ANNALS OF SURGERY

A£-- _ Vol. 223, No. 3,253-260

©1996 Lippincott-Raven Publishers

The Impact of Disease Pattern,

Surgical Management, and Individual
Surgeons on the Risk for
Relaparotomy for Recurrent Crohn's
Disease
Stefan Post, M.D., Christian Herfarth, M.D., F.A.C.S., Erkko B6hm, M.D.,
Gundi Timmermanns, M.D., Hardy Schumacher, M.D., Guido Schurmann, M.D.,
and Markus Golling, M.D.

From the Department of Surgery, University of Heidelberg, Heidelberg, Germany

Objective
The authors provide a multivariate analysis of a large single-center experience with limited surgery
for Crohn's disease.

Summary Background Data

During the past decade, the aim of surgery for Crohn's disease has shifted from radical operation,
achieving inflammation-free margins of resection, to "minimal surgery," intended to remove just
grossly inflamed tissue or performing strictureplasties.

Methods
Seven hundred ninety-three cases of resection and/or strictureplasty in 689 individuals with
histologically verified Crohn's disease were followed for a mean period of 50 months (range, 5-
166 months). Two different end points were analyzed: 1) any relaparotomy for recurrent (or
persistent) Crohn's disease and 2) relaparotomy for site-specific recurrence. More than 30
variables of patient/disease characteristics and surgical management were included in a
proportional hazard model.

Results
Five parameters were associated independently with the risk for relaparotomy: increased risk
coincided with young age at onset of disease, involvement of jejunum, enterocutaneous fistula, or
performed strictureplasty, and decreased risk followed ileocecal resection. Site-specific nsks of
reoperation were calculated on the basis of 1260 intestinal resections or anastomoses performed in
these patients. Young age at onset, duodenal and jejunal involvement, presence of enterocutaneous
or perianal fistula, and a single surgeon (of 23) were associated significantly with increased rsk of
regional recurrence but not strictureplasty or inflammation at margins of resection.

Conclusions
Limited surgery for Crohn's disease is not associated with increased risk of regional recurrence
requiring reoperation. However, patients with juvenile onset, proximal small bowel disease, and
some types of fistulae are at a considerable risk of experiencing early surgical recurrence.

253
254 Post and Others Ann. Surg. - March 1996

Crohn's disease primarily is considered to be a gener- since October 1981. Ofthis registry, all 793 laparotomies
alized disease of the entire gastrointestinal tract, for performed up until January 1994 for symptomatic, his-
which no curative therapy currently is available.' After tologically verified Crohn's disease were included in the
surgical therapy, almost all patients will experience re- current analysis, provided at least one intestinal resec-
currence if they are observed for a sufficiently long tion or strictureplasty had been undertaken. Mean fol-
time.2'3 Early endoscopic reappearance in most patients low-up was 50 months (range, 5-166 months). For pa-
after "curative" resection4'5 indicates that some degree of tients not seen at routine outpatient visits, information
recurrence (better: recrudescence6) may represent a uni- on reoperations at other hospitals was collected by
versal feature of this disease. Wide margins of resection mailed questionnaires or telephone interviews.
may offer no advantage.7 Some evidence suggests that a
radical surgical approach may even facilitate recur-
rence.8 Thus, there has been a shift in surgical interven- End Points
tion to a more conservative-even "minimal"-ap- The time of relaparotomy due to recurrent (or persis-
proach.9 Using this nonradical and noncurative ap- tent) symptomatic Crohn's disease served as an end
proach, histologic or endoscopic/radiologic definitions point. Except for reoperations due to perioperative com-
of recurrence have become meaningless. Studies on the plications or planned reinterventions, such as closure of
rates of recurrence have to rely either on clinical symp- enterostomy, any type of relaparotomy for severe colitis,
toms, which are nonspecific and hard to quantify, or on symptomatic intestinal stenosis, or abdominal fistulous
the rate of reoperation, which-although easily quanti- disease was included, regardless of histologic proof of re-
fiable-fails to count those patients with a clinical recur- currence. In addition, all relaparotomies for obviously
rence not requiring reoperation.6 persistent Crohn's disease, e.g., in excluded rectal seg-
Regardless of the definitions and actual patterns of re- ments after temporary ileostomy, were considered "re-
currence, surgery performed for proper indications al- currences." In case of repetitive intestinal resection or
most is invariably rehabilitating for people disabled by strictureplasty during relaparotomy, patients could be
the ravages of Crohn's disease.3 After resectional ther- entered again into the study. This occurred in 104 cases.
apy, most patients regret not undergoing surgery ear- Thus, the study population consisted of 689 individuals
lier.'0 As long as the perioperative risk is minimal,' '-'3 (preliminary data analysis confirmed that results were
indications to surgery have to be guided by the chances not different if these repeat cases were excluded). Calcu-
ofsymptomatic relief and also by the individual perspec- lations were performed for two different end points, ei-
tive of a recurrence-free interval of long duration. ther relaparotomy for any intestinal recurrence or relap-
Surgeons treating this disease should be aware of any fac- arotomy for regional recurrence. For this purpose, "re-
tors that might influence the rate of recurrence to facili- gion" was defined as within ± 10 cm of previous
tate proper decision making regarding the indication for resection/anastomosis. Defining this region in the small
surgery and the type of intervention. When comparing intestine was somewhat arbitrary because of the lack of
resectional versus nonresectional therapy, site-specific well-defined landmarks (as in the case of the jejunoileal
rather than patient-related rates of recurrence should be border). Calculations of regional recurrence were based
calculated. 14 To achieve these goals, the current study in- on 1260 different surgical sites of operation (any anasto-
tended to provide a multivariate analysis of numerous mosis or resected bowel segment was considered to rep-
perioperative variables based on a large single-center ex- resent a separate site at risk of recurrence). The disease
perience with limited surgery for Crohn's disease. pattern at the time of surgery was evaluated according to
the best available evidence of clinical, endoscopic, radio-
SUBJECTS AND METHODS graphic, and intraoperative findings. The pattern of dis-
ease was considered independent of therapeutic proce-
Study Cases dures (e.g., presence of jejunal disease in Table 1 does
Although data analysis was performed retrospectively, not imply that during operation either jejunal resection
those pertaining to the perioperative period and the pa- orjejunal strictureplasty was performed).
tients history were obtained from a prospective registry
that includes all patients with Crohn's disease admitted Surgical Technique
to the Department of Surgery, University of Heidelberg,
Resections were performed at the border of macro-
scopically detectable disease, with no attempt made to
Address reprint requests to Prof. Stefan Post, Abteilung fur Allgem- achieve histologically disease-free margins. Since 1984,
einchirurgie, Klinikum der Georg-August-Universitat, Robert- strictureplasty has been used for smaller ileal and jejunal
Koch-Str. 40, 37075 Gottingen, Germany. skip lesions. In many cases of severe perianal or recto-
Accepted for publication June 21, 1995. vaginal fistulous disease, stool deviation by loop ileos-
 Vol. 223.No. 3 Relaparotomy in Crohn's Disease 255

Table 1. PATIENT AND DISEASE CHARACTERISTICS AT TI. AE OF OPERATION

Frequency/Average Site-Specific
Parameter (for 793 laparotomies) Any Rec b Recurrencet
Female sex 458(57.7%) 0.92 (0.6. 3-i .27) 0.85(0.58-1.24)
Age 34 ±llyr(13-80)t 0.98 (0.9. .-0.996)§ 0.98(0.96-1.00)
Age at onset of disease 25 ±llyr(3-79)t 0.96 (0.9z [.0.98)¶** 0.96 (0.94-0.98)¶#
Duration of symptomatic disease 9 ± 6 yr (0-36)t 1.02 (0.9. )-1.05) 1.03(1.00-1.06)§
No. of previous laparotomies for CD 0.8 ± 1.2 (O-7)t 1.23 (1
Duodenal disease 28 (3.5%) 2.03 (1 .1( )-3.76)§ 2.42 (1 .36-4.33)II#
Jejunal disease 71 (9.0%) 2.83 (1 .8. ) 2.75 (1 .82-4.14)¶tt
heal disease 639(80.4%) 1.25 (0.8( )-1 .95) 1.16(0.66-2.04)
Colonic disease 467(58.9%) 0.88 (0.6z .-1 .22) 0.79 (0.54-1.15)
Rectal disease 137 (17.3%) 1.26 (0.8. kl.90) 0.91 (0.54-1.56)
Abdominal abscess 113(14.2%) 1.09 (0.7( )-1 .69) 0.75(0.43-1.32)
Perianal fistula 124(15.7%) 1.28 (0.8. 3-1 .99) 1 .63 (0.99.2.67)**
Enterocutaneous fistula 91 (11.5%) 1.91 (1.2( .-2.91)II# 2.67 (1.74-4.09)¶tt
Interenteric fistula 247(31.1%) 0.86 (0.6( )-1 .22) 0.55 (0.36-0.83)11 tt
Retroperitoneal fistula 0.61 (0.2. )-1.31) 0.71 (0.29-1.73)
Enterovesical fistula 42(5.3%) 0.50 (0.2( )-1.21) 0.47 (0.17-1.27)
Enterogenital fistula 47(5.9%) 0.68 (0.3( )-1 .53) 0.46(0.15-1.45)
Extraintestinal manifestations 71 (9.0%) 1.73(1.01 -2.98)§ 1.20(0.56-2.59)
* Results are risk ratios (with 95% confidence limits in parentheses) for univariate calculations based on the sample of 7 93 laparotomies.
t Results are risk ratios for univariate calculations based on the sample of 1 260 sites of intestinal reconstruction (95% c onfidence limits in parentheses).
1: Mean ± standard deviation (range).
§p<O.05.
Hp 0.01.
'p < 0.001 in univariate analysis.
#p<O.05.
<0 01
ttP < 0.001 in multivariate analysis.

tomy was performed in preference to primary proctec- out the inpatient period, with increased doses given pen-
tomy. Only in rare instances was enterostomy used for operatively. During part of the study period (1985-
anastomotic 15 All anastomoses were per- 1 99 1 ), patients without steroids received a routine 3-
formed in a single-layer, end-to-end fashion, with 0in- month course of prophylactic corticosteroid therapy
terrupted vertical mattress sutures using absorbable su- starting on the day ofsurgery (although being beyond the
ture material (polyglactin 9 1 0 or polyglycolic acid, 3-0 or scope of the current study, this policy did not show any
4-0). To avoid fistula formation, intra-abdominal drains influence on the rate ofrecurrence). Additional preoper-
were never used close to anastomoses, and only rarely ative drug therapy with salicylates or immunosuppres-
in cases of(usually extraperitoneal) abscesses. Whenever sive agents was restarted with the resumption of oral in-
feasible, omentoplasty was performed in the presence of take. After discharge, decisions on drug therapy were
fistulae or abscesses. There was no change in surgical made by the gastroenterologist caring for the patient.
technique throughout the study period. Data on further medical therapy during the years of fol-
low-up were incomplete and were not included in the
Surgeons current analysis.
Operations included in this study were performed by
23 surgeons all of whom followed the same guidelines
of surgical management as listed above. Eight of these Histologic Margin
surgeons had contributed more than 50 cases and were
analyzed separately, whereas the remaining 1 5 surgeons Mild, nonspecific inflammatory changes at the mar-
were grouped together. gins ofresected specimens were discounted. However, all
cases with significant inflammation at margins were in-
Medical Therapy cluded as affected by disease, regardless of specificity.
Patients on long-term corticosteroid therapy (> 3 Furthermore, both margins ofstrictureplasties were con-
weeks preoperatively) were continued on them through- sidered inflamed, even without histologic verification.
256 Post and Others Ann. Surg. . March 1996

Table 2. SURGICAL MANAGEMENT

Frequency/Average Site-Specific
Parameter (for 793 laparotomies) Any Recurrence* Recurrencet
Emergency laparotomy 33(4.2%) 0.96 (0.43-2.19) 0.46(0.11-1.86)
No. of anastomoses 1.7 ± 1.4 (0-15)t 1.20 (1.10-1.32)§ 1.14 (1.06-1.22)§
Histologic inflammation at margin of resection 490 (61.8%) 1.30 (0.93-1.83) 0.97 (0.65-1.43)
Length of resected specimen 32 ± 22 cm (0-158 cm)t 0.99 (0.98-0.999)11 1.0 (0.99-1.01)
Strictureplasty 82 (10.3%) [204 sites] 2.73 (1.76-4.25)§¶ 1.57 (0.96-2.57)
Jejunal resection 18 (2.3%) 1.70 (0.68-4.08) 1.84 (0.58-5.81)
Ileal resection** 130(16.4%) 1.85(1.29-2.65)§ 1.21 (0.72-2.02)
Ileocecal resection/right hemicolectomy 348 (43.9%) 0.42 (0.29-0.60)§# 0.65 (0.41-1.02)
Segmental resection of colon or rectum** 119 (15.0%) 1.21 (0.80-1.81) 1.05 (0.59-1.88)
Subtotal colectomy with ileorectal anastomosis 85 (10.7%) 0.97 (0.57-1.66) 1.64 (0.88-3.07)
Proctectomy/proctocolectomy 34 (4.3%) 0.55 (0.18-1.74) 0.82 (0.20-3.33)
Resection of previous anastomosis 147 (18.5%) 1.47 (0.99-2.17) 1.04 (0.57-1.89)
Closure of enteric fistula 115 (14.5%) 0.73 (0.42-1.27) 0.32 (0.12-0.88)11
Intestinal bypass 11 (1.4%) 1.50 (0.55-4.07) 2.54 (0.80-8.06)
Results are risk ratios (with 95% confidence limits in parentheses) for univariate calculations based on the sample of 793 laparotomis.
t Results are risk ratios for univariate calculations based on the sample of 1260 sites of intestinal reconstruction (95% confidence limits in parentheses).
$ Mean ± standard deviation (range).
§ p < 0.001 in univariate analysis.
1 p < 0.05 in univariate analysis.
T p < 0.05 in multivariate analysis.
# p < 0.001 in multivariate analysis.
Excluding ileocecal resections and right hemicolectomies.

Statistical Analysis gression analysis (including all parameters listed in Ta-

bles 1, 2, and 3) was performed by stepwise variable se-
Differences were considered significant at p < 0.05 lection, with p < 0.05 for entry and removal of variables
(two-sided). Distributions of reoperation-free survival from the model. Calculations were performed by the
were estimated by the product-limit method.16 Risk ra-
SAS procedures LIFETEST and PHREG (SAS Institute,
tios and confidence intervals were based on estimates Cary, NC).
from the proportional hazards model. 7 Multivariate re-

RESULTS
Table 3. IMPACT OF INDIVIDUAL In approximately half of the cases (47%), the operation
SURGEONS
analyzed was not the first laparotomy for Crohn's dis-
No. of Any Site-Specific ease, but the second (28%), third (10%), or fourth and
Parameter Operations Recurrence* Recurrencet more (9%). Because of complex and multifocal manifes-
tations of the disease, more than one intestinal anasto-
Surgeon A 166 0.80 (0.54-1.19) 0.86 (0.55-1.35)
Surgeon B 101 0.96 (0.64-1.45) 0.60 (0.34-1.06)
mosis was performed on 39% of the patients, three or
Surgeon C 75 1.54 (0.88-2.70) 2.20 (1.23-3.92)t more anastomoses were performed on 16% of the pa-
Surgeon D 69 1.14 (0.62-2.11) 0.82 (0.33-2.02) tients. At the time of surgery, the disease was localized to
Surgeon E 67 1.49 (0.69-3.21) 2.49 (1.14-5.42)§11 the small intestine in 38% of the patients and to the large
Surgeon F 63 1.61 (0.74-3.50) 1.44 (0.52-3.98) bowel in 17% of the patients; 45% of the patients had
Surgeon G 54 1.32 (0.65-2.68) 1.36 (0.55-3.36)
Surgeon H 52 2.84 (1.30-6.20)t 1.51 (0.47-4.85) both small and large bowel involvement. Details of pat-
All other surgeons 146 0.96 (0.67-1.39) 1.13 (0.75-1.73) tern are listed in Table 1.
Indications for laparotomy were multifactorial in
* Results are risk ratios (with 95% confidence limits in parentheses) for univariate
most cases. Besides various types of fistulae or abscesses,
calculations based on the sample of 793 laparotomies.
t Results are risk ratios for univariate calculations based on the sample of 1260 sites indications included symptomatic intestinal stenoses not
of intestinal reconstruction (95% confidence limits in parentheses). responding to medical treatment (72%), colitis refractory
t p < 0.01 in univariate analysis. to conservative management (11%), blind loop syn-
§ p < 0.05 in univariate analysis.
|| p < 0.01 in multivariate analysis.
drome (1.9%), and ureteral obstruction (1.4%). Laparot-
omy was performed in an emergency setting in 4.2% of
Vol. 223 - No. 3 Relaparotomy in Crohn's Disease 257

gou! likelihood of relaparotomy (multivariate risk ratio: 0.97

per year of age [95% confidence interval: 0.95-0.99] both
0 50 for any and regional recurrence).
0

X 40
any recurrence
Pattern of Disease (Table 1)
4-
0 30
Regardless of the type of operation performed, in-
volvement of the proximal intestinal tract (duodenum
_20 and jejunum) was found to be a significant risk factor.
Enterocutaneous fistulae increased the risk of relaparot-
E rO
regional recurrence omy for any as well as regional recurrence, whereas the
presence of an interenteric fistula anywhere in the abdo-
o- men decreased the risk of site-specific recurrence. Extra-
O 2 4 6 8 lO0 intestinal disease (ocular, articular, or skin symptoms)
years since operation
Figure 1. Kaplan-Meier estimate of the overall incidence of relaparotomy
contributed to the risk of nonregional recurrence only.
for recurrent Crohn's disease at any site vs. regional recurrence (within + Other types of fistulae or septic manifestations of disease
10 cm of previous surgery). were of no importance regarding the risk for relaparot-
omy.
In multivariate analysis, jejunal involvement and the
the cases because of toxic colon (n = 7), free intestinal presence of enterocutaneous fistula were associated sig-
perforation (n = 14), severe bleeding (n = 4), or ad- nificantly with the risk of any recurrence requiring relap-
vanced complete ileus (n = 8). Perioperative complica- arotomy (risk ratio 1.8 [95% confidence interval, 1.0-
tions in the current series were similar to those reported 3.0] and 1.8 [95% confidence interval, 1.1-2.8], respec-
earlier": uncomplicated course in 89.4%, minor compli- tively). Both parameters also were important for the risk
cations in 5.8%, relaparotomy for complications in 4.4%, of regional recurrence (multivariate risk ratio 2.4 [95%
and mortality 0.4%. confidence interval, 1.5-3.8] and 3.2 [95% confidence
interval, 2.0-5.2], respectively). In addition, operative
sites in patients with duodenal involvement (multivari-
Risk of Recurrence (Fig. 1) ate risk ratio 1.9 [95% confidence interval, 1.0-3.5]) or
An almost linear relationship could be detected be-
perianal fistulous disease (multivariate risk ratio 2.2
tween the time elapsed since operation and the overall
[95% confidence interval, 1.3-3.6]) carried an increased
incidence of reoperations due to recurrent Crohn's dis- risk of regional recurrence, whereas the presence of an
ease at any site (24% and 41% after 5 and 10 years, re-
interenteric fistula improved prognosis (multivariate risk
ratio 0.4 [95% confidence interval, 0.3-0.6]).
spectively) or regional recurrence (13% and 26% of pa-
tients and 1 1% and 21% of sites after 5 and 10 years, re-
spectively). Thus, more than half of the recurrences -
60
requiring reoperation were at the very same site. There
was no apparent trend of diminishing or increasing risk E
o 50-
throughout the 10-year follow-up period. 0 c 15 yrs.
o
Q 1 5-25 yrs.
5) 40-
Patient Characteristics (Table 1) 0
0
The risk for relaparotomy decreased with increasing -a 30-

age at time of operation (2% risk reduction per year of

age) as well as with age at onset ofdisease (4% risk reduc- ai 20-
I > 25 yrs.
tion per year of age; Fig. 2). The latter factor and the
duration of symptomatic disease also were associated 10-
with a significantly increased risk of site-specific recur-
rence. Patients with previous surgery for Crohn's disease
were more prone to undergo relaparotomy for recurrent 0 2 4 6
years since
.8 10
operation
disease at any site or at an identical site (23% and 18% Figure 2. Kaplan-Meier estimate of the overall incidence of relaparotomy,
increase of risk per each previous laparotomy, respec- depending on the age of first symptoms of Crohn's disease, demonstrat-
tively). Multivariate analysis revealed only age at onset ing a significantly increased risk for younger age of onset (log rank analy-
of symptomatic disease to contribute significantly to the sis: p = 0.0001).
258 Post and Others Ann. Surg. * March 1996

- 25- characteristics of "radical" or "minimal" surgery were

predictive of regional recurrence.
c:
w0 20-
S
Surgeons (Table 3)
0
o 10
one margin inflamed j
Despite standardized surgical technique, considerable
variations in the relative risks for relaparotomy were ob-
3 3 15-
S
5- served among the individual surgeons analyzed. How-
_
u) ever, this was mainly because of variations in the case
Inflammation on both marglns
mix, as illustrated by Surgeon H. His patients experi-
enced an almost threefold increase of risk for relaparot-
Inflammation
no
omy, which was explained easily by a substantial propor-
~. .. .. .. .. tion of high-risk patients, as illustrated by a lack of sig-
0 2 4 6 8 .10
years since operation nificance in multivariate statistics. A significant effect
Figure 3. Lack of influence of histologic inflammation at the margins of could be verified by multivariate analysis (p = 0.008) in
resection on the site-specific risk for relaparotomy (log-rank analysis: p = only one case (regional recurrence for Surgeon E).
0.87).

DISCUSSION
Surgical Management (Table 2) Overall recurrence rates found in the current study are
well in accordance with reports published previously,
Emergency procedures due to complete intestinal ob- some with a more prolonged follow-up, but most ofthem
struction, severe bleeding, toxic colon, or free perfora- on smaller series'3"18 (for overview of published series up
tion were not associated with the risk of recurrence. Pres- to 1990 see reference 6). In addition, it is not surprising
ence of histologic evidence of inflammation at one (ob-
that most of the recurrences observed were located at
served in 43.3% ofthe cases) or both margins of resection previous anastomoses, usually just proximal to it.4"'9 We
(in 18.5%) was without any consistent or significant were unable to identify the biphasic pattern of recurrence
effect (Fig. 3). After operations with multiple anastomo- previously considered to represent two different types of
ses, there was a marked increase in the risk for relaparot-
the disease.20'2'
omy for any as well as regional recurrence (20% and 14%
Although population-based studies suggested juvenile
increase of risk per each anastomosis, respectively). On onset to be associated with a high frequency of surgical
the other hand, extensive resections were associated with intervention,22 this is the first study to verify that mani-
a lower risk of any recurrence ( 1% risk reduction per each
festation of disease at a young age is associated indepen-
centimeter of resected specimen), but not with the risk of dently with a worse prognosis. This detrimental effect
site-specific recurrence. persisted into adulthood, i.e., age at operation or dura-
When the operation performed included a stricture- tion of symptoms was less important than age at first
plasty, it was associated with an increased risk for any symptoms. However, it cannot be excluded that differ-
recurrence. However, analysis of site-specific recurrence ences may diminish for follow-up periods beyond the
revealed that this was not because of an inherent risk of first 10 years after operation.23
the procedure itself, but because patients requiring strict- Disease pattern generally is analyzed according to the
ureplasties usually have proximal small bowel disease, traditional subdivision into colonic disease, ileocolic dis-
which often recurs at another site. Out of the various ease, and that limited to the small bowel.6'24 Probably
types of resections analyzed, proximal ileal resections because of small numbers, previous studies have been
were associated with increased risk of any, but not site- unable to demonstrate the prognostic importance of in-
specific, recurrences. Ileocecal resections (usually for ter- volvement of proximal segments of small intestine, as
minal ileal disease) carried a lower risk of any recurrence, found in the current series. In the majority of cases, prox-
and simple closure ofenteric fistulae (usually only at the imal disease also was associated with multiple-site in-
target opening ofthe fistula but not at the origin) carried volvement, e.g., just 5 of 71 patients with jejunal disease
a low risk of regional recurrence. did not have more distal manifestations. If included in
In multivariate statistics, strictureplasty and ileocecal the analysis, multiple-site involvement (like multiple
resection turned out to be significantly associated with anastomoses) was associated with poor prognosis, as re-
the risk for relaparotomy for any recurrence (risk ratio ported by others.2"'24 However, this was not an indepen-
1.8 [95% confidence interval, 1.1-3.0] and 0.4 [95% con- dent risk factor in multivariate analysis. As suggested
fidence interval, 0.3-0.6], respectively). None of the previously for clinical but not surgical recurrence,25 ex-
Vol. 223 * No. 3 Relaparotomy in Crohn's Disease 259

traintestinal manifestations of disease were found to be much small bowel, such as in terminal ileal disease.
independent predictors of surgical recurrence. However, strictureplasty appears to be as safe as resec-
In a study of 770 patients, Greenstein et al.26 intro- tional therapy when site-specific rates of recurrence are
duced the concept of two clinical forms of Crohn's dis- considered, which is well in accordance with a previous
ease, i.e., perforating (in case of any fistula, abscess, or study from Birmingham.'4 Because of the complexity
free perforation) and nonperforating disease. They pro- and variability of clinical manifestations, performing
vided evidence that patients with perforating disease randomized trials in surgery for Crohn's disease rarely is
tend to experience earlier recurrence and that recur- attempted.'9 Even without such a definite confirmation,
rences follow the same pattern as primary manifesta- "minimal" or "limited" surgery for Crohn's disease
tions. In accordance with data from the Cleveland should be considered to represent a well-established stan-
Clinic,27 we are unable to provide support for this classi- dard of treatment. The observation that out of 689 pa-
fication. We found enterocutaneous fistulae and con- tients included in the current series, only 1 developed
comitant perianal fistulae, but no other form of abscess overt short bowel syndrome after numerous laparoto-
or fistula or true perforation to be associated with early mies may provide additional support for this concept.
recurrence. Interenteric fistulae were even linked to an The problem of recurrence after surgical therapy for
improved prognosis. In addition, there was no trend to- Crohn's disease will not be resolved surgically. However,
ward repetition of clinical patterns in case of recurrence. up until now, no drug has been available with proven
Previously, we had demonstrated that perforating dis- efficacy for the maintenance of remission.32 The current
ease is unrelated to perioperative complications.'2 data may help to identify subgroups of patients most in
The surgeon risk factor has been well established in need of adjuvant medical therapy, such as those with ju-
studies on the rate of recurrence after resection of colo- venile onset, proximal small bowel involvement, and en-
rectal cancer.28 No reports on this issue have been pub- terocutaneous fistulae.
lished for Crohn's disease. Our observation that even
with a standardized technique in a single center, differ-
ences are detectable between individual surgeons should Acknowledgments
encourage inclusion ofthis variable in future studies, es- The authors thank Mrs. A. Golling for her help in preparation of
pecially in those with multicenter involvement.8 the manuscript and H. Bindewald for initiating the registry of Crohn's
Probably the most interesting findings in the current disease in Heidelberg.
study are related to the various aspects of "minimal sur-
gery."9 For many years, it has been accepted by most References
surgeons performing resections in Crohn's disease that
there is no need to achieve histologically free margins.6 7 1. Podolsky DK. Medical progress: Inflammatory bowel disease (sec-
ondoftwoparts). NEnglJ Med 1991; 325:1008-1016.
However, a recent study of 164 patients with a 3-year 2. Lock MR, Farmer RG, Fazio VW, et al. Recurrence and reopera-
follow-up indicated that multiple anastomoses in in- tion for Crohn's disease. N Engl J Med 1981; 304:1586-1588.
flamed tissue may carry a very high risk of early symp- 3. Sachar DB. The problem of postoperative recurrence of Crohn's
tomatic recurrence.'8 Using different end points, longer disease. Med Clin North Am 1990; 74:183-188.
follow-up, and a larger sample, the current study does 4. Rutgeerts P, Geboes K, Vantrappen G, et al. Predictability of the
not support these data. Performing anastomoses in postoperative course of Crohn's disease. Gastroenterology 1990;
99:956-963.
bowel segments with histologic signs of specific or non- 5. Olaison G, Smedh K, Sjbdahl R. Natural course of Crohn's disease
specific inflammation at the resection margin did not in- after ileocolic resection: endoscopically visualised ileal ulcers pre-
fluence the risk for reoperation, particularly not for site- ceding symptoms. Gut 1992; 33:331-335.
specific recurrence. Although the number of anastomo- 6. Williams JG, Wong WD, Rothenberger DA, Goldberg SM. Recur-
rence of Crohn's disease after resection. Br J Surg 1991; 78:10-19.
ses was associated significantly with increased rate of 7. Pennington L, Hamilton SR, Bayless TM, Cameron JL. Surgical
recurrence in univariate statistics, this was almost exclu- management of Crohn's disease: influence of disease at margin of
sively due to numerous anastomoses in cases of multiple resection. Ann Surg 1980; 192:311 -318.
strictureplasties for proximal skip lesions. 8. Ewe K, Herfarth C, Malchow H, Jesdinsky HJ. Postoperative re-
Despite recent reports on extensive experiences with currence of Crohn's disease in relation to radicality of operation
and sulfasalazine prophylaxis: a multicenter trial. Digestion 1989;
strictureplasties,29-3' none of the multivariate studies on 42:224-232.
recurrence of Crohn's disease have included substantial 9. Lee ECG. Aim of surgical treatment of Crohn's disease. Gut 1984;
numbers of this procedure.'3"18'25.Our data suggest that 25:217-222.
strictureplasty is associated with an increased risk for re- 10. Scott NA, Hughes LE. Timing of ileocolonic resection for symp-
operation related to the underlying pattern of disease tomatic Crohn's disease-the patient's view. Gut 1994; 35:656-
657.
rather than the procedure itself. Strictureplasty probably 11. Heimann TM, Greenstein AJ, Mechanic L, Aufses AH Jr. Early
should not be the first choice in cases in which a single complications following surgical treatment for Crohn's disease.
resection may be performed without sacrifice of too Ann Surg 1985; 201:494-498.
260 Post and Others Ann. Surg. March 1996

12. Post S, Betzler M, von Ditfurth B, et al. Risks of intestinal anasto- tion and recurrence in Crohn's colitis and ileocolitis: crude and
moses in Crohn's disease. Ann Surg 1991; 213:37-42. cumulative rates. N Engl J Med 1975; 293:685-690.
13. Michelassi F, Balestracci T, Chappell R, Block GE. Primary and 24. Farmer RG, Whelan G, Fazio VW. Long-term follow-up of
recurrent Crohn's disease. Experience with 1379 patients. Ann patients with Crohn's disease: relationship between the clinical
Surg 1991; 214:230-238. pattern and prognosis. Gastroenterology 1985; 88:1818-
14. Sayfan J, Wilson DA, Allan A, et al. Recurrence after stricture- 1825.
plasty or resection for Crohn'sdisease. BrJ Surg 1989; 76:335-338. 25. Cottone M, Rosselli M, Orlando A, et al. Smoking habits and re-
15. Post S, Herfarth C, Schumacher H, et al. Experience with ileos- currence in Crohn's disease. Gastroenterology 1994; 106:643-648.
tomy and colostomy in Crohn's disease. Br J Surg 1995; 82:1629- 26. Greenstein AJ, Lachman P, Sachar DB, et al. Perforating and non-
1633. perforating indications for repeated operations in Crohn's disease:
16. Kaplan EL, Meier P. Nonparameteric estimation from incomplete evidence for two clinical forms. Gut 1988; 29:588-592.
observations. J Am Stat Assoc 1958; 53:457-481. 27. McDonald PJ, Fazio VW, Farmer RG, et al. Perforating and non-
17. Cox DR. Regression models and life-tables. J R Stat Soc 1972; 34: perforating Crohn's disease: an unpredictable guide to recurrence
187-220. after surgery. Dis Colon Rectum 1989; 32:117-120.
18. Heimann TM, Greenstein AJ, Lewis B, et al. Prediction of early 28. Phillips RKS, Hittinger R, Blesovsky L, et al. Local recurrence fol-
symptomatic recurrence after intestinal resection in Crohn's dis- lowing 'curative' surgery for large bowel cancer: I. the overall pic-
ease. Ann Surg 1993; 218:294-298. ture. BrJ Surg 1984; 71:12-16.
19. Cameron JL, Hamilton SR, Coleman J, et al. Patterns of ileal re- 29. Fazio VW, Tjandra JJ, Lavery IC, et al. Long-term follow-up of
currence in Crohn's disease: a prospective randomized study. Ann strictureplasty in Crohn's disease. Dis Colon Rectum 1993; 36:
Surg 1992; 215:546-55 1. 355-361.
20. De Dombal FT, Burton I, GoligherJC. Recurrence ofCrohn's dis- 30. Dehn TCB, Kettlewell MG, Mortensen NJ, et al. Ten-year experi-
ease after primary excisional surgery. Gut 1971; 12:519-527. ence of strictureplasty for obstructive Crohn's disease. Br J Surg
21. Chardavoyne R, Flint GW, Pollack S, Wise L. Factors effecting 1989; 76:339-341.
recurrence following resection for Crohn's disease. Dis Colon Rec- 31. Spencer MP, Nelson H, Wolff BG, Dozois RR. Strictureplasty for
tum 1986; 29:495-502. obstructive Crohn's disease: the Mayo experience. Mayo Clin Proc
22. Sedgwick DM, Barton JR, Hamer-Hodges DW, et al. Population- 1994; 69:33-36.
based study of surgery in juvenile onset Crohn's disease. Br J Surg 32. Salomon P, Kombluth A, Aisenberg J, Janowitz HD. How
1991; 78:171-175. effective are current drugs for Crohn's disease: a metaanalysis. J
23. Greenstein AJ, Sachar DB, Pasternak BS, Janowitz HD. Reopera- Clin Gastroenterol 1992; 14:211-215.