiolo-
orrhage, or other etiologies that could be ruled out with gists did not predict postoperative ER utilization in mul-
outpatient imaging. tivariable analysis. The ASA is a relatively broad catego-
We found the second most common reason to present rization and is not neurosurgery specific, and we propose
to the ER within 30 days of a neurosurgical procedure that a new scoring system is required to predict ER uti-
was in fact for nonsurgical medical conditions. Many of lization and/or hospital readmission after cranial neuro-
these events, such as urinary tract infections, constipa- surgery. This subject requires additional study and vali-
tion, pneumonia, atrial fibrillation, and venous thrombo- dation. We also found that older patients were somewhat
embolism, were likely due to preexisting conditions that less likely to return to the ER. Elderly patients may have a
were exacerbated by surgery. This “catch-all” category higher threshold for ER return, or perhaps lacked physical
comprises both serious and relatively less serious medical mobility to present to our institution’s ER and instead had
complaints, many of which fall outside the expertise of the postoperative complications managed at a local ER or by
neurosurgeon. Therefore, postoperative collaboration with their primary physician.
hospitalist services and/or the patient’s primary care pro- We found an association between decreased LOS dur-
vider may lead to improvements in this regard. ing the index surgical admission and increased subsequent
risk of ER return within 30 days, and this association was
Risk Factors for ER Use Within 30 Days of Cranial maintained in multivariable analysis; nevertheless, the
Neurosurgery absolute difference (average 0.1 days) was quite small.
There was a correlation between cranial neurosurgical We presume that any noncontrolled confounders, such as
procedure types and an increased utilization of the hospi- complexity of surgery within a procedure class or medical
tal ER within 30 days of surgery. On multivariable analy- illness not otherwise categorized by analyzed variables,
sis, patients undergoing Ommaya reservoir placement and would actually increase both the index LOS and the rate of
those with cranial shunt placement/revision had signifi- postoperative ER return. Therefore, this finding may rep-
cantly increased ER utilization odds ratios. Conversely, resent the downside of the drive to reduce hospital LOS
patients with DBS electrode placement had a much lower and discharge patients as soon as possible after cranial sur-
rate of ER utilization after surgery. We believe these find- gery—an inadvertent increase in ER utilization.
ings reflect the intrinsic medical conditions and patient
health status behind these interventions. Patients under- Hospital Readmission as a Consequence of Postoperative
going Ommaya reservoir placement at our institution are ER Visitation
usually suffering from metastatic cancer or disseminated Variables that were associated in multivariable analy-
coccidioidomycosis, and these conditions place them at sis with an increased risk of inpatient readmission once a
high risk for substantial healthcare resource utilization. patient reached the hospital ER within 30 days of cranial
Anecdotally, we believe that recent CSF diversion tends to surgery included preexisting chronic kidney disease, the
increase ER utilization due to the fear of shunt failure, and procedure type of brain biopsy, and various chief com-
our results from the present study seem to confirm this. In plaints: wound infection, altered mentation without focal
contrast, patients undergoing elective deep brain stimulat- deficit, fever without surgical wound infection, and new
ing electrode placement are carefully selected with sub- or worsening focal neurological deficit. None of these
940 J Neurosurg Volume 135 • September 2021
 
*49 *
 Hardesty et al.

findings represented to us a “magic bullet” category on inpatient readmission. These patients may benefit the most
which to intervene to reduce conversion of ER visits into from a postoperative care pathway that can educate and
readmissions. We noted a discrepancy between the rela- intervene on an outpatient basis, without requiring patient
tive simplicity of brain biopsy as a procedure type and a emergency department utilization. We found that ground-
heightened rate of hospital readmission if brain biopsy pa- level falls and pain control as chief complaints were as-
tients utilize the ER after surgery. Our interpretation of sociated with reduced ER-to-hospital readmission rates.
this finding is that most of those hospital readmissions are Although ground-level falls were associated with a sig-
not due to surgical morbidity from the biopsy, but due to nificantly lower rate of readmission after ER utilization
the underlying patient factors and disease processes that in both univariable and multivariable (OR 0.204) analysis,
prompted a brain biopsy. Reducing conversions of ER many of these visits are likely appropriate after a periop-
visitations into hospital readmissions within 30 days of erative mechanical trauma, and fall as a chief complaint
cranial surgery (and expedited admission or discharge for was a relatively rare incidence (2.6% of all chief com-
high- and low-risk patients) is a topic worthy of significant plaints). The reduction of falls in postoperative patients,
further study. especially in the elderly, is a worthy goal, but based on
our data will not significantly impact ER visitation after
Study Strengths and Limitations surgery compared to more common chief complaints. In
The strengths of the present study include a large patient contrast, pain control was the most common chief com-
population undergoing a diverse set of procedure types, plaint for ER utilization (30.7%) and was also associated
as well as granular individual patient data due to our uni- with a significantly reduced risk of readmission after ER
fied inpatient and ER medical record system. The primary visitation in both univariable and multivariable (OR 0.594)
weakness of our study is that it is a single-institutional se- analysis. This provides further justification for a robust
ries, and so generalizability to other medical centers may postoperative pain control protocol, as well as impetus for
be limited. Our study was only able to capture ER visits to a standardized protocol to provide expedited disposition
our own ER facility. Patients from other cities or states who for these patients once in the ER after appropriate screen-
sought care at local facilities are therefore not captured. ing for serious complications such as hydrocephalus, intra-
Our rates of ER utilization reported here are subsequently cranial hemorrhage, or infection.
less than the true rates. The alternative to this study design
would be a state- or nationwide database study, but these Conclusions
de-identified data sets are quite limited by the granular- We report a postoperative ER utilization rate of 13.1%
ity of the data. There does not presently exist a publicly after cranial neurosurgery procedures. The most common
available database that includes the same level of detail reason for return to the ER postoperatively was pain con-
regarding individual patient ER utilization compared to trol. Risk factors for ER visits were identified based on
our present study. Our multivariable analysis examining multivariable analysis including procedure type of Omma-
relevant factors of patients readmitted from the ER (n = ya reservoir placement and CSF shunt placement/revision;
480) produced relatively wide confidence intervals; large factors associated with less postoperative ER use included
studies are required to detect statistically significant dif- DBS electrode placement, older patient age, and longer in-
ferences within rare events when patients are split among dex admission LOS. We also identified chief complaints,
so many procedure types and chief complaints. Of great procedural types, and patient variables that correlated with
socioeconomic interest would be the cost associated with increased or decreased inpatient readmission upon ER vis-
postoperative ER utilization; however, our hospital elec- itation. These findings will direct future quality improve-
tronic medical record did not allow for autonomous bill- ment initiatives and reduce ER utilization via prospective
ing (charges or reimbursement) data acquisition and so we implementation of care pathways for high-risk patients.
were unable to study this as a factor. Future studies should Further protocols for pain management postoperatively,
examine such costs further. Last, we only analyzed the especially in those patients with preexisting migraines, are
first ER visit encounter within 30 days following surgery, also warranted.
and patients rarely presented to the ER more than once
during this time period. These multivisit patients may be
an appropriate target for further outpatient intervention to References
lessen their healthcare resource utilization.   1. Han S, Smith TS, Gunnar W. Descriptive analysis of 30-day
readmission after inpatient surgery discharge in the Veterans
Health Administration. JAMA Surg. 2014;​149(11):​1162–1168.
Future Directions   2. Leppin AL, Gionfriddo MR, Kessler M, et al. Preventing
Some returns to the emergency department within 30 30-day hospital readmissions:​a systematic review and meta-
days are inarguably medically appropriate, such as in the analysis of randomized trials. JAMA Intern Med. 2014;​174(7):​
setting of pulmonary embolism, intracranial hemorrhage, 1095–1107.
or CSF shunt failure. These findings often require read-   3. Jencks SF, Williams MV, Coleman EA. Rehospitalizations
mission to the hospital ward or intensive care unit. The among patients in the Medicare fee-for-service program. N
Engl J Med. 2009;​360(14):​1418–1428.
reduction of these serious perioperative complications   4. Bohl MA, Ahmad S, Jahnke H, et al. Delayed hyponatremia
is a worthy quality improvement initiative but is not the is the most common cause of 30-day unplanned readmission
focus of the present paper. Herein, we sought to identify after transsphenoidal surgery for pituitary tumors. Neurosur-
chief complaints or procedure classes that are associated gery. 2016;​78(1):​84–90.
with discharge from the ER rather than conversion to an   5. Buchanan CC, Hernandez EA, Anderson JM, et al. Analysis

J Neurosurg Volume 135 • September 2021 941

 
*49 *
Hardesty et al.

of 30-day readmissions among neurosurgical patients:​surgi- Author Contributions

cal complication avoidance as key to quality improvement. J Conception and design: Hardesty, Mooney, Little. Acquisition
Neurosurg. 2014;​121(1):​170–175. of data: Hardesty, Mooney, Hendricks, Catapano, Brigeman,
  6. Cusimano MD, Pshonyak I, Lee MY, Ilie G. A systematic Bohl, Sheehy. Analysis and interpretation of data: Hardesty,
review of 30-day readmission after cranial neurosurgery. J Mooney, Little. Drafting the article: Hardesty, Little. Critically
Neurosurg. 2017;​127(2):​342–352. revising the article: all authors. Reviewed submitted version
  7. Dickinson H, Carico C, Nuño M, et al. Unplanned readmis- of manuscript: all authors. Approved the final version of the
sions and survival following brain tumor surgery. J Neuro- manuscript on behalf of all authors: Hardesty. Statistical analysis:
surg. 2015;​122(1):​61–68. Hardesty. Administrative/technical/material support: Little. Study
  8. Marcus LP, McCutcheon BA, Noorbakhsh A, et al. Incidence supervision: Hardesty, Little.
and predictors of 30-day readmission for patients discharged
home after craniotomy for malignant supratentorial tumors in Supplemental Information
California (1995-2010). J Neurosurg. 2014;​120(5):​1201–1211.
  9. Moghavem N, Morrison D, Ratliff JK, Hernandez-Boussard Previous Presentations
T. Cranial neurosurgical 30-day readmissions by clinical Portions of this paper were presented as a socioeconomic ses-
indication. J Neurosurg. 2015;​123(1):​189–197. sion podium talk at the 2017 AANS Annual Meeting in Los
10. Taylor BE, Youngerman BE, Goldstein H, et al. Causes and Angeles, CA.
timing of unplanned early readmission after neurosurgery.
Neurosurgery. 2016;​79(3):​356–369. Correspondence
Douglas A. Hardesty: The Ohio State University Wexner Medical
Center, Columbus, OH. [email protected].
Disclosures
Dr. Bohl reports ownership of SurgiSTUD, LLC, being a patent
holder in Dignity Health, and being a consultant for DePuy-
Synthes Spine. Dr. Little reports ownership of Kogent and being a
consultant for SPIway.

942 J Neurosurg Volume 135 • September 2021

 
*49 *
 CLINICAL ARTICLE
J Neurosurg 135:943–948, 2021

Gender diversity in United States neurosurgery training

programs
Katelyn Donaldson, MD,1 Katherine E. Callahan,1 Aaron Gelinne, MD,2 Wyll Everett,1
S. Elizabeth Ames, MD,3 Ellen L. Air, MD, PhD,4 and Susan R. Durham, MD, MS5
1
University of Vermont College of Medicine; 3Department of Orthopaedics and Rehabilitation and 5Division of Neurosurgery,
University of Vermont College of Medicine, Burlington, Vermont; 4Department of Neurosurgery, Henry Ford Health System,
Detroit, Michigan; and 2Department of Neurosurgery, University of North Carolina, Chapel Hill, North Carolina

OBJECTIVE  Neurosurgery continues to be one of the medical specialties with the lowest representation of females in
both the resident and faculty workforce. Currently, there are limited available data on the gender distribution of faculty
and residents in Accreditation Council for Graduate Medical Education (ACGME)–accredited neurosurgery training
programs. This information is critical to accurately measure the results of any effort to improve both the recruitment and
retention of women in neurosurgery. The objective of the current study was to define the current gender distribution of
faculty and residents in ACGME-accredited neurosurgery training programs.
METHODS  Data publicly available through institutional and supplemental websites for neurosurgical faculty and
residents at ACGME-accredited programs were analyzed for the 2017–2018 academic year. Data collected for faculty
included gender, age, year of residency graduation, academic rank, h-index, American Board of Neurological Surgery
certification status, and leadership positions. Resident data included gender and postgraduate year of training.
RESULTS  Among the 109 ACGME-accredited neurosurgical residency programs included in this study, there were
1350 residents in training, of whom 18.2% were female and 81.8% were male. There are 1320 faculty, of whom 8.7%
were female and 91.3% were male. Fifty-eight programs (53.2%) had both female faculty and residents, 35 programs
(32.1%) had female residents and no female faculty, 4 programs (3.7%) had female faculty and no female residents, and
6 programs (5.5%) lacked both female residents and faculty. Six programs (5.5%) had incomplete data. Female faculty
were younger, had lower h-indices, and were less likely to be board certified and attain positions of higher academic rank
and leadership.
CONCLUSIONS  This study serves to provide a current snapshot of gender diversity in ACGME-accredited neurosur-
gery training programs. While there are still fewer female neurosurgeons achieving positions of higher academic rank
and serving in leadership positions than male neurosurgeons, the authors’ findings suggest that this is likely due to the
small number of women in the neurosurgical field who are the farthest away from residency graduation and serves to
highlight the significant progress that has been made toward achieving greater gender diversity in the neurosurgical
workforce.
https://thejns.org/doi/abs/10.3171/2020.7.JNS192647
KEYWORDS  gender; neurosurgery; diversity

W
omen accounted for nearly one-half of all medi- the most recent report by the American Association of
cal school graduates in the US in 2018, and fe- Medical Colleges, joining orthopedic surgery as the low-
male residents currently account for nearly half est among all medical specialties.1 The gender gap in neu-
of the total resident workforce in all specialties in US resi- rosurgery widens over the course of a neurosurgery career,
dency programs.1,2 Despite this, the number of female resi- as women currently compose only 7.4% of all practicing
dents and attending physicians in neurosurgery remains board-certified neurosurgeons within the US.3 Female
one of the lowest among all medical specialties. Women neurosurgeons have been shown to have a higher rate of
account for 17.7% of neurosurgery residents according to attrition during residency and a lower rate of board certi-

ABBREVIATIONS  ABNS = American Board of Neurological Surgery; ACGME = Accreditation Council for Graduate Medical Education.
SUBMITTED  January 7, 2020.  ACCEPTED  July 31, 2020.
INCLUDE WHEN CITING  Published online January 29, 2021; DOI: 10.3171/2020.7.JNS192647.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  943

*49 *
Donaldson et al.

fication than males, both factors that profoundly influence TABLE 1. Distribution of female neurosurgical residents and
the gender distribution of the neurosurgical workforce.4,5 faculty
Efforts to improve the gender gap in neurosurgery have Total No. of F (%) No. of M (%)
recently been undertaken, as the benefits of diversity in
the workforce are well described in both medicine and Residents 1350 246 (18.2) 1104 (81.8)
other disciplines, particularly business.6–9 A diverse work- Faculty 1320 115 (8.7) 1205 (91.3)
force contributes a broader range of ideas, opinions, and Program directors 109 7 (6.4) 102 (93.6)
strategies that can lead to superior problem-solving and Department chair/division chief 109 3 (2.8) 106 (97.2)
productivity.10 Work environments with greater gender di-
versity typically have more inclusive cultures and, subse-
quently, reduced employee turnover and improved reten-
tion.10 Moreover, employees working within an inclusive
and equitable environment demonstrate greater satisfac- Data Analysis
tion, engagement, and performance.10 The recruitment and There were 109 ACGME-accredited neurosurgery
retention of a diverse neurosurgical workforce will serve residency programs training Doctor of Medicine resi-
to benefit the specialty of neurosurgery and the patients dents identified at the time of data collection. One pro-
that it serves. gram training Doctor of Osteopathic Medicine residents
There are limited data available on the gender distribu- was not included. Six programs had incomplete data,
tion of neurosurgery residents and faculty in Accreditation and available data were used as appropriate. The follow-
Council for Graduate Medical Education (ACGME)–ac- ing methods were used for the comparisons of male and
credited neurosurgery residency programs. It is necessary female faculty. A two-sample t-test was used to compare
to understand the current gender landscape in neurosur- gender and age. The mean number of years since resi-
gery residency programs to accurately measure the effect dency graduation was compared by gender and academic
of any effort to improve gender diversity in the neurosur- rank using ANOVA. The gender by academic rank inter-
gery workforce. The objective of the current study was to action term was included to test whether gender differenc-
provide baseline data of the gender distribution of faculty es were dependent on academic rank. Logistic regression
and residents in ACGME-accredited neurosurgery resi- models were used to assess the effect of gender on both
dency programs during the 2017–2018 academic year. academic rank and ABNS certification. Since academic
rank has three levels, a cumulative logit model was fit with
Methods the lowest academic rank as the reference level. The ef-
Data Collection fect of gender and academic rank on h-index was assessed
using ANOVA with gender and academic rank and their
This study was granted IRB exemption approval by interaction in the model. Since academic rank, board cer-
the University of Vermont Committee on Human Sub- tification, and h-indices are affected by years since resi-
jects Research Protection. Data collection began in the dency graduation, additional analyses were run for these
fall of 2017. At that time, there were 110 ACGME-ac- measures, with years since residency graduation included
credited neurosurgery residency programs identified via as a covariate in the models. Years since residency gradua-
the ACGME website (https://apps.acgme.org/ads/Public/ tion were missing for 355 of the faculty. To ensure that the
Programs/Search). Data on neurosurgery faculty and reduced sample data were consistent with the full sample,
residents were collected from publicly available residency unadjusted analyses were replicated in the reduced sample
program and academic medical institution websites. Data of faculty with years since residency graduation, which
not found on institutional websites were augmented using produced similar results to those of the full sample. All
alternative secondary publicly available websites such as analyses were performed using SAS version 9 statistical
Healthgrades (www.healthgrades.com) and Vitals (www. software (SAS Institute). Statistical significance was de-
vitals.com). Data collected for residents included gender, termined based on α = 0.05.
as determined by name, photo, and profile, and postgradu-
ate year. Data collected for faculty members included each
physician’s gender, as determined by name, photo, and pro- Results
file; age; years since residency graduation; academic rank; Program Characteristics
h-index; and American Board of Neurological Surgery A total of 109 ACGME-accredited neurosurgery resi-
(ABNS) certification status. The genders of the residency dency programs were included for analysis (Table 1). There
program director and of the department chair or division were 1350 neurosurgery residents in training, of whom
chief were also collected for each program. Faculty were 246 (18.2%) were female and 1104 (81.8%) were male.
defined as physicians who had completed a neurosurgery Fifty-eight programs (53.2%) had both female faculty and
residency and held the academic rank of assistant, associ- residents, 35 programs (32.1%) had female residents and
ate, or full professor. Adjunct faculty were not included in no female faculty, 4 programs (3.7%) had female faculty
the analysis. Board certification status was obtained from and no female residents, and 6 programs (5.5%) lacked
the ABNS website (https://abns.org/find-a-neurosurgeon/). both female residents and faculty. Six programs (5.5%)
The h-index was determined (https://www.scopus.com/ had incomplete data. Seven programs (6.4%) had female
freelookup/form/author.uri) to measure neurosurgery fac- residency program directors. Three programs (2.8%) had
ulty research publication productivity and impact. female department chairs or division chiefs.

944 J Neurosurg Volume 135 • September 2021

 
*49 *
 Donaldson et al.

TABLE 2. Gender differences among faculty: unadjusted analysis

Male Female p Value
Total faculty, n (%) 1205 (91.3) 115 (8.7)
Age in yrs, mean ± SE 52.1 ± 0.3 44.8 ± 0.8 <0.001
Yrs since residency graduation, LS means ± SE*
  All academic ranks 17.0 ± 0.3 12.9 ± 1.1 <0.001
  Assistant professor 9.3 ± 0.5 4.9 ± 1.5
  Associate professor 15.4 ± 0.6 13.2 ± 1.8
  Full professor 26.2 ± 0.5 20.6 ± 2.3
Academic rank, n (%)† <0.001
  Assistant professor 437 (38.9) 60 (54.1)
  Associate professor 281 (25.0) 31 (27.9)
  Full professor 405 (36.1) 20 (18.0)
ABNS certification, n (%) 943 (78.3)   69 (60.0) <0.001
h-index, LS means ± SE*
  All academic ranks 18.6 ± 0.4 14.8 ± 1.3 0.005
  Assistant professor 9.0 ± 0.6 6.4 ± 1.7
  Associate professor 15.2 ± 0.7 12.7 ± 2.2
  Full professor 31.7 ± 0.6 25.4 ± 2.8
LS = least-squares.
* Least-squares means and standard error from ANOVA and ANCOVA models.
† Data were missing for 1123 males and 111 females.

Academic Neurosurgery Faculty Characteristics using ANOVA models, gender differences among the fac-
There were 1320 neurosurgery faculty, of whom 115 ulty were no longer significant. We did not find that female
(8.7%) were female and 1205 (91.3%) were male (Table 2). faculty were less likely to be in a higher academic rank
The mean age of female faculty was 44.8 ± 0.8 (mean ± than male faculty (OR 1.0, 95% CI 0.6–1.7; p = 0.90) or
SE) years, which was significantly lower than that of male less likely to be ABNS certified than male faculty (OR
faculty, who had a mean age of 52.1 ± 0.3 years (t = 5.86, 0.6, 95% CI 0.3–1.1; p = 0.09). In addition, the h-index
df = 1044; p < 0.001). Female faculty also had a signifi- did not differ significantly between female (16.1 ± 1.6) and
cantly lower mean number of years since residency gradu- male (19.1 ± 0.5) faculty when adjusting for years since
ation (12.9 ± 1.1 years) than males (17.0 ± 0.3 years) across residency graduation (F = 3.26, df = 1, 875; p = 0.07). The
all academic ranks (F = 12.96, df = 1, 908; p < 0.001). For overall effect of academic rank continued to be significant
both male and female faculty, the mean number of years on h-index (F = 38.91, df = 2, 875; p < 0.001), with the h-
since residency graduation increased with increasing aca- index increasing with increasing academic rank for both
demic rank (F = 69.8, df = 2, 908; p < 0.001). The gender genders. The gender by academic rank interaction was not
by academic rank interaction was not significant (F = 0.74, significant (F = 0.20, df = 2, 875; p = 0.82) (Table 3).
df = 2, 908; p = 0.48). Using the available data, 54% (n =
60) of female faculty were assistant professors, 28% (n = Discussion
31) were associate professors, and 18% (n = 20) were full Gender Diversity in the Academic Neurosurgical
professors. Likewise, 39% (n = 437) of male faculty were Workforce
assistant professors, 25% (n = 281) were associate profes- The current study demonstrates continued progress on
sors, and 36% (n = 405) were full professors. Unadjusted eliminating gender disparity in the neurosurgical work-
analysis of gender differences among the faculty revealed force as female residents now account for 18.2% of the
that female faculty were less likely to hold positions of total neurosurgery resident workforce, which is more than
higher academic rank (OR 0.5, 95% CI 0.3–0.7; p < 0.001), triple the number of female neurosurgery residents in
were less likely to obtain ABNS certification (OR 0.4, training in the early 1980s.11 In addition, the percentage
95% CI 0.3–0.6; p < 0.001), and had a lower mean h-index of female faculty at academic neurosurgical programs is
(14.8 ± 1.3) compared with males (18.6 ± 0.4) across all now 8.7%, which also demonstrates an increase, although
academic ranks (F = 7.78, df = 1, 1169; p = 0.005). There at a somewhat slower pace since the late 1990s, at a time
was no significant gender by academic rank interaction to when females accounted for only 5% of practicing neuro-
suggest that the difference between female and male fac- surgeons.12 While the increasing number of female faculty
ulty in h-index differed by academic rank (F = 0.69, df = neurosurgeons and residents over the past several decades
2, 1169; p = 0.69) (Table 2). is encouraging, the rate of increase lags behind most other
When adjusting for years since residency graduation specialties, including many surgical subspecialties such

J Neurosurg Volume 135 • September 2021 945

 
*49 *
Donaldson et al.

TABLE 3. Gender differences among neurosurgery faculty adjusted for years since residency graduation
Male (n = 835) Female (n = 79) OR (95% CI) p Value
Academic rank, n (%) Male: ref 0.90
Female: 1.0 (0.6–1.7)
  Assistant professor 302 (36.2) 38 (48.1)
  Associate professor 203 (24.3) 25 (31.6)
  Full professor 330 (39.5) 16 (20.3)
ABNS certification, n (%) 679 (81.3) 50 (63.3) Male: ref 0.09
Female: 0.6 (0.3–1.1)
h-index, LS mean ± SE*
  All academic ranks 19.1 ± 0.5 16.1 ± 1.6 0.07
  Assistant professor 10.7 ± 0.9 8.4 ± 2.2
  Associate professor 15.6 ± 0.9 13.5 ± 2.6
  Full professor 30.9 ± 0.8 26.4 ± 3.2
* Least-squares means and standard error from ANOVA and ANCOVA models.

as general surgery and otolaryngology, where 40.1% and translate to the distribution of female faculty physicians
36.2% of the resident workforce, respectively, are female.1 by academic rank. Females currently account for 34% of
Over the past 3 decades, efforts to increase the number of associate professors and 21% of full professors across all
females applying to neurosurgery programs appear to have medical specialties.17 The gender gap in academic rank is
been successful, with the number of female applicants to even wider in surgical fields, in which females constitute
neurosurgical training programs increasing from an aver- 13.8% of associate professor and 7% of full professor po-
age of 13% in the 1990s to an all-time high of nearly 30% sitions among academic surgical faculty.18 In the current
in the 2019 match.13,14 study, we found that females accounted for 12.1% of as-
With more women applying to and entering neurosur- sistant professors, 9.9% of associate professors, and 4.7%
gery residency programs, strategies should continue to of full professors of neurosurgery, which is similar to the
be developed to maintain the number of females in the gender profile of many other surgical subspecialties. It is
neurosurgical workforce by reducing attrition of female important to note that in the current study, when adjusted
neurosurgeons both during residency and while in prac- for years after residency graduation, there were no dif-
tice. Previous studies have demonstrated an increased rate ferences between genders in the proportions of females
of attrition among female neurosurgeons compared with to males in each academic rank. A similar finding was
their male counterparts.4,5 Increased attrition of female also reported in a 2019 study by Dossani et al. in which
neurosurgeons would have a detrimental effect on the fu- they assessed gender disparities in academic rank within
ture of the neurosurgical workforce. Many factors, such neurosurgery and found fewer women in higher academic
as lifestyle, work hours, and emotional burden, have been ranks; as with the current study, these differences between
suggested as contributing to a higher attrition rate among genders were eliminated when adjusted for years in prac-
female neurosurgeons in all stages of their careers; how- tice.19 Taken together, data from both of these studies sug-
ever, there are currently very few objective data available gest that the current gender differences in academic rank
on this subject.12,15 A survey of European neurosurgeons among neurosurgical faculty may be explained in part by
found no significant differences between genders regard- a “pipeline effect,” which is representative of the few fe-
ing work hours, time in the operating room, or overall males who entered the neurosurgical workforce decades
administrative burden. Only the time spent on scientific ago and are now in advanced stages of an academic career,
work was greater in men than women.16 Organized neu- and that gender differences in academic rank among neu-
rosurgeons may disappear as more and more females enter
rosurgery, particularly the Section on Women in Neu-
the neurosurgical workforce.
rosurgery of the American Association of Neurological Leadership follows academic promotion, and with
Surgeons and the Congress of Neurological Surgeons, has fewer women in the higher academic ranks, it follows that
played a critical role in addressing barriers to recruitment there will be fewer women in leadership positions. Hav-
and retention of females in neurosurgery in an effort to ing females in visible leadership positions is one strategy
improve gender diversity in the neurosurgical workforce. that has been proposed to improve female faculty recruit-
ment and retention. Neurosurgery as a specialty has rec-
Academic Rank and Leadership Among Faculty ognized the need for greater female representation among
Gender disparities exist in academic progression across leadership in the field, and, over the past 15 years, there
the medical professions, and these differences are more has been a considerable increase in the number of females
acutely highlighted within surgical fields. While females within leadership positions in both academic programs
are increasingly represented in medical school classes and organized neurosurgery.20 There were no female de-
and residency programs, this narrative often does not partment chairs/division chiefs of neurosurgery residency

946 J Neurosurg Volume 135 • September 2021

 
*49 *
 Donaldson et al.

programs until 2004, when Dr. Karin Muraszko was ap- sional benchmarks of a neurosurgical career. Previous
pointed the first female department chair at the University studies have reported lower ABNS certification rates for
of Michigan.12 At the time of data collection for this study female neurosurgeons than for male neurosurgeons, al-
in 2017–2018, there were only 3 female department chairs/ though the reasons for this reported gender difference are
division chiefs (University of Michigan, University of Ver- not well defined.5 It is unclear whether this is due to fewer
mont, and University of California, Los Angeles), with the females choosing to become board certified, either by per-
latter two being named in 2017. Since the time the data sonal choice or having exceeded the time limit to apply
were collected for this study, 2 additional female depart- for board certification due to taking time off for maternity
ment chairs have been named (University of Nebraska and leave or having a reduced working schedule. It is also un-
University of Indianapolis), now bringing the total to 5. In clear whether there are any gender-specific differences in
addition, there are now 8 female residency program direc- the pass/fail rates of the ABNS certification examination.
tors, a number that also has been steadily climbing over The ABNS has issued a statement on family and medi-
the past decade. cal leave to address potential delays in ABNS certification
In organized neurosurgery, Dr. Shelly Timmons was due to taking time off for maternity leave or working a
named as the first female president of the American As- reduced schedule permitting extension on a case-by-case
sociation of Neurological Surgeons in 2018, and in 2019, basis.23 In the present study, we also found that female
Dr. Karin Muraszko was named the first female president neurosurgery faculty were less likely than male neuro-
of the Society of Neurological Surgeons in the nearly surgery faculty to be ABNS certified; however, when ad-
100-year history of the society. More females than ever justed for years since residency graduation, there was no
before now serve on the various executive committees of longer any significant difference in ABNS certification
national neurosurgical organizations, and efforts are be- between genders. Again, this suggests that gender differ-
ing undertaken to promote diversity among the speaking ences in ABNS certification rates are likely representative
faculty at national neurosurgical meetings. As more and of the increased number of females in early stages of their
more female neurosurgeons take on leadership roles in careers who are not yet eligible for ABNS certification.
both academic and organized neurosurgery, they serve
not only as important role models and mentors for female Limitations
medical students and neurosurgery residents but represent The current study was limited by the variability and ac-
neurosurgery on the national stage as an increasingly gen- curacy of data available on institutional websites. The data
der-balanced specialty of medicine. presented are a single snapshot of the academic neurosur-
gical workforce during late 2017 and 2018 and are subject
Academic Productivity Among Faculty to many changes with time. This highlights the need to
Both clinical and academic productivity define an aca- maintain a more complete database of resident and physi-
demic neurosurgical career. Academic productivity is of- cian characteristics to better determine the changing gen-
ten defined by the number and quality of one’s academic der landscape within the field.
publications. A 2016 study by Mueller et al. examined gen-
der disparities in scholarly productivity of US academic Conclusions
surgeons and found that female surgeons who were assis-
tant and full professors had significantly lower h-indices This study serves to provide a snapshot of gender di-
than their male counterparts of the same academic rank.21 versity in ACGME-accredited neurosurgery training
Sing et al. found that, although female representation in programs and highlights the significant progress that has
academic spine research doubled over the past 4 decades, been made toward achieving greater gender diversity in
females were half as likely as males to continue publish- the neurosurgical workforce. While there are still fewer
ing after 5 years and published half as many articles as se- female neurosurgeons achieving positions of higher aca-
nior authors.22 In the current study, the unadjusted analysis demic rank and serving in leadership positions than male
also found that the h-index for female neurosurgeons was neurosurgeons, this study suggests that these differences
lower than that for male neurosurgeons at all academic are likely due to the small number of women in neurosur-
ranks; however, after adjusting for years since residency gery in the later years of an academic career. Continued
graduation, the difference was no longer statistically sig- study of the barriers to recruitment and retention of fe-
nificant. This suggests that gender differences in academic males in neurosurgery are warranted to continue the trend
productivity among neurosurgeons are likely due to more toward greater gender diversity in the neurosurgical work-
females being at the earlier stages of their academic ca- force.
reers with correspondingly lower h-indices. As more fe-
males enter advanced stages of an academic career in the Acknowledgments
decades to come, further analysis of gender differences We would like to thank Peter Callas and Joan Skelly for sta-
in academic productivity should be undertaken to iden- tistical support and analysis. We would like to acknowledge Jayne
tify and eliminate any barriers to academic productivity Manigrasso for her contributions to this project.
among female neurosurgeons.
References
ABNS Certification   1. ACGME Residents and Fellows by Sex and Specialty, 2017.
ABNS certification is the one of the defining profes- American Association of Medical Colleges. Accessed

J Neurosurg Volume 135 • September 2021 947

 
*49 *
Donaldson et al.

September 2, 2020. https:​//www.aamc.org/data/workforce/ 17. Girod S, Fassiotto M, Grewal D, et al. Reducing implicit gen-
reports/492576/2-2-chart.html der leadership bias in academic medicine with an educational
  2. Table B-2. 2:​Total Graduates by U.S. Medical School and intervention. Acad Med. 2016;​91(8):​1143–1150.
Sex, 2013-2014 through 2017-2018. American Association of 18. Blumenthal KG, Huebner EM, Banerji A, et al. Sex differ-
Medical Colleges. Accessed September 2, 2020. https:​//www. ences in academic rank in allergy/immunology. J Allergy
aamc.org/download/321532/data/factstableb2-2.pdf Clin Immunol. 2019;​144(6):​1697–1702.e1.
  3. Table 1.3 Number and percentage of active physicians by sex 19. Dossani RA, Terrel D, Kosty JA, et al. Gender disparities in
and specialty. American Association of Medical Colleges. academic rank achievement in neurosurgery:​a critical as-
Accessed September 2, 2020. https:​//www.aamc.org/data/ sessment. J Neurosurg. 2020;133(6):1922–1927.
workforce/reports/492560/1-3-chart.html 20. Lautenberger DM, Raezer CL, Sloane RA. The State of
  4. Agarwal N, White MD, Pannullo SC, Chambless LB. Analy- Women in Academic Medicine:​The Pipeline and Pathways
sis of national trends in neurosurgical resident attrition. J to Leadership. Association of American Medical Col-
Neurosurg. 2019;​131(5):​1668–1673. leges;​2014. Accessed September 25, 2020. https:​//www.
  5. Lynch G, Nieto K, Puthenveettil S, et al. Attrition rates in hopkinsmedicine.org/women_science_medicine/_pdfs/
neurosurgery residency:​analysis of 1361 consecutive resi- the%20state%20of%20women%20in%20academic%20
dents matched from 1990 to 1999. J Neurosurg. 2015;​122(2):​ medicine%202013-2014%20final.pdf
240–249. 21. Mueller CM, Gaudilliere DK, Kin C, et al. Gender disparities
  6. Butkus R, Serchen J, Moyer DV, et al. Achieving gender in scholarly productivity of US academic surgeons. J Surg
equity in physician compensation and career advancement:​ Res. 2016;​203(1):​28–33.
a position paper of the American College of Physicians. Ann 22. Sing DC, Jain D, Ouyang D. Gender trends in authorship of
Intern Med. 2018;​168(10):​721–723. spine-related academic literature—a 39-year perspective.
  7. Dezsö CL, Ross DG. Does female representation in top man- Spine J. 2017;​17(11):​1749–1754.
agement improve firm performance? A panel data investiga- 23. Statement on family and medical leave. American Board of
tion. Strateg Manage J. 2012;​33(9):​1072–1089. Neurological Surgery. Accessed September 25, 2020. https:​//
  8. Hossain M, Farooque OA, Momin MA, Almotairy O. abns.org/family-and-medical-leave/
Women in the boardroom and their impact on climate change
related disclosure. Soc Responsib J. 2017;​13(4):​828–855.
  9. Zhuwao S, Ngirande H, Ndlovu W, Setati ST. Gender diver- Disclosures
sity, ethnic diversity and employee performance in a South The authors report no conflict of interest concerning the materi-
African higher education institution. SA J Hum Resour Man- als or methods used in this study or the findings specified in this
ag. 2019;​17(4):​a1061. paper.
10. Badal S. The business benefits of gender diversity. Gallup.
January 20, 2014. Accessed September 25, 2020. https:​// Author Contributions
www.gallup.com/workplace/236543/business-benefits-
gender-diversity.aspx Conception and design: Donaldson, Gelinne, Everett, Durham.
11. Abosch A, Rutka JT. Women in neurosurgery:​inequality Acquisition of data: Donaldson, Gelinne, Everett, Durham.
redux. J Neurosurg. 2018;​129(2):​277–281. Analysis and interpretation of data: Donaldson, Gelinne, Everett,
12. Spetzler RF. Progress of women in neurosurgery. Asian J Durham. Drafting the article: Callahan, Donaldson, Gelinne,
Neurosurg. 2011;​6(1):​6 –12. Everett, Durham. Critically revising the article: Callahan,
13. Table C-2. Residency Applicants from U.S. MD-Granting Ames, Air, Durham. Reviewed submitted version of manuscript:
Medical Schools to ACGME-Accredited Programs by Spe- Callahan, Ames, Air, Durham. Statistical analysis: Durham.
cialty and Sex, 2019-2020. American Association of Medical Administrative/technical/material support: Ames, Air, Durham.
Colleges 2019 Physician Data Book. November 12, 2020. Study supervision: Durham.
Accessed September 25, 2020. https:​//www.aamc.org/system/
files/2019-12/2019_FACTS_Table_C-2.pdf Supplemental Information
14. Durham SR, Donaldson K, Grady MS, Benzil DL. Analysis Previous Presentations
of the 1990-2007 neurosurgery residency match:​does ap- This paper was presented in poster form at AANS Virtual 2020.
plicant gender affect neurosurgery match outcome? J Neuro-
surg. 2018;​129(2):​282–289.
15. Benzil DL, Abosch A, Germano I, et al. The future of neu- Correspondence
rosurgery:​a white paper on the recruitment and retention of Katherine E. Callahan: Robert Larner MD College of Medicine at
women in neurosurgery. J Neurosurg. 2008;​109(3):​378–386. the University of Vermont, Burlington, VT. katherine.callahan@
16. Steklacova A, Bradac O, de Lacy P, Benes V. E-WIN Project med.uvm.edu.
2016:​evaluating the current gender situation in neurosurgery
across Europe—an interactive, multiple-level survey. World
Neurosurg. 2017;​104:​48–60.

948 J Neurosurg Volume 135 • September 2021

 
*49 *
 CLINICAL ARTICLE
J Neurosurg 135:949–954, 2021

Resident evaluations in the age of competency-based

medical education: faculty perspectives on minimizing
burdens
Jessica E. Rabski, MD, MEd,1,2 Ashirbani Saha, PhD,1 and
Michael D. Cusimano, MD, FRCSC, DABNS, PhD, MHPE1,3
1
Injury Prevention Research Office, St. Michael’s Hospital, Toronto; 2Department of Surgery, Division of Neurosurgery, University
of Ottawa; and 3Department of Surgery, Division of Neurosurgery, University of Toronto, Ontario, Canada

OBJECTIVE  Competency-based medical education (CBME), an outcomes-based approach to medical education, con-
tinues to be implemented across many postgraduate medical education programs worldwide, including a recent introduc-
tion into Canadian neurosurgical training programs (July 2019). The success of this educational paradigm shift requires
frequent faculty observation and evaluation of residents performing defined tasks of the specialty. A main challenge in-
volves providing residents with frequent performance evaluations and feedback that are feasible for faculty to complete.
This study aims to define what is currently happening and what changes are needed to make CBME successful for the
certification of neurosurgeons’ competence.
METHODS  A 55-item questionnaire was emailed nationwide to survey Canadian neurosurgical faculty.
RESULTS  Fifty-two complete responses were received and achieved a distribution highly correlated with the number of
faculty neurosurgeons practicing in each Canadian province (Pearson’s r = 0.94). Two-thirds (35/52) of faculty reported
currently taking a median of 10 minutes to complete evaluation forms at the end of a resident’s rotation block. Regard-
less of the faculty’s province of practice (p = 0.50) or years of experience (p = 0.06), they reported 3 minutes (minimum
1 minute, maximum 10 minutes, interquartile range [IQR] 3 minutes) as a feasible amount of time to spend completing
an evaluation form following an observation of a resident’s performance of an entrustable professional activity (EPA).
If evaluation forms took 3 minutes to complete, 85% of respondents (44/52) would complete EPA evaluations weekly or
daily. The faculty recommended 5 minutes as a feasible amount of time to provide oral feedback (minimum 1 minute,
maximum 20 minutes, IQR 3.25 minutes), which was significantly higher (p = 0.00099) than their recommended amount
of time for completing evaluation forms. The majority of faculty (71%) stated they would prefer to access resident evalu-
ation forms through a mobile application compared to a paper form (12%), an evaluation website (8%), or through a URL
link sent via email (10%; p = 0.0032).
CONCLUSIONS  To facilitate the successful implementation of CBME into a neurosurgical training curriculum, resident
EPA assessment forms should take 3 minutes or less to complete and be accessible through a mobile application.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201688
KEYWORDS  competency-based medical education; national survey; feasibility; resident evaluations; neurosurgery;
resident training

M astery learning is based on the premise that mastery is reassessed and achieved. This system involves a
learners must achieve a level of mastery or com- continuous cycle of practice and learning, close evaluation,
petence in prerequisite knowledge, skill, or at- and delivery of constructive student performance feedback
titude (e.g., able to tie a surgical knot) as defined by educa- until the learner accomplishes mastery of that learning ob-
tional objectives, before moving forward to learn increas- jective and is able to move on to the next stage of learn-
ingly complex information, skills, or behaviors.1 Should ing. Rather than mandating a specific amount of time for
a learner not achieve a desired level of mastery, they are learning a task (e.g., a 6-year residency), it is based on the
given formative and nonpunitive feedback, and additional mastery of a set of particular goals (such as being able to
supports in learning and opportunities for practice until manage a patient with a traumatic brain injury [TBI]).

ABBREVIATIONS  CBME = competency-based medical education; EPA = entrustable professional activity; IQR = interquartile range; TBI = traumatic brain injury.
SUBMITTED  May 7, 2020.  ACCEPTED  July 7, 2020.
INCLUDE WHEN CITING  Published online December 11, 2020; DOI: 10.3171/2020.7.JNS201688.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  949

*50 *
Rabski et al.

Mastery learning was further popularized in a 1978 to be highly variable, and is often simply implied in as-
WHO report by Dr. McGaghie et al. as an attempt to sessment validation studies.18–20 Neurosurgery faculty are
increase accountability and transparency of postgradu- unquestionably busy and in high demand for their clini-
ate medical training to the public.2 The report called for cal services, which can often be at odds with the demand
the development of a competency-based curriculum with to complete resident performance assessments, especially
a mastery goal within medical education. After this re- as a result of their lower physician-to-population ratios
port, a number of postgraduate medical education training (1:58,890 in the US and 1:100,000 in Canada).21,22 As such,
programs across Europe, Australasia, and North America defining what feasible means to busy neurosurgeons is
gradually introduced different models of a competency- highly relevant to helping faculty balance their strenuous
based medical education (CBME) framework into their clinical practice duties with their delivery of performance
medical curriculum.3–6 Modeled from behaviorist learn- assessments and quality feedback to residents.
ing principles,7 this outcomes-based approach to residency The purpose of this study was to define, from a neu-
education aims to ensure that necessary competencies that rosurgeon’s perspective, the feasibility of resident EPA
reflect societal and patient needs are attained throughout assessments within a CBME residency curriculum frame-
residency training by designing learning experiences that work. Identifying what feasibility of performing these
emphasize observable abilities.8 A cornerstone of this evaluations of residents means to neurosurgical faculty
educational model is that there is a continuous cycle of could help in the development and faculty acceptance of
frequent observations of resident performance of specified EPA assessment tools. Developing feasible means through
competencies that undergo evaluation and receive con- which faculty can evaluate residents within a CBME
structive feedback from staff physicians and educators. framework while continuing to meet the demands of busy
As such, staff physicians must have the necessary time to practice settings would be critical in the successful imple-
observe resident behaviors and performances of skills to mentation of CBME into residency programs.
provide quality feedback in a transparent and accountable
way. Methods
CBME learning outcomes or desired competences
can be operationalized by connecting them to authentic Questionnaire Composition and Administration
tasks required of independent medical practice. In some A 55-item questionnaire that took 15–20 minutes to
applications of CBME, entrustable professional activities complete was sent to all Canadian neurosurgery faculty
(EPAs) are used to identify authentic tasks of a medical with publicly available email addresses using a Survey-
discipline that can be assigned to a resident and, conse- Monkey platform. Standard online survey guidelines
quently, observed by a supervising physician to ensure a were followed (two reminder emails sent following initial
targeted level of performance has been attained.9,10 The email invitation) and survey responses were collected over
direct documentation of these observed performances by a 4-month period.23 Participation was voluntary and no
teachers of students lends a sense of accountability and identifiable information was collected to ensure partici-
transparency to the process.11 An example of a neurosur- pant confidentiality. Faculty were able to stop the survey at
gery EPA within the context of TBI treatment would be any point and resume at a more convenient time; however,
performing a “craniotomy for complex craniofacial trau- once submitted, the individual survey link would become
ma including injuries of the orbit, zygoma, and frontal inactive to prevent duplicate responses. Three waves of
sinus.”12 “Milestones” are linked to EPAs and help docu- emailed requests for responses were sent. This paper re-
ment a resident’s level of skill and progression toward ports on a subset of results regarding feasible expectations
independent practice.13 Using the previous EPA example, and targets for resident evaluations within the Canadian
milestones could therefore range from assisting with Competence by Design neurosurgical residency curricular
setup, opening, and closing for neurotrauma procedures framework. Approval for the study was provided by the
(level 1), to performing advanced procedures for patients local Research Ethics Board. Survey questions relevant to
with TBI (level 5).12 this study are included as Supplementary Material.
The success of this learning framework relies on fre-
quent and timely observations, evaluations and document- Statistical Analysis
ed feedback on resident behaviors, and performances by Descriptive analysis was performed and the propor-
staff physicians. Due to the demanding and busy nature of tion of responses for all survey questions was calculated.
medical practice, it is easy to imagine that CBME would Medians and ranges were calculated for questions involv-
need to be highly feasible to succeed. Faculty have ex- ing numerical responses. Categorical responses were ana-
pressed important concerns regarding the amount of time lyzed using the Fisher exact test and numerical responses
required to complete resident assessments and the use of were analyzed using the Wilcoxon rank-sum test for sub-
inefficient technology systems for accessing assessment group analysis related to the respondent’s provinces, re-
tools as they increase faculty workload and burden, and spondent’s experiences (up to 10 years and more than 10
reduce available time for clinical duties.14–17 What is per- years), and current and preferred ways of providing feed-
ceived to be feasible can differ considerably across medi- back. An exact binomial test was conducted to analyze
cal environments (e.g., clinics, emergency rooms, operat- the mode of filling out the evaluation forms. A p value
ing rooms, etc.) and medical specialties (e.g., family medi- < 0.05 was used for statistical significance. All analyses
cine, emergency medicine, neurosurgery, etc.). As such, were conducted using the R statistical program (https://
defining feasibility of resident assessment tools continues www.r-project.org/).

950 J Neurosurg Volume 135 • September 2021

 
*50 *
 Rabski et al.

TABLE 1. Province-based demographics of respondents’ neurosurgical practices (n = 52)

Province (no. of publicly available Median No. of Faculty Median No. of Residents Median No. of Residents
neurosurgeon email addresses/total no. of % of Survey Neurosurgeons Practicing on Neurosurgery Service Supervised by Respondent
practicing neurosurgeons)* Respondents at Respondent’s Hospital at Respondent’s Hospital per Day (range)
Alberta (13/38) 8 15 6 2 (1–2)
British Columbia (26/49) 15 8 3 1 (1–4)
Manitoba (10/12) 12 10 4 4 (1–5)
Maritime Provinces (10/13) 6 9 5 1 (1–2)
Ontario (71/117) 40 10 6 2 (1–4)
Quebec (42/80) 19 13 3 1 (1–2)
The total number of faculty neurosurgeons practicing in each province was estimated using the Canadian Medical Association 2018 Neurosurgery Profile report.21
* Emails were also sent to Saskatchewan (n = 9) and New Brunswick (n = 4) but no responses were received.

Results pleting a median of 2 (minimum 1, maximum 10, IQR 2.5)

Practice Demographics resident evaluation forms per rotation block, month, week,
or day. No significant differences existed between prov-
We received a total of 67 questionnaire responses and inces or by faculty experience in terms of current resident
of these, 52 responses (77.6%) were considered complete evaluation practices.
(> 75% of questions completed) and included in the analy-
sis. Respondents took a median time of 19 minutes (mini- Feasible Targets for Completing Resident Evaluation
mum 6 minutes, maximum 696 minutes, interquartile Forms
range [IQR] 10.5 minutes) to complete the questionnaire.
The distribution of respondents by province and clinical To be feasible, the ideal amount of time reported to
practice demographics are provided by province in Table complete an evaluation form following an observation of
1. The number of respondents from each province was a resident’s performance of an EPA during a single patient
highly correlated with the total number of faculty neuro- interaction was a median time of 3 minutes (minimum 1
surgeons practicing in the corresponding province (Pear- minute, maximum 10 minutes, IQR 3 minutes), regardless
son’s r = 0.94) and the number of survey invitation emails of province of practice (p = 0.50) or years of experience
sent to that provincial faculty (Pearson’s r = 0.98). With (p = 0.06). Taking respondents’ suggested feasible amount
the exception of respondents from Manitoba, respondents of time for evaluation form completion into consideration,
reported being responsible for the supervision of a median their corresponding postulated ideal frequencies of evalu-
of 1–2 residents per day; this did not differ based on fac- ation form completion are depicted in Table 3. In contrast
ulty experience. to what is prevalent in practice, respondents stated that if
With regard to respondents’ faculty experience, 40% evaluation forms took the feasible amount of time they
(21/52) had been in a neurosurgical faculty position for recommended to complete, 54% (28/52) and 31% (16/52)
less than 10 years and 60% (31/52) for 11 or more years. of respondents would complete EPA evaluation forms
Although neurosurgery faculty participation in this study weekly and daily, respectively.
did not vary significantly across provinces with regard to Respondents’ recommended feasible completion rates
years of experience (p = 0.93), the highest number of re-
sponses (40%, 21/52) were from Ontario, which has 35%
of the practicing Canadian neurosurgeons.21 The number TABLE 2. Summary of respondents’ current practices of resident
of residents supervised daily did not differ based on fac- evaluation form completion
ulty experience.
Median No. of Median No.
Current Resident Evaluation Practices Mins Spent of Evaluation
Form Completion % of Completing an Forms
Responses regarding neurosurgical faculty’s current Frequency* Respondents Evaluation Form Completed
resident evaluation form completion practices are summa-
rized in Table 2: 79% (41/52) of respondents use a single Daily 2 (n = 1) 2 1
frequency of evaluation form completion, whereas 21% Weekly 27 (n = 14) 5 2
(11/52) use two or more. Two-thirds (67%, 35/52) of re- Monthly 23 (n = 12) 5 2
spondents fill out resident evaluation forms at the end of a At the end of a 67 (n = 35) 10 2
resident’s rotation block. Approximately a quarter of fac- rotation block
ulty complete weekly (27%) or monthly (23%) evaluations; When prompted 2 (n = 1) — —
only 1 respondent indicated conducting daily evaluations. by resident (by
Those who completed residents’ evaluation forms at the email or verbally)
end of a rotation block took a median of 10 minutes to
do so, whereas those doing monthly or weekly evaluations None 2 (n = 1) — —
took a median of 5 minutes. Respondents reported com- * Respondents could choose multiple frequencies.

J Neurosurg Volume 135 • September 2021 951

 
*50 *
Rabski et al.

TABLE 3. Respondents’ suggested feasible frequencies for completing written evaluations and providing oral feedback on resident EPA
performances
Median Feasible Frequency of
Median Feasible Frequency of Evaluation Evaluation Form Completion to Median Feasible Median Feasible Frequency
Form Completion to Integrate into Busy Support Resident Competence No. of Residents of Providing Oral Feedback
Frequency Medical Practice (% of respondents)* (% of respondents)† Able to Supervise (% of respondents)
Per day 2 (31) 1 (21) 3 2 (35)
Per week 3 (54) 3 (54) 2 2 (48)
Per month 3 (8) 2 (12) 3 1 (10)
Per rotation block 7 (4) 3 (8) 3 3 (4)
Per 3 months 1 (2) 1 (2) 2 1 (2)
* One respondent did not answer both parts of the corresponding question.
† Two respondents did not answer both parts of the corresponding question.

for resident evaluation forms remained unchanged (p = cantly (p < 0.05, Fisher’s exact test). Although 40% (21/52)
0.77, Fisher’s exact test) when asked to consider optimal of faculty currently wait until the end of a resident’s rota-
evaluation form completion rates required for promoting tion block to complete evaluations, 83% (43/52) indicated
competency of residents. Respondents stated that a me- their preferred choice for completing an evaluation form
dian of 2 residents could realistically be evaluated if the on a resident’s performance of an EPA is immediately af-
respondents’ proposed feasible amount of time and corre- ter the activity or later that day at a convenient time (Table
sponding frequency for evaluation form completion were 4). However, 77% (40/52) of respondents currently and
considered. No significant variation (p = 0.83) was noted 87% (45/52) of respondents prefer to provide oral feedback
based on respondents’ years of experience in practice, but on a resident’s performance of an EPA immediately after
a lower median number of residents was noted in Ontario the activity or later that day at a convenient time. As such,
(n = 2) compared to other provinces (n = 3; p = 0.045, Wil- both the current and preferred frequencies of providing
coxon rank-sum test). oral feedback were similar (p = 0.97, Fisher’s exact test).

Feasible Targets for Providing Residents With Oral Access Preferences

Feedback Finally, to facilitate the completion of resident EPA eval-
Respondents indicated that the feasible amount of time uation forms into their busy schedules, 71% (37/52) stated
to provide oral feedback on an observation of a resident’s they would prefer to access a resident evaluation form of an
performance of an EPA during a single patient interaction EPA through a mobile application compared to paper form
was a median of 5 minutes (minimum 1 minute, maxi- (12%), an evaluation website (8%), or through a URL link
mum 20 minutes, IQR 3.25 minutes). Respondents stated sent via email (10%; p = 0.0032, exact binomial test).
that if providing oral feedback actually took the amount of
time they recommended, 48% (25/52) and 35% (18/52) of
respondents would provide it weekly and daily, respective- Discussion
ly (Table 3). The feasible amount of time needed to pro- This study is the first to report on faculty perspectives
vide oral feedback was significantly higher (p = 0.00099, regarding the feasibility of EPA performance assessments
Wilcoxon rank-sum test) than their recommended amount of neurosurgery residents within a CBME residency edu-
of time for filling out evaluation forms. cation framework. Faculty reported that an evaluation
Respondents’ current and preferred opportunities for form on a resident’s performance of an EPA, of a single
completing resident EPA evaluation forms differed signifi- patient encounter, would need to take 3 minutes to com-

TABLE 4. Summary of current and preferred ways of completing evaluations and providing oral feedback on resident
EPA performances
Evaluation Form Completion Oral Feedback Provision
Timing of Feedback Current Practice Preferred Practice Current Practice Preferred Practice
Immediately after the activity 11 (21) 25 (48) 27 (52) 36 (69)
Later that day at a convenient time 12 (23) 18 (35) 13 (25) 9 (17)
Later on that week 7 (14) 3 (6) 2 (4) 2 (4)
Before the end of a resident’s block 21 (40) 4 (8) 9 (17) 3 (6)
Other 1 (2) 2 (4) 1 (2) 2 (4)
All data given as number of responses (%).

952 J Neurosurg Volume 135 • September 2021

 
*50 *
 Rabski et al.

plete to make CBME feasible. This result was drastically this study has provided some feasible targets for CBME
lower than the proposed evaluation form completion times postgraduate medical curriculum developers and educa-
of 20 minutes to 1 hour by studies involving other nonneu- tors in an attempt to address concerns raised by faculty.
rosurgical specialties.24,25 In addition, by accommodating Potential future study directions would continue to utilize
faculty time preferences for resident EPA assessment form faculty perspectives and expertise to determine potential
completion, an improvement from monthly/end of a rota- feasible elements or features that could be incorporated
tion block to daily or weekly EPA form completion rates into the designing of resident EPA assessment forms. Any
was identified, with no restriction in the faculty’s ability to proposed design of a resident EPA assessment form would
supervise a similar number of residents. This highlights require validation to ensure faculty participation.34
the importance of making EPA assessments that are de-
signed around professional practice, and as such, mini- Limitations
mize clinical disruption as much as possible to promote We depended on publicly available email addresses
their uptake by faculty.24 Acknowledging the importance from university or hospital faculty websites, and as such,
and value of time for faculty neurosurgeons facilitates could be prone to selection bias favoring faculty heavily
a seamless integration of regular and formative resident involved in teaching, administration, or research. Howev-
EPA assessments into complex clinical environments, er, the completed responses achieved a distribution highly
thereby ultimately supporting CBME’s goals. correlated to both the total number of faculty neurosur-
Faculty also indicated that it was more feasible to spend geons practicing in the corresponding province and the
time providing feedback when it was done so orally, and number of survey invitation emails sent to that provincial
no significant differences existed between their actual neurosurgical faculty. Our small sample size may have re-
and preferred opportunities for providing oral feedback. sulted in missing minor differences for some subanalyses
This may suggest that providing oral feedback rather than of questions regarding frequencies, but statistical power
documented feedback integrates more efficiently into the was achieved for subanalysis of numerical responses.
day-to-day clinical practices of neurosurgeons; however, Although we had a significant number of responses,
although oral feedback has been shown to improve surgi- these results represented those of academic neurosurgeons
cal outcomes of surgeons,26 without associated documen- in Canada and may not be representative of all neurosur-
tation of the feedback, it is perceived as less accountable geons in Canada or elsewhere. However, compared to other
and transparent within CBME. As such, developing strate- subspecialties, Canadian neurosurgery has one of the low-
gies within a CBME curriculum to recognize and assign est ratios of number of physicians per 100,000 population
value to oral feedback, in addition to documented feed- and fewer neurosurgeons per capita than the US,21,22 so the
back, should be further explored. results of this study likely represent the reasonable limits
The study revealed that in addition to targeting a fea- on what busy neurosurgeons would accept as feasible in
sible EPA form completion time, faculty expressed sig- terms of resident evaluation within a CBME framework.
nificant preference for accessing resident evaluation forms
through a mobile application. This preference for EPA as-
sessments to be conducted on mobile platforms compared Conclusions
to online evaluation websites could be due to their sug- Findings from this study indicated that neurosurgery
gested user-friendly nature and ability to store data and resident EPA assessment forms should be designed to take
function offline.27,28 As such, additional efforts in creating less than 3 minutes to complete if faculty are expected to
mobile application platforms for documenting and track- complete such evaluations within the context of CBME.
ing resident evaluations may be beneficial to improve fea- Accomplishing this goal would facilitate faculty complet-
sibility and overall faculty participation. ing resident performance evaluations on a daily or weekly
The conceptualization of CBME learning principles by basis rather than on an end-of rotation basis, thereby fa-
organizing assessments around discipline-specific EPAs cilitating the CBME goal of frequent formative feedback.
was originally proposed for its ability to integrate into Finally, developing mobile application platforms to docu-
the natural flow of clinical practice.29 Although research- ment evaluations and feedback of resident performances
ers recognize that documenting clinical observations and would enable the successful implementation of CBME.
conversations must be both efficient and student/faculty
friendly to be feasible, there continues to be minimal evi- References
dence supporting CBME’s feasible integration into busy   1. Bloom BS. Learning for mastery. UCLA Evaluation Com-
medical practices.29,30 In addition, there are limited stud- ment-CSEIP. 1968;​1(2).
ies exploring specialty-specific faculty perspectives on   2. McGaghie WC, Miller GE, Sajid AW, Telder TW. Compe-
CBME’s feasible integration into diverse clinical prac- tency-Based Curriculum Development in Medical Educa-
tices. Some of these studies have acknowledged the poten- tion. World Health Organization;​1978. Accessed August 24,
tial risk of CBME to reduce the faculty’s available time to 2020. http:​//whqlibdoc.who.int/php/WHO_PHP_68.pdf
spend on their clinical practices and/or academic research   3. Carraccio C, Englander R, Van Melle E, et al. Advancing
efforts.31,32 Others have demonstrated that the feasibility competency-based medical education:​a charter for clinician-
educators. Acad Med. 2016;​91(5):​645–649.
of resident assessments within clinical contexts is heav-   4. Parent F, Jouquan J, De Ketele JM. CanMEDS and other
ily impacted by the time they require to complete, which, “competency and outcome-based approaches” in medical
in turn, results in poor faculty participation and up to education:​clarifying the ongoing ambiguity. Adv Health Sci
40% of staff refusing to complete them.14,33 In response, Educ Theory Pract. 2013;​18(1):​115–122.

J Neurosurg Volume 135 • September 2021 953

 
*50 *
Rabski et al.

  5. Biemans H, Wesselink R, Gulikers J, et al. Towards com- 24. Norcini JJ, Blank LL, Arnold GK, Kimball HR. The mini-
petence-based VET:​dealing with the pitfalls. J Vocat Educ CEX (clinical evaluation exercise):​a preliminary investiga-
Train. 2009;​61(3):​267–286. tion. Ann Intern Med. 1995;​123(10):​795–799.
  6. Frank JR, Snell L, Englander R, Holmboe ES. Implementing 25. Downs SH, Black N. The feasibility of creating a checklist
competency-based medical education:​moving forward. Med for the assessment of the methodological quality both of ran-
Teach. 2017;​39(6):​568–573. domised and non-randomised studies of health care interven-
  7. Hodge S. The origins of competency-based training. Aust J tions. J Epidemiol Community Health. 1998;​52(6):​377–384.
Adult Learn. 2007;​47(2):​179–209. 26. Maruthappu M, Trehan A, Barnett-Vanes A, et al. The
  8. Maudsley RF, Dagg P, Wilson J, et al. Addressing societal impact of feedback of surgical outcome data on surgical
health needs. In:​Competence by Design:​Reshaping Cana- performance:​a systematic review. World J Surg. 2015;​39(4):​
dian Medical Education. Royal College of Physicians and 879–889.
Surgeons of Canada, 2014:​8–17. 27. Yoo J-H. The meaning of information technology (IT) mo-
  9. ten Cate O. Entrustability of professional activities and com- bile devices to me, the infectious disease physician. Infect
petency-based training. Med Educ. 2005;​39(12):​1176–1177. Chemother. 2013;​45(2):​244–251.
10. Cate OT. A primer on entrustable professional activities. Ko- 28. Ventola CL. Mobile devices and apps for health care profes-
rean J Med Educ. 2018;​30(1):​1–10. sionals:​uses and benefits. P T. 2014;​39(5):​356–364.
11. Pichette J, Watkins EK. Competency-Based Education:​ 29. Ten Cate O, Graafmans L, Posthumus I, et al. The EPA-based
Driving the Skills-Measurement Agenda. Higher Education Utrecht undergraduate clinical curriculum:​development and
Quality Council of Ontario;​October 2018. Accessed August implementation. Med Teach. 2018;​40(5):​506–513.
24, 2020. http:​//www.heqco.ca/en-ca/Research/ResPub/ 30. Blades ML, Glaze S, McQuillan SK. Resident perspectives
Pages/Competency-based-Education-Driving-the-Skills- on competency-by-design curriculum. J Obstet Gynaecol
measurement-Agenda.aspx Can. 2020;​42(3):​242–247.
12. Selden N, Abosch A, Bambakidis N, et al. Neurological 31. Hanley M, Shearer C, Livingston P. Faculty perspectives on
Surgery Milestones. The Accreditation Council for Graduate the transition to competency-based medical education in an-
Medical Education. Second Revision:​March 2018. Accessed esthesia. Can J Anaesth. 2019;​66(11):​1320–1327.
August 24, 2020. https:​//www.acgme.org/Portals/0/PDFs/ 32. Holmboe ES, Ward DS, Reznick RK, et al. Faculty develop-
Milestones/NeurologicalSurgeryMilestones.pdf ment in assessment:​the missing link in competency-based
13. Pangaro L, ten Cate O. Frameworks for learner assessment medical education. Acad Med. 2011;​86(4):​460–467.
in medicine:​AMEE Guide No. 78. Med Teach. 2013;​35(6):​ 33. Tomiak A, Braund H, Egan R, et al. Exploring how the new
e1197–e1210. Entrustable Professional Activity assessment tools affect the
14. Bindal T, Wall D, Goodyear HM. Trainee doctors’ views on quality of feedback given to medical oncology residents. J
workplace-based assessments:​Are they just a tick box exer- Cancer Educ. 2020;​35(1):​165–177.
cise? Med Teach. 2011;​33(11):​919–927. 34. Diaz Voss Varela DA, Malik MU, Thompson CB, et al. Com-
15. Barrett A, Galvin R, Scherpbier AJJA, et al. Is the learn- prehensive assessment of thyroidectomy skills development:​
ing value of workplace-based assessment being realised? A a pilot project. Laryngoscope. 2012;​122(1):​103–109.
qualitative study of trainer and trainee perceptions and expe-
riences. Postgrad Med J. 2017;​93(1097):​138–142.
16. Menon S, Winston M, Sullivan G. Workplace-based assess- Disclosures
ment:​attitudes and perceptions among consultant trainers The authors report no conflict of interest concerning the materi-
and comparison with those of trainees. Psychiatr Bull. 2012;​ als or methods used in this study or the findings specified in this
36(1):​16–24. paper.
17. Abdelaal A. Procedure-based assessments:​the past, the pres-
ent and the future. Postgrad Med J. 2020;​96(1131):​7–8. Author Contributions
18. Naeem N. Validity, reliability, feasibility, acceptability and
educational impact of direct observation of procedural skills Conception and design: Rabski, Cusimano. Acquisition of data:
(DOPS). J Coll Physicians Surg Pak. 2013;​23(1):​77–82. Rabski. Analysis and interpretation of data: all authors. Draft-
19. Sharma VK, Krishna M, Lepping P, et al. Validation and ing the article: Rabski, Saha. Critically revising the article: all
feasibility of the Global Mental Health Assessment Tool— authors. Reviewed submitted version of manuscript: all authors.
Primary Care Version (GMHAT/PC) in older adults. Age Approved the final version of the manuscript on behalf of all
Ageing. 2010;​39(4):​496–499. authors: Rabski. Statistical analysis: Saha. Administrative/techni-
20. Hadley C, Lam SK, Briceño V, et al. Use of a formal assess- cal/material support: Rabski. Study supervision: Cusimano.
ment instrument for evaluation of resident operative skills
in pediatric neurosurgery. J Neurosurg Pediatr. 2015;​16(5):​ Supplemental Information
497–504. Online-Only Content
21. Canadian Medical Association. Neurosurgery Profile. August Supplemental material is available with the online version of the
2018. Accessed August 24, 2020. https:​//www.cma.ca/sites/ article.
default/files/2019-01/neurosurgery-e.pdf Supplementary Material. https://thejns.org/doi/suppl/​10.3171/​
22. Association of American Medical Colleges. Physician Spe- 2020.7.JNS201688.
cialty Data Report—Number of People per Active Physician
by Specialty, 2017. Accessed August 24, 2020. https:​//www. Correspondence
aamc.org/data-reports/workforce/interactive-data/number-
Jessica Rabski: The Ottawa Hospital, Ottawa, ON, Canada.
people-active-physician-specialty-2017
[email protected].
23. Saleh A, Bista K. Examining factors impacting online survey
response rates in educational research:​perceptions of gradu-
ate students. J Multidiscip Eval. 2017;​13(29):​63–74.

954 J Neurosurg Volume 135 • September 2021

 
*50 *
 HISTORICAL VIGNETTE
J Neurosurg 135:955–961, 2021

The mediums of dissemination of knowledge and

illustration in neurosurgery: unraveling the evolution
Bharath Raju, MD, MCh,1 Fareed Jumah, MD,1 Vinayak Narayan, MD,1 Anika Sonig, BA,2
Hai Sun, MD,1 and Anil Nanda, MD, MPH1
1
Department of Neurosurgery, Rutgers-Robert Wood Johnson Medical School and University Hospital, New Brunswick, New
Jersey; and 2Rice University, Houston, Texas

The earliest evidence of man’s attempts in communicating ideas and emotions can be seen on cave walls and ceilings
from the prehistoric era. Ingenuity, as well as the development of tools, allowed clay tablets to become the preferred
method of documentation, then papyrus and eventually the codex. As civilizations advanced to develop structured sys-
tems of writing, knowledge became a power available to only those who were literate. As the search to understand the
intricacies of the human brain moved forward, so did the demand for teaching the next generation of physicians. The
different methods of distributing information were forced to advance, lest the civilization falls behind. Here, the authors
present a historical perspective on the evolution of the mediums of illustration and knowledge dissemination through the
lens of neurosurgery. They highlight how the medium of choice transitioned from primitive clay pots to cutting-edge vir-
tual reality technology, aiding in the propagation of medical literature from generation to generation across the centuries.
https://thejns.org/doi/abs/10.3171/2020.7.JNS201053
KEYWORDS  history; papyrus; codex; Vesalius; illustration; animation; virtual reality

M
edical literature and knowledge have been grow- we investigate the history of the medium of illustrations
ing exponentially over the last decade. Techno- and the dissemination of knowledge of anatomy and neu-
logical advancement ushered in the era of big roanatomy. For better comprehension, we have divided the
data and artificial intelligence. Technology has enabled evolution into three stages: a pre-Renaissance era (before
the dissemination of information virtually, becoming ac- 1500 ad), from the Mesopotamian civilization up to the
cessible to all. However, this was not the situation even Italian Renaissance and the Age of Exploration, in which
a few decades ago. It is crucial to understand how these physical mediums dominated; an early modern era (after
technologies for dissemination of knowledge evolved, as 1500 ad), from the beginning of the Renaissance up to
it allows one to make more sense of the current world and the 19th century, during which there was a slow rise of
allows us to exercise critical thinking. technologies; and the contemporary era (1900 ad –present
Writing and illustration started as cave wall carvings day), a period of dramatic increase in technology leading
and symbols on clay pots. The genesis of languages and to virtual mediums.
alphabets led to the blossoming of a variety of mediums,
including papyrus, parchment, and paper. But these were
only accessible to a few literates and were considered a Physical Mediums of the Pre-Renaissance
sign of power. Along with language, people realized the Era: Before 1500 ad
importance of illustrations in communicating knowledge.
Clay Tablets and Stones
With the humanistic approach of the Renaissance, the con-
tent of books expanded, and the illustration of anatomy be- Writing on clay or stone is the earliest known form of
came a disruptive technology. The advent of printing tech- dissemination, dating back to the Mesopotamian civiliza-
nology and later photography made books and illustrations tion approximately 3000 years bc. The oldest known med-
easily reproducible, allowing rapid circulation and access ical inscriptions written by an unknown Sumerian physi-
for many. The further technological revolution made cian were excavated in the ruins of Nippur in modern-day
three-dimensional (3D) viewing of anatomy possible, pav- Iraq.1 They were written in small cuneiform characters on
ing the way for virtual reality (VR) and simulations. Here, a 5-inch-long quadrilateral tile of wet clay using a stylus,

ABBREVIATIONS  AR = augmented reality; VR = virtual reality; 3D = three-dimensional.

SUBMITTED  April 1, 2020.  ACCEPTED  July 6, 2020.
INCLUDE WHEN CITING  Published online December 4, 2020; DOI: 10.3171/2020.7.JNS201053.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  955

*50 *
Raju et al.

FIG. 1. Fragment of a clay tablet with cuneiform inscription describing

a medical treatment. From https://www.metmuseum.org/art/collection/
search/321680. Public domain.

then baked and dried in the sun to make them durable.

Long texts were inscribed on multiple tablets that were FIG. 3. The Edwin Smith Papyrus is the oldest surgical document
sequenced by a combination of numbers and unique char- describing the management of head injury. From https://en.wikipedia.
acters (Fig. 1). As of now, more than 500,000 clay tablets org/wiki/Edwin_Smith_Papyrus. Public domain. Figure is available in
have been recovered.2 Currently preserved in The British color online only.
Museum, the Dying lioness from Nineveh relief, carved on
alabaster stone, represents probably the oldest illustration stone cuneiform writings dwindled, and by the 6th century
of paraplegia dating back to 650 bc (Fig. 2).3,4 Following bc, papyrus became the predominant medium of illustra-
the introduction of Aramaic, one of the earliest alphabetic tion and knowledge dissemination.5
languages in human history, the use of clay tablets and
Papyrus
Ancient Egyptians were the first to use papyrus for
writing and illustration as far back as 2500 bc, which con-
tinued into Roman Egypt (30 bc–640 ad). “Papyrus” is
a Greek word originating from the Egyptian word “pa-
puro” meaning “the royal.” It was made from the plant
Cyperus papyrus, a natural vegetation on the Nile Delta.
The process of cultivation and production was expensive
and needed skilled labor, which explains why it was ex-
clusively under the control of the royal, or the pharaoh.5
As papyrus was expensive and limited, people would first
practice on clay, wood, or stone before writing on it. An-
cient Egyptian scribes, the select few who knew how to
read and write at the time, carried wooden toolkits con-
taining pens made of reed or bamboo stems and black and
red paint cake. Red was used for the names of evil spirits
or demons and for incantations, as the Egyptians believed
magic was effective together with medicine, as evident in
the Ebers Papyrus.3,6
The Kahun Gynaecological Papyrus, Edwin Smith Pa-
FIG. 2. The Dying lioness relief wall panel from Ashurbanipal’s (650 bc) pyrus (Fig. 3), and Ebers Papyrus are three important an-
palace at Nineveh depicts a lioness dragging its hind limbs because cient medical texts whose original authors are unknown.7
of a spinal cord injury from arrows sticking in its back. This picture is a They were instead named after people who owned or do-
cast of the original relief preserved in The British Museum. Photography
by Cornell University Library, Annetta Alexandridis, and Verity Platt. nated them to museums, such as the German Egyptolo-
Reproduced with permission from Cornell University Library. Figure is gist Georg Moritz Ebers (1837–1898). The Ebers Papyrus,
available in color online only. written 3500 years ago, is the longest and most complete

956 J Neurosurg Volume 135 • September 2021

 
*50 *
 Raju et al.

papyrus filled with spells and incantations to expel the

“evils of disease,” as well as herbal remedies for around
700 conditions including neurological disorders. The
Edwin Smith Papyrus, a less magical and more struc-
tured papyrus, is the oldest surgical document addressing
trauma management, likely written for surgeons in field
hospitals. It describes the history, examination, diagnosis,
treatment, and outcome of 48 cases, including head and
spine injuries.5,6 The magnificent survival of these papyri
over thousands of years was in part due to Egypt’s dry
climate, which prevented decomposition, and the practice
of entombing scientific texts over the dead as the Book of
the Dead.3
The Greek physician Hippocrates (460–370 bc) was the
first to write medical papyri free of spells and incantations.
His ancient collection of around 60 manuscripts, known as FIG. 4. Nag Hammadi codices made of parchment (leather), like the
the Hippocratic Corpus, contains various writings rang- Galenic corpus. From http://thetextualmechanic.blogspot.com/2018/05/
galen-on-parchment-codex.html. Public domain.
ing from Hippocrates’ personal scattered notes to detailed
case descriptions such as epilepsy, which he called the
“sacred disease.” He is perhaps most well-known for au- the codices from the Greek libraries and enslaved their
thoring the ethical code, which we now know as the Hip- scribes to translate the texts into Latin, which explains
pocratic Oath, part of the Papyrus Oxyrhynchus (275 ad).8 why Roman literature is an imitation of the Greeks’. Later
during the reign of the Byzantine Empire, the copying of
Codex codices by students ensured the survival of Greek litera-
Codex, which was introduced in the early Christian era, ture, as did textual criticism and translations into Arabic
had marked a revolution in the dissemination of knowl- and other languages. A chain of civil wars that eventu-
edge. It consisted of multiple sheets of papyrus (and later ally brought the Roman Empire to its knees resulted in
leather and vellum) bound together, where both sides could the loss of most Greek and Roman manuscripts, with a
be used to scribe, and is essentially the predecessor of the few remnants ultimately finding refuge in religious abbeys
modern-day book. Inspired by the design of the Roman and churches. None of the complete original manuscripts
wax tablets, which were reusable writing surfaces made of the Greeks and Romans exist today. However, efforts of
of wax lined with a wooden frame, the codex was initially scholars like Aldo Manuzio (1456–1516 ad) helped to pre-
used for religious texts.3 It was a revolutionary change at serve Greek medical literature through compiling, editing,
the time, when a nation’s cumulative wealth of codices be- and printing the works of physicians such as Hippocrates,
came a sign of its power and prosperity. The Egyptians Aristotle, and Dioscorides.9,13
had been using leather for writing as early as 2450 bc,
but it was not used regularly, as the papyrus was widely Physical Mediums of the Early Modern Era:
available at the time. At around 190 bc, a cold war erupted After 1500 ad
between Ptolemy V Epiphanes of Egypt and Eumenes II
of Pergamon, Greece. Fearing the collections of codices Books
in Pergamon might outstrip those of Alexandria, Ptolemy Paper
withheld the export of papyrus, which in turn motivated By the 14th century, the codex was made predominant-
the Greeks to utilize leather/parchment as an alternative ly from paper. The advent of printing technology ushered
(Fig. 4).9 Unlike the more fragile papyrus, parchment was in the era of the book, which is essentially a codex that is
more flexible and durable, which enabled the production printed rather than handwritten. The Chinese used paper
of codices with a greater number of sheets. Further finer as early as 105 ad. Ts’ai Lun (48–121 ad), a Chinese inven-
processing and thinning of the parchment yielded vellum, tor and politician, is credited with inventing paper by us-
which eventually replaced papyrus as the predominant ing the inner bark of a mulberry tree and bamboo fibers.10
medium of writing and dissemination.3,10 Paper reached Europe from China through Arab traders.
The Library of Pergamon, arguably the most impor- By the 14th century, manually operated paper mills were
tant library in the ancient world, had around 200,000 established in Europe, the first in Germany near Nurem-
volumes and received many contributions in the form of berg.14 The Frenchman Louis-Nicolas Robert (1761–1828)
codices from prominent anatomists and physicians such fully mechanized the process of papermaking, which was
as Galen (129–216 ad), Herophilus (330–260 bc), Pedanius later improved by John Dickinson.10
Dioscorides (40–90 ad), Erasistratus (304–250 bc), Rufus
(70–110 ad), Soranus, and Aretaeus of Ephesus. By dictat- Print
ing to scribes, Galen produced more than 500 manuscripts Printing underwent an important series of innovations,
and treatises, called the Galenic corpus, which included from the primitive surface printing techniques to our
20 medical references that no longer existed in the Egyp- modern-day offset printing technology. Woodblock print-
tian civilization.11–13 ing, also known as surface printing, was invented in Japan
After conquering the Greeks, the Romans confiscated around 770 ad and was key in the spread of Buddhism in

J Neurosurg Volume 135 • September 2021 957

 
*50 *
Raju et al.

FIG. 5. Left: Johannes de Ketham used woodblock printing for the first time in his book Fasciculus Medicinae. From https://www.
historyofinformation.com/image.php?id=3257. Public domain. Right: Woodblock-printed illustration of anatomy from De Humani
Corporis Fabrica by Andreas Vesalius. From https://commons.wikimedia.org/w/index.php?curid=458936. Public domain. Figure is
available in color online only.

Asia through the printing of religious prayers and litera- around 1450, when Johannes Gutenberg (1400–1468 ad),
ture.15 The technique involved pasting hand-drawn images a German blacksmith at the time, invented the printing
or text on a block of wood, then carving the excess wood press, which launched the printing revolution in Europe.
around. The woodblock was then inked and pressed onto a He introduced the first mechanical movable typeface,
sheet of paper to print the elevated markings.4,14 which employs small metal letters fitted to a frame that
In 1491, anatomist Johannes de Ketham used wood- is pressed against paper.14 Over time, paper completely re-
block printing for the first time in his book Fasciculus Me- placed vellum, and printed books replaced codex.
dicinae to create six illustrations of “five-figure series” and This rapid progress in printing resulted not only in the
anatomical fugitive sheets (Fig. 5 left).16,17 Jacopo Beren- widespread publication of anatomical, medical, and surgi-
gario da Carpi (1450–1530) was the first anatomist to print cal texts, but also in numerous feats of plagiarism. Flem-
multidimensional pictures of anatomy with descriptions ish surgeon Thomas Geminus (1510–1562), in his book
in his book Commentario Super Anatomia Mundini in Compendiosa Totius Anatomie Delineatio, used copper
1521.18 Andreas Vesalius (1514–1564), the founder of mod- engraving to plagiarize and reproduce 40 pictures of Ves-
ern anatomy and medicine, published a 7-volume series of alius’ Fabrica.15,21
books on human anatomy, De Humani Corporis Fabrica, In 1798, Alois Senefelder, of Munich, introduced lithog-
in 1543 (Fig. 5 right). It was the first medical book to use raphy, a technique that exploits the property of oil and wa-
a publication production team and contained 663 pages of ter immiscibility to print on paper from a stone or rubber
printed text and > 400 original images on 85 plates printed surface.22 By the second part of the 19th century, lithogra-
using woodcuts.19 These woodblocks, created by artists of phy became the principal means of printing. During this
the Venetian School in Italy, were transported to Switzer- time, there were numerous improvements to the process
land to be used in printing by Johannes Oporinus (1507– through the introduction of colored lithographs, commer-
1568), making him the first medical publisher in history. cial lithography, and offset printing. The latter method, in
The third and seventh volumes of the series focused on which an image is printed on a rubber cylinder and then
the CNS. The first edition, dedicated to the Holy Roman transferred onto paper, became the most popular.22 In the
Emperor Charles V, was silk bound and vellum paged and modern era of medical illustration using lithography, Max
is the only hand-colored version of the book.3,20 Brödel (1870–1941) was a pioneer. Oftentimes known as
A more advanced printing technique that followed the father of modern medical illustration, he contributed
woodcutting is intaglio, which used copper engravings to neurosurgery by collaborating with Harvey Cushing
instead of wood to create more durable, clearly defined and illustrating precise surgical anatomy from the sur-
prints. The next major leap in printing technology was geon’s operative perspective. He is also credited with in-

958 J Neurosurg Volume 135 • September 2021

 
*50 *
 Raju et al.

FIG. 6. Photographs of Cushing’s patients with different neurosurgical conditions. From http://photos.yale.edu/directory/dir_single_
collection.php?collection_id=128. Photography by Terry Dagradi. Yale University, Harvey Cushing/John Hay Whitney Medical Library.

venting the “carbon dust” illustration technique. In 1911, to create higher quality images, and the method was called
Brödel started and directed the Department of Medical the “daguerreotype process.” However, reprinting was not
Illustration at the Johns Hopkins School of Medicine.23,24 possible using this technique until the introduction of
Continuous improvements in digital printing technology negatives 20 years later.27 French physician and cytologist
are trying to replace offset printing. Alfred François Donné (1801–1878) employed this tech-
nique to study medical specimens. In 1844, Donné, along
Anatomical Models with photographer Jean Bernard Foucault, published the
Though it is outside the scope of this review, 3D models first photomicrograph of leukemia cells and platelets,
deserve brief mention here. These 3D models evolved as a demonstrating the scope of this instrument in medical
medium of dissemination and learning, especially anato- practice.28,29 With further refinements and widespread
my. They were created from metal, wax, clay, ivory, or pa- use, photographing patients, procedures, and specimens
became routine. Later in 1912, Harvey Cushing (1869–
per and came in handy when medical students struggled to
1939), the father of modern neurosurgery, built a registry
acquire and dissect cadavers within a day or two to avoid
of pathological brain specimens, microscopic slides, and
decomposition—given the absence of embalming at the photographic plates of his patients (Fig. 6). In 1979, years
time. Interested bibliophiles are advised to review papers after Cushing’s era, his registry was shifted to the sub-
written by Ballestriero and by Ortug and Yuzbasioglu on basement of a student dormitory at Yale University. After
wax and papier-mâché models, respectively.25,26 1994, collaborative efforts and funds restored the regis-
try, currently displayed in The Cushing Center at Yale
Photography University. The collection contains over 10,000 films and
Up until the advent of photography, creating medical glass plate negative photographs of brain specimens and
images was highly dependent on artists. In 1826, Joseph of Cushing’s patients during various stages of brain tumor
Niépce (1765–1833) produced the first photograph using management.30,31 Cushing’s photographic collections es-
the prototype, the camera obscura. He, along with Louis sentially represent the evolution of modern neurosurgery
Daguerre (1787–1851), improved the photographic process from late 1800 to 1936.

J Neurosurg Volume 135 • September 2021 959

 
*50 *
Raju et al.

Physical and Virtual Learning Mediums of the (Realistic Operative Workstation for Educating Neurosur-
Contemporary Era: After 1900 ad gical Apprentices), an anatomical simulator developed in
the United Kingdom, realistically simulated all the layers
Surgical Motion Picture of the head to allow training on neurosurgical procedures
The introduction of the surgical motion picture in the and craniotomy.40 Later, procedural simulators for critical
early 20th century gave life to medical images, revolution- care, craniotomy, ventriculostomy, spinal and vascular sur-
izing the means of knowledge dissemination. The first gery, and endoscopy were developed for surgical training
surgical video on record was of Dr. Robert Fitzpatrick and education.39
performing a cesarean section in 1915. In the following Augmented reality (AR) and VR are modern technolo-
decades, a chain of innovations improved the process of gies often used in conjunction with robotic surgery. While
filmmaking by incorporating color and audio into the sur- AR is used for perception enhancement, VR along with
gical motion pictures.32 James Rudolph Jaeger (1895–1968) robotics is used to simulate the surgical environment to
was the first neurosurgeon to produce and popularize the enhance learning.41 “VR” was a term coined by computer
neurosurgical motion picture for teaching. In 1959, he es- scientist Jaron Lanier who described it as “a collection of
tablished Rudolph Jaeger Productions, which produced technologies that allow people to interact efficiently with
21 neurosurgical motion picture documentaries that were 3D computerized databases in real time using their natu-
widely presented at scientific meetings and conventions at ral senses and skills.”42 The earliest virtual reality product
the time.33 Nowadays, the widespread use of high-definition was the Sensorama Simulator constructed by American
cameras within surgical microscopes and endoscopes have cinematographer and inventor Morton Heilig in 1962. In
made surgical videos a cornerstone of medical education. 2007, VR and simulation were integrated into a robotic
surgical system. Nowadays, many VR interfaces are key
Digital Illustration, Animation, and 3D Printing tools for teaching neuroanatomy, neurosurgical training,
Medical animation has facilitated the understanding and quantifying surgeons’ manual dexterity and compe-
of anatomy and the teaching of basic and advanced con- tency.39 AR in the neurosurgical field has progressed so
cepts of medicine and surgery. It was debuted in 1932 by rapidly in the last 2 decades, and the details of its history
Clarke and Hoshall in the Journal of Biological Photog- are beyond the scope of this review. Alaraj’s Comprehen-
raphy, where a two-dimensional image incorporated into sive Healthcare Simulation: Neurosurgery provides an
a film was physically moved to create the illusion of mo- extensive account of the topic.39
tion.23,34 Later, the advent of photography gave birth to cell
animation, where hand-drawn images are photographed Connecting the Dots
and sequenced frame by frame, creating an animated
film (https://en.wikipedia.org/w/index.php?title=Medical_ Imagine writing a book or a paper like our predecessors
animation&oldid=946202838). In 1960, William Fetter did 2000 years ago. Taking on such a project would have
created the first computer-generated illustration of the been someone’s whole life’s work. As we glimpse through
human form, later developing it into a 3D image, which the evolution of mediums of dissemination, it is evident
paved the way for computer-generated 3D medical anima- how much progress we have made and where we are head-
tion.35 However, it was not until 1975 that the first medi- ing. We are probably in one of those transitions noted
cal animation was published in the journal Science, which throughout history, but this time the book is at the brink
demonstrated a complex intracellular molecular process.36 of extinction. Contemplating history, it is obvious that the
Since then, there has been enormous growth in the medi- future is the digital book. Though it offers the convenience
cal animation industry, which today accounts for an annu- of carrying an entire library with you, how safe is it? Are
al revenue of $15 billion.23 With the availability of comput- we prepared to rely on the privacy and security of cloud
erized 3D models, 3D digitizing arms, and scanners, 3D platforms? Can we store this exponentially increasing
printing technology has evolved. It has varied applications knowledge on a physical medium? Only the future holds
from learning neuroanatomy to presurgical planning and the answers.
training of surgeons. Interested bibliophiles are encour-
aged to review the paper by Pucci et al.37 Conclusions
The knowledge of a civilization can exist for only as
Simulation, Virtual Reality, and Augmented Reality
long as the surviving documentation of that knowledge.
Simulation is the “imitation of some real thing, state of Any evidence of the abilities harbored by a civilization
affairs, or process.” Although it began as a military tool is limited by the structural integrity of the mediums with
to develop strategy, problem-solving, and decision-making which it is recorded. As humanity made leaps in develop-
skills, it ultimately made its way into the medical field. ment, so did the opportunity to build upon that which had
In the latter half of the 20th century, manikins were used already been created. This technological revolution paved
as a simulation to teach medical skills like resuscitation the way for the exponential growth of medical literature,
and suturing. A cardiopulmonary resuscitation manikin and it allows us to continue to advance, building upon the
called “Resusci Anne” was developed to teach mouth-to- gifts given to us by our predecessors.
mouth breathing,38,39 and in the 1960s, the first comput-
er-controlled manikin, Sim One, simulated a real human
look and possessed many physiological responses such References
as reacting to administered drugs.39 In 2012, ROWENA   1. Thorwald J. Science and Secrets of Early Medicine:​Egypt,

960 J Neurosurg Volume 135 • September 2021

 
*50 *
 Raju et al.

Mesopotamia, India, China, Mexico, Peru. Thames & Hud- 27. Davenport A. The History of Photography:​An Overview.
son;​ 1962. University of New Mexico Press;​1999.
  2. Senthilkumar SA, Rai B, Gunasekaran A. Big Data in health- 28. Diamantis A, Magiorkinis E, Androutsos G. Alfred Francois
care management:​a review of literature. Am J Theor Appl Donné (1801-78):​a pioneer of microscopy, microbiology and
Bus. 2018;​4(2):​57–69. haematology. J Med Biogr. 2009;​17(2):​81–87.
  3. Parker S. Kill or Cure:​An Illustrated History of Medicine. 29. Olvido KC. Reenactment, Reconstruction, Recovery:​Nine-
DK;​ 2013. teenth-Century Photographs in the History of Surgery. Uni-
  4. Aminoff MJ, Boller F, Swaab DF, eds. Handbook of Clinical versity of California;​2017.
Neurology. Vol 95. Elsevier;​2009. 30. Harvey Cushing/John Hay Whitney Medical Library. Ac-
  5. Mark JJ. Egyptian papyrus. Ancient History Encyclope- cessed August 13, 2020. https:​//library.medicine.yale.edu
dia. Accessed August 13, 2020. https:​//member.ancient.eu/ 31. Wahl CJ, Tubbs RS, Spencer DD, Cohen-Gadol AA. Harvey
Egyptian_Papyrus Cushing as a book collector, bibliophile, and archivist:​the
  6. Haas LF. Papyrus of Ebers and Smith. J Neurol Neurosurg precedence for the genesis of the Brain Tumor Registry. His-
Psychiatry. 1999;​67(5):​578. torical vignette. J Neurosurg. 2009;​111(5):​1091–1095.
  7. Hughes JT. The Edwin Smith Surgical Papyrus:​an analysis 32. Updegraff HL. The rise of the surgical motion picture. Am J
of the first case reports of spinal cord injuries. Paraplegia. Surg. 1940;​48(3):​523–524.
1988;​26(2):​71–82. 33. Whiteley W. J. Rudolph Jaeger. Accessed August 13, 2020.
  8. Iniesta I. Hippocratic corpus. BMJ. 2011;​342:​d688. http:​//www.mecfilms.com/rudolph.htm
  9. Pearce JMS. Early contribution of Alexandria medical school 34. Clarke CD, Hoshall EM. Illustration:​Its Technique and Ap-
to the anatomy, physiology and pathology of the nervous sys- plication to the Sciences. John D Lucas Co;​1939.
tem. Rev Neurol (Paris). 2019;​175(3):​119–125. 35. Weiner DD. Computer animation—an exciting new tool for
10. Goedvriend GJM. Papermaking past and present. Endeavour. educators. IEEE Trans Educ. 1971;​14(4):​202–209.
1988;​12(1):​38–43. 36. Collins DM, Cotton FA, Hazen EE Jr, et al. Protein crys-
11. Tubbs RI, Gonzales J, Iwanaga J, et al. The influence of an- tal structures:​quicker, cheaper approaches. Science. 1975;​
cient Greek thought on fifteenth century anatomy:​Galenic 190(4219):​1047–1053.
influence and Leonardo da Vinci. Childs Nerv Syst. 2018;​ 37. Pucci JU, Christophe BR, Sisti JA, Connolly ES Jr. Three-
34(6):​1095–1101. dimensional printing:​technologies, applications, and limita-
12. Gurunluoglu R, Gurunluoglu A, Williams SA, Cavdar S. The tions in neurosurgery. Biotechnol Adv. 2017;​35(5):​521–529.
history and illustration of anatomy in the Middle Ages. J 38. Rosen KR. The history of medical simulation. J Crit Care.
Med Biogr. 2013;​21(4):​219–229. 2008;​23(2):​157–166.
13. Calkins CM, Franciosi JP, Kolesari GL. Human anatomical 39. Alaraj A. Comprehensive Healthcare Simulation:​Neurosur-
science and illustration:​the origin of two inseparable disci- gery. Springer International Publishing;​2018.
plines. Clin Anat. 1999;​12(2):​120–129. 40. Ashpole R. Introducing Rowena:​a simulator for neurosurgi-
14. Childress D. Johannes Gutenberg and the Printing Press. cal training. Bull R Coll Surg Engl. 2015;​97(7):​299–301.
Twenty-First Century Books;​2008. 41. Ferro M, Brunori D, Magistri F, et al. A portable da Vinci
15. Hind AM. History of Engraving and Etching. Dover Publica- simulator in virtual reality. Paper presented at:​Third IEEE
tions;​ 2011. International Conference on Robotic Computing (IRC);​Feb-
16. Smith SB. From Ars to Scientia:​the revolution of anatomic ruary 25–27, 2019;​Naples, Italy.
illustration. Clin Anat. 2006;​19(4):​382–388. 42. Albani JM, Lee DI. Virtual reality-assisted robotic surgery
17. Ganseman Y, Broos P. Leonardo da Vinci and Andreas Vesa- simulation. J Endourol. 2007;​21(3):​285–287.
lius;​the shoulder girdle and the spine, a comparison. Acta
Chir Belg. 2008;​108(4):​477–483.
18. Donaldson IM. Jacopo Berengario da Carpi:​the first anat- Disclosures
omy book with a complete series of illustrations. J R Coll The authors report no conflict of interest concerning the materi-
Physicians Edinb. 2008;​38(4):​375. als or methods used in this study of the findings specified in this
19. Russell GA. Vesalius and the emergence of veridical repre- paper.
sentation in Renaissance anatomy. Prog Brain Res. 2013;​203:​
3–32. Author Contributions
20. Sakai T. Historical evolution of anatomical terminology from
ancient to modern. Anat Sci Int. 2007;​82(2):​65–81. Conception and design: Raju. Acquisition of data: Raju. Drafting
21. Lanska DJ, Lanska JR. Medieval and Renaissance anato- the article: Raju, Jumah, Sonig. Critically revising the article:
mists:​the printing and unauthorized copying of illustrations, Jumah, Sun. Reviewed submitted version of manuscript: Nanda,
and the dissemination of ideas. Prog Brain Res. 2013;​203:​ Raju, Jumah, Narayan, Sonig. Administrative/technical/material
33–74. support: Narayan. Study supervision: Nanda, Sun.
22. Lithography. Encyclopedia Britannica. Accessed August 13,
2020. https:​//www.britannica.com/technology/lithography Supplemental Information
23. A history of medical animation and scientific storytelling. Previous Presentations
Accessed August 13, 2020. https:​//random42.com/a-history- Portions of this paper were presented at the AANS Virtual 2020
of-scientific-storytelling E-Poster Hall.
24. Patel SK, Couldwell WT, Liu JK. Max Brödel:​his art, legacy,
and contributions to neurosurgery through medical illustra- Correspondence
tion. J Neurosurg. 2011;​115(1):​182–190.
Anil Nanda: Rutgers-Robert Wood Johnson Medical School, New
25. Ballestriero R. Anatomical models and wax Venuses:​art
Brunswick; Rutgers New Jersey Medical School, Newark, NJ.
masterpieces or scientific craft works? J Anat. 2010;​216(2):​
[email protected].
223–234.
26. Ortug A, Yuzbasioglu N. Tracing the papier mache anatomi-
cal models of Ottoman Turkish medicine and Louis Thomas
Jerôme Auzoux. Surg Radiol Anat. 2019;​41(10):​1147–1154.

J Neurosurg Volume 135 • September 2021 961

 
*50 *
 HISTORICAL VIGNETTE
J Neurosurg 135:962–968, 2021

Historical overview of the Department of Neurosurgery at

Yonsei University College of Medicine in the Republic of
Korea
Dong Ah Shin, MD, PhD, and Dong Kyu Chin, MD, PhD
Department of Neurosurgery, Yonsei University College of Medicine, Seoul, Republic of Korea

Originally founded in 1885, Gwanghyewon later became the Severance Hospital (named after philanthropist Louis Sever-
ance, who supported and funded the construction of a modern hospital) and Yonsei University College of Medicine. The
Department of Neurosurgery at Severance Hospital was established in 1957, and its residency program began in 1961.
Currently, the Department of Neurosurgery has 34 professors and 17 fellows; specialties include vascular, functional,
pediatric, tumor, skull base, and spine neurosurgery. With its state-of-the-art neurosurgical facilities and services, the
Department of Neurosurgery has developed into a department of excellence within the Yonsei University Health System.
In this vignette, the authors present a historic overview of the Department of Neurosurgery.
https://thejns.org/doi/abs/10.3171/2020.8.JNS201076
KEYWORDS  history; Gwanghyewon; Severance Hospital; Yonsei University College of Medicine; Yonsei University
Health System

T
he late 19th century in Korea was marked by Cho- ing a coup d’état in 1884. The injury was so deep that it
sun’s transition from a medieval to a modern dy- was difficult to treat the wound with traditional Korean
nasty. The Chosun Dynasty, which ruled the Ko- medicine. At that time, experts specializing in Chinese
rean Peninsula for 518 years, was occupied by Japan in oriental medicine, including herbal medications, acupunc-
1910. The Japanese occupation ended with the conclusion ture, and moxibustion, were in charge of public health in
of World War II in 1945. In 1948, the Republic of Korea Chosun. Under the care of Horace Allen, Young-ik Min
was formally established. This period also involved enor- recovered completely within 3 months. King Gojong was
mous social and economic changes due to both internal deeply impressed by the modern medicine practiced by
and external pressures. This vigorous transformation Horace Allen and consequently supported the establish-
brought forth modern medicine in Korea. The Gwanghy- ment of Gwanghyewon in 1885, as Chosun’s first modern
ewon (House of Extended Grace) was originally founded hospital.1,5 Soon after its establishment, the hospital was
in 1885, and subsequently converted into the Severance renamed Chejungwon (or House of Civilized Virtue), and
Hospital and Yonsei University College of Medicine.1–3 In in 1886, it became Chejungwon Medical School, which
1957, the Department of Neurosurgery was established at was Korea’s first Western-style medical school.5 Oliver
Severance Hospital.4 In this vignette, we present a review R. Avison, a missionary doctor to Korea who was origi-
of the department’s history by collating and describing nally from the University of Toronto, became the fourth
historical events. director of Chejungwon in 1893 (Fig. 1B). For 50 years,
Avison worked tirelessly to bring modern medical prac-
tices and training to Korea. On September 3, 1904, the
The Birth of Severance Hospital hospital was renamed Severance Hospital in honor of the
Horace Newton Allen, an American physician, was benevolent American philanthropist Louis Henry Sever-
a missionary from a Presbyterian mission in New York ance of the Standard Oil fortune (Figs. 1C and 2). The first
(Fig. 1A). He successfully treated Queen Myeongseong’s 7 graduates in 1908 received doctor’s licenses issued by
nephew, Young-ik Min, who was severely injured dur- the Korean government. The role of traditional medicine

ABBREVIATIONS  WFNS = World Federation of Neurosurgical Societies.

SUBMITTED  April 24, 2020.  ACCEPTED  August 5, 2020.
INCLUDE WHEN CITING  Published online January 22, 2021; DOI: 10.3171/2020.8.JNS201076.

962 J Neurosurg  Volume 135 • September 2021 ©AANS 2021, except where prohibited by US copyright law

*51 *
 Shin and Chin

FIG. 1. A: Horace Newton Allen was an American Presbyterian missionary who studied medicine at Miami University. He intro-
duced modern medical treatment to Korea through Chejungwon. B: Oliver R. Avison was a Canadian missionary and physician.
He stayed in Korea from June 1892 to November 1935. He served as the Fourth General Director of Jejungwon. He laid the foun-
dations of Yonsei University. C: Louis Henry Severance made a significant financial contribution to Chejungwon. His contribution
was used to build a new Severance Hospital. After Severance’s death, his son, John Severance, also made additional financial
contributions to the hospital. Permission to use this image was obtained from the Dong-Eun Medical Museum, which belongs to
Yonsei University. Figure is available in color online only.

has decreased since the introduction of Western medicine. pitals (Severance Hospital, Gangnam Severance Hospital,
During the Japanese occupation, public health was rapidly Yongin Severance Hospital, and the Dental Hospital).
modernized by the government and missionaries. Japan
eliminated oriental medicine from the public health sys- The Beginning of Neurosurgery at Severance
tem and built Western-style hospitals, mainly for the care
of Japanese residents in Korea. In contrast, missionary Hospital
doctors were actively spreading the knowledge of Western According to the First Annual Report of Chejungwon
medicine for Korean habitats as a tool for religious work. Severance Hospital (formerly Gwanghyewon), many pa-
Severance Hospital evolved into the current Yonsei tients treated there in 1885 experienced neurological
University Health System. This system consists of three problems, including epilepsy, migraine, delirium, sciatica,
colleges (Medicine, Dentistry, and Nursing) and four hos- back pain, and scoliosis.6 Oliver R. Avison, who became
the director of the Chejungwon in 1894, treated neurologi-
cal patients who had these and other ailments. Before the
establishment of a neurosurgical specialty at Severance
Hospital, neurosurgery was usually performed by general
surgeons.4,7 In 1926, Yong Sul Lee performed operations
for several cases of spinal cord tumors.4 He published the
article, “Treatment for tubercular spine,” in 1926.8 In 1932,
Alfred Ludlow treated a case of cervical cord injury.4
“Three cases of cerebral concussion” was published by Dr.
Peter Lew in 1934.9 “On the brain tumors” was published
by Dr. Chung Chul Lee in 1935.10 The first brain surgery
at Severance Hospital was performed by Ki Sup Lee in
1943.7,11

Establishment of the Department of

Neurosurgery at Yonsei University
During the Korean War, many Korean surgeons had the
opportunity to observe the advanced neurosurgical and
management skills of foreign neurosurgeons.12,13 The Ko-
rean War provided medical education to Korean doctors
posted at dispatched hospitals, including the Mobile Army
FIG. 2. Severance Hospital in 1904, which was built in front of Sungny-
Surgical Hospital and the hospital ship Jutlandia. Some
emun. Chejungwon was renamed as Severance Hospital on September Korean surgeons also went abroad to study advanced neu-
3, 1904. Permission to use this image was obtained from the Dong-Eun rosurgical practice in greater depth.13 Tai Jun Moon was
Medical Museum, which belongs to Yonsei University. one such pioneer (Fig. 3).12,14 In 1954, he visited the De-

J Neurosurg Volume 135 • September 2021 963

 
*51 *
Shin and Chin

FIG. 3. Dr. Tai Joon Moon established the Department of Neurosurgery

at Yonsei University, the first such department in Korea. He pioneered
neurosurgery in Korea, established the Korean Board of Neurosurgery,
and served as President of the Korean Neurosurgical Society, the Ko- FIG. 4. Dr. Hun Jae Lee (November 11, 1921–March 13, 1983) is a
rean Medical Association, and the World Medical Association. He was key figure in the history of the Yonsei Department of Neurosurgery. He
also elected to serve as a senator in the Korean Assembly. Permission established the subspecialty system within the Department of Neurosur-
to use this image was obtained from Tai Joon Moon’s family. Figure is gery, which remains one of his most significant legacies. Reprinted with
available in color online only. permission from Shin DA, Choi JU, Kim KS, Park HW. A great teacher of
neurosurgery in Korea: Hun Jae Lee (1921–1983). Yonsei Med J. 2016
May;57(3):539–541. Copyright: Yonsei University College of Medicine
2016; CC BY-NC 3.0 (https://creativecommons.org/licenses/by-nc/3.0/).
partment of Neurosurgery at Thomas Jefferson Univer-
sity Hospital in Philadelphia, Pennsylvania, where he was
trained under the guidance of Rudolph Jaeger for 3 years.15
In 1957, he returned to Korea and started an independent the Korean war, he joined the army as a surgeon, where he
neurosurgical service at Severance Hospital, which was gained considerable experience in neurosurgery at the Mo-
equivalent to a neurosurgery division. In 1961, the Depart- bile Army Surgical Hospital. He then underwent resident
ment of Neurosurgery was founded at Severance Hospi- training at the University of Michigan Medical School
tal, and Tai Jun Moon served as the founding chairman. in Ann Arbor, Michigan, between 1955 and 1957.12,18 He
In the same year, a resident training program was also earned his American Board of Neurological Surgery ac-
introduced. Dr. Kyu Woong Choi became the first resi- creditation in 1958. He was invited to Yonsei University to
dent.4 The residency program has, to date, produced more serve as the Chairman of the Department of Neurosurgery
than 170 neurosurgeons who provide exemplary service in 1966. Hun Jae Lee also served as one of the founding
at various institutions across the country. In 1964, direct members of the Korean Neurological Society (established
intracranial aneurysm clipping was performed by Tai Jun in 1961). Due to Lee’s focus on the future and his power-
Moon for the first time in the Republic of Korea.16 ful sense of intuition, the department developed markedly
during the 17 years of his service, which concluded with
The Evolution of the Neurosurgery Program him passing away in 1981.13
Hun Jae Lee also pioneered the establishment of sub-
at Yonsei University: The Lee Years specialty divisions, including vascular, functional, pedi-
Hun Jae Lee, the second chairman of the department atric, and spine neurosurgery. By the early 1970s, highly
(1966–1981), made several significant contributions to the specialized experts managed each subspecialty, as this
department and aided its further growth (Fig. 4).17 During system enabled faculty members to focus on their primary

964 J Neurosurg Volume 135 • September 2021

 
*51 *
 Shin and Chin

field of expertise.17 When selecting residents, Hun Jae Lee derwent training at Queen Square, and SickKids Hospital
always chose one resident from remote cities of the coun- at the University of Toronto, Canada, before returning to
try. His intention was to train residents across the coun- Korea in 1977.2,4 Under his leadership, Severance Hospital
try.17 As a result, Yonsei Neurosurgery alumni have been became a prominent center for treatment of patients with
occupying important positions in major hospitals across pediatric brain tumors, moyamoya disease, spasticity, pe-
the nation. diatric epilepsy, dysraphism, other congenital anomalies,
and pediatric neurotrauma.2,28–31 In 1985, he first intro-
Maturation of the Department’s Subspecialty duced a radioisotope cisternogram grading system.32 In
addition, in 1989, he performed the first neuroendoscopy
System for hydrocephalus in Korea.33 He was the first neurolo-
Kyu Chang Lee became the third chairman (1981– gist in the world to treat hypothalamic hamartomas using
1988), succeeding Hun Jae Lee. He returned to Korea in neuroendoscopy.30 Joong Uhn Choi also hosted the 19th
1975 after studying neurophysiology for 3 years at the International Society of Pediatric Neurosurgery meeting
Max Planck Institute in Germany, under the guidance held in Seoul in 1991.2
of Rolf Hassler. During his time in Germany, he visited In 1989, Kyu-Sung Lee (the seventh chairman, 2003–
Switzerland to learn the microsurgical aneurysm clip- 2007) introduced the surgery for skull base tumors in
ping technique from Gazi Yaşargil. This experience led Korea. He was the first neurologist in Korea to perform
him to become a neurovascular surgeon. Upon his return cavernous sinus surgery and use the transpetrosal and far-
to Yonsei University, he introduced the classic pterional lateral transcondylar approaches. In 1993, he started the
transsylvian approach, the concept of cisternal dissection, first cadaver dissection workshop for skull base surgery
and the technique of microvascular anastomosis in Ko- in Korea at Yonsei University to teach Korean neurosur-
rea.16,19,20 Kyu Chang Lee had clipped 1000 intracranial geons. With his efforts, the neurooncology division be-
aneurysms by 1992, and 2000 by 1999. He also presided came the Neuro-oncology and Skull Base Surgery Center
over the Sixth International Workshop on Cerebrovascular in 2005.4 He was the president of the 14th Asian Austral-
Surgery in Seoul in 2000. asian Congress of Neurological Surgeons, and the presi-
The stereotactic and functional neurosurgery subdi- dent of the 13th Asian Oceanian Congress of Skull Base
vision was established by Sang Sup Chung, the fourth Surgery held in Jeju Island, Korea, in 2015. He also served
departmental chairman (1988–1994). After completing as the chair of the World Federation of Neurosurgical So-
his residency at Severance Hospital in 1970, he went to cieties (WFNS) Skull Base Surgery Committee. With his
Edinburgh University to study stereotactic and func- tireless efforts, the WFNS cadaver dissection course has
tional neurosurgery under John Gillingham and Edward been held in conjunction with the Korean Brain Tumor
Hitchcock.4,21 After returning to Severance Hospital, he Society since 2008. He is the honorary president of the
performed various functional neurosurgeries, including Asian Australasian Society of Neurological Surgeons.
those for movement disorders, epilepsy, and pain. He per- Currently, 34 professors and 17 fellows work in the
formed radiosurgeries as well.22 He was also an expert in Department of Neurosurgery. Their fields of expertise
microvascular decompression for the treatment of hemifa- include vascular, functional, pediatric, tumor, skull base,
cial spasms and trigeminal neuralgia, and he operated on and spinal neurosurgeries (Fig. 5). Last year alone, 98
more than 2000 patients with those conditions.4,23 He was research papers from this department were published in
the President of the Third Asian Society of Stereotactic, international journals. In the same year, a total of KRW
Functional, and Computer-Assisted Neurosurgery meet- 5,541,002,224 ($4,523,267 US) was received as funding
ing, held in Seoul in 1999. support by the department. Our department has the largest
The modern history of spinal surgery at Yonsei Uni- number of surgery cases in Korea and is involved in more
versity began with Young Soo Kim (the department’s fifth than 9000 cases of elective surgery (Fig. 6). Inspired by
chairman, 1994–2001).24 He used a surgical microscope the good will of Horace Newton Allen, Oliver R. Avison,
for a lumbar discectomy in 1975, which was the first micro- and Louis Henry Severance, the Yonsei University Health
scopic discectomy performed in Korea.25,26 In 1980, with System (Fig. 7) educates doctors from developing coun-
a scholarship from the British Council, he underwent fur- tries. Each year, more than 10 visiting scholars join the
ther training for treating spinal cord injury at Stoke Man- Department of Neurosurgery. We are actively involved in
deville Hospital in the United Kingdom. When Gangnam the teaching and training of neurosurgeons, as we have
Severance Hospital was established in 1983, he was ap- learned from our history and experience that education
pointed as the head of the neurosurgery department. With is the best way to share knowledge that will eventually
Young Soo Kim’s enthusiasm, the Yonsei Spine Institute lead to the evolution of neurosurgery. Our department
was founded in 2003, and this institute developed into has played a role in the activities of WFNS since 1989. In
the Gangnam Severance Spine Hospital in 2005. Young 2013, the 15th WFNS World Congress of Neurosurgery
Soo Kim introduced chemonucleolysis for the treatment was held in Seoul. Just as Severance Hospital started as
of herniated lumbar discs in 1984, and he had performed Allen’s mission, Seung Kon Huh was sent by Severance
surgery for more than 3000 such cases by 1990.27 He was Hospital Mission Centre to Tanzania in 2017. Presently,
also the president of the Fourth International Neurotrau- he works with local doctors, assisting them in the areas of
ma Society meeting held in Seoul in 1997. education, research, and patient treatment, as well as per-
Joong Uhn Choi (the sixth chairman, 2001–2003) es- forming other missionary activities. Since our founding,
tablished the Division of Pediatric Neurosurgery. He un- the Korean Neurosurgical Society has grown significantly,

J Neurosurg Volume 135 • September 2021 965

 
*51 *
Shin and Chin

FIG. 5. A photograph of faculty members outside the restored Gwanghyewon on the campus of Yonsei University, taken January
2, 2016. Copyright owned by Yonsei University. Published with permission. Figure is available in color online only.

FIG. 6. Left: The number of operations performed by different subspecialties in 2016 at Yonsei University College of Medicine.
Right: The total number of operations performed in each of the last 5 years.

966 J Neurosurg Volume 135 • September 2021

 
*51 *
 Shin and Chin

FIG. 7. An aerial view of Severance Hospital taken in March 2018. Photograph taken by Yonsei University. Permission to use this
image was obtained from Yonsei University. Figure is available in color online only.

and as of 2019, it consists of 3169 board-certified mem- possible. We would like to thank Editage for English-language
bers, 85 training institutions, 5 regional societies, and 12 editing.
subspecialty societies nationwide.
References
Conclusions   1. Yeo IS. Severance Hospital:​bringing modern medicine to
In this paper we present a general overview of the early Korea. Yonsei Med J. 2015;​56(3):​593–597.
history of neurosurgery and current status of the Depart-   2. Di Rocco C. Joon-Uhn Choi:​The 2005–2006 ISPN Presi-
dent. Childs Nerv Syst. 2007;​23(9):​925–928.
ment of Neurosurgery at Yonsei University. This article   3. Park HW, Yeo IS, Choi JU. The Department of Neurosurgery
describes the clinical, academic, and scientific develop- at Seoul National University:​past, present, and future. Neu-
ment of the department, including its present activities rosurgery. 2003;​52(5):​1240–1241.
and future mission. The tradition of collaborative, benevo-   4. Department of Neurosurgery, Yonsei University. 50 Years
lent, and cooperative efforts exercised by its founders and History of Department of Neurosurgery in Yonsei University.
chairpersons will endure, as we continue to serve society ML Communications;​ 2007:​63–71.
with the goal of freeing our patients from disease and suf-   5. Park HW, Park YJ, Yeo IS, Kim IS. Medical education in Je
fering. Joon Won. Article in Korean. Uisahak. 1999;​8(1):​25–44.
  6. Allen HN, Heron JW. First Annual Report of the Korean
Government Hospital, Seoul. R Meiklejohn & Co, 1886.
Acknowledgments   7. Park KB, Roh YH, Lee-Park O, Park S. History of neurosur-
We thank those who reviewed the manuscript, assisted in gery in Democratic People’s Republic of Korea. World Neu-
historical research, or provided photographs. The publication rosurg. 2015;​84(3):​855–859.
committee members included Jong Hee Chang, Won Seok Chang,   8. Lee YS. Treatment for tubercular spine. The Severance Bul-
Moo Sung Kang, Eui Hyun Kim, Yong Bae Kim, Keun Young letin. 1926;​8:​1–4.
Park, Kyu Won Shim, and Seong Yi. All of them are significant   9. Lew P. Three cases of cerebral concussion. Korean Med J.
contributors to this paper. We are especially grateful to Joong 1934;​4:​74–84.
Uhn Choi, Sang Sup Chung, Kyu-Sung Lee, and Keun Su Kim, 10. Lee CC. On the brain tumors. Korean Med J. 1935;​5:​65.
whose support and encouragement made the writing of this paper 11. The Korean Neurosurgical Society. 40 Years History of the

J Neurosurg Volume 135 • September 2021 967

 
*51 *
Shin and Chin

Korean Neurosurgical Society. Jungang Munhwasa;​2002:​ 27. Kim YS, Chin DK, Yoon DH, et al. Predictors of successful
3–258. outcome for lumbar chemonucleolysis:​analysis of 3000 cases
12. Lee HJ. 90 years history of Korean Medical Science. The during the past 14 years. Neurosurgery. 2002;​51(5)(suppl):​
KMA News. 1975:​11. Accessed October 2, 2020. http:​//pdfs. S123–S128.
doctorsnews.co.kr/1975/19750206/19750206_11.pdf 28. Choi JU, Kim DS, Chung SS, Kim TS. Treatment of germ
13. Sim BS. Perspectives in international neurosurgery:​neuro- cell tumors in the pineal region. Childs Nerv Syst. 1998;​14(1-
surgery in the Republic of Korea (South Korea). Neurosur- 2):​41–48.
gery. 1979;​5(1 Pt 1):​75. 29. Choi JU, Kim DS, Kim EY, Lee KC. Natural history of moy-
14. Hwang SN. History of Korean neurosurgery. World Neuro- amoya disease:​comparison of activity of daily living in sur-
surg. 2015;​84(2):​192–196. gery and non surgery groups. Clin Neurol Neurosurg. 1997;​
15. Chalouhi N, Osterholm J, Jabbour P, et al. History of the 99(2)(suppl 2):​S11–S18.
Department of Neurosurgery at Thomas Jefferson University 30. Choi JU, Yang KH, Kim TG, et al. Endoscopic disconnection
Hospital. Neurosurgery. 2013;​73(4):​709–718. for hypothalamic hamartoma with intractable seizure. Report
16. Lee KC. Development of cerebrovascular surgery in Korea. of four cases. J Neurosurg. 2004;​100(5)(Suppl Pediatrics):​
Korean J Cerebrovasc Surg. 2005;​7:​5 –11. 506–511.
17. Shin DA, Choi JU, Kim KS, Park HW. A great teacher of 31. Shim KW, Park EK, Lee YH, et al. Treatment strategy for in-
neurosurgery in Korea:​Hun Jae Lee (1921–1983). Yonsei tracranial primary pure germinoma. Childs Nerv Syst. 2013;​
Med J. 2016;​57(3):​539–541. 29(2):​239–248.
18. Kahn EA. Journal of a Neurosurgeon. Charles C Thomas;​ 32. Matsumoto S, Tamaki N. Hydrocephalus. In:​Matsumoto S,
1972. Tamaki N, eds. Hydrocephalus:​Pathogenesis and Treat-
19. Lee KC. The present and future of cerebrovascular surgery. J ment. Springer Science & Business Media;​2012.
Korean Neurosurg Soc. 2002;​32(4):​295–299. 33. Choi JU. Atlas of Practical Neuroendoscopy. Yonsei Univer-
20. Lee KC. The role of neurosurgeons in the era of endovascular sity Press;​2001.
treatment of cerebral aneurysms. Neurol Med Chir (Tokyo).
2002;​42(5):​242–243.
21. Hwang SN. History of Korean stereotactic and functional Disclosures
neurosurgery. Neurosurgery. 2005;​56(2):​406–409. The authors report no conflict of interest concerning the materi-
22. Chung SS. History of stereotactic surgery in Korea. In:​Lo- als or methods used in this study or the findings specified in this
zano M, Gildenberg PL, Tasker RR, eds. Textbook of Stereo- paper.
tactic and Functional Neurosurgery. Springer Berlin Heidel-
berg;​ 2009:​171–178. Author Contributions
23. Chung SS, Chang JH, Choi JY, et al. Microvascular decom-
pression for hemifacial spasm:​a long-term follow-up of 1,169 Conception and design: Shin. Acquisition of data: Shin. Analysis
consecutive cases. Stereotact Funct Neurosurg. 2001;​77(1-4):​ and interpretation of data: Shin. Drafting the article: Shin.
190–193. Critically revising the article: Chin. Reviewed submitted version
24. Kim YS. A clinical study of intervertebral disc surgery based of manuscript: Chin. Administrative/technical/material support:
on 1500 cases. J Korean Neurosurg Soc. 1973;​2(1):​71–82. Chin. Study supervision: Chin.
25. Kim YS, Chung SS, Lee KC, Lee HJ. Experience with mi-
crosurgery in the herniations of lumbar, cervical and thoracic Correspondence
intervertebral discs. J Korean Neurosurg Soc. 1979;​8(2):​ Dong Kyu Chin: Gangnam Severance Hospital, Yonsei University
299–306. College of Medicine, Seoul, Korea. [email protected].
26. Kim YS, Lee KC. Microsurgical operation of herniated lum-
bar disc. J Korean Neurosurg Soc. 1982;​11(4):​515–521.

968 J Neurosurg Volume 135 • September 2021

 
*51 *
 CLINICAL ARTICLE
J Neurosurg 135:969–976, 2021

Predicting the consistency of intracranial meningiomas

using apparent diffusion coefficient maps derived from
preoperative diffusion-weighted imaging
Kenya Miyoshi, MD,1 Tsukasa Wada, MD,1 Ikuko Uwano, PhD,2 Makoto Sasaki, MD,2
Hiroaki Saura, MD,1 Shunrou Fujiwara, PhD,1 Fumiaki Takahashi, PhD,3 Eiki Tsushima, PhD,4 and
Kuniaki Ogasawara, MD1
1
Department of Neurosurgery and 2Division of Ultrahigh Field MRI, Institute for Biomedical Sciences, and 3Division of Medical
Engineering, Department of Information Science, Iwate Medical University School of Medicine, Morioka; and 4Department of
Physical Therapy, Hirosaki University School of Health Science, Hirosaki, Japan

OBJECTIVE  The consistency of meningiomas is a critical factor affecting the difficulty of resection, operative complica-
tions, and operative time. The apparent diffusion coefficient (ADC) is derived from diffusion-weighted imaging (DWI) and
is calculated using two optimized b values. While the results of comparisons between the standard ADC and the consis-
tency of meningiomas vary, the shifted ADC has been reported to be strongly correlated with liver stiffness. The purpose
of the present prospective cohort study was to determine whether preoperative standard and shifted ADC maps predict
the consistency of intracranial meningiomas.
METHODS  Standard (b values 0 and 1000 sec/mm2) and shifted (b values 200 and 1500 sec/mm2) ADC maps were
calculated using preoperative DWI in patients undergoing resection of intracranial meningiomas. Regions of interest
(ROIs) were placed within the tumor on standard and shifted ADC maps and registered on the navigation system. Tumor
tissue located at the registered ROI was resected through craniotomy, and its stiffness was measured using a durometer.
The cutoff point lying closest to the upper left corner of a receiver operating characteristic (ROC) curve was determined
for the detection of tumor stiffness such that an ultrasonic aspirator or scissors was always required for resection. Each
tumor tissue sample with stiffness greater than or equal to or less than this cutoff point was defined as hard or soft tu-
mor, respectively.
RESULTS  For 76 ROIs obtained from 25 patients studied, significant negative correlations were observed between stiff-
ness and the standard ADC (ρ = −0.465, p < 0.01) and the shifted ADC (ρ = −0.490, p < 0.01). The area under the ROC
curve for detecting hard tumor (stiffness ≥ 20.8 kPa) did not differ between the standard ADC (0.820) and the shifted
ADC (0.847) (p = 0.39). The positive predictive value (PPV) for the combination of a low standard ADC and a low shifted
ADC for detecting hard tumor was 89%. The PPV for the combination of a high standard ADC and a high shifted ADC for
detecting soft tumor (stiffness < 20.8 kPa) was 81%.
CONCLUSIONS  A combination of standard and shifted ADC maps derived from preoperative DWI can be used to pre-
dict the consistency of intracranial meningiomas.
https://thejns.org/doi/abs/10.3171/2020.6.JNS20740

M
KEYWORDS  meningioma; consistency; apparent diffusion coefficient; durometer; diagnostic technique

eningiomas are the most common primary brain mas are reportedly easier to resect and have a lower in-
tumors, and resection is the first-choice treat- cidence of operative complications and shorter operative
ment for these tumors. Tumor consistency is a time than stiffer meningiomas.1–4 Preoperative prediction
critical factor affecting the difficulty of resection, opera- of tumor consistency thus allows the selection of an ap-
tive complications, and operative time; softer meningio- propriate surgical strategy for resection.

ABBREVIATIONS  ADC = apparent diffusion coefficient; AUC = area under the ROC curve; DWI = diffusion-weighted imaging; NPV = negative predictive value; PPV =
positive predictive value; ROC = receiver operating characteristic; ROI = region of interest; SPGR = spoiled gradient–recalled acquisition in the steady state.
SUBMITTED  March 10, 2020.  ACCEPTED  June 30, 2020.
INCLUDE WHEN CITING  Published online November 13, 2020; DOI: 10.3171/2020.6.JNS20740.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 135 • September 2021
  969

*52 *
Miyoshi et al.

Many investigators have assessed meningioma con- Preoperative MRI and Data Analyses
sistency using preoperative MRI. Whereas some studies MRI studies were performed using a 3-T MR scanner
showed a negative correlation between T2-weighted sig- (Discovery MR750, GE Healthcare) with an 8-channel
nal intensity and meningioma consistency,3–8 others did head coil within 7 days before surgery. Patients under-
not find this correlation.2,9,10 Hard meningiomas report- went DWI with different b values (0, 200, 1000, and 1500
edly had higher fractional anisotropy values on diffusion sec/mm2) based on a single-shot spin-echo echo-planar
tensor imaging than soft meningiomas.9–11 Furthermore, sequence (repetition time/echo time = 4000/72 msec; field
the results of comparisons between the standard apparent of view = 240 mm2; matrix = 128 × 128; slice thickness =
diffusion coefficient (ADC) and the consistency of menin- 3 mm without interslice gaps; motion probing gradient = 3
giomas vary: the standard ADC was lower12 or higher13 in orthogonal directions, average = 4; number of slices = 38).
hard meningiomas or was not correlated with meningioma Subsequently, patients underwent gadolinium-enhanced
consistency.8 The variation in results among these studies 3D T1-weighted imaging with spoiled gradient–recalled
may depend on differences in methodological factors. In acquisition in the steady state (SPGR) for the navigation
particular, the accuracy of the subjective grading system system.
of tumor consistency defined based on a difference in de- Two kinds of ADC maps were calculated using two
vices used for tumor resection12,13 or a visual analog scale pairs of diffusion-weighted images obtained with two
during surgery8 should be verified based on the stiffness different b values: a pair of 0 and 1000 sec/mm2 values
of tumor tissue measured objectively using a durometer.14 for the standard ADC map and a pair of 200 and 1500
Meningioma consistency is also heterogeneous within a sec/mm2 for the shifted ADC map, which was previously
tumor.15 The consistency should be determined in each reported as the optimal pair of b values for assessing body
region within a tumor rather than in the whole of a tu- stiffness.16 One to five ROIs, each with a diameter of 10
mor. For this purpose, using a navigation system, regions mm, were placed within homogeneous regions in the tu-
of interest (ROIs) in preoperative MR images should be mor on the standard and shifted ADC maps by one neu-
matched with locations for resection in an operative field roradiologist. Locations of ROIs on these two ADC maps
as accurately as possible. were set identically. The ADC value was then determined
A recent study showed that the shifted ADC calculat- in each ROI. Furthermore, each ROI set on the standard
ed using two optimized b values was strongly and nega- and shifted ADC maps was registered to the 3D-SPGR
tively correlated with stiffness of the liver in patients with images for the navigation system.
chronic liver diseases.16 The standard ADC is calculated
as though water molecules move freely, but movement of Intraoperative Procedures
water molecules is actually limited due to tissue structures
After incorporating the 3D-SPGR images including
such as cell membranes.17 The shifted ADC has been de-
the registered ROIs with a navigation system (Curve,
vised to weight these limitations.16 Shifted ADC maps can
Brainlab AG), the patient underwent craniotomy under
be obtained using a conventional MRI system without
general anesthesia. One senior neurosurgeon blinded to
special or invasive devices.
the ADC values sharply resected tumor tissue located on
The purpose of the present prospective cohort study
the navigation system at a registered ROI with a diameter
was to determine whether standard and shifted ADC
maps derived from preoperative diffusion-weighted imag- of 8 mm without electrical cauterization. Immediately af-
ing (DWI) predict the consistency of intracranial menin- ter, to quantitatively assess the consistency of the resected
giomas. tumor tissue, an investigator blinded to ADC values mea-
sured the tumor tissue stiffness as pressure (kPa) using a
durometer (NEUTONE TDM-Z2 PM, Try-All Corp.).18,19
Methods The device used a presser foot that gauges stiffness by
Patients measuring how far the material indents the foot, with a
For this prospective observational study, patients were higher pressure indicating a harder material.18,19
enrolled if they met the following inclusion criteria: intra- Furthermore, based on the difference in devices used
cranial meningioma with a maximum diameter of ≥ 30 for tumor resection, the same neurosurgeon classified the
mm diagnosed on conventional MRI; no previous craniot- consistency of tumor tissues surrounding the ROI into the
omy, radiation therapy, or chemotherapy for brain lesions; following 5 grades (defined as device-dependent grades):
and age > 16 years. Patients meeting the following criteria grade 1, the tumor is easily removed with only a suction
were excluded: no 3-T MRI performed, due to implantable tube; grade 2, the tumor cannot be removed with only a
electronic devices or implantable metals, which are con- suction tube, and a combination with an ultrasonic aspira-
traindications for 3-T MRI; intratumoral hemorrhage on tor is required; grade 3, an ultrasonic aspirator is required
MRI; and informed consent not obtained. Only patients throughout tumor removal; grade 4, the tumor cannot be
who were diagnosed as having meningiomas by histologi- removed with only an ultrasonic aspirator, and a combi-
cal examination after resection through craniotomy met nation with scissors is required; and grade 5, scissors are
the final eligibility requirements for this study. required throughout tumor removal.
Our hospital’s institutional ethics committee assessed
and approved the study protocol. Written, informed con- Statistical Analysis
sent was obtained from each patient or a family member Data are expressed as means ± standard deviations.
before enrollment in the study. Spearman’s rank correlation coefficient was used to as-

970 J Neurosurg Volume 135 • September 2021

 
*52 *
 Miyoshi et al.

the presence or absence of statistical dependence due to

multiple sampling of individual tumors, the ROI popula-
tion obtained was analyzed using the bootstrap method.
Briefly, one ROI from one to five ROIs obtained for each
tumor was randomly selected, and samples with different
combinations of ROIs were generated. Spearman’s rank
correlation coefficient and the p value between two pa-
rameters were calculated for each sample. The mean and
95% CI values of Spearman’s rank correlation coefficients
and the p values for all samples were calculated and com-
pared with Spearman’s rank correlation coefficients and
the p values for the original ROI population.

Results
Over the course of 15 months, 29 patients satisfied the
inclusion criteria. Four patients were excluded, 3 of whom
did not undergo 3-T MRI due to having implantable elec-
tronic devices or implantable metals, and 1 patient who
had an intratumoral hemorrhage found on MRI. The re-
FIG. 1. Relationships between the stiffness of each tumor tissue sample maining 25 patients successfully underwent tumor resec-
measured using a durometer and the device-dependent grade deter- tion through craniotomy and were diagnosed as having
mined based on differences in the devices used for tumor resection as meningiomas by histological examination. These patients
follows: grade 1, the tumor is easily removed with only a suction tube; were finally included in the present study.
grade 2, the tumor cannot be removed with only a suction tube, and a The age range of the 25 patients (10 men and 15 wom-
combination with an ultrasonic aspirator is required; grade 3, an ultra- en) was 41–84 years (mean age 61 ± 12 years). The tumor
sonic aspirator is required throughout tumor removal; grade 4, the tumor
cannot be removed with only an ultrasonic aspirator, and a combination was located at the supratentorial convexity in 7 patients,
with scissors is required; and grade 5, scissors are required throughout the parasagittal sinus in 6 patients, the falx cerebri in 3 pa-
tumor removal. The dotted horizontal line denotes the cutoff point ob- tients, the tuberculum sellae in 2 patients, the infratentori-
tained from the ROC curve for the stiffness to detect device-dependent al convexity in 2 patients, the sphenoid ridge in 2 patients,
grade ≥ 3. the olfactory groove in 1 patient, the tentorium cerebelli
in 1 patient, and the planum sphenoidale in 1 patient. All
meningiomas with Simpson grades ≤ 3 were successfully
sess correlations between two parameters. The difference resected. Histological examinations diagnosed meningo-
in these coefficients between two parameters was assessed thelial-type meningioma in 13 patients, transitional type
with the 95% confidence interval (CI). The cutoff point in 6 patients, fibrous type in 4 patients, psammomatous
lying closest to the upper left corner of a receiver operat- type in 1 patient, and clear cell type in 1 patient.
ing characteristic (ROC) curve for the stiffness to detect For the 25 patients studied, a total of 78 ROIs were set
device-dependent grade ≥ 3 was determined. Each tumor on MR images. However, the ADC was not obtained due
tissue with stiffness ≥ 3 or < 3 at this cutoff point was to strong distortion on diffusion-weighted images in the
defined as hard or soft tumor, respectively. ROC curves ROI in the tumor located at the tentorium cerebelli. Tumor
were created to assess the accuracies of the standard ADC tissue in another ROI was not intraoperatively obtained
and the shifted ADC for detecting hard tumor. The area due to the deep location of the targeted ROI. These 2 ROIs
under the ROC curve (AUC) was used to determine the were excluded from subsequent analyses, and the remain-
ability to discriminate between hard tumor and soft tumor. ing 76 ROIs were finally analyzed. Four, 4, 9, 5, and 3 pa-
The method of Pepe and Longton20 was used for pairwise tients had 5, 4, 3, 2, and 1 ROIs for analyses, respectively,
comparison of the AUCs for the standard ADC and the in each tumor. Overall, 40, 17, 14, 3, and 2 ROIs were me-
shifted ADC. A p value < 0.05 was considered significant ningothelial type, transitional type, fibrous type, clear cell
for all tests. Differences of sensitivity, specificity, positive type, and psammomatous type, respectively. The stiffness
predictive values (PPVs), and negative predictive values of the tumor tissue resected in each ROI was successfully
(NPVs) among the standard ADC, the shifted ADC, and measured using a durometer in all 76 ROIs. Tumor tis-
a combination of these two were assessed using 95% CIs sues surrounding 17, 28, 24, 6, and 1 ROIs were intraop-
calculated using binomial distributions. Last, taking ac- eratively classified into device-dependent grades 1, 2, 3, 4,
count for the correlation between different measures on and 5, respectively. Thirteen (59%) of 22 patients with ≥ 2
the same tumor, the relationship between the two parame- ROIs in a tumor had ROIs with different device-dependent
ters was evaluated using generalized linear mixed-effects grades in the tumor.
models with a subject random intercept. Device-dependent A significant positive correlation was observed be-
grade was assumed to be a multinomial variable with a cu- tween the device-dependent grade and the stiffness of
mulative logit link. Stiffness on a durometer, the standard each tumor tissue measured using a durometer (ρ = 0.715,
ADC, and the shifted ADC were assumed to be a nor- 95% CI 0.584–0.810; p < 0.01) (Fig. 1). The cutoff point
mal variable with an identity link. In addition, to elucidate lying closest to the upper left corner of the ROC curve for

J Neurosurg Volume 135 • September 2021 971

 
*52 *
Miyoshi et al.

FIG. 2. Relationships between device-dependent grade and the standard ADC (left) or the shifted ADC (right).

the stiffness to detect a device-dependent grade ≥ 3 was Based on Fig. 3, which shows the relationships among
20.8 kPa (Fig. 1). With that cutoff point, the sensitivity, the standard ADC, shifted ADC, and stiffness, ROC
specificity, PPV, and NPV were 90%, 76%, 72%, and 92%, curves for the former two for detecting hard tumor (stiff-
respectively. Thus, 40 tumor tissues with stiffness ≥ 20.8 ness ≥ 20.8 kPa) were calculated as shown in Fig. 4. AUCs
kPa and 36 tumor tissues with stiffness < 20.8 kPa were for the standard ADC and the shifted ADC were 0.820
defined as hard and soft tumors, respectively. (95% CI 0.716–0.899) and 0.847 (95% CI 0.746–0.919), re-
Significant negative correlations were observed be- spectively. The AUCs did not differ between the standard
tween the device-dependent grade and the standard ADC ADC and the shifted ADC (difference between AUCs,
(ρ = −0.519, 95% CI −0.667 to −0.333; p < 0.01) and the 0.027; p = 0.39). Table 1 shows the sensitivity, specificity,
shifted ADC (ρ = −0.557, 95% CI −0.695 to −0.379; p < PPV, and NPV for the standard and shifted ADCs with
0.01) (Fig. 2). The correlation coefficients did not differ be- the cutoff point lying closest to the upper left corner of
tween these two ADCs. Significant negative correlations the ROC curve for detecting hard tumor. The sensitivity
were observed between the stiffness and the standard and NPV were greater for the standard ADC than for the
ADC (ρ = −0.465, 95% CI −0.624 to −0.267; p < 0.01) and shifted ADC, and the specificity and PPV were greater for
the shifted ADC (ρ = −0.490, 95% CI −0.645 to −0.298; the shifted ADC than for the standard ADC, though a sig-
p < 0.01) (Fig. 3). The correlation coefficients also did not nificant difference between these two ADCs was observed
differ between these two ADCs. only for specificity.

FIG. 3. Relationships between stiffness on a durometer and the standard ADC (left) or the shifted ADC (right). The dotted vertical
lines denote the cutoff points obtained from ROC curves for the stiffness used to detect device-dependent grade ≥ 3. The dotted
horizontal lines denote the cutoff points obtained from the ROC curves for the hard tumor defined based on the stiffness.

972 J Neurosurg Volume 135 • September 2021

 
*52 *
 Miyoshi et al.

and the standard ADC (p < 0.01), between the device-de-

pendent grade and the shifted ADC (p < 0.01), between
the stiffness on a durometer and the standard ADC (p <
0.01), and between the stiffness on a durometer and the
shifted ADC (p < 0.01).
When the original ROI population (76 ROIs) was an-
alyzed using the bootstrap method, the total number of
samples with different combinations of 25 ROIs (25 tu-
mors in total) was theoretically > 50 billion (54 × 44 × 39
× 25 × 13). Therefore, just 10,000 samples with different
combinations of 25 ROIs were randomly selected and ana-
lyzed (Table 2). Spearman’s rank correlation coefficients
and the p values between any two parameters for the origi-
nal ROI population were approximately equal to the mean
and 95% CI of such values obtained from 10,000 samples.

Discussion
The present study demonstrated that a combination of
standard and shifted ADC maps derived from preopera-
tive DWI predicts the consistency of intracranial menin-
giomas.
ROIs ≥ 2 were obtained in 22 of 25 tumors studied.
FIG. 4. ROC curves used to compare accuracy between the standard This multiple sampling of individual tumors might cause
ADC and the shifted ADC for the detection of hard tumor. statistical overestimation or dependency. Therefore, tak-
ing into account the correlation between different mea-
sures on the same tumor, the relationship between the
Figure 5 shows the relationships among the standard two parameters was evaluated using generalized linear
ADC, the shifted ADC, and the consistency of the tumor mixed-effects models with a subject random intercept.
tissue. A strong positive correlation was observed between As a result, significant relationships between both param-
the former two (ρ = 0.894, 95% CI 0.837–0.932; p < 0.01). eters were observed, suggesting the absence of statistical
The PPV for the combination of standard ADC ≤ 0.880 × overestimation caused by the multiple sampling. Further,
10−3 and shifted ADC ≤ 0.680 × 10−3 for detecting hard tu- the bootstrap method was used to determine the pres-
mor was 89% (95% CI 77%–100%). The PPV for the com- ence or absence of statistical dependency in our original
bination of standard ADC > 0.880 × 10−3 and shifted ADC ROI population (76 ROIs). It was found that Spearman’s
> 0.680 × 10−3 for detecting soft tumor was 81% (95% CI rank correlation coefficients and the p values between any
67%–96%). The PPV for the combination of standard ADC two parameters for the original ROI population were ap-
≤ 0.880 × 10−3 and shifted ADC > 0.680 × 10−3 for detect- proximately equal to the mean and 95% CI of such values
ing hard tumor was 50% (95% CI 29%–71%). There was obtained from the 10,000 samples. These finding suggest
only one ROI with a combination of conventional ADC > the absence of statistical dependency in the original ROI
0.880 × 10−3 and shifted ADC ≤ 0.680 × 10−3. population.
Gadolinium-enhanced 3D T1-weighted SPGR images Meningioma consistency is heterogeneous within a tu-
and standard and shifted ADC maps in each patient with mor.15 Actually, in the present study, more than half of the
hard or soft meningiomas are shown in Fig. 6. patients with ROIs ≥ 2 within a tumor had ROIs showing
Generalized linear mixed-effects models with a sub- different device-dependent grades within the tumor. The
ject random intercept showed a significant relationship correlation coefficient between the subjective device-de-
between the device-dependent grade and the stiffness on a pendent grading system and the stiffness of tumor tissue
durometer (p < 0.01), between the device-dependent grade measured using a durometer corresponded with that re-

TABLE 1. Sensitivity, specificity, PPV, and NPV for the standard ADC and the shifted ADC for the
detection of hard meningiomas
Standard Shifted Statistical Significance
ADC ADC (by comparison of 95% CIs)
Sensitivity (95% CI) 88% (77–98%) 63% (47–78%) No
Specificity (95% CI) 61% (45–77%) 92% (83–100%) Yes
PPV (95% CI) 71% (59–84%) 89% (78–100%) No
NPV (95% CI) 81% (67–96%) 69% (56–82%) No
Cutoff point (mm2/sec) 0.880 × 10−3 0.680 × 10−3

J Neurosurg Volume 135 • September 2021 973

 
*52 *
Miyoshi et al.

ported in a previous study.14 In addition, using a navigation

system, ROIs in preoperative MR images were matched
with locations for resection on an operative field as ac-
curately as possible. Thus, the standard ADC and shifted
ADC values were negatively correlated with the device-
dependent grade or the stiffness of tumor tissue measured
using a durometer.
MR elastography can obtain a body stiffness map.
Several studies have demonstrated that stiffness on MR
elastography was correlated more strongly with menin-
gioma consistency than T1- or T2-weighted signal inten-
sity.2,14 However, because stiffness on MR elastography is
calculated based on the phase shift made by intentional
mechanical vibration obtained from data with and without
this vibration, this method needs a special device to apply
vibration to the brain.2,14 Whether the mechanical vibra-
tion affects the brain also remains unknown. The corre-
lation coefficient of the shifted ADC was approximately
equal to that between stiffness on this MR elastography
and the device-dependent grade or the stiffness measured
using a durometer.14 Thus, the shifted ADC may predict
FIG. 5. Relationships among the standard ADC, the shifted ADC, and the consistency of meningioma with the same accuracy as
the consistency of tumor tissue. Closed and open circles denote hard MR elastography.
and soft tumor tissues, respectively. The dotted horizontal and vertical Because ADC is a continuous variable, hard or soft tu-
lines denote the cutoff points obtained from the ROC curves for dis- mor was determined based on stiffness as a continuous
criminating between hard and soft tumor tissues defined based on the variable to assess the accuracy of the standard ADC or
stiffness. the shifted ADC for discrimination between hard tumor
and soft tumor. For this determination, a hard tumor was
clinically defined as requiring an ultrasonic aspirator or

FIG. 6. Gadolinium-enhanced 3D T1-weighted images with SPGR (left column), standard (middle column), and shifted (right
column) ADC maps in a 68-year-old woman (upper row) and a 49-year-old man (lower row) with a hard and soft meningioma,
respectively. Red closed circles denote ROIs placed within the tumor. The ADC values on both standard and shifted ADC maps
are decreased in the hard meningioma and increased in the soft meningioma compared with the cerebral tissue contralateral to
the tumor.

974 J Neurosurg Volume 135 • September 2021

 
*52 *
 Miyoshi et al.

TABLE 2. Spearman’s rank correlation coefficients and p values between any two parameters from the original ROI population (76 ROIs)
and the means and 95% CIs of such values obtained from 10,000 samples with different combinations of 25 ROIs
Spearman’s Rank Correlation Coefficient Spearman’s Rank Correlation Coefficient p Value
10,000 Samples w/ Different 10,000 Samples w/ Different
Combinations of 25 ROIs Combinations of 25 ROIs
Original ROI Original ROI
95% CI Population 95% CI Population
Mean Lower Limit Upper Limit (76 ROIs) Mean Lower Limit Upper Limit (76 ROIs)
Device-dependent grade and 0.746 0.745 0.746 0.715 <0.01 <0.01 <0.01 <0.01
stiffness on durometer
Device-dependent grade and −0.600 −0.605 −0.599 −0.519 <0.01 <0.01 <0.01 <0.01
standard ADC
Device-dependent grade and −0.633 −0.633 −0.636 −0.557 <0.01 <0.01 <0.01 <0.01
shifted ADC
Stiffness on durometer and −0.537 −0.538 −0.537 −0.465 0.01 0.01 0.01 <0.01
standard ADC
Stiffness on durometer and −0.578 −0.578 −0.577 −0.490 <0.01 <0.01 <0.01 <0.01
shifted ADC

scissors throughout tumor removal. Based on the ROC work (perfusion).26,27 The latter reportedly contributes to
analysis, a hard tumor was defined as having stiffness ≥ the standard ADC.4,12 Furthermore, the standard ADC is
20.8 kPa. This value was considerably greater than the calculated as water molecules move according to Gaussian
value of > 6 kPa used in a study of MR elastography.14 (normal) diffusion, and the signal decays monoexponen-
However, that study did not explain the reason for using tially.16 However, these conditions are actually limited in
that definition. Our subsequent analysis demonstrated that biological tissue. The shifted ADC is calculated by using
the standard ADC or the shifted ADC discriminated be- two key b values optimized to reflect Gaussian diffusion,
tween hard tumor and soft tumor with moderate accuracy, non-Gaussian diffusion, and intravoxel incoherent mo-
and this discriminative ability did not differ between the tion.16 The finding that specificity and the PPV for detect-
standard ADC and the shifted ADC. It also showed that ing hard tumor were greater for the shifted ADC than for
approximately 90% of hard tumors or soft tumors showed the standard ADC suggested that soft tumor is excessively
a low standard ADC or a high shifted ADC, respectively, affected by tumor cellularity and the amount of fibrous
but the PPV for detecting hard tumor or soft tumor differed tissue. In contrast, the finding that sensitivity and the NPV
among combinations of the standard ADC and the shifted for detecting hard tumor were greater for the standard
ADC. Overall, ≥ 80% of tumors with a low standard ADC ADC than for the shifted ADC suggested that hard tumor
and a low shifted ADC were hard tumors, and ≥ 80% of is affected by microcirculation of blood in the tumor, as
tumors with a high standard ADC and a high shifted ADC well as tumor cellularity and the amount of fibrous tissue.
were soft tumors. These findings suggest the need to pre- A pair of b values (200 and 1500 sec/mm2) was used
pare an ultrasonic aspirator and/or scissors for resection of for the shifted ADC map. This was previously reported as
tumors with a low standard ADC and a low shifted ADC. the optimal pair of b values for assessing body stiffness.16
In particular, when the endoscopic endonasal skull base The optimal combination of b values is theoretically organ
approach for resection of anterior skull base meningiomas specific (e.g., body vs brain) but, according to the existing
is planned, the combination of these two ADCs gives im- literature, it should be very similar for all human tissues.16
portant information. In contrast, half of the tumors with a As a limitation other than the combinations of the stan-
low standard ADC and a high shifted ADC were soft, and dard ADC and the shifted ADC mentioned above, the
the other half were hard. Thus, consistency of tumors with ADC was not obtained due to strong distortion on diffu-
this condition might be unpredictable, which is a limita- sion-weighted images in an ROI in the tumor located at the
tion of the present method. tentorium cerebelli. Magnetic susceptibility differences
Consistency of a meningioma generally depends on the among brain tissue, bone, and air-filled sinuses cause geo-
fibrous component content and the cell density.21,22 A hard metrical distortion in MR images, in particular, diffusion-
tumor is rich in fibrous components and has high cell den- weighted images.28 ADC maps may not be generated un-
sity that results in strong intercellular adhesion.22 While der such conditions.
the standard ADC correlates with tumor cellularity and
the amount of fibrous tissue,23–25 it is also affected by intra-
voxel incoherent motion and non-Gaussian diffusion.26,27 Conclusions
In an intravoxel incoherent motion model, biological tissue The present study demonstrated that a combination of
contains two distinct environments: molecular diffusion of standard and shifted ADC maps derived from preopera-
water in the tissue (sometimes referred to as “true diffu- tive DWI predicts the consistency of intracranial menin-
sion”) and microcirculation of blood in the capillary net- giomas.
J Neurosurg Volume 135 • September 2021 975
 
*52 *
Miyoshi et al.

Acknowledgments 16. Le Bihan D, Ichikawa S, Motosugi U. Diffusion and intra-

voxel incoherent motion MR imaging–based virtual elastog-
This work was partly supported by Grants-in-Aid for Strategic raphy:​a hypothesis-generating study in the liver. Radiology.
Medical Science Research from the Ministry of Education, Cul- 2017;​285(2):​609–619.
ture, Sports, Science and Technology of Japan (S1491001) and for 17. Iima M, Kataoka M, Kanao S, et al. Intravoxel incoherent
Scientific Research from the Japan Society for the Promotion of motion and quantitative non-gaussian diffusion MR imaging:​
Science (JP18K09002). evaluation of the diagnostic and prognostic value of several
markers of malignant and benign breast lesions. Radiology.
References 2018;​287(2):​432–441.
18. Uchiyama T, Nagaoka M. Durometer. Article in Japanese.
  1. Jääskeläinen J. Seemingly complete removal of histologically Biomechanisms Japan. 2016;​40:​97–102.
benign intracranial meningioma:​late recurrence rate and 19. Kogo H, Miyabara H, Okawa H, et al. Effect of body flex-
factors predicting recurrence in 657 patients. A multivariate ibility on the lumbar muscle stiffness of community-dwelling
analysis. Surg Neurol. 1986;​26(5):​461–469. elderly. Rigakuryoho Kagaku. 2015;​30(4):​605–608.
  2. Murphy MC, Huston J III, Glaser KJ, et al. Preoperative as- 20. Pepe MS, Longton G. Standardizing diagnostic markers to
sessment of meningioma stiffness using magnetic resonance evaluate and compare their performance. Epidemiology.
elastography. J Neurosurg. 2013;​118(3):​643–648. 2005;​16(5):​598–603.
  3. Sitthinamsuwan B, Khampalikit I, Nunta-aree S, et al. Pre- 21. Maiuri F, Iaconetta G, de Divitiis O, et al. Intracranial menin-
dictors of meningioma consistency:​a study in 243 consecu- giomas:​correlations between MR imaging and histology. Eur
tive cases. Acta Neurochir (Wien). 2012;​154(8):​1383–1389. J Radiol. 1999;​31(1):​69–75.
  4. Yao A, Pain M, Balchandani P, Shrivastava RK. Can MRI 22. Muthupillai R, Rossman PJ, Lomas DJ, et al. Magnetic reso-
predict meningioma consistency?:​a correlation with tumor nance imaging of transverse acoustic strain waves. Magn
pathology and systematic review. Neurosurg Rev. 2018;​41(3):​ Reson Med. 1996;​36(2):​266–274.
745–753. 23. Castillo M, Smith JK, Kwock L, Wilber K. Apparent dif-
  5. Hoover JM, Morris JM, Meyer FB. Use of preoperative mag- fusion coefficients in the evaluation of high-grade cerebral
netic resonance imaging T1 and T2 sequences to determine gliomas. AJNR Am J Neuroradiol. 2001;​22(1):​60–64.
intraoperative meningioma consistency. Surg Neurol Int. 24. Gauvain KM, McKinstry RC, Mukherjee P, et al. Evaluating
2011;​2:​142. pediatric brain tumor cellularity with diffusion-tensor imag-
  6. Ortega-Porcayo LA, Ballesteros-Zebadúa P, Marrufo-Melé- ing. AJR Am J Roentgenol. 2001;​177(2):​449–454.
ndez OR, et al. Prediction of mechanical properties and sub- 25. Guo AC, Cummings TJ, Dash RC, Provenzale JM. Lym-
jective consistency of meningiomas using T1-T2 assessment phomas and high-grade astrocytomas:​comparison of water
versus fractional anisotropy. World Neurosurg. 2015;​84(6):​ diffusibility and histologic characteristics. Radiology. 2002;​
1691–1698. 224(1):​177–183.
  7. Smith KA, Leever JD, Chamoun RB. Predicting consistency 26. Le Bihan D, Breton E, Lallemand D, et al. Separation of dif-
of meningioma by magnetic resonance imaging. J Neurol fusion and perfusion in intravoxel incoherent motion MR
Surg B Skull Base. 2015;​76(3):​225–229. imaging. Radiology. 1988;​168(2):​497–505.
  8. Watanabe K, Kakeda S, Yamamoto J, et al. Prediction of hard 27. Le Bihan D, Breton E, Lallemand D, et al. MR imaging of
meningiomas:​quantitative evaluation based on the magnetic intravoxel incoherent motions:​application to diffusion and
resonance signal intensity. Acta Radiol. 2016;​57(3):​333–340. perfusion in neurologic disorders. Radiology. 1986;​161(2):​
  9. Kashimura H, Inoue T, Ogasawara K, et al. Prediction of me- 401–407.
ningioma consistency using fractional anisotropy value mea- 28. Le Bihan D, Poupon C, Amadon A, Lethimonnier F. Arti-
sured by magnetic resonance imaging. J Neurosurg. 2007;​ facts and pitfalls in diffusion MRI. J Magn Reson Imaging.
107(4):​784–787. 2006;​24(3):​478–488.
10. Romani R, Tang WJ, Mao Y, et al. Diffusion tensor magnetic
resonance imaging for predicting the consistency of intra-
cranial meningiomas. Acta Neurochir (Wien). 2014;​156(10):​ Disclosures
1837–1845.
11. Tropine A, Dellani PD, Glaser M, et al. Differentiation of Dr. Ogasawara reports receiving a consigned research fund from
fibroblastic meningiomas from other benign subtypes using Nihon Medi-Physics Co., Ltd.
diffusion tensor imaging. J Magn Reson Imaging. 2007;​25(4):​
703–708. Author Contributions
12. Yogi A, Koga T, Azama K, et al. Usefulness of the apparent Conception and design: Ogasawara, Wada. Acquisition of
diffusion coefficient (ADC) for predicting the consistency of data: Miyoshi, Uwano, Saura. Analysis and interpretation of
intracranial meningiomas. Clin Imaging. 2014;​38(6):​802–807. data: Ogasawara, Miyoshi, Uwano, Saura. Drafting the article:
13. Phuttharak W, Boonrod A, Thammaroj J, et al. Preopera- Ogasawara, Miyoshi. Critically revising the article: Sasaki,
tive MRI evaluation of meningioma consistency:​a focus Fujiwara. Reviewed submitted version of manuscript: Fujiwara.
on detailed architectures. Clin Neurol Neurosurg. 2018;​169:​ Approved the final version of the manuscript on behalf of all
178–184. authors: Ogasawara. Statistical analysis: Fujiwara, Takahashi,
14. Hughes JD, Fattahi N, Van Gompel J, et al. Higher-resolution Tsushima. Study supervision: Wada, Sasaki.
magnetic resonance elastography in meningiomas to deter-
mine intratumoral consistency. Neurosurgery. 2015;​77(4):​ Correspondence
653–659. Kuniaki Ogasawara: Iwate Medical University, Morioka, Japan.
15. Zada G, Yashar P, Robison A, et al. A proposed grading [email protected].
system for standardizing tumor consistency of intracranial
meningiomas. Neurosurg Focus. 2013;​35(6):​E1.

976 J Neurosurg Volume 135 • September 2021

 
*52 *
 J Neurosurg 135:977, 2021

ERRATUM

Holmes tremor: a delayed complication

after resection of brainstem cavernomas
TO THE READERSHIP: An error appeared in the ar-
ticle by Cenzato et al. (Cenzato M, Colistra D, Iacopino G,
et al. Holmes tremor: a delayed complication after resec-
tion of brainstem cavernomas. J Neurosurg. Published on-
line December 11, 2020. doi:10.3171/2020.7.JNS201352).
The legend to Figure 6 was incomplete. My drawing
was inspired by a photograph of a cadaveric specimen pre-
pared by Dr. Kaan Yağmurlu and published previously. To
acknowledge this, the figure legend has been updated (see
the figure here).
The article has been corrected online as of March 12,
2021. FIG. 6. Anatomical relationships between HT pathways and the subfa-
cial SEZ. Freehand drawing showing how the central tegmental tract
Davide Colistra, MD (red), within which the ROT runs, is at the level of the floor of the fourth
Great Metropolitan Hospital Niguarda, Milan, Italy ventricle just below the subfacial area (yellow), which is the SEZ used in
case 4. Copyright for drawing: Davide Colistra; published with permis-
sion. The drawing is based on a photograph of a cadaveric dissection
CORRESPONDING ARTICLE  See pp 693–703. prepared by Kaan Yağmurlu, MD, which appears in The Rhoton Collec-
INCLUDE WHEN CITING  tion; with permission (CC BY-NC-SA 4.0 [http://creativecommons.org/
Published online March 12, 2021; DOI: 10.3171/2021.2.JNS201352a. licenses/by-nc-sa/4.0]). Figure is available in color online only.
©AANS 2021, except where prohibited by US copyright law

J Neurosurg Volume 135 • September 2021 977

 
*52 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*53 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*53 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*53 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*53 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*53 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*53 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*53 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*53 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*53 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*53 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*53 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*53 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*53 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*53 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*53 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*54 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*54 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*54 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*54 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*54 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*54 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*54 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*54 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*54 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*54 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*54 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*54 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*54 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*54 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*54 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*55 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*55 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*55 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*55 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*55 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*55 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*55 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*55 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*55 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*55 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*55 *
 J Neurosurg 135:978–988, 2021

Neurosurgical Forum
LETTERS TO THE EDITOR

Bleeding risk after cavernous ancy might be explained by the presence of a remnant,
which was not visible due to the detection limitations of
malformation surgery: remnant or early postoperative MRI (EPMRI). Indeed, the EPMRI
recurrence? has a sensitivity for CM remnant detection of 66.67%,
specificity of 76.74%, positive predictive value of 16.67%,
and negative predictive value of 97.06%,8 demonstrating
TO THE EDITOR: We read with great interest the ar-
that EPMRI is not reliable in detecting postoperative CM
ticle by Garcia et al.1 (Garcia RM, Oh T, Cole TS, et al.
Recurrent brainstem cavernous malformations following remnants due to its high risk of false-positive and false-
primary resection: blind spots, fine lines, and the right- negative results.
angle method. J Neurosurg. Published online November Second, the authors declare a postoperative CM rem-
20, 2020. doi:10.3171/2020.6.JNS201555) regarding man- nant rate (PCRR) of 6.6%.1 This value is significantly
agement and surgical treatment of recurrent brainstem lower than the PCRR range reported in the literature
cavernous malformations (BSCMs) following primary (10%–19%).2–4 Surely the senior author’s experience in
microsurgical resection (PMR). In this article, the authors the surgical management of BSCMs helped to reduce the
highlighted the pivotal role of the right-angle method in PCRR; this is also supported by the 3.8% reduction of the
reducing or being aware of blind spots, underlining the PCRR between the first and the second half of the series.
importance of meticulous inspection of the resection cav- However, it is likely that the low sensitivity of the EPMRI
ity, and at the same time the need for advanced surgical also contributed to reducing the PCRR rate.
experience for an effective and safe surgical management Third, we noted that the authors did not analyze
of brainstem lesions. We greatly commend the authors for Zabramski type and perilesional hemosiderin ring (PHR).
this fascinating and comprehensive study; their results Indeed, the presence of a PHR and Zabramski type II CM
will be useful to aid in making balanced treatment deci- are both significantly associated with bleeding risk of the
sions in cases of BSCMs. However, after a detailed analy- postsurgical CM remnant.2–4,9,10 If one or both of these fac-
sis, we would like to discuss some important issues that tors are present, the surgeon should aim for complete re-
the article raises. moval of the BSCM, even if this increases the risk of neu-
First, we would like to focus on the term “recurrent” rological damage, while aiming to avoid the permanence
used by the authors. Etymologically, recurrent comes of a postoperative BSCM remnant with high bleeding risk.
from the Latin re-currere, i.e., to happen again. In this Finally, we are extremely impressed by the authors’
sense, the “recurrent BSCM” indicates a completely re- data and with the results that have been so honestly de-
moved BSCM that reappears after a variable time span, scribed: this is not common in the literature, where in
which is free of disease. In the literature, in reference to many cases remnants do not appear to exist. On the con-
the outcomes of cavernous malformation (CM) surgery, trary, we think that before CM surgery, it is necessary to
the term “remnant” is more established than “recur- clearly explain to the patient the possibility of recurrence
rent.”2–4 We believe that in the event of a BSCM reap- or remnant after surgery. The authors’ data are, therefore,
pearance after surgery, the use of the term “remnant” is of paramount importance.
more consistent with the subjective admission of the high
risk of permanent residual disease from the PMR, as ad- Edoardo Agosti, MD
mitted by the authors themselves. From a careful analy- Francesco Doglietto, MD, PhD
sis of the provided data, there is evidence of the possible Marco M. Fontanella, MD
persistence of remnant BSCM after PMR in the 14 cases University of Brescia, Italy
analyzed, even if not identifiable at 48-hour MRI. Specifi-
cally, the mean volume of the 14 recurrent BSCMs was References
3209 mm3; considering the mean time of 127 days from   1. Garcia RM, Oh T, Cole TS, et al. Recurrent brainstem
surgery until recurrence, an average daily growth rate of cavernous malformations following primary resec-
25 mm was found. This average daily growth rate appears tion:​blind spots, fine lines, and the right-angle method.
excessive when compared to the data in the literature on J Neurosurg. Published online November 20, 2020. doi:​
the dynamic morpho-volumetry of CMs.5–7 This discrep- 10.3171/2020.6.JNS201555

978 J Neurosurg Volume 135 • September 2021

 
*56 *
 Neurosurgical forum

  2. Cenzato M, Stefini R, Ambrosi C, Giovanelli M. Post-opera- complete resection is critical for a microsurgical cure in
tive remnants of brainstem cavernomas:​incidence, risk fac- patients with BSCMs who are deemed surgical candidates.
tors and management. Acta Neurochir (Wien). 2008;​150(9):​ Agosti et al. raise the concern that the frequency that
879–887.
  3. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical
BSCM remnants are reported in the literature after pri-
management, and functional outcome of brainstem cavern- mary resection was significantly higher than the rate pub-
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. lished in our series. Although some studies have reported a
  4. Fontanella MM, Agosti E, Zanin L, et al. Cerebral cavern- remnant rate greater than 10%,2,3 there are arguably more
ous malformation remnants after surgery:​a single-center reports suggesting a remnant rate of less than 10%.4–6 We
series with long-term bleeding risk analysis. Neurosurg Rev. believe the 6.6% remnant rate proposed in our series is
Published online November 19, 2020. doi:​10.1007/s10143- consistent with other previously published series.
020-01436-7 In our earlier analysis of patient outcomes, we did
  5. Clatterbuck RE, Moriarity JL, Elmaci I, et al. Dynamic
nature of cavernous malformations:​a prospective magnetic evaluate hemosiderin staining as a potential predictor of
resonance imaging study with volumetric analysis. J Neuro- outcome, but we did not find multivariate statistical sig-
surg. 2000;​93(6):​981–986. nificance to support this factor as a predictor of outcome
  6. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural because nearly all lesions demonstrated some visible he-
history of familial cavernous malformations:​results of an mosiderin staining intraoperatively. The Zabramski classi-
ongoing study. J Neurosurg. 1994;​80(3):​422–432. fication has aided neurosurgeons in the basic classification
  7. Porter RW, Detwiler PW, Spetzler RF, et al. Cavernous mal- of cerebral cavernomas.7 However, we find that clinical
formations of the brainstem:​experience with 100 patients. J
Neurosurg. 1999;​90(1):​50–58.
and anatomical factors are more helpful in determining
  8. Chen B, Göricke S, Wrede K, et al. Reliable? The value the optimal moment to intervene. More than 47% of pa-
of early postoperative magnetic resonance imaging after tients experienced one hemorrhagic event in our series,
cerebral cavernous malformation surgery. World Neurosurg. and 28% experienced a second event. The median time
2017;​103:​138–144. from the last hemorrhagic event was 28 days. In our ex-
  9. Lanzino G, Spetzler RF, eds. Cavernous Malformations of perience, other factors, such as the size of the cavernoma,
the Brain and Spinal Cord. Thieme Medical Publishers;​ pial presentation, and time since the last hemorrhage, are
2007. more important classifying factors given the eloquence of
10. Agosti E, Flemming KD, Lanzino G. Symptomatic cavern-
ous malformation presenting with seizure without hemor- brainstem anatomy.
rhage:​analysis of factors influencing clinical presentation. Resection of BSCMs should be undertaken with the
World Neurosurg. 2019;​129:​e387–e392. intent to remove the lesion completely and not harm the
patient. The delicate balance of these two objectives is
Disclosures particularly difficult in the brainstem, much more so than
The authors report no conflict of interest. in the cerebrum. As our knowledge of brainstem anatomy,
safe entry zones, and exposures afforded by skull base
Correspondence surgical approaches has increased, it has become routine
Edoardo Agosti: [email protected]. to curatively resect BSCMs and achieve favorable long-
term outcomes. We hope our experience and adoption of
INCLUDE WHEN CITING  insights such as the right-angle method will help others
Published online March 12, 2021; DOI: 10.3171/2020.12.JNS204192.
obtain excellent microsurgical results.

Response Roxanna M. Garcia, MD, MS, MPH

We were pleased to read the review and insightful com- Northwestern University, Chicago, IL
mentary provided by Agosti et al. Previously, we proposed Taemin Oh, MD
a grading system to predict outcomes of patients after re- University of California, San Francisco, CA
section of BSCMs.1 As a continuation of this work, we
highlight the surgical challenges for the small subset of Tyler S. Cole, MD
patients needing reoperation for remnant lesions. We rec- Benjamin K. Hendricks, MD
ognize the inconsistent and sometimes confusing seman- Michael T. Lawton, MD
tics of “recurrent” and “residual” used in the literature. Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center,
We agree with Agosti et al. that, from a pathophysiological Phoenix, AZ
standpoint, CMs that were “recurrent” within this series
were, in fact, “remnants” or “residuals” following PMR. Acknowledgments
BSCMs do not recur after complete resection, but a small Dr. Roxanna Garcia served as the StrokeNet research fellow
remnant can be difficult to detect on postoperative MRI. from 2018 to 2019 and was a Fogarty Global Health Trainee
Only 2 of our patients had lesions that were immediately from 2019 to 2020. Research reported in this publication was
apparent after primary surgical intervention, and 3 ad- supported under the StrokeNet award no. U24 NS107233-01 and
ditional patients demonstrated a remnant on postopera- the Fogarty International Center and National Institute of Mental
tive surveillance MRI. Therefore, the detection of small Health, of the NIH, under award no. D43 TW010543. The content
is solely the responsibility of the authors and does not necessarily
remnants is challenging, but it is critical to postoperative represent the official views of the NIH.
surveillance and is made easier by the progressive growth The authors thank the staff of Neuroscience Publications at
of, or hemorrhage by, the remnant (which we have termed Barrow Neurological Institute for assistance with manuscript
“recurrence”). Early aggressive surgery with the goal of preparation.

J Neurosurg  Volume 135 • September 2021 979

*56 *
Neurosurgical forum

References References
  1. Garcia RM, Ivan ME, Lawton MT. Brainstem cavern-   1. Lilja-Cyron A, Andresen M, Kelsen J, et al. Intracranial
ous malformations:​surgical results in 104 patients and a pressure before and after cranioplasty:​insights into intracra-
proposed grading system to predict neurological outcomes. ​ ​
nial physiology. J Neurosurg. 2020;133(5):1548–1558.
Neurosurgery. 2015;​76(3):​265–268.   2. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on
  2. Ziyal IM, Ozgen T. Natural history of brainstem cavernous cerebrospinal fluid hydrodynamics in patients with the syndrome
malformations. Neurosurgery. 2001;​49(4):​1023–1024. ​ ​
of the trephined. Acta Neurochir (Wien). 1984;70(1-2):21–30.
  3. Bruneau M, Bijlenga P, Reverdin A, et al. Early surgery for   3. Luce JM, Huseby JS, Kirk W, Butler J. A Starling resistor
brainstem cavernomas. Acta Neurochir (Wien). 2006;​148(4):​ regulates cerebral venous outflow in dogs. J Appl Physiol
405–414. ​ ​
Respir Environ Exerc Physiol. 1982;53(6):1496–1503.
  4. Li D, Yang Y, Hao S-Y, et al. Hemorrhage risk, surgical   4. Piechnik SK, Czosnyka M, Richards HK, et al. Cerebral ve-
management, and functional outcome of brainstem cavern- nous blood outflow:​a theoretical model based on laboratory
ous malformations. J Neurosurg. 2013;​119(4):​996–1008. ​ ​
simulation. Neurosurgery. 2001;49(5):1214–1223.
  5. Wang C-C, Liu A, Zhang J-T, et al. Surgical management of   5. Barami K, Sood S. The cerebral venous system and the pos-
brain-stem cavernous malformations:​report of 137 cases. tural regulation of intracranial pressure:​implications in the
Surg Neurol. 2003;​59(6):​444–454. management of patients with cerebrospinal fluid diversion.
  6. Ferroli P, Sinisi M, Franzini A, et al. Brainstem cavernomas:​ ​ ​
Childs Nerv Syst. 2016;32(4):599–607.
long-term results of microsurgical resection in 52 patients.   6. Barami K. Cerebral venous overdrainage:​an under-recog-
Neurosurgery. 2005;​56(6):​1203–1214. nized complication of cerebrospinal fluid diversion. Neuro-
  7. Zabramski JM, Wascher TM, Spetzler RF, et al. The natural ​ ​
surg Focus. 2016;41(3):E9.
history of familial cavernous malformations:​results of an   7. Barami K, Sood S, Ham SD, Canady AI. Postural changes in
ongoing study. J Neurosurg. 1994;​80(3):​422–432. intracranial pressure in chronically shunted patients. Pediatr
​ ​
Neurosurg. 2000;33(2):64–69.
INCLUDE WHEN CITING    8. Doepp F, Schreiber SJ, von Münster T, et al. How does the
Published online March 12, 2021; DOI: 10.3171/2021.1.JNS204224. blood leave the brain? A systematic ultrasound analysis of
cerebral venous drainage patterns. Neuroradiology. 2004;​
©AANS 2021, except where prohibited by US copyright law
​
46(7):565–570.

Disclosures
Intracranial physiology and ICP The author reports no conflict of interest.

Correspondence
TO THE EDITOR: I read with great interest the ar- Kaveh Barami: [email protected].
ticle by Lilja-Cyron et al.1 (Lilja-Cyron A, Andresen M,
Kelsen J, et al. Intracranial pressure before and after cra- INCLUDE WHEN CITING 
nioplasty: insights into intracranial physiology. J Neuro- Published online April 16, 2021; DOI: 10.3171/2020.12.JNS204282.
surg. 2020;133[5]:1548–1558).
With regard to changes in intracranial pressure (ICP) Response
before and after cranioplasty, cerebral venous physiology
may explain the postural changes between supine and sit- We thank Dr. Barami for the interest in our recent paper
ting positions as well as the discrepancy between findings and for sharing his thoughts on our observations regard-
in this study and that of Fodstad et al., as the authors men- ing ICP before and after cranioplasty (following emer-
tion.2 In this scenario, an evolved protective mechanism gency decompressive craniectomy). We certainly agree
that prevents siphoning of venous blood from the cranium that cerebral venous outflow (through various routes) as
known as the “Starling resistor” may become nonfunc- well as jugular vein collapse to prevent siphoning in the
tional during craniectomy, resulting in posture-related upright position are key aspects in understanding normal
overdrainage of venous blood from the venous outflow ICP physiology. When the normal integrity of the cranium
tracts.3 This seems to be related to CSF and sagittal si- is broken (e.g., following decompressive craniectomy) the
nus pressure dissociation as proposed by the mathematical “rules are changed” and homeostatic regulatory/physi-
model in the study of Piechnik et al.,4 causing intracranial ological mechanisms become defective. In the study by
hypotension depending on varying circumstances (such Fodstad et al.,1 the smaller cranial defect in this patient
as head elevation, position, individual venous anatomy, cohort likely explains the “less abnormal” ICP before cra-
and compensatory mechanisms).5–7 It is plausible that af- nioplasty compared to our observations, although differ-
ter cranioplasty the Starling resistor becomes functional ences in individual venous anatomy may play a role when
again. With regard to individual venous anatomy, Doepp comparing these rather small patient cohorts. As previous-
et al. in a systematic ultrasound and MRI study of cranial ly shown by Dr. Barami, application of the Starling resis-
venous outflow tracts have shown that there is predomi- tor concept in the context of CSF diversion can elegantly
nantly jugular drainage in 72% of healthy volunteers.8 In explain the possible effects on cerebral venous outflow.2
22%, the jugular drainage equals the nonjugular drainage, In the setting of decompressive craniectomy, in which not
and in 6% the drainage pattern is nonjugular.8 These find- only CSF pressure is changed, the normal intracranial
ings suggest that in the general population there are varia- physiology is challenged even more.
tions in the anatomy of cranial venous outflow. Following decompressive craniectomy, our data indi-
cate that the normal postural ICP drop of 5–10 mm Hg
Kaveh Barami, MD, PhD is absent, resulting from both a lower ICP in the supine
Kaiser Permanente Northern California, Sacramento, CA position and a less negative ICP in the upright position.3

980 J Neurosurg  Volume 135 • September 2021

*56 *
 Neurosurgical forum

Furthermore, the intracranial pulse wave amplitude is se- Myhr J, Moen KG, et al. Association of cause of injury
verely diminished. When cranioplasty is performed, nor- and traumatic axonal injury: a clinical MRI study of
mal intracranial physiology appears to be re-established moderate and severe traumatic brain injury. J Neurosurg.
regarding both postural ICP changes and intracranial 2020;133[5]:1559–1567). The authors reported their inves-
pulsatility. Applying the Starling resistor concept in this tigation of causal mechanistic effects on the incidence of
setting, sustained posture-related venous overdrainage fol- traumatic axonal injury (TAI). Moe et al. accurately ana-
lowing decompressive craniectomy (related to individual lyzed the role of each known cause of TAI and the un-
venous anatomy) might explain some cases of the poorly derlying pathophysiology; however, some concerns need
defined condition termed “syndrome of the trephined” or to be highlighted. It is well known that TAI is one of the
“sunken skin flap syndrome,”4,5 although other factors such main causes of impairment during the posttraumatic phase
as changes in cerebral blood/CSF flow as well as impaired in patients who have no detectable intracranial lesions on
cerebral metabolism probably also play important roles.6 CT,2 even patients with mild traumatic brain injury. Sever-
Intracranial pulsatility is believed to be an important driv- al phenomena explain how the strong acceleration-decel-
er of intracranial fluid movements (including glymphatic eration and rotational-angular acceleration forces in high-
flow),7,8 and normalization of this factor following cranio- impact trauma lead to progressive changes in the axons,
plasty probably is equally important in this regard. resulting in deformation of the brain tissue.2
An external injury involves shearing forces that stimu-
Alexander Lilja-Cyron, MD, PhD late the formation of axon retraction balls, which result
Marianne Juhler, MD, DMSc from a swelling phenomenon at the end of the axonal axis
Rigshospitalet, Copenhagen, Denmark due to external shear force and tension that lead to the final
breakage of the axon. During this process the permeabil-
ity of the axon membrane changes and large amounts of
References Ca2+ enter the cells, reversing the flux of plasma transport
  1. Fodstad H, Love JA, Ekstedt J, et al. Effect of cranioplasty on and activating the cysteine protein signal pathway and cas-
cerebrospinal fluid hydrodynamics in patients with the syn-
drome of the trephined. Acta Neurochir (Wien). 1984;​70(1-2):​
pase-3. These events lead to the degradation of the axonal
21–30. cytoskeleton network. During this progressive phenom-
  2. Barami K, Sood S. The cerebral venous system and the pos- enon, axons usually maintain their morphology several
tural regulation of intracranial pressure:​implications in the hours after injury, and for that reason it is crucial to per-
management of patients with cerebrospinal fluid diversion. form MRI in patients days after trauma.3
Child’s Nerv Syst. 2016;​32(4):​599–607. Although we understand the complex nature of re-
  3. Lilja-Cyron A, Andresen M, Kelsen J, et al. Long-term effect porting hundreds of neuroimaging results and thank the
of decompressive craniectomy on intracranial pressure and authors for their investigation of this important topic,
possible implications for intracranial fluid movements. Neu-
rosurgery. 2020;​86(2):​231–240. we have to note the highly heterogeneous rating of the
  4. Sedney CL, Dillen W, Julien T. Clinical spectrum and radio- samples reported by Moe et al.1 and the possible conse-
graphic features of the syndrome of the trephined. J Neurosci quences. The fact that there is substantial agreement be-
Rural Pract. 2016;​6(3):​438–441. tween raters (linear Cohen’s kappa 0.74) should be inter-
  5. Honeybul S, Ho KM, Lind CRP, Gillett GR. Validation of the preted with caution given that some of the pitfalls in this
CRASH model in the prediction of 18-month mortality and study may have decreased the interrater reliability to an
unfavorable outcome in severe traumatic brain injury requir- unknown degree. Having more than two raters may have
ing decompressive craniectomy. J Neurosurg. 2014;​120(5):​
1131–1137.
masked valuable ratings in the analysis, and thus the gen-
  6. Ashayeri K, Jackson EM, Huang J, et al. Syndrome of the eralized kappa may not have captured the range of poten-
trephined:​a systematic review. Neurosurgery. 2016;​79(4):​ tial agreement or disagreement between assessments by
525–533. multiple raters.4
  7. Wagshul ME, Eide PK, Madsen JR. The pulsating brain:​a
review of experimental and clinical studies of intracranial Mónica Patricia Herrera-Martinez, MD1,3
pulsatility. Fluids Barriers CNS. 2011;​8(1):​5.
  8. Plog BA, Nedergaard M. The glymphatic system in central
Ezequiel García-Ballestas, MD1,3
nervous system health and disease:​past, present, and future. Ivan David Lozada-Martínez, MS2–4
Annu Rev Pathol Mech Dis. 2018;​13(1):​379–394. Luis Rafael Moscote-Salazar, MD1–4
Mohammed Al-Dhahir, MD5
INCLUDE WHEN CITING  1
Center of Biomedical Research (CIB), Faculty of Medicine,
Published online April 16, 2021; DOI: 10.3171/2021.2.JNS204324.
University of Cartagena, Colombia
©AANS 2021, except where prohibited by US copyright law 2
Medical-Surgical Research Center, University of Cartagena, Colombia
3
Latinamerican Council of Neurocritical Care (CLaNi), Cartagena,
Colombia

Traumatic axonal injury: causes and

4
Colombian Clinical Research Group in Neurocritical Care,
University of Cartagena, Colombia
effects 5
Strong Memorial Hospital, University of Rochester, MN

References
TO THE EDITOR: We read with great interest the
  1. Moe HK, Limandvik Myhr J, Moen KG, et al. Association of
article published by Moe et al.1 (Moe HK, Limandvik cause of injury and traumatic axonal injury:​a clinical MRI

J Neurosurg  Volume 135 • September 2021 981

*56 *
Neurosurgical forum

study of moderate and severe traumatic brain injury. J Neuro- classification of TAI grade. More details can be found in
surg. 2020;​133(5):​1559–1567. the Methods and Results sections of that paper.4 Hence,
  2. Kim M, Ahn JS, Park W, et al. Diffuse axonal injury (DAI) we hope our response has clarified some of the concerns
in moderate to severe head injured patients:​Pure DAI vs.
non-pure DAI. Clin Neurol Neurosurg. 2018;​171:​116–123.
raised by Herrera-Martinez et al.
  3. Ma J, Zhang K, Wang Z, Chen G. Progress of research on
diffuse axonal injury after traumatic brain injury. Neural Hans Kristian Moe, MD, PhD1
Plast. 2016;​2016:​9746313. Anne Vik, MD, PhD1,2
  4. O’Leary S, Lund M, Ytre-Hauge TJ, et al. Pitfalls in the use Turid Follestad, MSc, PhD1
of kappa when interpreting agreement between multiple rat-
ers in reliability studies. Physiotherapy. 2014;​100(1):​27–35. Toril Skandsen, MD, PhD1,2
Asta Kristine Håberg, MD, PhD1,2
Disclosures Kent Gøran Moen, MD, PhD1,3
The authors report no conflict of interest. 1
Norwegian University of Science and Technology (NTNU),
Trondheim, Norway
Correspondence 2
St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Ivan David Lozada-Martínez: [email protected]. 3
Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

INCLUDE WHEN CITING 

Published online May 7, 2021; DOI: 10.3171/2021.1.JNS204452.
References
  1. Johnson VE, Stewart W, Smith DH. Axonal pathology in
traumatic brain injury. Exp Neurol. 2013;​246:​35–43.
Response   2. Toth A, Kovacs N, Tamas V, et al. Microbleeds may expand
acutely after traumatic brain injury. Neurosci Lett. 2016;​617:​
We thank authors Herrera-Martinez et al. for their in- 207–212.
terest in our paper on the association of cause of injury   3. Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI
with traumatic axonal injury (TAI). The authors elaborate study of traumatic axonal injury in patients with moderate
more thoroughly on the underlying cellular mechanisms and severe traumatic brain injury. J Neurol Neurosurg Psy-
behind TAI than what was within the scope of our study. chiatry. 2012;​83(12):​1193–1200.
While TAI detectability on histopathological examina-   4. Moe HK, Follestad T, Andelic N, et al. Traumatic axonal
injury on clinical MRI:​association with the Glasgow Coma
tion has been reported to peak at 24 hours,1 little is known Scale score at scene of injury or at admission and prolonged
about how the indirect signs of TAI visualized on clini- posttraumatic amnesia. J Neurosurg. Published online Octo-
cal MRI develop over the first few hours and days after ber 23, 2020. doi:​10.3171/2020.6.JNS20112
injury. One recent study indicated that traumatic micro-   5. de Vet HCW, Mokkink LB, Terwee CB, et al. Clinicians are
bleeds detected on susceptibility-weighted imaging (SWI) right not to like Cohen’s κ. BMJ. 2013;​346:​f2125.
were more visible 1 week after injury.2 We also know from   6. de Vet HCW, Dikmans RE, Eekhout I. Specific agreement on
previous studies that almost all non-hemorrhagic TAI le- dichotomous outcomes can be calculated for more than two
sions (detected on FLAIR or diffusion-weighted imaging raters. J Clin Epidemiol. 2017;​83:​85–89.
[DWI]) have disappeared at 3 months after injury.3 We INCLUDE WHEN CITING 
want to emphasize that in the present study we performed Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21156.
MRI at a median of 8 days (IQR 4–17 days) post-injury, so
it is unlikely that the MRIs were performed too early and ©AANS 2021, except where prohibited by US copyright law
thus could have affected the results in our study.
Herrera-Martinez et al. also argue that the “substantial
agreement between raters . . . should be interpreted with
caution,” and that “having more than two raters may have On the right side of history:
masked valuable ratings in the analysis.” We acknowl- expanding diversity within
edge the possibility that a misunderstanding might have
occurred. The linearly weighted Cohen’s kappa (not the neurosurgery
generalized kappa) was calculated between two raters in a
previous paper,3 namely K.G.M. (who was able to consult TO THE EDITOR: As the US grows increasingly di-
with neuroradiologists K.A.K. and J.R. when needed) and verse, it is crucial to develop a healthcare system in which
M.F. Both K.G.M. and M.F. were blinded to the scores of shared ethnic background, language, gender, and sexual
the other rater. Interrater analyses were not performed for orientation foster connections between physicians and
J.R. (n = 66) since all of his scores were also evaluated by their patients. The article by Corley and colleagues1 is in-
K.A.K. or M.F. The two latter neuroradiologists had been strumental in improving medical care through increased
part of the abovementioned interrater analyses.3 Moreover, diversity, and we believe now is the time to wholeheartedly
in a recent larger study including the Trondheim cohort of embrace and embody the changing facets of American so-
patients with moderate and severe traumatic brain injury, ciety (Corley J, Kim E, Philips CA, et al. One hundred
interrater analyses were performed for multiple raters.4 In years of neurosurgery: contributions of American women.
this more recent study, the positive and negative agree- J Neurosurg. 2021;134[2]:337–342).
ment5,6 for the presence or absence of TAI were 0.90 (95% Homogeneity restricts the practice of medicine. Igno-
CI 0.77–0.95) and 0.69 (95% CI 0.42–0.84), respectively, rance toward a patient’s social and cultural background can
and the intraclass correlation coefficient was 0.78 for the cause oversight of relevant parameters in patient pathology.

982 J Neurosurg  Volume 135 • September 2021

*56 *
 Neurosurgical forum

On a societal level, the “homogeneity of members’ social Bethany J. Thach, BA

background and ideology [can result in] groupthink.”2 This Donald K. E. Detchou, BA
psychological phenomenon leads to ideas that lack nuance; Perelman School of Medicine at the University of Pennsylvania,
therefore, a homogenous group of physicians runs an in- Philadelphia, PA
creased risk of coming to erroneous conclusions. In medi-
cine, this can mean inaccurate diagnoses that put patients’ References
lives at risk. Additionally, homogenous groups of medical   1. Corley J, Kim E, Philips CA, et al. One hundred years of neu-
students, physicians, and nurse practitioners hinder innova- rosurgery:​contributions of American women. J Neurosurg.
tion and progress in the medical field.3 Advocacy for more 2021;​134(2):​337–342.
diverse medical institutions allows healthcare leaders to   2. Janis IL. Groupthink. IEEE Eng Manage Rev. 2008;​5(1):​36.
avoid herd behavior. This is only attainable by welcoming   3. Howard SD, Lucas TH. The Supreme Court’s recent decision
perspectives from many walks of life. is a call for increased diversity in neurosurgery. Letter. Neu-
Academic institutions have worked to foster medical rosurgery. 2020;​87(6):​E715–E716.
student interest in neurosurgery, with efforts including cre-   4. Lubelski D, Xiao R, Mukherjee D, et al. Improving medi-
cal student recruitment to neurosurgery. J Neurosurg. 2020;​
ating neurosurgery interest groups, encouraging clinical 133(3):​848–854.
exposure and research opportunities, and providing con-   5. Detchou DK, Glauser G, Dimentberg R, et al. The Frazier
tinued mentorship.4–6 Still, analysis of underrepresented Scholar Program at Penn Neurosurgery:​an adaptable model
minority student performance indicates that both objective for nurturing early interest in neurosurgery for current and
and subjective measures of student critique are commonly aspiring medical students. Letter. J Neurosurg. 2021;​134(3):​
difficult areas for minority students to excel in due to im- 1015–1017.
plicit bias, lack of financial support and/or mentorship, and   6. Dixon A, Silva NA, Sotayo A, Mazzola CA. Female medical
sexual harassment, among others.7,8 Given the emphasis on student retention in neurosurgery:​a multifaceted approach.
World Neurosurg. 2019;​122:​245–251.
measures of medical student performance in the neurosur-   7. Benzil DL, Muraszko KM, Soni P, et al. Toward an un-
gery match, such as the United States Medical Licensing derstanding of sexual harassment in neurosurgery. J
Examination step scores, clerkship grades, and number of Neurosurg. Published online November 10, 2020. doi:​
research products, gender and ethnic minorities may ad- 10.3171/2020.6.JNS201649
ditionally benefit from mentorship driven by residents and   8. Abosch A, Rutka JT. Women in neurosurgery:​inequality
attending physicians. redux. J Neurosurg. 2018;​129(2):​277–281.
We propose the implementation of opportunities that pro-   9. Patel SI, Rodríguez P, Gonzales RJ. The implementation of
vide youth from underrepresented groups with the chance an innovative high school mentoring program designed to
to pursue their interest in medicine. Establishment of scien- enhance diversity and provide a pathway for future careers in
healthcare related fields. J Racial Ethn Health Disparities.
tific interest at early educational stages may increase future 2015;​2(3):​395–402.
pools of diverse applicants interested in neurosurgery, and 10. Robinson MA, Douglas-Vail MB, Bryce JN, van Zyl TJ.
the medical field as a whole.9,10 At the medical student level, Medical school outreach and mentorship for rural secondary
we recommend that the AANS be utilized to amplify the school students:​a pilot of the Southwestern Ontario Medical
call for diversity in one of the most scientifically engaging Mentorship Program. Can J Rural Med. 2017;​22(2):​62–67.
and versatile specialties in medicine. Within the field, there
is much to accomplish in the areas of increasing minorities Disclosures
in academic positions, promotion of women in leadership The authors report no conflict of interest.
positions, and defeating barriers affecting minority medical
students. Specific approaches to action may include creat- Correspondence
ing a diversity section in the AANS or requesting diversity Donald K. E. Detchou: [email protected].
initiatives through AANS medical student chapters.
The complexities of American diversity call for an in- INCLUDE WHEN CITING 
Published online April 23, 2021; DOI: 10.3171/2021.1.JNS21176.
trospective look at the healthcare system. Physicians and
medical providers should reflect the diverse nature of their
patients for myriad reasons. Notably, the quality of patient Response
care is heavily reliant on both proper communication and Tissot et al.’s letter calls forth the need to substantially
understanding of patients’ lived experiences to provide advance inclusivity and accentuate clear intentions to re-
comfort and reassurance. examine the fundamental aims that the field of neurosur-
gery should prioritize, namely, promoting diversity. While
Marianne I. J. Tissot, BA we agree with the overall goals outlined in the letter, we
Perelman School of Medicine at the University of Pennsylvania, hope to extend beyond objectives and implement action-
Philadelphia, PA able and systemic approaches. 
Andre E. Boyke, MS Over the past decade, the average Black, Hispanic, and
Albert Einstein College of Medicine, New York, NY Native American neurosurgical resident populations were
4.4%, 4.1%, and 0.2%, respectively.1 Other concerns such
Alvin Onyewuenyi, MPH as improving the gender distribution among female faculty
Chicago Medical School, Rosalind Franklin University of Medicine and
and residents,2 as well as establishing accommodations for
Science, Chicago, IL
individuals with disabilities,3 are necessary to take into
Gregory Glauser, MBA account to promote an equitable neurosurgical environ-
Evalyn S. Mackenzie, BA ment. Initially released in 2018 by White Coats for Black

J Neurosurg  Volume 135 • September 2021 983

*56 *
Neurosurgical forum

Lives, the Racial Justice Report Card (RJRC) exists as a References

15-point metric to critically assess the institutional climate   1. Yaeger KA, Munich SA, Byrne RW, et al. Trends in United
for those Black, Indigenous, and People of Color (BIPOC). States neurosurgery residency education and training over the
This effort focuses on representation and recruitment, in- last decade (2009–2019). Neurosurg Focus. 2020;48(3):E6.
stitutional safety, and resource allocation. The RJRC can   2. Donaldson K, Callahan KE, Gelinne A, et al. Gender
serve as an effective tool toward solidifying institutional diversity in United States neurosurgery training pro-
grams. J Neurosurg. Published online January 29, 2021.
changes in neurosurgery. doi:10.3171/2020.7.JNS192647
Attracting young and talented underrepresented mi-   3. Kim EE, Klein AL, Lartigue JW, et al. Diversity in neurosur-
norities (URMs) will ensure that neurosurgery continues gery. World Neurosurg. 2021;145:197–204.
to progress in equitable clinical advancements. Research   4. Muncan B, Majumder N, Tudose N. From high school to
findings have demonstrated that adolescents with early hospital: how early exposure to healthcare affects adolescent
clinical exposure to healthcare have an increased interest career ideas. Int J Med Educ. 2016;7:370–371. 
in pursuing a medical career.4 Partnering with communi-   5. Avent CM, Boyce AS, Servance LD, et al. Implementation of
a biomedical engineering research experience for African-
ties to offer operating room tours or research mentorship American high school students at a tier one research univer-
would improve students’ preparedness to excel in academ- sity. J Biomech Eng. 2018;140(8):0847011–0847018.
ics5 and cultivate a diverse applicant pool of future neuro-   6. Bae GH, Lee AW, Park DJ, et al. Ethnic and gender diversity
surgeons. in hand surgery trainees. J Hand Surg. 2015;40(4):790–797. 
While pipeline programs help mitigate the workforce   7. Takeshita J, Wang S, Loren AW, et al. Association of racial/
gap of URMs in medicine, enhance research productiv- ethnic and gender concordance between patients and physi-
ity,6 and increase patient satisfaction,7 there must be an in- cians with patient experience ratings. JAMA Network Open.
crease in the funding available for neurosurgeons to con- 2020;3(11):e2024583. 
  8. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and
duct racial- and gender-focused research to ensure that all surgical outcomes: it is not all black and white. Ann Surg.
patients have access to equitable care. National societies 2008;248(4):647–655. 
should encourage the development of studies that evalu-   9. Parmeshwar N, Stuart ER, Reid CM, et al. Diversity in plastic
ate the impact of race and ethnicity on neurosurgical out- surgery: trends in minority representation among applicants
comes,8 residency application trends,9 career trajectories and residents. Plast Reconstr Surg. 2019;143(3):940–949. 
for underrepresented persons, or recommendations to im- 10. Lee EQ, Chukwueke UN, Hervey-Jumper SL, et al. Barriers
prove race and gender inclusion in clinical trials.10 to accrual and enrollment in brain tumor trials. Neuro Oncol.
2019;21(9):1100–1117. 
Of course, renewed policies and research goals will
have limited success without a culture change to one that INCLUDE WHEN CITING 
fosters allyship. Perhaps the greatest modern example of Published online April 23, 2021; DOI: 10.3171/2021.2.JNS21185.
this phenomenon is the #HeForShe movement, which has
heralded a dynamism of men showing support for women ©AANS 2021, except where prohibited by US copyright law
with a new organized purpose. Additionally, the emer-
gence of the term “upstander” has also changed the cul-
ture of medicine and emboldened people to stand up for
discriminated and harassed minority persons.  A new approach for local tumor
These practices need to permeate the neurosurgery control
workplace through education, advocacy, and establishing
entryways for other progressive like-minded people. Ad- TO THE EDITOR: We read with great interest the pa-
ditionally, allyship not only involves relationships between per by Schipmann et al.1 (Schipmann S, Müther M, Stög-
the group majority and the group minority; partnerships bauer L, et al. Combination of ALA-induced fluorescence-
should be forged with groups of all protected classes of guided resection and intraoperative open photodynamic
people: from racial and gender minorities; to those of the therapy for recurrent glioblastoma: case series on a prom-
lesbian, gay, bisexual, transgender, and queer or questioning ising dual strategy for local tumor control. J Neurosurg.
(LGBTQ) community; and those who identify as disabled 2021;134[2]:426–436). The authors present a combined
or who struggle with mental or physical illness. Ultimately, 5-aminolevulinic acid (5-ALA) and photodynamic ther-
we all stand to gain from a diverse workplace, policies and apy (PDT) approach in the treatment of recurrent high-
infrastructure that are inclusive, research that is forward grade gliomas (HGGs). Their experience has proved to be
thinking, and a culture that is protective for everyone. an innovative and safe method for local tumor control, an
effort for which they should be commended. In particular,
Antoinette J. Charles, BS they brought the discussion to bear on a crucial issue of
Duke University School of Medicine, Durham, NC such 5-ALA use in the treatment of recurrent HGG, which
Julia B. Duvall, BS is currently done on a case-by-case basis and without a
Harvard Medical School, Boston, MA proper consensus.2
In the literature and in general practice, 5-ALA’s in-
Alexis O. Umoye, BS disputable role has been well underlined for newly diag-
University of California, Davis School of Medicine, Sacramento, CA
nosed HGG cases, whereas its significance and its pos-
Jacquelyn Corley, MD sible pitfalls in recurrent HGGs are de facto less clear.
Duke University School of Medicine, Durham, NC As is widely acknowledged, an accurate differentiation
The Gender Equity Initiative in Global Surgery, Boston, MA of glioma recurrence from treatment-induced changes

984 J Neurosurg  Volume 135 • September 2021

*56 *
 Neurosurgical forum

is of paramount importance because it can change the   4. Zikou A, Sioka C, Alexiou GA, et al. Radiation necrosis,
patient’s management. Against this background, we want pseudoprogression, pseudoresponse, and tumor recurrence:​
to emphasize 5-ALA heterogeneity in the “recurrent set- imaging challenges for the evaluation of treated gliomas.
Contrast Media Mol Imaging. 2018;​2018:​6828396.
ting” and the need to properly address this feature in or-   5. Ellingson BM, Chung C, Pope WB, et al. Pseudoprogres-
der to better integrate it with other techniques such as sion, radionecrosis, inflammation or true tumor progression?
PDT. 3 Challenges associated with glioblastoma response assessment
As a matter of fact, the presence of inflammatory tissue, in an evolving therapeutic landscape. J Neurooncol. 2017;​
as in the case of peritumoral reactive inflammation, pseu- 134(3):​495–504.
doprogression, or radiation-induced necrosis, may influ-   6. Miyatake SI, Kuroiwa T, Kajimoto Y, et al. Fluorescence
ence the intraoperative fluorescence detection. Therefore, of non-neoplastic, magnetic resonance imaging-enhancing
surgeons must be aware that not everything that glitters tissue by 5-aminolevulinic acid:​case report. Neurosurgery.
2007;​61(5):​E1101–E1104.
is gold—a critical awareness of 5-ALA potentialities and   7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo-
drawbacks in recurrent gliomas is necessary. rescence behavior of reactive tissue changes following glio-
With specific focus on pseudoprogression and radiation blastoma treatment with radiation and chemotherapy. Acta
necrosis,4 we believe it is important to point out 5-ALA Neurochir (Wien). 2015;​157(2):​207–214.
heterogeneous behavior in such cases in order to properly
design future studies in which investigators are able to Disclosures
properly select patients who, during a second surgery with The authors report no conflict of interest.
5-ALA, become eligible for PDT, and to avoid resections
improperly exceeding planned limits. Correspondence
Pseudoprogression, which constitutes a strong reac- Grazia Menna: [email protected].
tion to effective therapy and is associated with damage
INCLUDE WHEN CITING 
to the endothelium, is linked to a high responsiveness to Published online May 7, 2021; DOI: 10.3171/2021.2.JNS21409.
5-ALA. Different studies have suggested the presence of a
peritumoral inflammatory state, and an increased reactive
mitotic activity could explain these false-positive results. Response
This is relevant in order to avoid patients’ exposure to un- With great interest have we read the letter to the editor
necessary treatment5 and to tailor resection, especially in from Menna et al. regarding our publication on open PDT,
eloquent areas. Furthermore, there are currently no data on and we thank the authors for their interest in our research.
PDT selectivity in this context. We agree that accurate detection of tumor recurrence is
A different consideration has to be made for radiation of paramount importance for patient management, and we
necrosis. The paper points out that PDT has high selectiv- are aware that it is not always possible to reliably differ-
ity for tumor cells, having shown no effect in a patient op- entiate tumor progression from treatment-related changes
erated on for suspected recurrence and with 5-ALA posi- with conventional imaging techniques.1
tivity, but in whom histological analysis revealed the lesion However, with our use of advanced imaging modali-
to be radionecrosis. However, we want to highlight how ties and the Response Assessment in Neuro-Oncology
5-ALA behavior in radiation necrosis has been linked to (RANO) criteria2 before allocating a glioma patient with
conflicting evidence—further studies are needed to better suspicion of tumor recurrence or progression to further
clarify this issue in order to provide an improved selection surgical treatment (including open PDT), we reach a high
of candidates for the dual approach.6,7 certainty for the detection of tumor tissue and for distin-
In thanking the author for providing such interesting guishing recurrent tumor from pseudoprogression or ra-
food for thought, we wish that future studies on the topic diation necrosis.
will take our suggestion. These advanced imaging modalities include MR spec-
Grazia Menna, MD troscopy, reaching a sensitivity and specificity of 91% and
Alessandro Olivi, MD 95%, respectively;3 MR perfusion;1 and PET studies us-
Giuseppe Maria Della Pepa, MD ing 11C-MET or 18F-FET as tracers.4–6 With the use of 18F-
Institute of Neurosurgery, Fondazione Policlinico Universitario Agostino FET PET, a sensitivity of up to 100% and specificity of
Gemelli IRCCS, Catholic University, Rome, Italy 91% can be expected.6 When in doubt, the combination
of several imaging modalities can help to increase sensi-
References tivity and help in making clinical decisions. These data
  1. Schipmann S, Müther M, Stögbauer L, et al. Combination of together with our experience in daily routine use of these
ALA-induced fluorescence-guided resection and intraopera- advanced imaging modalities have led to the fact that dis-
tive open photodynamic therapy for recurrent glioblastoma:​ tinguishing between tumor progression or recurrence and
case series on a promising dual strategy for local tumor posttherapeutic changes is no longer a limiting issue. Be-
control. J Neurosurg. 2021;​134(2):​426–436. cause progression and ultimately death are inevitable in
  2. Birzu C, French P, Caccese M, et al. Recurrent glioblastoma:​ malignant gliomas, with no generally accepted treatment
from molecular landscape to new treatment perspectives. options after progression, early detection of progression
Cancers (Basel). 2020;​13(1):​47.
  3. La Rocca G, Sabatino G, Menna G, et al. 5-aminolevulinic by using multiple methods is an integral part of modern
acid false positives in cerebral neuro-oncology:​not all that neuro-oncological management.
is fluorescent is tumor. A case-based update and literature In addition, histological assessment of intraoperative
review. World Neurosurg. 2020;​137:​187–193. frozen sections, a routine procedure in our department,

J Neurosurg  Volume 135 • September 2021 985

*56 *
Neurosurgical forum

does also help to confirm the diagnosis of recurrent tumor rescence behavior of reactive tissue changes following glio-
before application of PDT. blastoma treatment with radiation and chemotherapy. Acta
With this degree of patient selection, we have a very low Neurochir (Wien). 2015;​157(2):​207–214.
  8. Lau D, Hervey-Jumper SL, Chang S, et al. A prospective
risk of operating on patients without real tumor progres- Phase II clinical trial of 5-aminolevulinic acid to assess
sion and underlying pseudoprogression. This lowers the the correlation of intraoperative fluorescence intensity and
remaining uncertainty of the reported higher false-positive degree of histologic cellularity during resection of high-grade
fluorescence rate of 5-ALA in recurrent HGGs. In addi- gliomas. J Neurosurg. 2016;​124(5):​1300–1309.
tion, several studies have shown a high positive predictive   9. Nabavi A, Thurm H, Zountsas B, et al. Five-aminolevulinic
value for the use of 5-ALA in recurrent gliomas.7–9 The acid for fluorescence-guided resection of recurrent malig-
specificity of 5-ALA is related to the intensity of fluores- nant gliomas:​a phase II study. Neurosurgery. 2009;​65(6):​
cence. Areas with strong fluorescence are accompanied by 1070–1077.
10. Chohan MO, Berger MS. 5-aminolevulinic acid fluorescence
a high predictive value for the presence of tumor tissue. guided surgery for recurrent high-grade gliomas. J Neuroon-
In areas with weak fluorescence, the false-positive rate col. 2019;​141(3):​517–522.
is higher. However, histologically these areas are charac-
terized by an abundance of gliotic tissue in combination INCLUDE WHEN CITING 
with inflammatory and reactive cells, but with only a little Published online May 7, 2021; DOI: 10.3171/2021.3.JNS21420.
normal brain.8–10 Consequently, resection of these areas or ©AANS 2021, except where prohibited by US copyright law
treatment with PDT is not likely to cause functional im-
pairment.
We agree that there are currently no data on PDT se-
lectivity in the context of tumor recurrence. However, the Don’t blame the resilient
fact that the one patient who was included in our study and
in whom histology revealed that radiation necrosis had no corticotroph
effect of PDT on postoperative MRI suggests a certain se-
lectivity of this treatment strategy. TO THE EDITOR: We read with interest the article by
In summary, overtreatment of patients with pseudopro- Lee et al.1 (Lee HC, Yoon HK, Kim JH, et al. Comparison
gression can generally be avoided. With modern imaging of intraoperative cortisol levels after preoperative hydro-
modalities and intraoperative frozen sections, a high cer- cortisone administration versus placebo in patients without
tainty for the detection of real tumor progression can be adrenal insufficiency undergoing endoscopic transsphe-
achieved. This is particularly important because patients noidal removal of nonfunctioning pituitary adenomas: a
with recurrent HGGs have a dismal prognosis and our sug- double-blind randomized trial. J Neurosurg. 2021;134[2]:​ ​
gested treatment strategy of re-resection and open PDT of- 526–534).
fers a promising treatment option. Historical misunderstandings may result in the unnec-
essary fear of addisonian crises and subsequent blanket use
Stephanie Schipmann, MD, PhD of steroids in pituitary surgery. Despite the obvious utility
Walter Stummer, MD, PhD of glucocorticoid therapy,2,3 its widespread use has led to
University Hospital Münster, Germany tragic postoperative deaths.4,5 Furthermore, the perceived
risk of hypopituitarism after pituitary operations has led
References to variable perioperative cortisol replacement paradigms
  1. van Dijken BRJ, van Laar PJ, Smits M, et al. Perfusion MRI largely unsupported by high-quality evidence.6
in treatment evaluation of glioblastomas:​clinical relevance of Recently, two randomized trials supported withhold-
current and future techniques. J Magn Reson Imaging. 2019;​ ing perioperative steroids in patients without preopera-
49(1):​11–22. tive adrenal insufficiency undergoing pituitary surgery.1,7
  2. Chukwueke UN, Wen PY. Use of the Response Assessment
in Neuro-Oncology (RANO) criteria in clinical trials and
Lee and colleagues1 randomized patients to placebo or a
clinical practice. CNS Oncol. 2019;​8(1):​CNS28. single dose of 100 mg of intravenous hydrocortisone 30
  3. van Dijken BRJ, van Laar PJ, Holtman GA, van der Hoorn minutes prior to induction of anesthesia, and their find-
A. Diagnostic accuracy of magnetic resonance imaging ings provide valuable insight into the resilience of the hu-
techniques for treatment response evaluation in patients with man corticotroph. The treatment group had higher serum
high-grade glioma, a systematic review and meta-analysis. cortisol levels throughout the operation, reflective of drug
Eur Radiol. 2017;​27(10):​4129–4144. administration, followed by lower cortisol levels on all 3
  4. Bashir A, Mathilde Jacobsen S, Mølby Henriksen O, et al. postoperative days. Interestingly, adrenocorticotropic hor-
Recurrent glioblastoma versus late posttreatment changes:​
diagnostic accuracy of O-(2-[18F]fluoroethyl)-L-tyrosine mone (ACTH) was only suppressed at induction. Dose-de-
positron emission tomography (18F-FET PET). Neuro Oncol. pendent pharmacokinetics suggest that a 100-mg dose is
2019;​21(12):​1595–1606. expected to have a plasma half-life greater than 1.7 hours
  5. Deuschl C, Kirchner J, Poeppel TD, et al. 11C-MET PET/MRI due to increased apparent distribution volume.8 We are
for detection of recurrent glioma. Eur J Nucl Med Mol Imag- curious to know whether the lower postoperative cortisol
ing. 2018;​45(4):​593–601. levels could be reflective of a longer biological half-life
  6. Galldiks N, Dunkl V, Stoffels G, et al. Diagnosis of pseudo- (i.e., an extended suppressive effect), or is the study simply
progression in patients with glioblastoma using O-(2-[18F]
fluoroethyl)-l-tyrosine PET. Eur J Nucl Med Mol Imaging.
underpowered to detect confounding by different rates of
2015;​42(5):​685–695. early adrenal insufficiency?
  7. Kamp MA, Felsberg J, Sadat H, et al. 5-ALA-induced fluo- This apparent paradox between rapid pituitary corti-

986 J Neurosurg  Volume 135 • September 2021

*56 *
 Neurosurgical forum

cotroph recovery and blunted adrenal response from the erative steroids in patients undergoing transsphenoidal resec-
administration of a single dose of hydrocortisone deserves tion for pituitary disease: randomized prospective clinical
more attention, as does the fundamental historic confusion trial to assess safety. Neurosurgery. 2019;85(2):E226–E232.
between atrophy of the adrenal cortex and secondary ad-   8. Toothaker RD, Welling PG. Effect of dose size on the
pharmacokinetics of intravenous hydrocortisone during
renal insufficiency as they relate to perioperative mortal- endogenous hydrocortisone suppression. J Pharmacokinet
ity.4,5 Although an addisonian crisis can precipitate from Biopharm. 1982;10(2):147–156.
any cause of long-term adrenal cortex atrophy, the human   9. Campbell HJ. The effects of partial hypophysectomy in the
corticotroph has a unique ability to sustain hypothalamus- rabbit. J Physiol. 1959;149(2):394–409.
pituitary-adrenal (HPA) function in the face of direct in- 10. Arvat E, Di Vito L, Lanfranco F, et al. Stimulatory effect of
jury. adrenocorticotropin on cortisol, aldosterone, and dehydro-
An important goal of selective adenomectomy, the most epiandrosterone secretion in normal humans: dose-response
common operation in pituitary surgery, is preservation of study. J Clin Endocrinol Metab. 2000;85(9):3141–3146.
pituitary function. However, some injury to normal gland
can occur. Therefore, how much residual healthy pituitary Disclosures
gland is necessary to support normal HPA axis function? The authors report no conflict of interest.
Landmark studies in 1959 by H. J. Campbell from Maud- Correspondence
sley Hospital, London, included serial partial hypophy-
Michael P. Catalino: [email protected].
sectomies in rabbits and compared adrenal ascorbic acid
content and depletion, as well as lymphopenic response to INCLUDE WHEN CITING 
stress.9 Campbell found that only 1% of residual anterior Published online May 14, 2021; DOI: 10.3171/2021.2.JNS21473.
pituitary was required for a normal adrenal cortex response
to stress, and that only 10% was required for both normal
ascorbic acid concentrations and histological appearance Response
of the adrenal gland. Likewise, studies in humans support First, we thank Dr. Catalino and coworkers for their
a rapid and potent stimulatory effect of ACTH on adrenal great enthusiasm about our recent article, in which we
cortisol production.10 compared intraoperative serum cortisol and ACTH con-
In conclusion, for patients with normal preoperative centrations after preoperative administration of 100 mg of
adrenal function, endogenous ACTH secretion at induc- hydrocortisone (HC group) versus placebo (C group) in pa-
tion should provide adequate support for the first 24 hours, tients without adrenal insufficiency undergoing endoscopic
which explains why preoperative supplementation is not transsphenoidal pituitary surgery. Our results showed that
routinely necessary. Thereafter, recovery room and daily serum cortisol levels were significantly higher in the group
fasting morning cortisol evaluations are adequate to safely HC throughout the operation, but no patient showed intra-
diagnose early ACTH deficiency in the perioperative set- operative hypotension due to adrenal insufficiency. For that
ting. The pituitary surgeon and endocrinologist can rest reason, we totally agree with Catalino et al.’s statement
assured that the resilient corticotroph will not let us, or our that for patients with normal preoperative adrenal function
patients, down. undergoing pituitary surgery, preoperative hydrocortisone
supplementation is not routinely necessary.
Michael P. Catalino, MD, MSc With respect to early postoperative cortisol levels, they
Carolyn S. Quinsey, MD wonder why morning cortisol levels were relatively low in
G. Stephen DeCherney, MD, MPH the group HC from 1 to 3 days postoperatively. The exact
University of North Carolina, Chapel Hill, NC underlying mechanism of early adrenal insufficiency after
pituitary surgery has not been fully revealed.1–3 Three pos-
References sibilities may be considered to explain this finding. First,
  1. Lee HC, Yoon HK, Kim JH, et al. Comparison of intra- long-lasting feedback suppression of the HPA axis caused
operative cortisol levels after preoperative hydrocortisone by exogenous steroid administration may be responsible for
administration versus placebo in patients without adrenal relatively low cortisol levels during the early postoperative
insufficiency undergoing endoscopic transsphenoidal removal period. Second, serum cortisol levels and the incidence of
of nonfunctioning pituitary adenomas: a double-blind ran- adrenal insufficiency during the early postoperative period
domized trial. J Neurosurg. 2021;134(2):526–534. were secondary outcome measures in our study. Moreover,
  2. Matson DD, Crigler JF Jr. Managment of craniopharyngioma sample size was calculated based on immediate postoper-
in childhood. J Neurosurg. 1969;30(4):377–390.
  3. Liew SY, Akker SA, Guasti L, Pittaway JFH. Glucocorticoid
ative cortisol level. Therefore, there is a possibility that our
replacement therapies: past, present and future. Curr Opin study is underpowered to detect significant differences in
Endocr Metab Res. 2019;8:152–159. both secondary outcome measurements between the two
  4. Fraser CG, Preuss FS, Bigford WD. Adrenal atrophy and groups. Third, in our study, early adrenal insufficiency was
irreversible shock associated with cortisone therapy. JAMA. shown on postoperative day 2 or 3, not on postoperative
1952;149(17):1542–1543. day 1. Early adrenal insufficiency after pituitary surgery
  5. Lewis L, Robinson RF, Yee J, et al. Fatal adrenal cortical in- may be reflective of temporary or permanent HPA axis
sufficiency precipitated by surgery during prolonged continu- suppression by surgical manipulation of normal pituitary
ous cortisone treatment. Ann Intern Med. 1953;39(1):116–126.
  6. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary gland. Indeed, 20% of patients in our study demonstrated
surgery: guidelines for perioperative assessment and manage- early (3 days) or delayed (3 months) ACTH deficiency af-
ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750. ter surgery, although the human pituitary gland has a wide
  7. Sterl K, Thompson B, Goss CW, et al. Withholding periop- range of ability for functional resilience.

J Neurosurg  Volume 135 • September 2021 987

*56 *
Neurosurgical forum

Taken together, our study suggests that in patients un- References

dergoing pituitary surgery, preoperative preserved HPA   1. Pofi R, Gunatilake S, Macgregor V, et al. Recovery of the
axis can provide adequate endogenous pituitary ACTH hypothalamo-pituitary-adrenal axis after transsphenoidal
and adrenal cortisol secretions intraoperatively as well as adenomectomy for non-ACTH-secreting macroadenomas. J
for at least the first 24 hours postoperatively, which ex- Clin Endocrinol Metab. 2019;104(11):5316–5324.
plains why preoperative steroid supplementation is not rou-   2. Inder WJ, Hunt PJ. Glucocorticoid replacement in pituitary
surgery: guidelines for perioperative assessment and manage-
tinely necessary in such patients. However, to adequately ment. J Clin Endocrinol Metab. 2002;87(6):2745–2750.
diagnose and treat early postoperative ACTH deficiency,   3. de Vries F, Lobatto DJ, Bakker LEH, et al. Early postopera-
daily fasting morning cortisol evaluation should be done tive HPA-axis testing after pituitary tumor surgery: reli-
postoperatively. Moreover, a large-scale prospective study ability and safety of basal cortisol and CRH test. Endocrine.
is needed to clarify the exact mechanism of postoperative 2020;67(1):161–171.
adrenal insufficiency and to help patient stratification in
individuals with preoperative normal adrenal function INCLUDE WHEN CITING 
Published online May 14, 2021; DOI: 10.3171/2021.3.JNS21487.
who undergo pituitary surgery.
©AANS 2021, except where prohibited by US copyright law
Hyung-Chul Lee, MD, PhD
Yong Hwy Kim, MD, PhD
Hee-Pyoung Park, MD, PhD
Seoul National University Hospital, Seoul National University
College of Medicine, Seoul, Korea

988 J Neurosurg Volume 135 • September 2021

 
*56 *