Received: 20 July 2020 Revised: 28 November 2020 Accepted: 2 December 2020

DOI: 10.1002/alz.12283

RESEARCH ARTICLE

The diagnostic and prognostic capabilities of plasma

biomarkers in Alzheimer’s disease

Joel Simrén1,2,† Antoine Leuzy3,† Thomas K. Karikari1 Abdul Hye4

Andréa Lessa Benedet5 Juan Lantero-Rodriguez1 Niklas Mattsson-Carlgren3,6,7
Michael Schöll1,8,9 Patrizia Mecocci10 Bruno Vellas11 Magda Tsolaki12
Iwona Kloszewska13 Hilkka Soininen14 Simon Lovestone15 Dag Aarsland4,16
for the AddNeuroMed consortium Oskar Hansson3,17 Pedro Rosa-Neto5
Eric Westman18,19 Kaj Blennow1,2 Henrik Zetterberg1,2,9,20,‡
Nicholas J. Ashton1,4,8,‡
1
Department of Psychiatry and Neurochemistry, The Sahlgrenska Academy at the University of Gothenburg, Mölndal, Sweden
2
Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Gothenburg, Sweden
3
Clinical Memory Research Unit, Lund University, Malmö, Sweden
4
Department of Old Age Psychiatry, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK
5
Translational Neuroimaging Laboratory, McGill University, Montréal, Canada
6
Department of Neurology, Skåne University Hospital, Lund, Sweden
7
Wallenberg Centre for Molecular Medicine, Lund University, Lund, Sweden
8
Wallenberg Centre for Molecular and Translational Medicine, University of Gothenburg, Gothenburg, Sweden
9
Department of Neurodegenerative Disease, UCL Queen Square Institute of Neurology, London, UK
10
Department of Medicine, Institute of Gerontology and Geriatrics, University of Perugia, Perugia, Italy
11
CHU Toulouse, Toulouse, France
12
Aristotle University of Thessaloniki, Thessaloniki, Greece
13
Medical University of Lodz, Lodz, Poland
14
Institute of Clinical Medicine, University of Eastern Finland, Kuopio, Finland
15
University of Oxford, Oxford, UK
16
Centre for Age-Related Medicine, Stavanger University Hospital, Stavanger, Norway
17
Memory Clinic, Skåne University Hospital, Malmö, Sweden
18
Division of Clinical Geriatrics, Department of Neurobiology, Care Sciences and Society, Karolinska Institute, Stockholm, Sweden
19
Department of Neuroimaging, Centre for Neuroimaging Sciences, Psychology and Neuroscience, King’s College London, Institute of Psychiatry, London, UK
20
UK Dementia Research Institute at UCL, London, UK

Correspondence
Dr. Nicholas J. Ashton, Department of Psychi- Abstract
atry and Neurochemistry, The Sahlgrenska
Academy at the University of Gothenburg,
Introduction: This study investigated the diagnostic and disease-monitoring potential
Mölndal, Sweden. of plasma biomarkers in mild cognitive impairment (MCI) and Alzheimer’s disease (AD)
Email: [email protected]
dementia and cognitively unimpaired (CU) individuals.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2021 The Authors. Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association

Alzheimer’s Dement. 2021;1–12. wileyonlinelibrary.com/journal/alz 1

2 SIMRÉN ET AL .

†
First authors Joel Simrén and Antoine Leuzy
contributed equally to this work. Methods: Plasma was analyzed using Simoa assays from 99 CU, 107 MCI, and 103 AD
‡
Senior authors Henrik Zetterberg and
Nicholas J. Ashton contributed equally to
dementia participants.
this work. Results: Phosphorylated-tau181 (P-tau181), neurofilament light, amyloid-β
(Aβ42/40), Total-tau and Glial fibrillary acidic protein were altered in AD demen-
Funding information
Swedish Research Council, Grant/Award Num- tia but P-tau181 significantly outperformed all biomarkers in differentiating AD
ber: #2017-00915; Alzheimer Drug Discovery
dementia from CU (area under the curve [AUC] = 0.91). P-tau181 was increased
Foundation (ADDF), USA, Grant/Award Num-
ber: #RDAPB-201809-2016615; Swedish in MCI converters compared to non-converters. Higher P-tau181 was associated
Alzheimer Foundation, Grant/Award Num-
with steeper cognitive decline and gray matter loss in temporal regions. Longitudinal
ber: #AF-742881; Hjärnfonden, Sweden,
Grant/Award Number: #FO2017-0243; change of P-tau181 was strongly associated with gray matter loss in the full sample
Swedish government and the County Coun-
and with Aβ measures in CU individuals.
cils; ALF-agreement, Grant/Award Number:
#ALFGBG-715986; European Union Joint Discussion: P-tau181 detected AD at MCI and dementia stages and was strongly asso-
Program for Neurodegenerative Disorders,
ciated with cognitive decline and gray matter loss. These findings highlight the poten-
Grant/Award Number: JPND2019-466-236;
European Research Council, Grant/Award tial value of plasma P-tau181 as a non-invasive and cost-effective diagnostic and prog-
Number: #681712; Swedish State Support
nostic biomarker in AD.
for Clinical Research, Grant/Award Number:
#ALFGBG-720931; Alzheimer Drug Discov-
ery Foundation (ADDF), USA, Grant/Award KEYWORDS
Number: #201809-2016862; UK Dementia Alzheimer’s disease, magnetic resonance imaging, phosphorylated tau, plasma biomarkers
Research Institute at UCL

1 INTRODUCTION cohorts that more closely resemble those from primary care settings
(eg, not predefined by in vivo measures). In the present study, we tested
Alzheimer’s disease (AD) has until recently been diagnosed based the diagnostic and prognostic performance of putative blood biomark-
solely on clinical symptomatology, with definitive neuropathological ers of AD pathology (Aβ42, Aβ42/Aβ40, total [T-tau], and P-tau181),
confirmation at post-mortem. Using cerebrospinal fluid (CSF) and axonal injury (neurofilament light [NfL]), and astrogliosis (glial fibrillary
positron emission tomography (PET)−based biomarkers, however, acidic protein [GFAp]), which included their baseline and longitudinal
amyloid beta (Aβ) and tau pathology,1,2 the pathological hallmarks of association with cognitive decline and grey matter (GM) atrophy in the
AD, can now be measured in vivo. These biomarkers are now con- clinical AddNeuroMed cohort.
sidered to provide supportive evidence for AD in recent diagnos-
tic criteria and have helped to refine the biological definition of the
disease.3–5 Owing to the perceived invasiveness of lumbar punctures 2 METHODS
for CSF and the high cost of PET imaging, neither technique is suitable
for widespread use in a primary care settings, where the majority of 2.1 Participants
dementia diagnoses are made. Both techniques, however, have greatly
guided the rapid discovery and validation of blood biomarkers for AD We included 309 participants from the AddNeuroMed study,20,21
pathology, measured by ultra-sensitive immunoassays or mass spec- which is a public-private partnership initiated to discover plasma
trometry methods. Recent studies have shown plasma measures to biomarkers of AD for use in clinical trials, including cognitively unim-
relate to Aβ deposition,6–8 astrogliosis9 and neurodegeneration.10–13 paired (CU) participants and patients with MCI (including a subset
Tau phosphorylated at threonine181 (P-tau181) may be the most clin- which progressed to AD dementia, n = 19), and individuals with AD
ically meaningful plasma biomarker in AD to date, and multiple stud- dementia. The participants in this study were selected based on the
ies performed in independent cohorts demonstrate its high specificity availability of both plasma samples and magnetic resonance imaging
for AD, strong associations with Aβ and tau PET, and high accuracy (MRI) imaging. All patients were recruited from local memory clinics at
for predicting progression from mild cognitive impairment (MCI) to AD participating sites (University of Kuopio, Finland; University of Peru-
dementia.14–19 gia, Italy; Aristotle University of Thessaloniki, Greece; King’s College
It is expected that blood biomarkers will advance the diagnostic London, United Kingdom; University of Lodz, Poland; and University
workup of individuals with cognitive complaints, early onset demen- of Toulouse, France), whereas CU participants were recruited from
tia, and atypical dementia presentations. By rapid indication of the non-related members of the patient’s families, caregiver’s relatives,
underlying pathology, blood biomarkers will permit for more informed or social centers for the elderly. The diagnosis of probable AD was
patient management and symptomatic treatment. Still, more evidence made according to Diagnostic and Statistical Manual of Mental Disor-
is needed to demonstrate the validity of blood biomarkers in clinical ders, Fourth Edition, (DSM-IV) and National Institute of Neurological,
SIMRÉN ET AL . 3

2.3 Voxel-based morphometry

RESEARCH IN CONTEXT
After intensity nonuniformity correction,26 anatomical images were
1. Systematic review: The authors reviewed the available
segmented into probabilistic GM, white matter (WM), and CSF.26,27
scientific literature on PubMed for articles examining
Voxel-based morphometry (VBM)28 was then performed on the struc-
plasma biomarkers in Alzheimer’s disease (AD). Recent
tural segmented GM images, nonlinearly resampled to the MNI-152
publications report diagnostic performance of individual
template, and smoothed with a Gaussian kernel of full width at half
plasma biomarkers, but no head-to-head studies compar-
maximum (FWHM) of 8 mm.
ing the diagnostic performance and prediction of disease
progression between the major putative plasma biomark-
ers. Therefore, in this study, we compare their diagnos- 2.4 Biochemical analyses
tic performance, their longitudinal trajectories, and how
they predict cognitive decline and gray matter (GM) loss. At the time of assessment, all blood samples were drawn by venipunc-
2. Interpretation: Our findings suggest that one of these ture and collected into ethylenediaminetetraacetic acid (EDTA) tubes.
biomarkers is superior—plasma tau phosphorylated at Participants were required to fast for at least 2 hours prior to col-
threonine181 (P-tau181) accurately detected AD at the lection. All samples were centrifuged at 2000 g for 10 min at 4◦ C.
dementia and mild cognitive impairment (MCI) stages and Plasma supernatant was collected, divided into aliquots, and then
was more strongly associated with cognitive decline and frozen at −80◦ C until further use.29 Plasma analysis were performed
GM loss compared to other plasma biomarkers. These on an HD-1 analyzer (Quanterix, Lexington, MA) at the Department
findings highlight the potential value of plasma P-tau181 of Psychiatry and Neurochemistry, University of Gothenburg between
as a non-invasive and cost-effective diagnostic and prog- May and August 2019. Commercially available Simoa assays were
nostic biomarker in AD. used to quantify GFAp, Aβ42, Aβ40, and T-tau (GFAP Discovery,
3. Future directions: Further validation studies will be #102336 and Advantage Neuro 3-plex, #101995). In-house Simoa
needed if these biomarkers are to be incorporated into assays were used to quantify NfL30 and P-tau181.17 More detailed
clinical practice. This includes implementing cut-offs and information on plasma biomarker assay performance is provided in
harmonizing inter-laboratory handling. Table S1.

2.5 Cognition

Communicative Disorders and Stroke–Alzheimer’s Disease and Related Global cognition was assessed using longitudinal Mini-Mental State
Disorders Association (NINCDS-ADRDA) criteria.22 MCI was defined Examination (MMSE), tested at an average [median] of 4 time points
according to the Petersen criteria.23 Conversion from MCI-AD was (interquartile range [IQR] 3–5], with data extending up to 4 years (aver-
defined as fulfilling MCI criteria at baseline, but at a later visit meeting age years 2.4, ± 0.9 years).
the criteria for probable AD dementia. Patients were excluded if
they had significant psychiatric or unstable somatic illness. Further
information on study design, enrollment, and inclusion and exclusion 2.6 Statistical analyses
criteria have been described elsewhere20,24 and is further described in
the Supplement. All participants gave written informed consent, with Demographic, clinical, and plasma biomarker findings were compared
the study approved by local ethical review boards in each participating between groups using Kruskal-Wallis, Mann-Whitney, or Fisher exact
country. tests. Diagnostic accuracy of plasma markers was assessed using
age-adjusted area-under-the-curve (AUC) values from receiver-
operating characteristic (ROC) analyses (AD dementia vs CU and MCI;
2.2 MRI acquisition and processing MCI vs CU; MCI converters vs non-converters; and MCI converters
vs CU). Differences in AUCs were evaluated using bootstrapping
T1-weighted structural MR images (baseline, 3 and 12 months) were (n = 1000).31–33 In a second step, we compared the best-performing
acquired using a sagittal 3D magnetization-prepared rapid gradient- plasma biomarker (based on AUC) to a model incorporating all plasma
echo (MP-RAGE) sequence, using six different 1.5T MR systems (four biomarkers using logistic regression for each of the five contrasts
General Electric [General Electric Healthcare, Milwaukee, WI]; one used in the ROC analyses. Models were then compared using the
Siemens [Siemens Medical Solutions, Erlangen, Germany]; and one Akaike information criterion (AIC),34,35 where a difference of ≥2
Picker [General Electric Healthcare]). All data were preprocessed points indicates a better model fit. Linear mixed-effects (LME) models
through the HiveDB database system.25 Full MRI details are described were used to assess (1) change in plasma measures over time and (2)
elsewhere.24 the association between plasma measures at baseline and cognition
4 SIMRÉN ET AL .

TA B L E 1 Baseline demographics and clinical characteristics

MCI
Characteristic CU Total Converters Non-converters AD P-value
No. 99 107 19 88 103
†
Age, years (SD) 73 (6.14) 74.47 (5.89) 73.21 (6.62) 74.74 (5.73) 76.35 (5.76) <.001
Sex, female/male (% females) 53/46 (53.5) 56/51 (52.2) 8/11 (42.1) 48/40 (54.5) 63/40 (61.2) .370
†‡
Education, years (SD) 11.23 (4.8) 8.97 (4.28) 9.26 (4.81) 8.91 (4.18) 7.82 (3.66) <.001
APOE ε4 status, pos./neg. (% pos.) 31/68 (31.3)† 39/68 (36.4)† 12/6 (66.7) 56/27 (32.5) 58/45 (56.3) <.01
‡† †
MMSE score, mean (SD) 29.07 (1.26) 27.21 (1.82) 26.58 (2.06) 27.34 (1.75) 21.07 (4.42) <.001
‡† †
CDR score, mean (SD) 0.04 (0.13) 0.50 (0.07) 0.53 (0.11) 0.49 (0.05) 1.05 (0.51) <.001
P-tau181, mean (pg/mL) (SD) 8.85 (4.48)‡ † 13.13 (6.21)† 17.13 (6.19)§ 12.26 (5.89) 19.43 (7.57) <.001
†‡ †
NfL, mean (pg/mL) (SD) 18.35 (8.68) 25.96 (15.56) 26.51 (9.26) 25.84 (16.65) 32.47 (15.29) <.001
Aβ42, mean (pg/mL) (SD) 10.07 (2.71) 10.31 (2.52) 10.67 (3.12) 10.24 (2.42) 9.43 (2.91) <.05
Aβ42/Aβ40, mean (SD) 0.037 (0.006)† 0.036 (0.006)† 0.035 (0.004) 0.037 (0.006) 0.032 (0.008) <.001
† †
T-tau, mean (pg/mL) (SD) 2.36 (1.07) 2.69 (1.08) 2.97 (0.93) 2.64 (1.10) 3.21 (2.48) <.001
GFAp, mean (pg/mL) (SD) 125.23 (73.76)† 147.81 (81.14)† 176.46 (107.2) 143.84 (77.24) 219.04 (136.1) <.001

Abbreviations: AD, AD dementia; CDR, cognitive dementia rating scale; CU, cognitively unimpaired. Data are given as mean (SD). Demographic factors
and clinical characteristics were assessed using Fisher exact test across the whole group, and then to compare between two groups. Kruskal-Wallis test
was used to assess continuous variables (MCI, CU, and AD), and Mann-Whitney test (MCI converters vs MCI non-converters). Significant differences in
plasma biomarker concentrations are after accounting for the effects of age, educational level, APOE ε4 status, and sex.; GFAp, glial fibrillary acidic pro-
tein; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NfL, neurofilament light; P-tau181, phosphorylated tau 181; T-tau, total tau;
Aβ42/40, amyloid-β42/40.
† = significantly different from AD, P < 0.05; ‡ = significantly different from MCI, P < 0.05. Significantly different from MCI non-converters, P < 0.05 = § (only
comparing with MCI converters).

(MMSE). These had subject-specific intercepts and slopes and included 3.2 Biomarker concentrations across diagnostic
the interaction between (continuous) time and plasma measures as groups
predictors (adjusted for age, sex, education, and apolipoprotein E
(APOE) ε4 genotype). We also evaluated interactions between plasma At baseline, group comparison revealed increased levels of P-
predictors and diagnosis; when significant, we performed subgroup tau181, NfL, T-tau, and GFAp (P < 0.001), and decreased Aβ42/Aβ40
analyses within diagnostic groups. These analyses were performed in (P < 0.001)—but not Aβ42—in AD dementia compared with CU (Fig-
R (v4.0.0; significance set at P < 0.05, two-sided). ure 1A). Significant differences were also observed between AD
For voxelwise analyses (plasma predictors with GM volume as the dementia and MCI for P-tau181, NfL, T-tau, Aβ42/Aβ40 (P < 0.001),
response variable), the same LME models as described earlier were and GFAp (P < 0.01). Only P-tau181 and NfL were shown to be
also implemented at the voxel level using VoxelStats,36 with findings increased in MCI compared with CU (P < 0.001). In addition, MCI
corrected for multiple comparisons using random field theory (RFT),37 patients who later converted to AD had significantly higher baseline
P < 0.05). concentrations of plasma P-tau181 than MCI patients with a stable dis-
ease state (P < 0.001) (Table 1, Figure 1B). This was not observed for
any other plasma biomarker.
3 RESULTS

3.1 Participants 3.3 Comparative diagnostic performance of

plasma biomarkers
Three hundred nine participants were included: 99 CU controls, 107
MCI (19 who converted to AD dementia at the second visit [12 months ROC curves demonstrating the diagnostic accuracy of plasma biomark-
(±1.3)] and 88 non-converters), and 103 AD dementia patients. Base- ers are shown in Figure 2. Across all comparisons, the highest AUC
line characteristics are summarized in Table 1. Information on follow- values were seen with plasma P-tau181; AD dementia versus CU
up timepoints can be found in Tables S2 and S3. Information on plasma [AUC = 0.91, 95% confidence interval [CI] 0.86-0.96; Figure 2A]; AD
biomarkers and participant characteristics can be found in the Supple- dementia versus MCI [AUC = 0.75, 95% CI 0.67-0.83; Figure 2B]; MCI
mentary results. versus CU [AUC = 0.71, 95% CI 0.63-0.79; Figure 2C]; MCI converters
SIMRÉN ET AL . 5

F I G U R E 1 Baseline comparisons of plasma biomarker concentrations across groups. (A) Plasma concentrations of P-tau181, NfL, Aβ42,
Aβ42/Aβ40, T-tau, and GFAp are shown for CU, MCI, and AD (B) Plasma concentrations of P-tau181, NfL, Aβ42, Aβ42/Aβ40, T-tau, and GFAp are
shown for MCI and MCI-AD patients. Abbreviations: Aβ42/40, amyloid-β42/40; AD, Alzheimer’s disease; CU, cognitively unimpaired; GFAp; glial
fibrillary acidic protein; MCI, mild cognitive impairment; NfL, neurofilament light; P-tau181, phosphorylated tau 181; T-tau, total tau.
MCI = patients who did not progress to any type of dementia; MCI-AD = patients who progressed to AD dementia. All P values are derived from a
univariate linear model, adjusted for the effects of age, sex, APOE ε4 status, and education

versus non-converters [AUC = 0.77, 95% CI, 0.61-0.84; Figure 2D]; and converters (Aβ42/Aβ40, P < 0.05), and MCI converters versus CU (NfL,
MCI converters versus CU [AUC = 0.87, 95% CI 0.74-0.98; Figure 2E]). P < 0.01). For the investigated contrasts, combining P-tau181 with any
In three of these five contrasts, the AUC for plasma P-tau181 was sig- additional biomarkers or APOE did not result in significantly higher
nificantly higher than that of the next best-performing plasma marker: AUC values. Using logistic regression, lower AIC values (indicating
AD dementia versus CU (NfL, P < 0.001), MCI converters versus non- better model fit) were observed for models with P-tau181 only—as
6 SIMRÉN ET AL .

F I G U R E 2 Discriminative performance of biomarkers across diagnostic groups. Receiver-operating characteristics (ROC) curves displaying
the performance of plasma P-tau181, NfL, Aβ42, Aβ42/Aβ40, T-tau, and GFAp to distinguish (A) Alzheimer’s disease (AD) dementia from
cognitively unimpaired (CU), (B) mild cognitive impairment (MCI) from CU, (C) AD dementia from MCI, (D) individuals with MCI at baseline who
later converted to AD dementia (MCI converters) from those who did not convert during the follow-up, and (E) MCI converters versus CU.
Abbreviations: Aβ42/40, amyloid-β42/40; AD, Alzheimer’s disease; AUC, area under the curve; CU, cognitively unimpaired; GFAp, glial fibrillary
acidic protein; MCI, mild cognitive impairment; NfL, neurofilament light; P-tau181, phosphorylated tau 181; T-tau, total tau. MCI = patients who
did not progress to any type of dementia; MCI-AD = patients who progressed to AD dementia.

compared to models combining all plasma biomarkers—across the five was observed for both measures in MCI (P-tau181: β = 4.40, 95% CI
contrasts investigated in the ROC analyses (Table S4). 1.20-7.71; P < 0.01; NfL: β = 3.79 95% CI 1.06-12.14, P < 0.01). No sig-
nificant findings were found for the remaining measures. Furthermore,
none of the biomarkers changed over time in CU.
3.4 Longitudinal trajectories of plasma
biomarkers
3.5 Plasma biomarkers and longitudinal cognitive
Change in plasma measures over time are shown in Figure. In decline
AD dementia, plasma P-tau181 (Figure S1A) and NfL (Figure S1B)
increased significantly over time (β = 5.34, 95% CI 3.19-7.32; P < 0.001; Figure 3 shows the association between baseline plasma measures on
β = 3.28, 95% CI 1.46-6.97, P < 0.01, respectively). A similar pattern longitudinal change in MMSE scores. Panels A-F illustrate estimates
SIMRÉN ET AL . 7

F I G U R E 3 Baseline plasma biomarker concentrations and longitudinal cognitive decline. Plasma measures was used as continuous predictors,
but for visualization purposes, the graphs show results for terciles (high, intermediate, and low concentration): (A) P-tau181, (B) NfL, (C) Aβ42, (D)
Aβ42/Aβ40, (E) T-tau, and (F) GFAp at baseline. Abbreviations: Aβ42/40, amyloid-β42/40; GFAp, glial fibrillary acidic protein; NfL, neurofilament
light; P-tau181, phosphorylated tau 181; T-tau, total tau. Estimated means and 95% confidence interval (CI) from linear mixed-effects models
adjusted for age, sex, educational level, diagnosis, and APOE ε4 genotype.

based on separate LME models for each predictor—adjusted for age, 3.6 Plasma biomarkers and longitudinal brain
sex, education, diagnosis, and APOE ε4—with cognitive trajectories atrophy
shown by plasma terciles (eg, high, moderate, and low). More abnormal
plasma measures were associated with an accelerated decline in MMSE Figure 4A shows voxelwise associations between plasma measures
scores. Because interactions between plasma measures and diagno- at baseline and longitudinal change in GM volume using the pooled
sis were significant, LME models were also applied within diagnostic cohort, adjusted for age, sex, APOE ε4, center, and diagnosis. Aside from
groups (Figure S2). In AD dementia, high P-tau181 at baseline was Aβ measures (Aβ42 > Aβ42/Aβ40), which showed some correlation
associated with low baseline MMSE scores (β = −0.34, 95% CI −0.39 to GM loss in the orbitofrontal cortex (P < 0.05; RFT corrected),
to −0.28; P < 0.001) and accelerated declines in MMSE (β = −0.23, 95% P-tau181 was the only measure that showed significant associations
CI −0.45 to −0.04; P < 0.001, Figure 3A). Plasma NfL and MMSE were (P < 0.05; RFT corrected) to GM volume after correcting for multiple
also significantly associated at baseline (β = −0.30, 95% CI −0.40 to comparisons. Using baseline plasma levels, higher P-tau181 was sig-
−0.28; P < 0.001) but showed only a trend level association longitudi- nificantly related to GM volume loss in the medial and lateral temporal
nally (β = −0.07, 95% CI −0.29 to 0.12; P = 0.05, Figure 3B). In MCI, lobe (P < 0.05; RFT corrected) (Figure 4A). Longitudinally, increasing
even though not significantly associated with baseline MMSE, elevated plasma P-tau181 was predominantly associated with decreasing
P-tau181 at baseline was associated with steeper declines in MMSE GM volume in the medial and lateral temporal lobe (Figure 4B), as
(β = −0.07, 95% CI −0.28 to −0.05; P < 0.001). Plasma Aβ42/Aβ40 well as in the posterior cingulate and opercula (P < 0.05; RFT cor-
levels and MMSE were significantly associated at baseline but showed rected). Findings by diagnostic subgroup (CU, MCI, AD) are shown
only a trend-level association over time (P = 0.07; Figure 3D). No in Figures S3-S5. Within the CU group, the strongest associations
significant associations were found between plasma biomarkers and between longitudinal change of Aβ42 and Aβ42/Aβ40 and GM volume
MMSE in CU participants. Similar findings were obtained when exam- change were seen in the posterior cingulate and prefrontal cortex.
ining the association between plasma biomarkers and annual change Within the AD group, baseline P-tau181, Aβ42/Aβ40, T-tau, and
in MMSE by diagnostic subgroup using linear regression (adjusted for GFAp were associated with GM volume change. Using longitudi-
age, sex, education and APOE ε4 genotype) (Table S5). The relation- nal measures, associations were comparatively sparse and largely
ships between baseline plasma biomarkers and longitudinal changes in seen in anterior temporal and orbitofrontal regions using P-tau181
MMSE by diagnostic subgroup are shown in Figure S2. and NfL.
8 SIMRÉN ET AL .

F I G U R E 4 Baseline and delta biomarker concentrations and longitudinal gray matter (GM) loss. T-statistical parametric maps (P < 0.05,
random field theory (RFT) corrected) showing the association between plasma measures at baseline and longitudinal gray matter loss (change GM
volume; A). B shows the association between longitudinal plasma (change plasma) and change GM volume. Abbreviations: Aβ42/40,
amyloid-β42/40; GFAp, glial fibrillary acidic protein; NfL, neurofilament light; P-tau181, phosphorylated tau 181; T-tau, total tau

4 DISCUSSION tive power of plasma P-tau181 for separating MCI and AD dementia
is relatively low when no information on Aβ is available.17 However,
In this study, we demonstrated that putative plasma biomarkers for in this study, the diagnostic accuracy was high for the separation of
AD pathology, neurodegeneration, and astrogliosis are significantly patients with cognitive impairment (MCI and AD) from CU controls.
altered in AD dementia as compared to CU and MCI. Only P-tau181 We demonstrate that some clinically classified MCI and AD demen-
and NfL were increased in MCI as compared to CU, and P-tau181 tia patients exhibit low plasma P-tau181 concentrations. Our previous
significantly increased in MCI converters versus non-converters. P- studies on plasma P-tau181 do indicate that dementia patients without
tau181 was the only plasma biomarker that could provide high diag- Aβ pathology present with low plasma P-tau181.15–18 Thus we would
nostic accuracy to distinguish AD dementia from CU (AUC = 0.91). Fur- speculate that an individual clinically diagnosed as having AD dementia
thermore, plasma P-tau181 was the only biomarker associated with or MCI but presenting with low plasma P-tau181 may be more indica-
cognitive decline. In analyses comparing baseline plasma measures tive of non-AD pathology than AD pathology in accordance with the
with longitudinal structural MRI, we observed that P-tau181, and to ATN system outlined in the National Institute on Aging–Alzheimer’s
a lesser extent Aβ42/Aβ40, correlated with GM loss. Furthermore, Association (NIA-AA) framework.3 In these cases, further clinical eval-
increasing concentrations of plasma P-tau181 were associated signif- uation may be needed to determine the cause of the cognitive decline.
icantly with GM loss, while this was not seen for the other biomarkers Indeed, we have shown that retrospective plasma P-tau181 values sig-
investigated. nificantly improved the clinical evaluation of AD dementia patients,
In the present study, P-tau181 was the only plasma biomarker to as it predicts neuropathologically confirmed AD with high accuracy.41
show high diagnostic accuracy to identify AD. This diagnostic accuracy, Furthermore, the association of plasma P-tau181 at baseline with both
however, was marginally lower than studies where Aβ and tau (the A cognitive decline and GM atrophy, particularly in brain regions vulner-
and T components of ATN) status3 had been determined using CSF or able to AD pathology, is in line with the idea that tau pathology is an
PET-based biomarkers15–17 (Table S6), where these diagnostic capa- important driver of cortical atrophy42,43 and verifies that plasma P-
bilities are comparable to that of CSF P-tau18138 but inferior to Tau tau181 is indeed tracking meaningful pathophysiological features of
PET.39,40 We also confirmed previous findings showing that the predic- AD. At the group level, baseline and longitudinal plasma levels and GM
SIMRÉN ET AL . 9

atrophy measures were more associated in the AD group, whereas the line Aβ42 and longitudinal changes in Aβ42/Aβ40, however, were not
association of higher P-tau181 and MMSE decline was stronger in the associated with GM loss in MCI or AD but confined to CU partici-
MCI group. These results indicate that plasma P-tau181 concentra- pants, largely in the posterior cingulate and prefrontal cortex. These
tions could be used to monitor individuals during a typically short clini- data show that although this Aβ immunoassay is relatively poor diag-
cal trial period and to predict short-term disease progression in clinical nostically, it can capture relevant pathophysiological aspects of AD
practice. Plasma P-tau181 was the best predictor of cognitive decline, continuum—which is suggested in our data to be strongest preclini-
corroborating the results of an earlier study, which found P-tau181 to cally. The group comparisons of T-tau also demonstrated low diagnos-
be better at predicting conversion to AD dementia in CU or MCI cases tic capability, despite being statistically significant, which is in line with
compared to other plasma biomarkers.15 Thus evidence is emerging previous reports.57,58 Yet, no associations between baseline and lon-
that P-tau181 is a promising biomarker with great potential to identify gitudinal measures of plasma T-tau were found with GM change. In a
and monitor AD in clinical routine and pharmaceutical trials. Yet, it similar manner, significant increases in plasma GFAp were also found in
must be recognized that further information on plasma P-tau181 is AD dementia patients but showed relatively poor diagnostic accuracy,
required to fulfill these applications: optimal preanalytical conditions, was not longitudinally altered, did not associate with cognitive decline,
optimal sample matrices type (eg, EDTA plasma or serum), the influence or was not related to GM loss in the whole sample. Yet, in the AD
of hemolysis, fasting versus non-fasting sample collection, centrifuga- group alone, a significant association was observed between baseline
tion conditions, and how freeze-thaw cycles affect precision. Further- T-tau and GFAp and GM volume change over time. This was stronger
more, although there are emerging data on longitudinal measures of for GFAp, primarily in the cuneus, posterior cingulate, and medial pre-
P-tau181,19,41,44 including this study, how P-tau181 concentrations frontal cortex.
fluctuate in serial sampling over a shorter period is not yet known. A previous study has demonstrated that P-tau181 was the best
Moderate diagnostic capabilities of NfL were observed in this study. plasma indicator of AD risk, which is comparable to CSF P-tau181,
NfL, which is one of the main constituents of the axonal cytoskele- and that the addition of other plasma biomarkers did not improve this
ton, has emerged as a clinically useful biomarker of general neuroax- association.15 We corroborate these findings by demonstrating that
onal injury, found in most neurodegenerative diseases,45–47 includ- plasma P-tau181 is superior predictor of disease state, in any compari-
ing AD,10 as well as in infectious,30 traumatic,48 neuroinflammatory,49 son, and that by adding Aβ42/Aβ40, T-tau, GFAp, or NfL to P-tau181 did
and vascular50 conditions.13 There was also a trend-level association not improve the diagnostic accuracy. Janelidze et al. also showed that
between baseline NfL and longitudinal cognitive decline. However, the combination P-tau181 and Aβ42/Aβ40 was the best predictor of
because the relationship between NfL and MMSE scores at baseline Aβ status. Although this analysis cannot be performed in this study, we
were highly significant, the failure to show the statistically significant provide clear evidence that baseline measure of plasma Aβ is the only
association longitudinally seen in other studies10 might be explained other plasma biomarker associated with AD-related brain changes,
by the number of participants with longitudinal MMSE, possibly in con- which are relatively stronger in unimpaired disease status. Therefore,
junction with the relatively short follow-up time of this study. Base- although plasma P-tau is currently the leading blood biomarker for
line or longitudinal increases of plasma NfL levels were not associated potential clinical application, the combination of P-tau and Aβ species
with GM volume loss, and this is likely due to NfL being a marker for could serve as a superior assessment of preclinical or prodromal AD
white matter damage in cognitively impaired individuals.51,52 Plasma pathology for clinical trials.
NfL may act as a complimentary biomarker to P-tau181 in clinical prac- The strengths of this study include the large sample size at base-
tice. When P-tau181 is seen to be negative, a positive NfL test would line, and the availability of longitudinal clinical assessment, MRI scans,
be indicative of a non-Alzheimer’s neurodegenerative disorder and the and serial plasma measurements. We have also assessed the most
patient should undergo further investigation (eg, fluorodeoxyglucose prominent putative blood biomarkers head-to-head, including plasma
[FDG]-PET or dopamine imaging). A negative test for both modalities P-tau181, which has proven to be a potential biomarker for clini-
would suggest that the individual is unlikely to have progressive neu- cal use. Potential limitations of this study include the use of Simoa-
rodegenerative disease (Parkinson disease being an exception).13,53 based measurements for Aβ42/Aβ40, which has been shown infe-
Although it has been possible to measure plasma Aβ for a number of rior to IPMS methods6,7 or the fully automated Elecsys immunoas-
years, early studies failed to consistently detect a difference between say (Roche Diagnostics)8 in terms of its ability to accurately detect Aβ
AD and CU.54 However, owing to technological advances in immuno- pathology and AD dementia. Similarly, we acknowledge that plasma P-
precipitation coupled with mass spectrometry (IPMS) and ultrasen- tau217 could not be added as a comparison in this study.59 Second,
sitive immunoassay techniques, a large number of studies have now this research cohort had exclusion criteria in place to enrich for AD,
demonstrated meaningful diagnostic accuracies of plasma Aβ measure- which likely resulted in a more homogenous patient selection than one
ments for cerebral amyloid and AD.6–8,55,56 Although we found lower would expected in a primary care setting. Third, we did not have a
levels of Aβ42/Aβ40 in AD dementia, as compared to MCI and CU, gold-standard biomarker (eg, CSF biomarkers, PET, or neuropathologi-
which is in line with previous findings,6,8,55 comparatively low diag- cal assessments) to confirm AD pathology in accordance with the ATN
nostic performance was observed across all comparisons. A signifi- system3 in MCI or AD dementia cases. This would have allowed for
cant association, however, was observed between baseline Aβ42/Aβ40 the examination of how plasma biomarkers, particularly P-tau181, per-
and Aβ42 with GM loss in the orbitofrontal cortex. Furthermore, base- form in suspected non-Alzheimer’s disease (SNAP) pathophysiology. In
10 SIMRÉN ET AL .

addition, there were no CU participants who converted to MCI, which FUNDING

could have further corroborated the prognostic value of P-tau181 at
the earliest stage. This is likely to be, at least in part, due to the rela- Anna Lisa and Brother Björnsson’s Foundation. Dr. Kaj Blennow is
tively high dropout rate in the MCI group. This also applies to the num- supported by the Swedish Research Council (#2017-00915), the
ber of participants with MCI who eventually converted to AD, which Alzheimer Drug Discovery Foundation (ADDF), USA (#RDAPB-
limited the potential to perform more advanced statistics in this group 201809-2016615), the Swedish Alzheimer Foundation (#AF-742881),
alone. Yet, we were still able to demonstrate that higher plasma P- Hjärnfonden, Sweden (#FO2017-0243), the Swedish state under
tau181 levels at MCI indicate a faster progression to AD dementia and the agreement between the Swedish government and the County
corroborates previous studies.15,60 Finally, the comparatively modest Councils, the ALF-agreement (#ALFGBG-715986), and European
associations observed between plasma biomarkers and atrophy mea- Union Joint Program for Neurodegenerative Disorders (JPND2019-
sures precluded generalizability of results at the diagnostic group level 466-236). Dr. Henrik Zetterberg is a Wallenberg Scholar supported
given our modest sample size. Further studies using larger sample sizes by grants from the Swedish Research Council (#2018-02532), the
are needed to properly characterize the relationship between plasma European Research Council (#681712), Swedish State Support for
markers and neurodegeneration across the AD continuum. Clinical Research (#ALFGBG-720931), the Alzheimer Drug Discovery
Foundation (ADDF), USA (#201809-2016862), and the UK Dementia
Research Institute at UCL.
5 CONCLUSIONS

An increasing body of evidence suggests that plasma biomarkers are AUTHOR CONTRIBUTIONS
diagnostically meaningful and are associated with clinical progression
in AD. In this study, we have shown the superior diagnostic and prog- Concept and design: Joel Simrén, Antoine Leuzy, Kaj Blennow, Hen-
nostic utility of plasma P-tau181 as compared to other putative plasma rik Zetterberg, Nicholas J. Ashton, and the AddNeuroMed consor-
biomarkers (Aβ42/Aβ40, NfL, T-tau, and GFAp) of AD. Plasma P-tau181 tium. Acquisition, analysis, or interpretation of data: Original data in this
has potential value in primary care and memory clinics settings as a study acquired by Joel Simrén, Antoine Leuzy, Thomas K. Karikari,
rapid and accurate tool in patient management. Furthermore, plasma Andréa Lessa Benedet, Kaj Blennow, Henrik Zetterberg, and Nicholas
P-tau181, potentially in combination with plasma Aβ42/40 preclini- J. Ashton. Drafting of the manuscript: Joel Simrén, Antoine Leuzy, Kaj
cally, will be of interest for monitoring disease progression in clinical Blennow, Henrik Zetterberg, and Nicholas J. Ashton. Critical revision
trials. of the manuscript for important intellectual content: All authors. Statisti-
cal analysis: Joel Simrén, Antoine Leuzy, Kaj Benedet, and Nicholas J.
Ashton. Obtained funding: Antoine Leuzy, Kaj Blennow, Henrik Zetter-
DISCLOSURES berg, and Nicholas J. Ashton. Administrative, technical, or material sup-
port: Abdul Hye, Eric Westman, and the AddNeuroMed consortium.
Drs. Joel Simrén, Antoine Leuzy, Thomas K. Karikari, Abdul Hye, Supervision: Henrik Zetterberg and Nicholas J. Ashton.
Andréa Lessa Benedet, Juan Lantero Rodriguez, Niklas Mattsson-
Carlgren, Michael Schöll, Patrizia Mecocci, Bruno Vellas, Magda REFERENCES
Tsolaki, Iwona Kloszewska, Simon Lovestone, Dag Aarsland, Pedro 1. Scholl M, Maass A, Mattsson N, et al. Biomarkers for tau pathology. Mol
Rosa-Neto, Eric Westman, and Nicholas J. Ashton report no disclo- Cell Neurosci. 2019;97:18-33.
2. Ashton NJ, Scholl M, Heurling K, et al. Update on biomarkers for amy-
sures. Dr. Hilkka Soininen has served in consulting or advisory board
loid pathology in Alzheimer’s disease. Biomark Med. 2018;12(7):799-
for ACImmune, MERCK, and Novo Nordisk. Dr. Oskar Hansson has 812.
acquired research support (for the institution) from Roche, Pfizer, GE 3. Jack CR Jr, Bennett DA, Blennow K, et al. NIA-AA Research Frame-
Healthcare, Biogen, AVID Radiopharmaceuticals, and Euroimmun, work: toward a biological definition of Alzheimer’s disease. Alzheimers
Dement. 2018;14(4):535-562.
and has received consultancy or speaker fees (paid to the institution)
4. Dubois B, Feldman HH, Jacova C, et al. Advancing research diagnos-
from Biogen and Roche. Dr. Kaj Blennow has served as a consultant tic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol.
or on advisory boards for Axon, Biogen, CogRx, Lilly, MagQu, Novartis, 2014;13(6):614-629.
and Roche Diagnostics. Dr. Henrik Zetterberg has served at scientific 5. McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of
advisory boards for Denali, Roche Diagnostics, Wave, Samumed, dementia due to Alzheimer’s disease: recommendations from the
National Institute on Aging-Alzheimer’s Association workgroups on
Siemens Healthineers, Pinteon Therapeutics, and CogRx; has given
diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement.
lectures in symposia sponsored by Fujirebio, Alzecure, and Biogen; and 2011;7(3):263-269.
is a co-founder of Brain Biomarker Solutions in Gothenburg AB (BBS), 6. Nakamura A, Kaneko N, Villemagne VL, et al. High performance
which is a part of the GU Ventures Incubator Program (outside submit- plasma amyloid-beta biomarkers for Alzheimer’s disease. Nature.
2018;554(7691):249-254.
ted work). Drs. Henrik Zetterberg and Kaj Blennow are co-founders
7. Schindler SE, Bollinger JG, Ovod V, et al. High-precision plasma beta-
of Brain Biomarker Solutions in Gothenburg, a GU Ventures-based amyloid 42/40 predicts current and future brain amyloidosis. Neurol-
platform company at the University of Gothenburg. ogy. 2019;93(17):e1647-e1659.
SIMRÉN ET AL . 11

8. Palmqvist S, Janelidze S, Stomrud E, et al. Performance of fully auto- 27. Cocosco CA, Zijdenbos AP, Evans AC. A fully automatic and
mated plasma assays as screening tests for Alzheimer disease-related robust brain MRI tissue classification method. Med Image Anal.
beta-amyloid status. JAMA Neurol. 2019;76(9):1060-1069. 2003;7(4):513-527.
9. Oeckl P, Halbgebauer S, Anderl-Straub S, et al. Glial fibrillary acidic 28. Ashburner J, Friston KJ. Voxel-based morphometry–the methods.
protein in serum is increased in Alzheimer’s disease and correlates Neuroimage. 2000;11(6 Pt 1):805-821.
with cognitive impairment. J Alzheimers Dis. 2019;67(2):481-488. 29. Sattlecker M, Khondoker M, Proitsi P, et al. Longitudinal protein
10. Mattsson N, Cullen NC, Andreasson U, Zetterberg H, Blennow K. changes in blood plasma associated with the rate of cognitive decline
Association between longitudinal plasma neurofilament light and neu- in Alzheimer’s disease. J Alzheimers Dis. 2016;49(4):1105-1114.
rodegeneration in patients with Alzheimer disease. JAMA Neurol. 30. Gisslen M, Price RW, Andreasson U, et al. Plasma concentration of the
2019;76(7):791-799. Neurofilament Light Protein (NFL) is a biomarker of CNS injury in HIV
11. Benedet AL, Ashton NJ, Pascoal TA, et al. Plasma neurofilament light infection: a cross-sectional study. EBioMedicine. 2016;3:135-140.
associates with Alzheimer’s disease metabolic decline in amyloid- 31. Hanley JA, McNeil BJ. A method of comparing the areas under receiver
positive individuals. Alzheimers Dement (Amst). 2019;11:679-689. operating characteristic curves derived from the same cases. Radiol-
12. Ashton NJ, Leuzy A, Lim YM, et al. Increased plasma neurofilament ogy. 1983;148(3):839-843.
light chain concentration correlates with severity of post-mortem neu- 32. Hanley JA, McNeil BJ. The meaning and use of the area under
rofibrillary tangle pathology and neurodegeneration. Acta Neuropathol a receiver operating characteristic (ROC) curve. Radiology.
Commun. 2019;7(1):5. 1982;143(1):29-36.
13. Ashton NJ, Janelidze S, Al Ahmad K, Leuzy A, Van der Ende EL, Karikari 33. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for
TK, Diagnostic value of plasma neurofilament light: A multicentre R and S+ to analyze and compare ROC curves. BMC Bioinformatics.
validation study. PREPRINT (Version 1) available at Research Square 2011;12:77.
[+https://doiorg/1021203/rs3rs-63386/v1+]. 31 August 2020. 34. Akaike H. A new look at the statistical model identification. IEEE Trans-
14. Mielke MM, Hagen CE, Xu J, et al. Plasma phospho-tau181 increases actions on Automatic Control. 1974;19(6):716-723.
with Alzheimer’s disease clinical severity and is associated with 35. Olofsen E, Dahan A. Using Akaike’s information theoretic criterion in
tau- and amyloid-positron emission tomography. Alzheimers Dement. mixed-effects modeling of pharmacokinetic data: a simulation study.
2018;14(8):989-997. F1000Res. 2013;2:71.
15. Janelidze S, Mattsson N, Palmqvist S, et al. Plasma P-tau181 in 36. Mathotaarachchi S, Wang S, Shin M, et al. VoxelStats: a MATLAB Pack-
Alzheimer’s disease: relationship to other biomarkers, differential age for Multi-Modal Voxel-Wise Brain Image Analysis. Front Neuroin-
diagnosis, neuropathology and longitudinal progression to Alzheimer’s form. 2016;10:20.
dementia. Nat Med. 2020;26(3):379-386. 37. Worsley KJ, Marrett S, Neelin P, Vandal AC, Friston KJ, Evans
16. Thijssen EH, La Joie R, Wolf A, et al. Diagnostic value of plasma phos- AC. A unified statistical approach for determining significant sig-
phorylated tau181 in Alzheimer’s disease and frontotemporal lobar nals in images of cerebral activation. Hum Brain Mapp. 1996;4(1):58-
degeneration. Nat Med. 2020;26(3):387-397. 73.
17. Karikari TK, Pascoal TA, Ashton NJ, et al. Blood phosphorylated tau 38. Mattsson N, Smith R, Strandberg O, et al. Comparing (18)F-AV-1451
181 as a biomarker for Alzheimer’s disease: a diagnostic perfor- with CSF t-tau and p-tau for diagnosis of Alzheimer disease. Neurology.
mance and prediction modelling study using data from four prospec- 2018;90(5):e388-e395.
tive cohorts. Lancet Neurol. 2020;19(5):422-433. 39. Leuzy A, Smith R, Ossenkoppele R, et al. Diagnostic performance of
18. Benussi A, Karikari TK, Ashton N, et al. Diagnostic and prognostic value RO948 F 18 tau positron emission tomography in the differentiation
of serum NfL and p-Tau181 in frontotemporal lobar degeneration. J Neu- of Alzheimer disease from other neurodegenerative disorders. JAMA
rol Neurosurg Psychiatry. 2020;91(9):960-967. Neurol. 2020;77(8):955-965.
19. Karikari TK, Benedet AL, Ashton NJ, et al. Diagnostic performance 40. Ossenkoppele R, Rabinovici GD, Smith R, et al. Discriminative
and prediction of clinical progression of plasma phospho-tau181 in the accuracy of [18F]flortaucipir positron emission tomography for
Alzheimer’s Disease Neuroimaging Initiative. Mol Psychiatry. 2020. Alzheimer disease vs other neurodegenerative disorders. JAMA.
20. Lovestone S, Francis P, Kloszewska I, et al. AddNeuroMed–the 2018;320(11):1151-1162.
European collaboration for the discovery of novel biomarkers for 41. Lantero Rodriguez J, Karikari TK, Suarez-Calvet M, et al. Plasma p-
Alzheimer’s disease. Ann N Y Acad Sci. 2009;1180:36-46. tau181 accurately predicts Alzheimer’s disease pathology at least
21. Lovestone S, Francis P, Strandgaard K. Biomarkers for disease modifi- 8 years prior to post-mortem and improves the clinical charac-
cation trials–the innovative medicines initiative and AddNeuroMed. J terisation of cognitive decline. Acta Neuropathol. 2020;140(3):267-
Nutr Health Aging. 2007;11(4):359-361. 278.
22. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan 42. La Joie R, Visani AV, Baker SL, et al. Prospective longitudinal atrophy
EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS- in Alzheimer’s disease correlates with the intensity and topography of
ADRDA Work Group under the auspices of department of health baseline tau-PET. Sci Transl Med. 2020;12(524).
and human services task force on Alzheimer’s Disease. Neurology. 43. Gordon BA, McCullough A, Mishra S, et al. Cross-sectional and longi-
1984;34(7):939-944. tudinal atrophy is preferentially associated with tau rather than amy-
23. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. loid beta positron emission tomography pathology. Alzheimers Dement
Mild cognitive impairment: clinical characterization and outcome. Arch (Amst). 2018;10:245-252.
Neurol. 1999;56(3):303-308. 44. Moscoso A, Grothe MJ, Ashton NJ, et al. Time course of phospho-
24. Simmons A, Westman E, Muehlboeck S, et al. MRI measures of rylated tau181 in blood across the Alzheimer’s disease spectrum.
Alzheimer’s disease and the AddNeuroMed study. Ann N Y Acad Sci. medRxiv. 2020. 2020.2007.2013.20152025.
2009;1180:47-55. 45. Rohrer JD, Woollacott IO, Dick KM, et al. Serum neurofilament light
25. Muehlboeck JS, Westman E, Simmons A. TheHiveDB image data man- chain protein is a measure of disease intensity in frontotemporal
agement and analysis framework. Front Neuroinform. 2014;7:49. dementia. Neurology. 2016;87(13):1329-1336.
26. Sled JG, Zijdenbos AP, Evans AC. A nonparametric method for auto- 46. Verde F, Steinacker P, Weishaupt JH, et al. Neurofilament light chain in
matic correction of intensity nonuniformity in MRI data. IEEE Trans serum for the diagnosis of amyotrophic lateral sclerosis. J Neurol Neu-
Med Imaging. 1998;17(1):87-97. rosurg Psychiatry. 2019;90(2):157-164.
12 SIMRÉN ET AL .

47. Rojas JC, Karydas A, Bang J, et al. Plasma neurofilament light chain 56. Verberk IMW, Slot RE, Verfaillie SCJ, et al. Plasma amyloid as pre-
predicts progression in progressive supranuclear palsy. Ann Clin Transl screener for the earliest Alzheimer pathological changes. Ann Neurol.
Neurol. 2016;3(3):216-225. 2018;84(5):648-658.
48. Shahim P, Zetterberg H, Tegner Y, Blennow K. Serum neurofilament 57. Mattsson N, Zetterberg H, Janelidze S, et al. Plasma tau in Alzheimer
light as a biomarker for mild traumatic brain injury in contact sports. disease. Neurology. 2016;87(17):1827-1835.
Neurology. 2017;88(19):1788-1794. 58. Zetterberg H, Wilson D, Andreasson U, et al. Plasma tau levels in
49. Siller N, Kuhle J, Muthuraman M, et al. Serum neurofilament light chain Alzheimer’s disease. Alzheimers Res Ther. 2013;5(2):9.
is a biomarker of acute and chronic neuronal damage in early multiple 59. Palmqvist S, Janelidze S, Quiroz YT, et al. Discriminative accuracy of
sclerosis. Mult Scler. 2019;25(5):678-686. plasma phospho-tau217 for Alzheimer disease vs other neurodegen-
50. Gattringer T, Pinter D, Enzinger C, et al. Serum neurofilament erative disorders. JAMA. 2020;324(8):772-781.
light is sensitive to active cerebral small vessel disease. Neurology. 60. Karikari TK, Benedet AL, Ashton NJ, et al. Diagnostic performance
2017;89(20):2108-2114. and prediction of clinical progression of plasma phospho-tau181
51. Moore EE, Hohman TJ, Badami FS, et al. Neurofilament relates to white in the Alzheimer’s disease neuroimaging initiative. medRxiv. 2020.
matter microstructure in older adults. Neurobiol Aging. 2018;70:233- 2020.2007.2015.20154237.
241.
52. Zetterberg H, Skillback T, Mattsson N, et al. Association of cere- SUPPORTING INFORMATION
brospinal fluid neurofilament light concentration with Alzheimer dis- Additional supporting information may be found online in the Support-
ease progression. JAMA Neurol. 2016;73(1):60-67.
ing Information section at the end of the article.
53. Hansson O, Janelidze S, Hall S, et al. Blood-based NfL: a biomarker
for differential diagnosis of parkinsonian disorder. Neurology.
2017;88(10):930-937.
54. Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers
for the diagnosis of Alzheimer’s disease: a systematic review and meta- How to cite this article: Simrén J, Leuzy A, Karikari TK, et al.
analysis. Lancet Neurol. 2016;15(7):673-684. The diagnostic and prognostic capabilities of plasma
55. Janelidze S, Stomrud E, Palmqvist S, et al. Plasma beta-amyloid biomarkers in Alzheimer’s disease. Alzheimer’s Dement.
in Alzheimer’s disease and vascular disease. Sci Rep. 2016;6:
2021;1-12. https://doi.org/10.1002/alz.12283
26801.