Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

Flora H. Duits; Kirsten E. J. Wesenhagen; Laura Ekblad; Emma Wolters; Eline A. J. Willemse; Philip Scheltens; Wiesje M. van der Flier; Charlotte E. Teunissen; Pieter Jelle Visser; Betty M. Tijms

doi:10.1186/s13195-020-00713-3

Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

Duits, Flora H.; Wesenhagen, Kirsten E. J.; Ekblad, Laura; Wolters, Emma; Willemse, Eline A. J.; , ; Scheltens, Philip; van der Flier, Wiesje M.; Teunissen, Charlotte E.; Visser, Pieter Jelle; Tijms, Betty M. 2021-01-04 00:00:00 Background: As Alzheimer’s disease (AD) pathology presents decades before dementia manifests, unbiased biomarker cut-points may more closely reflect presence of pathology than clinically defined cut-points. Currently, unbiased cerebrospinal fluid (CSF) tau cut-points are lacking. Methods: We investigated CSF t-tau and p-tau cut-points across the clinical spectrum using Gaussian mixture modelling, in two independent cohorts (Amsterdam Dementia Cohort and ADNI). Results: Individuals with normal cognition (NC) (total n = 1111), mild cognitive impairment (MCI) (total n = 1213) and Alzheimer’s disease dementia (AD) (total n = 1524) were included. In both cohorts, four CSF t- and p-tau distributions and three corresponding cut-points were identified. Increasingly high tau subgroups were characterized by steeper MMSE decline and higher progression risk to AD (cohort/platform-dependent HR, t-tau 1.9–21.3; p-tau 2.2–9.5). Limitations: The number of subjects in some subgroups and subanalyses was small, especially in the highest tau subgroup and in tau PET analyses. Conclusions: In two independent cohorts, t-tau and p-tau levels showed four subgroups. Increasingly high tau subgroups were associated with faster clinical decline, suggesting our approach may aid in more precise prognoses. Keywords: Alzheimer’s disease, CSF tau, Gaussian mixture modelling, Prognosis * Correspondence: k.wesenhagen@amsterdamumc.nl Part of the data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: https://adni.loni.usc.edu/wpcontent/uploads/ how_to_apply/ADNI_Acknowledgement_List.pdf. Flora H. Duits and Kirsten E. J. Wesenhagen contributed equally to this work. Department of Neurology, Alzheimer Center Amsterdam, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, the Netherlands Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 2 of 25 Background distributions) was higher than clinically based cut- Abnormal levels of amyloid-β 1-42 (Aβ42), total tau (t- points, resulting in more sensitive detection of prede- tau) and tau phosphorylated at threonine 181 (p-tau-181) mentia AD [13–16]. As of yet, however, it remains un- are biomarkers for the presence of Alzheimer’s disease clear whether it is also possible to detect unbiased cut- (AD) pathology in the brain [1], and part of established re- points in t-tau and p-tau levels. search criteria for AD across the cognitive continuum [2, High t-tau levels in the cerebrospinal fluid (CSF) are 3]. Classification schemes based on biomarkers depend on thought to reflect neuronal degeneration or injury, and cut-points, and different approaches exist to determine elevated t-tau levels can be found in the CSF in various such cut-points. The most often used traditional approach conditions involving neuronal death, for example after determines cut-points by optimizing the sensitivity and an acute stroke. In contrast, p-tau-181 is presumed to specificity to detect clinical AD-type dementia compared reflect the formation of phosphorylated tau in the brain to controls [4–6]. However, approaches that use clinical and to represent more specifically the formation of labels as outcomes may not be optimal, because clinical la- neurofibrillary tangles, one of the neuropathological hall- bels do not optimally reflect the absence or presence of marks of AD [17, 18]. As tau pathology is a hallmark of AD pathology: For example, almost 30% of cognitively in- AD, it can be hypothesized that similarly to amyloid, t- tact individuals in their seventies have AD pathology [7], and p-tau levels may be a mixture of values belonging to and up to 20% of clinical AD dementia cases do not show normal and affected individuals, from which unbiased AD pathology at neuropathological examination [8–11]. cut-points might be determined. As such, cut-point based on clinical labels can be biased. The objective of this study was to investigate whether Gaussian mixture modelling provides an approach to subgroups can be identified in CSF t- and p-tau levels determine cut-points independent of clinical information using Gaussian mixture modelling and to determine cut- [12]. This approach is based on the notion that the dis- points. We characterized tau subgroups in terms of clin- tribution of biomarker values in a population is a mix- ical and biological characteristics and longitudinal trajec- ture of values belonging to subpopulations, i.e. normal tories of cognitive decline. We repeated analyses in the and affected individuals. Previous studies using this ap- independent ADNI cohort to determine the robustness proach have found a bimodal distribution of Aβ42 levels, of the identified subgroups and tested stability of group of which the cut-point (i.e. the intersection of these membership by studying longitudinal changes in t-tau Table 1 Participant characteristics of the Amsterdam Dementia Cohort (ADC) and ADNI cohorts ADC ADNI Characteristic NC MCI AD dementia NC MCI AD dementia N = 740 N = 591 N = 1296 N = 371 N = 622 N = 228 a c a c MMSE, mean ± SD 28.2 ± 1.8 26.5 ± 2.4 20.5 ± 5 29.1 ± 1.2 27.7 ± 1.8 23.3 ± 2 a a a b Age, mean ± SD 59.6 ± 8.9 66.4 ± 8.2 66.2 ± 8.1 73.8 ± 5.9 72.4 ± 7.5 74.9 ± 8.1 a c a c Female, n (%) 306 (41.4%) 217 (36.7%) 674 (52%) 195 (52.6%) 255 (41%) 95 (41.7%) a c a c APOE e4 carrier, n (%) NC 258 (36.3%) 287 (52.4%) 791 (65.3%) 103 (27.8%) 307 (49.4%) 154 (67.5%) a c Innotest: T-tau (pg/ml), mean ± SD 296.4 ± 200.8 466.4 ± 303.6 716.6 ± 417.1 n.a. n.a. n.a. a c Innotest: P-tau (pg/ml), mean ± SD 48.4 ± 22.7 66.8 ± 33.6 87.6 ± 39.5 n.a. n.a. n.a. a c Innotest: Aβ42 (pg/ml), mean ± SD 1071.2 ± 246.9 859.1 ± 288.1 648.4 ± 166.6 n.a. n.a. n.a. a c Innotest: Abnormal Aβ42 (< 813 pg/ml), n (%) 124 (16.8%) 326 (55.2%) 1173 (90.5%) n.a. n.a. n.a. a c Luminex T-tau (pg/ml), mean ± SD n.a. n.a. n.a. 67.4 ± 32.8 90.4 ± 54.8 126.6 ± 61.4 a c Luminex P-tau (pg/ml), mean ± SD n.a. n.a. n.a. 32.4 ± 18.8 39.2 ± 23.7 51.6 ± 30.7 a c Luminex Aβ42 (pg/ml), mean ± SD n.a. n.a. n.a. 201.7 ± 51.9 171.2 ± 52.5 139.6 ± 38.8 a c Luminex Abnormal Aβ42 (< 192 pg/ml), n(%) n.a. n.a. n.a. 156 (42%) 403 (64.8%) 210 (92.1%) a c Elecsys T-tau (pg/ml), mean ± SD n.a. n.a. n.a. 238.5 ± 90 284.9 ± 126.7 370.2 ± 144.4 a c Elecsys P-tau (pg/ml), mean ± SD n.a. n.a. n.a. 21.9 ± 9.2 27.6 ± 14.4 36.9 ± 15.7 a c Elecsys Aβ42 (pg/ml), mean ± SD n.a. n.a. n.a. 1337 ± 647.9 1020.7 ± 554.8 694.6 ± 420.7 a c Elecsys Abnormal Aβ42 (< 880 pg/ml), n (%) n.a. n.a. n.a. 104 (28%) 325 (53%) 189 (85%) n.a. not available, NC cognitively normal, MCI mild cognitive impairment, AD Alzheimer’s disease Differs from NC with p < .05 Differs from MCI with p < .001 Differs from MCI and NC with p < .001 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 3 of 25 and p-tau levels. Finally, we compared subgroups on tau www.adni-info.org) was used for validation of the results. PET uptake that was available for a subset of individuals ADNI started in 2003 as a public-private collaboration in ADNI. under the supervision of Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI is to study Methods whether serial magnetic resonance imaging (MRI), posi- We investigated the existence of CSF t- and p-tau sub- tron emission tomography (PET), other biological markers groups in data from two independent clinical cohorts. The and clinical and neuropsychological measures can be com- memory clinic-based Amsterdam Dementia Cohort bined to measure the progression of mild cognitive im- (ADC) was used for testing our hypothesis [19], and the pairment (MCI) and early Alzheimer’s disease (AD). Alzheimer’s Disease Neuroimaging Initiative (ADNI; Please see www.adni-info.org for the latest information. a) Amsterdam Dementia Cohort Amsterdam Dementia Cohort 0 0 0 1000 2000 3000 0 100 200 300 T−tau pg/ml P−tau pg/ml b) ADNI Luminex ADNI Luminex 0 0 0 100 200 300 400 0 50 100 150 200 T−tau pg/ml P−tau pg/ml c) ADNI Elecsys ADNI Elecsys 0 0 200 400 600 800 25 50 75 100 T−tau pg/ml P−tau pg/ml Normal amyloid Abnormal amyloid Subtype 1 Subtype 2 Subtype 3 Subtype 4 Fig. 1 Tetramodal distributions in t-tau and p-tau levels in ADC and ADNI. Levels in ADC are shown in (a); levels in ADNI are shown in (b) for Luminex and (c) for Elecsys assay. Grey colours in the distributions reflect for those tau levels the number of individuals with abnormal amyloid levels Count Count Count Count Count Count Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 4 of 25 The institutional review boards of all participating institu- further assessed whether subgroups showed differences tions approved the procedures for this study. Written in- in cognitive decline, stratifying subjects based on their formed consent was obtained from all participants or baseline cognitive state (defined as NC, MCI or demen- surrogates. Participant selection and CSF analyses are tia). First, decline in MMSE (outcome) was assessed with summarized below; for more detailed methods and infor- linear mixed models using the R package “lmer4”,includ- mation on Apolipoprotein E (APOE) genotyping and PET ing the main terms time and tau subgroup, and inter- imaging, see Additional file 1. action terms time*tau subgroup. For individuals without dementia at baseline, Cox proportional hazards models Participants were used to compare the rate of progression from NC to In short, patients from the ADC who visited our mem- MCI or AD dementia and from MCI to AD dementia be- ory clinic between November 2000 and December 2016 tween tau subgroups. We ran 5 models: (1) without covar- were selected (n = 2724) if they had baseline CSF tau iates; (2) including age, sex and educational level; (3) measurements available and had subjective cognitive de- model 2 + amyloid status; (4) model 3 + baseline cognitive cline (considered as normal cognition (NC)), mild cogni- state; and (5) model 4 + APOE-e4 carriership (dichotom- tive impairment (MCI) or AD dementia. Participants ous). For the Cox proportional hazards models, data from from ADNI who had baseline CSF biomarkers available were selected (n = 1221) for the replication analyses if Table 2 Consistency of subgroup labelling between t-tau and they met the study-specific criteria of NC, MCI or de- p-tau (ADC and ADNI), and across platforms (ADNI) mentia. A subset of 619 individuals in ADNI (51%; 183 Biomarker: Platform ADC p-tau Innotest NC, 345 MCI and 91 with AD dementia) with available ADC T-tau: Innotest Subgroup 1 2 3 4 follow-up CSF measures were selected for longitudinal analyses. 1 960 83 0 0 2 140 661 58 0 CSF biomarkers 3 6 209 399 18 In ADC, CSF biomarkers (β-amyloid , hTAU-Ag, (1-42) 40 0 33 60 and phospo-tau 181P) were assessed with INNOTEST P-tau: Luminex (Fujirebio, Ghent, Belgium) on a routine basis as de- T-tau: Luminex Subgroup 1 2 3 4 scribed before [20]. In ADNI, CSF biomarkers were ana- lysed using a multiplex xMAP Luminex platform 1 210 108 13 1 (Luminex Corp) with immunoassay kit-based reagents 2 116 261 89 2 (INNO-BIA Alzbio3; Innogenetics) [21](n = 1213 partic- 3 3 159 138 8 ipants), and on Elecsys (Roche, Basel, Switserland) [21] 4 0 10 76 19 (n = 1193 participants, overlap with Luminex 98%). P-tau: Elecsys T-tau: Elecsys Subgroup 1 2 3 4 Statistical analysis Gaussian mixture modelling was used to identify cut- 1 290 0 0 0 points in the distribution of t-tau and p-tau values. First, 2 186 309 0 0 the number of distributions that best described the data 3 0 145 185 2 was determined with the R boot.comp function. This 40 0 14 61 function sequentially tests increasing number of compo- T-tau: Elecsys nents in the data using parametric bootstrapping of the T-tau: Luminex Subgroup 1 2 3 4 likelihood ratio (i.e. likelihood of x components vs. likeli- hood of having one more component, i.e. x + 1), until 1 256 70 1 0 the null hypothesis cannot be rejected anymore (p > 2 35 381 43 1 0.05, i.e. no improvement of additional component for 3 0 44 255 4 model fit). Then, we identified data-driven cut-points as 40 0 33 70 the points where the lines of two fitted Gaussian distri- P-tau: Elecsys butions intersected. Using these cut-points, we labelled P-tau: Luminex Subgroup 1 2 3 4 subjects according to tau subgroups. Next, within each cohort, we compared subgroups based on demographi- 1 287 36 0 0 cal, clinical and biological characteristics with ANOVA 2 167 296 63 2 or chi-square tests, when appropriate. For a subset of in- 3 21 119 121 48 dividuals with available repeated mini-mental state 40 1 13 12 examination (MMSE) and/or clinical follow-up, we Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 5 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 T-tau: Innotest ADC N All 1043 859 632 93 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 349 350–671 672–1380 > 1380 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 576 (78%) 135 (18%) 24 (3%) 5 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 262 (44%) 207 (35%) 118 (20%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 205 (16%) 517 (40%) 490 (38%) 84 (6%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 57 (9.9%) 47 (34.8%) 16 (66.7%) 4 (80%) 2.71E−12 6.49E−15 7.84E−05 4.00E−02 6.73E−01 9.90E−01 n (%) MCI 74 (28.2%) 140 (67.6%) 108 (91.5%) 4 (100%) 2.45E−16 6.64E−29 6.01E−02 1.31E−05 9.90E−01 9.90E−01 AD dementia 146 (71.2%) 472 (91.3%) 474 (96.7%) 81 (96.4%) 5.86E−11 1.08E−21 2.76E−05 2.96E−03 9.83E−01 9.90E−01 MMSE, mean ± NC 28.2 ± 1.8 28.2 ± 1.6 27.4 ± 2.1 27.8 ± 1.9 1.00E+00 1.34E−01 9.65E−01 1.61E−01 9.62E−01 9.62E−01 SD MCI 26.7 ± 2.4 26.6 ± 2.2 25.9 ± 2.8 25.2 ± 2.6 9.85E−01 2.53E−02 6.41E−01 7.27E−02 6.84E−01 9.47E−01 AD dementia 21.3 ± 4.6 20.8 ± 5 20.2 ± 5.1 18.8 ± 5.1 5.53E−01 3.94E−02 7.13E−04 2.82E−01 5.16E−03 8.41E−02 Age, mean ± SD NC 58.2 ± 8.5 64.2 ± 8 66.1 ± 9.7 65.8 ± 11.7 0.00E+00 4.83E−05 1.86E−01 7.42E−01 9.75E−01 1.00E+00 MCI 64 ± 8.4 67.9 ± 7.6 69.1 ± 7.1 69.9 ± 12 1.24E−06 5.96E−08 4.54E−01 5.28E−01 9.58E−01 9.97E−01 AD dementia 66.5 ± 8 66.5 ± 7.9 65.7 ± 8.3 65.7 ± 8.1 1.00E+00 6.10E−01 8.52E−01 3.68E−01 8.11E−01 1.00E+00 Female, n (%) NC 234 (40.6%) 57 (42.2%) 11 (45.8%) 4 (80%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 65 (24.8%) 83 (40.1%) 67 (56.8%) 2 (50%) 3.53E−03 1.71E−08 9.90E−01 3.21E−02 9.90E−01 9.90E−01 AD dementia 81 (39.5%) 263 (50.9%) 277 (56.5%) 53 (63.1%) 4.51E−02 3.63E−04 2.58E−03 4.94E−01 2.98E−01 9.90E−01 APOE e4 carrier, NC 183 (33.1%) 58 (45%) 15 (62.5%) 2 (40%) 8.88E−02 3.56E−02 9.90E−01 1.00E+00 9.90E−01 9.90E−01 n (%) MCI 92 (36.9%) 115 (59.6%) 77 (75.5%) 3 (75%) 2.15E−05 6.98E−10 9.90E−01 5.64E−02 1.00E+00 9.90E−01 AD dementia 113 (58.9%) 313 (65.3%) 310 (67.5%) 55 (67.1%) 8.18E−01 2.55E−01 9.90E−01 1.00E+00 1.00E+00 9.90E−01 T-tau: Luminex ADNI N All 335 468 310 108 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 54 55–95 96–171 > 171 s n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 154 (42%) 151 (41%) 63 (17%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 168 (27%) 249 (40%) 146 (23%) 59 (9%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 13 (6%) 68 (30%) 101 (44%) 46 (20%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 54 (35.1%) 55 (36.4%) 45 (71.4%) 2 (66.7%) 9.90E−01 1.34E−05 9.90E−01 3.58E−05 9.90E−01 9.90E−01 n (%) MCI 53 (31.5%) 155 (62.2%) 138 (94.5%) 57 (96.6%) 8.68E−09 9.22E−29 1.73E−16 2.12E−11 4.08E−06 9.90E−01 AD dementia 9 (69.2%) 59 (86.8%) 97 (96%) 45 (97.8%) 9.90E−01 1.70E−02 4.10E−02 3.26E−01 5.23E−01 9.90E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 6 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 MMSE, mean ± NC 29.1 ± 1.2 29 ± 1 29.1 ± 1.4 29 ± 1 9.87E−01 1.00E+00 1.00E+00 9.97E−01 1.00E+00 1.00E+00 SD MCI 28.3 ± 1.6 27.8 ± 1.8 27.3 ± 1.8 26.8 ± 1.9 2.27E−02 1.41E−05 6.60E−07 7.24E−02 1.66E−03 2.86E−01 AD dementia 24.4 ± 1.3 23.6 ± 2 23 ± 2 23.3 ± 2 5.75E−01 1.02E−01 2.87E−01 2.60E−01 8.11E−01 9.11E−01 Age, mean ± SD NC 72.5 ± 5.6 74 ± 5.8 76.3 ± 6.2 74.1 ± 3.4 9.69E−02 5.53E−05 9.64E−01 3.49E−02 1.00E+00 9.10E−01 MCI 70.7 ± 7.5 72.6 ± 7.6 73.7 ± 7.4 73.3 ± 7 5.69E−02 2.46E−03 1.15E−01 4.89E−01 9.34E−01 9.78E−01 AD dementia 78.4 ± 6.6 75.6 ± 8.7 75.1 ± 7.5 72.2 ± 8.6 6.66E−01 4.98E−01 6.62E−02 9.72E−01 1.08E−01 1.73E−01 Female, n (%) NC 81 (52.6%) 78 (51.7%) 33 (52.4%) 3 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 68 (40.5%) 89 (35.7%) 64 (43.8%) 34 (57.6%) 1.00E+00 9.90E−01 2.01E−01 8.23E−01 1.98E−02 6.12E−01 AD dementia 3 (23.1%) 18 (26.5%) 45 (44.6%) 29 (63%) 1.00E+00 9.90E−01 1.51E−01 1.58E−01 1.30E−03 3.44E−01 APOE e4 carrier, NC 32 (20.8%) 44 (29.1%) 26 (41.3%) 1 (33.3%) 7.19E−01 2.05E−02 9.90E−01 7.07E−01 1.00E+00 9.90E−01 n (%) MCI 43 (25.6%) 117 (47%) 103 (70.5%) 44 (74.6%) 1.01E−04 2.45E−14 4.77E−10 5.27E−05 1.46E−03 9.90E−01 AD dementia 7 (53.8%) 43 (63.2%) 75 (74.3%) 29 (63%) 9.90E−01 9.90E−01 1.00E+00 9.90E−01 9.90E−01 9.90E−01 T-tau: Elecsys ADNI N All 291 495 332 75 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 192 193–311 312–514 > 514 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 126 (35%) 164 (45%) 69 (19%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 152 (25%) 258 (42%) 157 (26%) 41 (7%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 13 (6%) 73 (33%) 106 (48%) 30 (14%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 40 (31.7%) 67 (40.9%) 40 (58%) 3 (75%) 8.49E−01 3.93E−03 9.90E−01 1.47E−01 9.90E−01 9.90E−01 n (%) MCI 52 (34.2%) 153 (59.3%) 146 (93%) 40 (97.6%) 9.25E−06 1.06E−25 1.23E−11 1.67E−12 2.77E−05 9.90E−01 AD dementia 8 (61.5%) 67 (91.8%) 100 (94.3%) 29 (96.7%) 6.33E−02 4.91E−03 6.03E−02 9.90E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 29.2 ± 1.1 29 ± 1.1 29 ± 1.4 29.2 ± 1 5.42E−01 8.40E−01 9.99E−01 9.95E−01 9.66E−01 9.79E−01 SD MCI 28.1 ± 1.7 27.9 ± 1.8 27.2 ± 1.8 27.1 ± 1.8 6.43E−01 1.70E−05 6.98E−03 2.46E−04 3.86E−02 9.98E−01 AD dementia 24.5 ± 1.5 23.5 ± 1.9 23 ± 2.1 23.5 ± 1.9 3.88E−01 7.21E−02 4.57E−01 3.92E−01 1.00E+00 6.83E−01 Age, mean ± SD NC 71.6 ± 5.5 74.4 ± 5.8 76.1 ± 6.2 76.7 ± 6 3.60E−04 1.86E−06 2.95E−01 1.56E−01 8.47E−01 9.96E−01 MCI 70.9 ± 7.9 72.4 ± 7.4 73.9 ± 7.4 71.7 ± 6.9 2.02E−01 3.72E−03 9.26E−01 2.48E−01 9.47E−01 3.81E−01 AD dementia 77.7 ± 7.3 75.6 ± 8.1 74.3 ± 8.1 73.5 ± 8.7 8.13E−01 4.81E−01 4.09E−01 7.40E−01 6.61E−01 9.69E−01 Female, n (%) NC 69 (54.8%) 80 (48.8%) 37 (53.6%) 4 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 7.74E−01 9.90E−01 MCI 55 (36.2%) 99 (38.4%) 72 (45.9%) 23 (56.1%) 9.90E−01 6.41E−01 2.01E−01 9.69E−01 2.90E−01 9.90E−01 AD dementia 3 (23.1%) 23 (31.5%) 51 (48.1%) 17 (56.7%) 9.90E−01 9.40E−01 5.40E−01 2.35E−01 1.86E−01 9.90E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 7 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 APOE e4 carrier, NC 26 (20.6%) 45 (27.4%) 29 (42%) 1 (25%) 9.90E−01 1.58E−02 1.00E+00 2.54E−01 9.90E−01 9.90E−01 n (%) MCI 41 (27%) 120 (46.5%) 105 (66.9%) 32 (78%) 8.40E−04 2.90E−11 3.90E−08 4.95E−04 2.03E−03 9.90E−01 AD dementia 5 (38.5%) 51 (69.9%) 75 (70.8%) 19 (63.3%) 3.67E−01 2.55E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Innotest ADC N All 1106 953 490 78 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 56 57–96 97–159 > 159 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 564 (76%) 153 (21%) 19 (3%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 272 (46%) 214 (36%) 98 (17%) 7 (1%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 270 (21%) 586 (45%) 373 (29%) 67 (5%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 61 (10.8%) 44 (28.8%) 16 (84.2%) 3 (75%) 3.22E−07 2.14E−18 6.88E−03 3.57E−05 8.98E−01 9.90E−01 n (%) MCI 78 (28.7%) 153 (71.5%) 89 (90.8%) 6 (85.7%) 9.08E−20 6.37E−25 2.78E−02 1.57E−03 9.90E−01 9.90E−01 AD dementia 204 (75.6%) 546 (93.2%) 358 (96%) 65 (97%) 4.83E−12 2.03E−13 1.07E−03 5.59E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 28.2 ± 1.8 28.1 ± 1.6 27.9 ± 2.1 27.5 ± 1.7 9.99E−01 9.53E−01 8.79E−01 9.71E−01 8.94E−01 9.68E−01 SD MCI 26.7 ± 2.3 26.6 ± 2.4 25.7 ± 2.7 27.1 ± 2.1 9.88E−01 1.99E−03 9.60E−01 7.13E−03 9.40E−01 3.93E−01 AD dementia 20.9 ± 5 20.7 ± 4.9 20.3 ± 5.2 19 ± 4.8 9.11E−01 4.53E−01 2.98E−02 7.26E−01 5.43E−02 2.07E−01 Age, mean ± SD NC 58.3 ± 8.6 63.1 ± 8.3 67.1 ± 6.5 68.3 ± 16.8 7.58E−09 7.08E−05 9.30E−02 2.07E−01 6.19E−01 9.94E−01 MCI 64.1 ± 8.2 68 ± 7.9 69.3 ± 6.8 70.2 ± 10.1 3.55E−07 1.67E−07 1.77E−01 5.48E−01 8.89E−01 9.91E−01 AD dementia 66.7 ± 7.8 66.1 ± 8.1 65.6 ± 8 67.4 ± 9.2 7.76E−01 3.62E−01 9.13E−01 8.01E−01 6.02E−01 3.47E−01 Female, n (%) NC 228 (40.4%) 67 (43.8%) 9 (47.4%) 2 (50%) 9.90E−01 1.00E+00 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 73 (26.8%) 85 (39.7%) 54 (55.1%) 5 (71.4%) 2.15E−02 4.97E−06 1.80E−01 9.45E−02 9.90E−01 9.90E−01 AD dementia 118 (43.7%) 302 (51.5%) 215 (57.6%) 39 (58.2%) 2.38E−01 3.89E−03 2.77E−01 4.48E−01 9.90E−01 9.90E−01 APOE e4 carrier, NC 181 (33.3%) 62 (43.1%) 14 (73.7%) 1 (25%) 2.22E−01 4.13E−03 9.90E−01 1.39E−01 9.90E−01 9.90E−01 n (%) MCI 90 (35%) 133 (66.2%) 60 (71.4%) 4 (66.7%) 4.10E−10 6.79E−08 9.90E−01 9.90E−01 9.90E−01 9.90E−01 AD dementia 148 (59.2%) 367 (66.6%) 228 (65.7%) 48 (75%) 3.09E−01 7.44E−01 1.74E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Luminex ADNI N All 329 538 316 30 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 23 24–46 47–99 > 99 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 138 (37%) 166 (45%) 62 (17%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 173 (28%) 266 (43%) 166 (27%) 13 (2%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 18 (8%) 106 (47%) 88 (39%) 14 (6%) n.t. n.t. n.t. n.t. n.t. n.t. Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 8 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 Abnormal ab42, NC 41 (29.7%) 66 (39.8%) 44 (71%) 3 (100%) 5.28E−01 6.79E−07 2.93E−01 3.11E−04 7.84E−01 9.90E−01 n (%) MCI 47 (27.2%) 184 (69.2%) 155 (93.4%) 13 (100%) 9.96E−17 4.99E−34 1.93E−06 3.28E−08 2.30E−01 9.90E−01 AD dementia 11 (61.1%) 98 (92.5%) 85 (96.6%) 14 (100%) 4.36E−03 1.28E−04 1.63E−01 9.90E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 29.1 ± 1.2 29 ± 1.2 29.1 ± 1 30 ± 0 9.35E−01 1.00E+00 5.35E−01 9.83E−01 4.60E−01 5.32E−01 SD MCI 28.2 ± 1.6 27.5 ± 1.9 27.5 ± 1.8 27 ± 1.5 2.41E−04 1.71E−03 7.75E−02 1.00E+00 7.42E−01 7.33E−01 AD dementia 23.6 ± 1.9 23.3 ± 2 23.3 ± 2.1 23.6 ± 1.8 9.56E−01 9.15E−01 1.00E+00 9.94E−01 9.80E−01 9.55E−01 Age, mean ± SD NC 73 ± 5.3 73.7 ± 6.5 75.4 ± 5.6 72.8 ± 2.5 7.76E−01 4.23E−02 1.00E+00 1.98E−01 9.94E−01 8.74E−01 MCI 71.7 ± 7.7 73 ± 7.8 72.6 ± 6.9 69.2 ± 7.8 2.58E−01 6.72E−01 6.78E−01 9.41E−01 2.92E−01 4.12E−01 AD dementia 80.6 ± 7.8 76 ± 7.4 72.9 ± 8.5 72.4 ± 5.3 9.87E−02 9.25E−04 1.76E−02 2.83E−02 3.57E−01 9.96E−01 Female, n (%) NC 69 (50%) 88 (53%) 33 (53.2%) 3 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 65 (37.6%) 108 (40.6%) 73 (44%) 8 (61.5%) 9.90E−01 9.90E−01 9.47E−01 9.90E−01 9.90E−01 9.90E−01 AD dementia 4 (22.2%) 43 (40.6%) 36 (40.9%) 10 (71.4%) 9.90E−01 9.90E−01 9.20E−02 9.90E−01 3.45E−01 3.92E−01 APOE e4 carrier, NC 28 (20.3%) 43 (25.9%) 29 (46.8%) 2 (66.7%) 9.90E−01 1.47E−03 9.90E−01 2.57E−02 9.90E−01 9.90E−01 n (%) MCI 49 (28.3%) 132 (49.6%) 115 (69.3%) 10 (76.9%) 8.90E−05 6.32E−13 5.36E−03 5.41E−04 6.08E−01 9.90E−01 AD dementia 7 (38.9%) 73 (68.9%) 61 (69.3%) 12 (85.7%) 1.71E−01 1.74E−01 1.24E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Elecsys ADNI N All 467 454 199 63 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 21 22–36 37–55 > 55 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 208 (57%) 127 (35%) 24 (7%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 241 (40%) 229 (38%) 105 (17%) 33 (5%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 27 (12%) 98 (44%) 70 (32%) 27 (12%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 63 (30.3%) 61 (48%) 23 (95.8%) 2 (66.7%) 9.91E−03 7.59E−09 1.00E+00 2.49E−04 1.00E+00 1.00E+00 n (%) MCI 83 (34.4%) 173 (75.5%) 102 (97.1%) 33 (100%) 5.17E−18 1.25E−25 2.03E−11 1.99E−05 1.75E−02 1.00E+00 AD dementia 18 (66.7%) 92 (93.9%) 68 (97.1%) 26 (96.3%) 2.61E−03 6.13E−04 8.52E−02 1.00E+00 1.00E+00 1.00E+00 MMSE, mean ± NC 29.1 ± 1.2 28.9 ± 1.2 29.5 ± 0.8 29 ± 1 9.35E−01 1.00E+00 5.35E−01 9.83E−01 4.60E−01 5.32E−01 SD MCI 28.1 ± 1.7 27.7 ± 1.8 27.2 ± 1.8 26.9 ± 1.9 2.41E−04 1.71E−03 7.75E−02 1.00E+00 7.42E−01 7.33E−01 AD dementia 24 ± 1.7 23.3 ± 2 23 ± 2.1 23.6 ± 1.8 9.56E−01 9.15E−01 1.00E+00 9.94E−01 9.80E−01 9.55E−01 Age, mean ± SD NC 72.5 ± 5.5 75.2 ± 6.3 77.3 ± 5.5 74.1 ± 3.4 7.76E−01 4.23E−02 1.00E+00 1.98E−01 9.94E−01 8.74E−01 MCI 70.9 ± 7.6 73.4 ± 7.6 73.4 ± 7.3 72.6 ± 7.1 2.58E−01 6.72E−01 6.78E−01 9.41E−01 2.92E−01 4.12E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 9 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 AD dementia 78.1 ± 7.7 75.2 ± 7.7 73.8 ± 8.4 72.9 ± 8.7 9.87E−02 9.25E−04 1.76E−02 2.83E−02 3.57E−01 9.96E−01 Female, n (%) NC 101 (48.6%) 75 (59.1%) 11 (45.8%) 3 (100%) 4.77E−01 1.00E+00 1.00E+00 1.00E+00 1.00E+00 1.00E+00 MCI 94 (39%) 89 (38.9%) 46 (43.8%) 20 (60.6%) 1.00E+00 1.00E+00 1.79E−01 1.00E+00 1.76E−01 8.25E−01 AD dementia 7 (25.9%) 35 (35.7%) 37 (52.9%) 15 (55.6%) 1.00E+00 1.84E−01 3.15E−01 2.39E−01 6.04E−01 1.00E+00 APOE e4 carrier, NC 45 (21.6%) 41 (32.3%) 14 (58.3%) 1 (33.3%) 2.51E−01 1.50E−03 1.00E−03 1.66E−01 1.00E+00 1.00E+00 n (%) MCI 72 (29.9%) 122 (53.3%) 79 (75.2%) 25 (75.8%) 2.56E−06 7.89E−14 3.92E−06 1.36E−03 1.48E−01 1.00E+00 AD dementia 10 (37%) 72 (73.5%) 52 (74.3%) 16 (59.3%) 5.81E−03 8.60E−03 1.00E+00 1.00E−03 1.00E+00 1.00E+00 All pairwise comparisons are Tukey HSD adjusted for multiple testing n.t. not tested Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 10 of 25 945 subjects was available (357 normal and 588 MCI sub- dementia subjects had lowest baseline MMSE scores, jects). Subsets of individuals in ADNI also had repeated highest proportion of APOE e4 carriers, lowest levels of CSF tau measures available, for which we tested changes Aβ42, and highest levels of tau. MCI subjects had values over time in tau subgroups, or had TAU PET available, in between NC subjects and AD dementia patients. for which we compared tau uptake according to Braak stages between subgroups. All analyses except for Cox Gaussian mixture modelling reveals four subgroups proportional hazard analyses were stratified for baseline Mixture modeling showed that four distributions (i.e. a cognitive state, and adjusted for age and sex, and cognitive tetramodel distribution) best fitted the data for both t- outcomes additionally for level of education [22]. In Cox tau and p-tau levels, with an optimal fit for four distribu- proportional hazard analyses, no stratification for baseline tions (log-likelihood ratio for 3 vs. 4 distributions, for t- cognitive state was performed due to small size of the tau: 97.2, and p-tau: 28.3, both p < 0.001, no further im- resulting groups; instead, baseline cognitive state was provement for 5 distributions: log-likelihood ratio for 5 added as additional covariate. All statistical analyses were vs 4 distributions, for t-tau: 3.9, and p-tau 15, both p > performed in R version 3.6.1 “Action of the Toes”,mixture 0.05; see Additional file 2 for fit statistics of all fitted modelling was performed with the mixtools package (ver- models, and Fig. 1a for a visualisation of the four distri- sion 1.1.0), estimated marginal means and trends were butions). In the ADC (using Innotest), this yielded three computed with the R package “emmeans” v1.4, and sensi- cutoffs (95% confidence interval (CI)), for t-tau—349 tivity and specificity analyses with epiR v.1.0-15. (304–382), 671 (582–834) and 1380 (1260–1505) pg/mL, and for p-tau—56 (46–60), 96 (71–121) and 159 (138– Results 240) (for n per subgroup defined by cut-points, see Ta- Patient characteristics bles 2 and 3). The first cut-points for t-tau (349 pg/mL) Table 1 shows baseline characteristics of the ADC and and p-tau (56 pg/mL) were comparable to the t-tau and ADNI cohorts. Compared to the ADC, subjects in the p-tau cut-points of 375 pg/ml and 52 pg/ml we previ- ADNI cohort were approximately 10 years older and had ously reported [5], and showed similar sensitivity and a lower prevalence of AD dementia and a higher preva- specificity performance to distinguish between clinical lence of MCI. In ADC, subjects with NC were about AD dementia and controls (see Table 4 for sensitivity 7 years younger compared to MCI and AD patients, and and specificity comparisons). Sensitivity and specificity the NC and the MCI subjects were more often male for distinguishing NC vs MCI were also comparable to than AD dementia subjects. In ADNI, MCI subjects those resulting from the clinical cut-point (Table 4). were youngest, and MCI and AD dementia subjects were T-tau and p-tau strongly correlated across the total more often male than NC. In both cohorts, AD group (r = .92, p < .001); however, when comparing Table 4 Sensitivity and specificity for clinical comparisons NC vs AD-type dementia NC vs MCI First cut-point Cut-point (literature) First cut-point Cut-point (literature) Dataset First cut- Cut-point Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity point (literature) T-tau ADC: 349 375 0.84 0.78 0.81 0.82 0.56 0.78 0.52 0.82 Innotest (0.8–0.86) (0.74–0.82) (0.78–0.84) (0.78–0.86) (0.52–0.6) (0.75–0.81) (0.47–0.56) (0.79–0.84) ADNI: 54 93 0.95 0.4 0.64 0.81 0.73 0.42 0.35 0.81 Luminex (0.9–0.98) (0.33–0.48) (0.55–0.72) (0.74–0.86) (0.69–0.76) (0.36–0.47) (0.31–0.39) (0.76–0.84) ADNI: 192 300 0.94 0.33 0.64 0.78 0.76 0.34 0.37 0.77 Elecsys (0.88–0.97) (0.26–0.4) (0.55–0.72) (0.71–0.84) (0.72–0.79) (0.29–0.39) (0.33–0.4) (0.72–0.81) P-tau ADC: 56 52 0.78 0.77 0.83 0.72 0.54 0.76 0.59 0.7 Innotest (0.75–0.81) (0.72–0.81) (0.8–0.86) (0.68–0.77) (0.5–0.58) (0.73–0.79) (0.55–0.63) (0.67–0.73) ADNI: 23 23 0.9 0.4 0.9 0.4 0.72 0.37 0.72 0.37 Luminex (0.84–0.95) (0.33–0.47) (0.84–0.95) (0.33–0.48) (0.68–0.76) (0.32–0.42) (0.68–0.76) (0.32–0.43) ADNI: 21 24 0.84 0.56 0.8 0.67 0.61 0.56 0.5 0.68 Elecsys (0.76–0.9) (0.48–0.63) (0.72–0.87) (0.6–0.74) (0.57–0.65) (0.51–0.61) (0.46–0.54) (0.63–0.72) MCI mild cognitive impairment, NC normal cognition Source: [5] Source: [21] Source: [23] Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 11 of 25 Fig. 2 T-tau and p-tau subgroup comparisons within each cohort, stratified for cognitive state. Left, ADC; middle, ADNI Luminex; right, ADNI Elecsys. Comparisons for MMSE are shown in (a), for age in (b), for proportion female in (c) and for proportion of APOE-e4 carriers in (d). See Table 3 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 12 of 25 classification of individuals based on t-tau and p-tau, point of t-tau (54 pg/mL) with the cut-point of 93 pg/mL concordance was somewhat lower (79%; Table 2). Apply- previously determined for ADNI [21], our new cut-point ing mixture modelling in ADNI showed that similar to for t-tau resulted in higher sensitivity to detect clinical the ADC, a tetramodal distribution best fitted the CSF t- AD dementia versus controls, at the cost of lower speci- tau and p-tau data (log-likelihood ratio for 3 vs 4 distri- ficity (Table 4). The first p-tau cut-point (23 pg/mL) was butions, for t-tau: 25.2 and for p-tau: 54.3, both with p < identical to the cut-point reported in the literature [21]. 0.05, no further improvement for 5 distributions: log- In ADNI, we further repeated analyses on the novel likelihood ratio for 5 vs 4 distributions, for t-tau: 11.2, Elecsys data as an analytical validation, and again observed and p-tau: 20.3, with p = 0.08 and p = 0.05, respectively). a tetramodal distribution for t-tau (log-likelihood ratio for The tetramodal distribution yielded three different cut- 3 vs 4 distributions, for t-tau: 12.4, and p-tau: 19.6, both points for t-tau measured with Luminex (95%CI)—54 with p < 0.05, no further improvement for 5 distributions: (42–68), 95 (68–125) and 171 (146–263) pg/mL respect- log-likelihood ratio for 5 vs 4 distributions, for t-tau: 7.1, ively (Fig. 1b), and for p-tau levels (95%CI)—23 (20–28), and p-tau 10.6, both p > 0.05). The tetramodal distribution 46 (38–57) and 99 (74–124). Comparing the first cut- yielded for t-tau the cut-points (95%CI)—192 (129–235), Fig. 3 Comparison of annual MMSE decline for t-tau and p-tau subgroups, stratified for cognitive state. Left, t-tau; right, p-tau; top, ADC; middle, ADNI Luminex; bottom, ADNI Elecsys. See Table 5 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 13 of 25 312 (239–405) and 515 (417–679), and for p-tau (95%CI)—21 (17–23), 36 (28–43) and 55 (47–68) (Fig. 1c). For the first t- tau cut-point measured with Elecsys (192 pg/mL) compared to the cut- point of 300 pg/mL previ- ously reported for ADNI [23], our new cut-point also yielded higher sensi- tivity but lower specificity for Elecsys t-tau (Table 4). For Elecsys p-tau, the first cut-point of 21 pg/ mL was comparable to a previously reported cut- point of 24 pg/mL [23] and yielded similar sensi- tivity and lower specifi- city estimates (Table 4). In ADNI, 81% of the sub- jects were labelled identi- cally using t-tau labels for Luminex and Elecsys, whereas between- platform correspondence for p-tau labelling was 60% (Table 2). For Lumi- nex, the correlation be- tween t-tau and p-tau was .67 (p <.001) and the correspondence of t-tau and p-tau labelling was 52%. For Elecsys, the cor- relation between t-tau and p-tau was .98 (p <.001), and the corres- pondence of subgroup la- belling was 71%. Clinical and biological characteristics of tau subgroups Gradually higher t- and p-tau subgroups in ADC were characterized by increasingly high preva- lence of abnormal amyl- oid, with the highest two t-tau and p-tau groups consisting for more than 94% of amyloid Table 5 Tau subgroup comparisons on MMSE at first visit and annual change rates Subgroup Baseline estimated marginal means ± SE of p values of pairwise comparisons effect subgroup between subgroups Bio- Cognitive Cohort: N per p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 marker state platform subgroup 1/2/3/4 T-tau NC ADC: Innotest 240/73/13/4 0.428 28.4 ± 0.1 28 ± 0.2 27.3 ± 0.4 23.8 ± 0.9 6.45E−02 1.07E−02 2.00E−07 1.15E−01 2.20E−06 ADNI: Luminex 151/151/60/2 0.231 28.8 ± 0.1 28.8 ± 0.1 28.6 ± 0.1 29.3 ± 0.6 8.90E−01 1.07E−01 5.10E−01 7.92E−02 5.25E−01 ADNI: Elecsys 124/163/67/3 0.725 28.8 ± 0.1 28.8 ± 0.1 28.6 ± 0.1 28.7 ± 0.5 7.25E−01 1.58E−01 8.06E−01 2.20E−01 8.65E−01 MCI ADC: Innotest 199/170/92/4 0.270 26.6 ± 0.2 25 ± 0.2 24.2 ± 0.3 24.2 ± 1.7 7.98E−09 1.13E−11 1.47E−01 2.99E−02 6.39E−01 ADNI: Luminex 162/237/145/56 0.0002 27.9 ± 0.2 26.4 ± 0.2 25.1 ± 0.2 24 ± 0.4 1.36E−06 4.46E−16 4.12E−17 3.32E−05 3.80E−08 ADNI: Elecsys 145/247/154/41 0.0001 27.7 ± 0.2 26.7 ± 0.2 25.1 ± 0.2 23.7 ± 0.5 9.96E−04 2.04E−14 4.56E−14 1.12E−07 3.71E−09 AD dementia ADC: Innotest 122/303/280/38 0.172 20.4 ± 0.4 19.7 ± 0.3 18.9 ± 0.3 17.1 ± 0.8 1.58E−01 3.00E−03 1.58E−04 3.87E−02 1.32E−03 ADNI: Luminex 13/63/95/43 0.031 23.9 ± 0.9 22 ± 0.4 21.6 ± 0.3 20.8 ± 0.5 5.04E−02 1.41E−02 2.35E−03 4.16E−01 5.82E−02 ADNI: Elecsys 13/68/99/28 0.008 25 ± 0.9 21.8 ± 0.4 21.3 ± 0.3 21.3 ± 0.6 1.00E−03 1.33E−04 6.41E−04 3.69E−01 4.77E−01 P-tau NC ADC: Innotest 237/76/15/2 0.120 28.4 ± 0.1 28.2 ± 0.2 27.5 ± 0.4 22.2 ± 1 2.84E−01 2.48E−02 1.10E−08 1.03E−01 3.91E−08 ADNI: Luminex 137/164/59/3 0.883 28.9 ± 0.1 28.7 ± 0.1 28.7 ± 0.1 29.1 ± 0.8 6.64E−02 1.13E−01 7.68E−01 7.89E−01 5.74E−01 ADNI: Elecsys 206/124/24/2 0.367 28.8 ± 0.1 28.7 ± 0.1 28.5 ± 0.2 29.2 ± 0.7 2.81E−01 1.70E−01 5.56E−01 4.27E−01 4.39E−01 MCI ADC: Innotest 208/173/79/5 0.575 26.3 ± 0.2 25.1 ± 0.2 24.3 ± 0.3 23.8 ± 1.4 6.13E−05 7.46E−08 7.69E−02 2.19E−02 3.41E−01 ADNI: Luminex 164/259/161/12 0.009 27.7 ± 0.2 26.3 ± 0.2 25 ± 0.2 23.8 ± 0.8 1.03E−06 8.81E−17 8.02E−06 1.20E−05 4.47E−03 ADNI: Elecsys 228/224/103/32 0.011 27.6 ± 0.2 26.1 ± 0.2 24.6 ± 0.3 23.7 ± 0.5 6.25E−08 3.34E−18 1.74E−12 8.99E−06 1.14E−05 AD dementia ADC: Innotest 149/348/216/30 0.250 20.2 ± 0.4 19.4 ± 0.2 19 ± 0.3 16.3 ± 0.9 7.19E−02 1.25E−02 4.35E−05 2.97E−01 6.82E−04 ADNI: Luminex 17/98/83/14 0.483 23.7 ± 0.8 21.5 ± 0.3 21.5 ± 0.4 21.8 ± 0.9 1.21E−02 1.30E−02 1.20E−01 9.54E−01 7.47E−01 ADNI: Elecsys 26/91/66/25 0.118 23.4 ± 0.7 21.8 ± 0.3 21.2 ± 0.4 21.2 ± 0.7 3.21E−02 4.37E−03 1.91E−02 2.39E−01 4.10E−01 a b c All analyses were stratified for cognitive state and adjusted for age, sex and level of education. Slope that differs from 0 is indicated with when p < .05, when p < .01 or when p < .001 p values of pairwise comparisons Time Subgroup Estimated slopes ± SE of p value of pairwise comparisons of slope NC cognitively normal, MCI mild cognitive impairment, AD Alzheimer’s disease between subgroups effect x Time subgroup differences between subgroups Bio- 3vs4 p value p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 3 vs 4 marker Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 14 of 25 Table 5 Tau subgroup comparisons on MMSE at first visit and annual change rates (Continued) p values of pairwise comparisons Time Subgroup Estimated slopes ± SE of p value of pairwise comparisons of slope between subgroups effect x Time subgroup differences between subgroups Bio- 3vs4 p value p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 3 vs 4 marker a c T-tau 2.54E−04 1.14E−07 1.34E−05 0 ± 0.1 −0.2 ± 0.1 −0.3 ± 0.2 −2.2 ± 0.5 6.16E−02 1.51E−01 2.65E−06 7.06E−01 1.97E−05 8.86E−05 b c 3.06E−01 3.12E−01 3.23E−03 0±0 −0.1 ± 0 − 0.3 ± 0 0.1 ± 0.3 1.38E−01 3.36E−04 5.15E−01 1.21E−02 3.55E−01 1.36E−01 b c 8.82E−01 9.20E−02 1.15E−01 0 ± 0 −0.1 ± 0 −0.2 ± 0 0 ± 0.2 2.32E−01 1.60E−02 9.47E−01 1.20E−01 7.42E−01 4.69E−01 a c c 9.71E−01 2.05E−03 8.94E−13 −0.2 ± 0.1 −1.0 ± 0.1 − 1.3 ± 0.1 −0.6 ± 1 2.48E−10 3.78E−11 6.98E−01 1.35E−01 6.68E−01 5.13E−01 a c c c 1.39E−02 1.74E−47 7.39E−17 −0.2 ± 0.1 −0.7 ± 0.1 −1.2 ± 0.1 − 1.6 ± 0.2 4.68E−05 3.56E−13 1.82E−14 7.80E−05 1.39E−07 1.69E−02 a c c c 9.62E−03 7.02E−43 7.36E−16 −0.2 ± 0.1 − 0.6 ± 0.1 −1.1 ± 0.1 − 1.8 ± 0.2 5.03E−04 2.90E−11 3.83E−14 6.21E−05 5.28E−09 9.50E−04 c c c c 2.61E−02 1.72E−71 1.73E−04 −1.7 ± 0.2 −2.0 ± 0.1 −2.3 ± 0.1 − 3.5 ± 0.4 1.49E−01 1.55E−02 1.95E−05 2.00E−01 1.92E−04 1.69E−03 a c c c 1.77E−01 3.33E−16 1.29E−01 −1.5 ± 0.8 −2.0 ± 0.4 − 2.4 ± 0.3 − 3.2 ± 0.5 6.20E−01 3.06E−01 6.60E−02 3.67E−01 3.53E−02 1.34E−01 c c c 9.33E−01 8.53E−13 3.68E−02 −0.3 ± 0.8 −2.2 ± 0.4 − 2.4 ± 0.3 − 3.0 ± 0.6 2.48E−02 9.79E−03 4.88E−03 6.24E−01 2.33E−01 3.65E−01 P-tau 3.65E−06 2.80E−07 3.36E−05 −0.1 ± 0.1 −0.1 ± 0.1 −0.2 ± 0.2 −2.5 ± 0.5 3.79E−01 4.50E−01 2.05E−06 7.62E−01 5.29E−06 2.55E−05 5.43E−01 2.10E−01 1.35E−01 0 ± 0 −0.1 ± 0 −0.2 ± 0.1 −0.1 ± 0.4 5.87E−02 4.56E−02 8.32E−01 4.88E−01 9.82E−01 8.85E−01 a c c c 3.15E−01 1.24E−01 8.11E−03 −1.3 ± 0.6 −2.1 ± 0.3 −2.7 ± 0.4 −3.0 ± 0.6 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 c c c b 7.18E−01 2.94E−09 5.39E−08 −0.4 ± 0.1 −0.9 ± 0.1 −1.2 ± 0.1 −2.0 ± 0.7 1.32E−05 1.08E−07 2.23E−02 6.44E−02 1.32E−01 2.75E−01 a c c c 1.95E−01 2.19E−27 3.33E−16 −0.2 ± 0.1 −0.7 ± 0.1 −1.3 ± 0.1 −2.0 ± 0.3 1.04E−04 2.19E−16 3.76E−07 1.31E−07 1.50E−04 6.60E−02 b b c c 1.32E−01 1.76E−46 5.50E−19 −0.2 ± 0.1 −0.8 ± 0.1 −1.4 ± 0.1 −1.8 ± 0.2 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 c c c c 4.03E−03 2.75E−65 9.95E−05 −1.8 ± 0.2 −2 ± 0.1 −2.4 ± 0.1 −3.7 ± 0.4 3.14E−01 5.30E−03 7.50E−05 2.36E−02 2.48E−04 6.23E−03 c c a 7.26E−01 1.62E−10 9.00E−02 −0.7 ± 0.7 −2.5 ± 0.3 −2.6 ± 0.3 −2.1 ± 0.8 2.00E−02 1.33E−02 2.06E−01 7.23E−01 6.68E−01 5.53E−01 a c c c 9.88E−01 1.88E−18 9.79E−02 −1.3 ± 0.6 −2.1 ± 0.3 −2.7 ± 0.4 −3.0 ± 0.6 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 15 of 25 abnormal participants. This relationship also held for associations of higher t- and p-tau subgroups with amyl- lower t-tau values, with a higher prevalence of abnormal oid, cognitive state, and demographic factors were amyloid in the second subgroup than the lowest tau sub- mostly reproduced in ADNI. group. T- and p-tau subgroups were also associated with cognitive state, with lower subgroups containing the Rates of cognitive decline over time depend on tau highest proportion of cognitively normal participants, subgroups while highest subgroups contained more demented par- We further studied whether subjects across tau sub- ticipants (Table 3). Therefore, we stratified subsequent groups differed in rates of cognitive decline, as measured comparisons between tau subgroups for cognitive state. with the MMSE stratified for cognitive state. In ADC, Average MMSE was lower for higher tau subgroups, tau subgroups were not associated with cognitive decline with the strongest effects observed in AD-type dementia in MCI or NC; however, in the dementia phase, higher (Fig. 2; Table 3). Tau subgroups also differed in demo- tau subgroups were characterized by faster cognitive de- graphic factors, including age (on average lower in the cline on MMSE (Fig. 3; Table 4). In ADNI, faster MMSE lowest tau subgroup in NC and MCI), sex (higher pro- decline with higher tau subgroups in dementia was portion of women in higher t-tau and p-tau subgroups), reproduced. While in ADC no association between tau and APOE e4 carriership (higher prevalence in higher t- subgroups and MMSE decline was found for participants tau and p-tau subgroups) (Fig. 2; Table 3). The Fig. 4 Proportional Hazard curves for progression to dementia in initially non-demented individuals in different tau subgroups. T-tau subgroups are shown in (a) and p-tau subgroups in (b); left, ADC; middle, ADNI Luminex; right, ADNI Elecsys. See Table 6 for statistical descriptions Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 16 of 25 Table 6 Cox proportional hazard models for time to progression in individuals with NC or MCI Model 1: no Model 2: + sex, Model 3: model Model 4: model Model 5: model 4+ covariates age, education 2 + amyloid status 3 + diagnosis APOE-e4 carriership HR (95%CI) p HR (95%CI) sex, p HR (95%CI) sex, p HR (95%CI) sex, age, p HR (95%CI) sex, age, edu, p age, edu age, edu, abeta edu, abeta, diagnosis abeta, diagnosis, APOE-e4 T-tau subgroups Subgroup 1 All Reference Subgroup 2 ADC: Innotest 6.3 (4.3, 9.2) 5.79E 6 (3.9, 9.2) 1.32E 3.4 (2.2, 5.3) 5.96E 3.4 (2.1, 5.3) 1.08E 3.3 (2.1, 5.3) 4.31E −21 −16 −08 −07 −07 ADNI: Luminex 2.1 (1.4, 3) 8.76E 2 (1.4, 3) 1.58E 1.7 (1.2, 2.5) 5.84E 1.6 (1.1, 2.4) 1.24E 1.5 (1, 2.2) 2.98E −05 −04 −03 −02 −02 ADNI: Elecsys 1.9 (1.3, 2.8) 9.27E 1.8 (1.2, 2.7) 2.44E 1.8 (1.2, 2.7) 2.56E 1.8 (1.2, 2.6) 3.71E 1.7 (1.1, 2.5) 9.74E −04 −03 −03 −03 −03 Subgroup 3 ADC: Innotest 14.6 (9.8, 6.14E 13.2 (8.3, 20.8) 3.31E 6.2 (3.8, 10) 1.96E 6.0 (3.7, 9.8) 6.96E 6.1 (3.7, 10.1) 1.86E 21.8) −39 −28 −13 −13 −12 ADNI: Luminex 5.2 (3.6, 7.6) 1.31E 4.9 (3.4, 7.2) 3.53E 2.8 (1.9, 4.1) 2.86E 2.6 (1.8, 3.9) 1.57E 2.4 (1.6, 3.6) 1.06E −18 −17 −07 −06 −05 ADNI: Elecsys 4.4 (3, 6.4) 6.15E 4 (2.7, 5.9) 2.05E 3.1 (2.1, 4.5) 1.88E 2.8 (1.9, 4.2) 1.91E 2.7 (1.8, 4) 1.60E −14 −12 −08 −07 −06 Subgroup 4 ADC: Innotest 21.3 (7.5, 9.81E 15 (4.4, 50.9) 1.43E 6.5 (1.9, 22.3) 2.93E 6.6 (1.9, 22.7) 2.70E 6.4 (1.9, 22.2) 3.30E 60.6) −09 −05 −03 −03 −03 ADNI: Luminex 6.7 (4.2, 10.7) 1.01E 6.6 (4.1, 10.6) 3.89E 3.4 (2.1, 5.6) 8.31E 2.8 (1.7, 4.5) 6.50E 2.6 (1.5, 4.2) 2.63E −15 −15 −07 −05 −04 ADNI: Elecsys 7.1 (4.3, 11.7) 2.96E 7.2 (4.3, 12) 2.37E 4.6 (2.7, 7.7) 6.71E 3.9 (2.3, 6.6) 2.99E 3.5 (2.1, 6.1) 3.59E −14 −14 −09 −07 −06 P-tau subgroups Subgroup 1 All Reference Subgroup 2 ADC: Innotest 4.4 (3.1, 6.2) 4.13E 3.9 (2.7, 5.7) 2.35E 2.1 (1.4, 3.2) 1.71E 2.1 (1.4, 3.1) 2.28E 2.1 (1.4, 3.2) 4.88E −17 −12 −04 −04 −04 ADNI: Luminex 2.2 (1.5, 3) 7.78E 2.1 (1.5, 3) 1.38E 1.5 (1.0, 2.1) 3.30E 1.4 (1.0, 2.0) 5.85E 1.4 (1, 2) 7.19E −06 −05 −02 −02 −02 ADNI: Elecsys 2.6 (1.9, 3.6) 9.65E 2.6 (1.9, 3.5) 3.83E 2.2 (1.6, 3.0) 1.73E 2.0 (1.5, 2.8) 1.22E 2 (1.4, 2.7) 4.34E −10 −09 −06 −05 −05 Subgroup 3 ADC: Innotest 10.2 (7, 14.8) 1.48E 8.4 (5.6, 12.8) 1.10E 3.7 (2.4, 5.8) 6.58E 3.6 (2.3, 5.6) 2.03E 3.5 (2.2, 5.6) 1.02E −33 −23 −09 −08 −07 ADNI: Luminex 4.4 (3.1, 6.3) 1.81E 4.3 (3.0, 6.2) 4.37E 2.3 (1.6, 3.3) 1.80E 2.2 (1.5, 3.2) 7.21E 2 (1.4, 3) 2.88E −16 −16 −05 −05 −04 ADNI: Elecsys 5.0 (3.6, 7.1) 7.09E 4.7 (3.3, 6.7) 4.35E 3.0 (2.1, 4.3) 4.19E 2.7 (1.9, 3.9) 1.32E 2.6 (1.8, 3.7) 8.22E −20 −18 −09 −07 −07 Subgroup 4 ADC: Innotest 9.5 (3.4, 26.5) 1.69E 4.6 (1.4, 15.5) 1.28E 2.1 (0.6, 7.2) 2.19E 2.1 (0.6, 7.1) 2.29E 4 (1.2, 13.8) 2.72E −05 −02 −01 −01 −02 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 17 of 25 Table 6 Cox proportional hazard models for time to progression in individuals with NC or MCI (Continued) Model 1: no Model 2: + sex, Model 3: model Model 4: model Model 5: model 4+ covariates age, education 2 + amyloid status 3 + diagnosis APOE-e4 carriership HR (95%CI) p HR (95%CI) sex, p HR (95%CI) sex, p HR (95%CI) sex, age, p HR (95%CI) sex, age, edu, p age, edu age, edu, abeta edu, abeta, diagnosis abeta, diagnosis, APOE-e4 ADNI: Luminex 5.0 (2.4, 10.6) 2.83E 5.8 (2.7, 12.4) 5.32E 2.5 (1.2, 5.5) 1.76E 2.3 (1.0, 4.9) 3.81E 2.1 (0.9, 4.5) 6.85E −05 −06 −02 −02 −02 ADNI: Elecsys 7.7 (4.7, 12.6) 3.47E 8.2 (4.9, 13.4) 1.24E 4.7 (2.8, 7.7) 3.52E 3.9 (2.3, 6.6) 2.51E 3.6 (2.1, 6.1) 2.11E −16 −16 −09 −07 −06 Amyloid status Amyloid abnormal ADC: Innotest (< 10.8 (7.3, 8.71E 8.9 (5.9, 13.4) 6.65E n.t. 7.8 (5.0, 12.0) 1.88E 7.2 (4.5, 11.5) 5.27E 813 pg/ml) 15.9) −33 −25 −20 −17 ADNI: Luminex (< 5.2 (3.8, 7.0) 2.29E 5.0 (3.6, 6.8) 4.91E n.t. 4.4 (3.2, 6.0) 1.79E 3.8 (2.7, 5.3) 4.16E 192 pg/ml) −25 −24 −20 −15 ADNI: Elecsys (< 4.2 (3.2, 5.5) 1.60E 4.1 (3.1, 5.3) 8.27E n.t. 3.5 (2.7, 4.6) 1.30E 3 (2.2, 4) 2.51E 880 pg/ml) −26 −25 −19 −13 Continuous predictors Continuous ab1-42 (z ADC: Innotest 0.3 (0.2, 0.4) 3.31E 0.3 (0.3, 0.4) 8.16E 0.6 (0.4, 0.8) 2.84E 0.6 (0.4,0.9) 4.17E 0.6 (0.4, 0.8) 4.75E score; HR per SD) −36 −27 −03 −03 −03 ADNI: Luminex 0.5 (0.4, 0.5) 1.71E 0.5 (0.4, 0.5) 6.62E 0.6 (0.5, 0.8) 7.12E 0.7 (0.5, 0.9) 2.19E 0.7 (0.6, 0.9) 1.79E −30 −23 −05 −03 −02 ADNI: Elecsys 0.4 (0.3, 0.5) 2.24E 0.4 (0.3, 0.5) 7.05E 0.6 (0.4, 0.8) 1.14E 0.6 (0.5,0.8) 1.56E 0.6 (0.5, 0.8) 5.41E −23 −22 −04 −04 −04 Continuous t-tau (z score; ADC: Innotest 1.6 (1.5, 1.7) 7.67E 1.9 (1.7, 2.1) 3.60E 1.5 (1.3, 1.7) 2.42E 1.5 (1.3, 1.7) 4.57E 1.6 (1.4, 1.8) 3.46E HR per SD) −58 −32 −11 −11 −11 ADNI: Luminex 1.7 (1.6, 1.9) 7.95E 1.7 (1.6, 1.9) 5.61E 1.4 (1.3, 1.6) 1.80E 1.4 (1.2, 1.5) 1.15E 1.3 (1.2, 1.5) 1.69E −32 −31 −11 −08 −07 ADNI: Elecsys 1.6 (1.5, 1.8) 8.36E 1.7 (1.5, 1.8) 1.39E 1.4 (1.3, 1.6) 2.41E 1.4 (1.3, 1.5) 1.51E 1.4 (1.2, 1.5) 4.90E −27 −25 −13 −10 −09 Continuous p-tau (z ADC: Innotest 1.8 (1.7, 2.0) 2.92E 1.7 (1.5, 1.9) 3.31E 1.4 (1.2, 1.6) 1.11E 1.4 (1.2, 1.6) 3.56E 1.5 (1.3, 1.7) 3.03E score; HR per SD) −42 −22 −07 −07 −08 ADNI: Luminex 1.5 (1.4, 1.6) 5.58E 1.6 (1.4, 1.7) 1.75E 1.3 (1.2, 1.5) 1.67E 1.3 (1.2, 1.4) 2.46E 1.3 (1.1, 1.4) 4.28E −24 −25 −08 −07 −06 ADNI: Elecsys 1.7 (1.5, 1.8) 3.36E 1.7 (1.5, 1.9) 1.06E 1.5 (1.3, 1.6) 1.72E 1.4 (1.3, 1.5) 9.62E 1.4 (1.2, 1.5) 3.05E −31 −29 −13 −11 −09 N.t. not tested Source: Tijms BM et al., Clinical Chemistry. 2018;64(3):576–585 Source: [21] Source: Hansson O et al., Alzheimer’s & Dementia. 2018;14(11):1470–1481 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 18 of 25 Table 7 Frequencies of individuals remaining or changing and still showed higher HRs for progression to AD-type subgroup over time from baseline (rows) dementia compared to the first tau subgroup (HR Biomarker: Platform Subgroup at last (95%CI) = 2.1 (1.4, 3.0), p < .001). measurement T-tau: Luminex Baseline subgroup 1 2 3 4 Longitudinal changes in tau concentrations in ADNI 1 110 35 1 0 Examining transitions over time to higher tau groups in ADNI, we observed that the majority of individuals for 2 22 188 42 0 both Luminex and Elecsys t-tau subgroups remained in 3 0 18 139 15 the same subgroup as first measured (Luminex: 472 40 1 13 35 (76% of 619); Elecsys: 443 (76% of 586); Table 7; see T-tau: Elecsys Baseline subgroup 1 2 3 4 Table 8 and Fig. 5 for continuous results). Of individuals 192 25 1 0 who changed, the majority shifted to one tau group 2 11 206 36 0 higher (Table 8). 3 0 17 147 13 Comparison with tau PET in ADNI 40 1 4 33 Finally, we compared CSF tau subgroups on tau PET up- P-tau: Luminex Baseline subgroup 1 2 3 4 take values available for 345 individuals (235 NC; 93 1 103 63 11 0 MCI; 28 dementia; of note, these included n = 232 new 2 24 164 91 5 CSF observations not included in mixture analyses). Fig- 3 2 23 100 13 ure 6 shows that tau PET uptake increased with higher t-tau and p-tau subgroups. For all Braak regions, the up- 40 1 6 6 take of the highest two tau subgroups was significantly P-tau: Elecsys Baseline subgroup 1 2 3 4 higher than the lowest two (or three) subgroups 1 185 33 0 0 (Table 9). The second lowest t-tau subgroup also 2 9 197 25 0 showed higher average tau uptake in Braak I/II brain 3 0 14 75 12 areas compared to subgroup 1, and the second lowest p- 40 0 4 33 tau subgroup in addition also to Braak III/IV and V/VI compared to subgroup 1. with MCI, in ADNI, higher tau subgroups in MCI were associated with MMSE decline (Table 5). Discussion Next, we tested for individuals without dementia (i.e. In this study, we used Gaussian mixture modelling to NC and MCI) whether tau-subgroups differed in terms determine unbiased cut-points for CSF tau levels. We of progression to MCI or AD-type dementia. In the identified three cut-points resulting in four different dis- ADC, 46/381 (12%) of NC patients showed clinical pro- tributions, and the cut-point between the lowest two gression either to MCI (n = 39) within 2.3 ± 1.6 years or subgroups corresponded closely to an existing clinically to AD-type dementia (n = 16) in 4.5 ± 4.0 years, and 178/ defined cut-point [21]. Furthermore, two additional tau 591 (30%) of MCI patients progressed to AD-type de- groups with highest t- and p-tau levels were discovered mentia in 2.4 ± 1.6 years. Across the total group of non- in the data. We similarly observed four distributions in demented subjects, hazard ratios (HRs) increased with the independent ADNI cohort, and despite differences increasing tau or p-tau subgroups compared to the low- between ADC and ADNI in cohort composition, tau est tau or p-tau subgroups (Fig. 4; Table 6). Repeating subgroups showed similar clinical and biological charac- analyses including covariates sex, age and education level teristics in both study cohorts. These findings suggest (model 2), amyloid status (model 3), baseline cognitive that t-tau and p-tau levels may not necessarily reflect state (model 4) and APOE-e4 carriership (model 5) gen- disease stage, but possibly different biological subtypes erated largely similar results for t-tau subgroups, al- of AD. though HRs were somewhat attenuated. Results were Tau is an intracellular protein playing an important largely consistent for ADNI albeit with somewhat lower role in microtubule assembly and stabilization in axons HR values (Table 6), where 65/371 (17.5%) NC showed [24]. Hyperphosphorylation disturbs its function, result- clinical progression either to MCI (n = 47) within 3 ± 9 ing in the formation of aggregates or neurofibrillary tan- years or to AD-type dementia (n = 18) in 8 ± 3 years, and gles, which is one of the hallmarks of AD pathology. 212/622 (34%) MCI individuals to AD-type dementia in Still, the precise factors influencing t- and p-tau CSF 4.1 ± 2.3 years. Of note is that in ADNI, individuals in levels remain unclear. Measures correlated highly, and the second Luminex t-tau subgroup had levels below the even though subgroup labelling showed moderate con- official cut-point defined by ADNI (i.e. 93 pg/ml [21]) cordance, t-tau and p-tau subgroups showed similar Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 19 of 25 Table 8 T-tau and p-tau subgroup comparisons on annual change in CSF t-tau and p-tau values Effect Interaction Annual change (se) for each subgroup p values pairwise comparisons in slope time subgroup x time differences between subgroups Biomarker: Fp value Fp value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 platform a c b T-tau: 14 1.74E−04 0.2 8.96E−01 1.72 (0.75) 2.03 (0.57) 2.2 (0.71) 1.08 (1.44) 7.45E−01 6.43E−01 6.94E−01 8.50E−01 Luminex a c c a T-tau: 37 2.55E−09 2.5 5.98E−02 3.34 (1.68) 4.8 (1.16) 8.71 (1.46) 8.36 (3.31) 4.74E−01 1.62E−02 1.77E−01 3.67E−02 Elecsys c c b c P-tau: 2 1.63E−01 15 2.93E−09 3.17 (0.72) 4.67 (0.6) 2.89 (0.88) −15.23 (2.95) 1.11E−01 8.07E−01 2.58E−09 9.58E−02 Luminex a c a P-tau: 0.005 2.85E−02 1.5 2.13E−01 0.26 (0.13) 0.48 (0.13) 0.49 (0.21) −0.22 (0.37) 2.12E−01 3.28E−01 2.25E−01 9.68E−01 Elecsys a b c CSF t- and p-tau values are in pg/ml. Baseline effects are reported in the last columns. Bold font highlights significant effects. Slope that differs from 0 is indicated with when p < .05, when p < .01 or when p < .001 SE standard error Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 20 of 25 Table 8 T-tau and p-tau subgroup comparisons on annual change in CSF t-tau and p-tau values (Continued) p values pairwise comparisons in slope Baseline Baseline estimated marginal means (SE) for p values pairwise comparisons in baseline estimates between differences between subgroups effect tau group each subgroup subgroups Biomarker: 2vs4 3 vs4 Fp value Subgroup Subgroup Subgroup Subgroup 1vs2 1vs3 1vs4 2vs3 2vs4 3vs4 platform 1 2 3 4 T-tau: 5.41E−01 4.86E−01 1318 7.48E−273 44.6 (1.89) 76.4 (1.43) 129.3 (1.73) 221.9 (3.24) 1.65E−36 1.83E 6.43E 1.68E−87 3.96E 3.88E Luminex −137 −206 −177 −96 T-tau: 3.11E−01 9.23E−01 1330 1.30E−261 162.4 (4.49) 254.1 (3.05) 392.8 (3.64) 616.8 (7.86) 7.82E−53 1.58E 1.28E 7.20E 1.89E 4.47E Elecsys −168 −215 −117 −184 −99 P-tau: 8.98E−11 6.95E−09 808 6.11E−222 21.8 (1.06) 39.1 (0.84) 65.5 (1.22) 109.5 (3.99) 9.63E−33 2.98E 1.14E−73 5.97E−56 9.88E−54 7.98E Luminex −101 −24 P-tau: 7.35E−02 9.28E−02 1456 3.10E−270 16.8 (0.36) 28.4 (0.34) 43.4 (0.52) 66.0 (0.90) 1.96E 7.62E 7.62E 1.64E−90 1.76E 6.98E Elecsys −181 −221 −221 −167 −78 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 21 of 25 ADNI Elecsys t-tau a) ADNI Luminex t-tau CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 300 700 0 0 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 250 600 0 0 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 250 600 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 250 600 01234567 01234567 01234567 01234567 01234567 01234567 Follow−up (years) Follow−up (years) b) ADNI Luminex p-tau ADNI Elecsys p-tau CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 100 100 80 80 60 60 40 40 20 20 0 0 01234567 01234567 01234567 01234567 01234567 01234567 Follow−up (years) Follow−up (years) Fig. 5 Changes over time in t-tau and p-tau levels, stratified for tau subgroup and cognitive state. Changes in t-tau levels are shown in (a) and in p-tau levels in (b). Left, ADNI Luminex; right, ADNI Elecsys. See Table 7 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia T−tau pg/ml P−tau pg/ml P−tau pg/ml T−tau pg/ml Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 22 of 25 Fig. 6 Comparison of t-tau and p-tau subgroups in tau PET uptake according to Braak stages. Tau PET uptake for t-tau subgroups are shown in (a)and for p-tau subgroups in (b). Left, ADNI Luminex; right, ADNI Elecsys. See Table 9 for statistical descriptions. SUVr, standardized uptake value ratio Table 9 T-tau and p-tau subgroup comparison on tau PET uptake T-tau T-tau subgroup estimated marginal means (SE) T-tau subgroup pairwise comparisons p value subgroup effect Tau PET Subgroup Fp value Subgroup Subgroup Subgroup Subgroup 1vs2 1vs3 1vs4 2vs3 2vs4 3vs4 SUVr 1 2 3 4 Braak I/II T-tau 28 3.69E 1.18 (0.016) 1.22 (0.013) 1.36 (0.018) 1.51 (0.052) 2.91E 1.91E 1.65E 8.36E 1.03E 4.78E −16 −02 −12 −09 −09 −07 −03 P-tau 37.6 5.14E 1.18 (0.012) 1.27 (0.014) 1.43 (0.024) 1.47 (0.049) 1.02E 1.97E 1.95E 3.40E 1.22E 4.67E−01 −21 −06 −18 −08 −08 −04 Braak III/IV T-tau 30 2.64E 1.13 (0.019) 1.17 (0.016) 1.34 (0.022) 1.56 (0.061) 5.19E−02 2.98E 6.58E 3.73E 3.08E 7.24E −17 −12 −11 −09 −09 −04 P-tau 31 1.04E 1.13 (0.014) 1.23 (0.017) 1.38 (0.029) 1.51 (0.059) 1.79E 1.50E 7.23E 7.87E 3.88E 4.08E −17 −05 −13 −10 −06 −06 −02 Braak V/VI T-tau 16 7.49E 1.05 (0.018) 1.09 (0.015) 1.21 (0.021) 1.32 (0.058) 1.50E−01 3.80E 1.01E 2.69E 8.12E 5.80E−02 −10 −08 −05 −06 −05 P-tau 14.2 1.04E 1.06 (0.014) 1.13 (0.016) 1.22 (0.028) 1.27 (0.056) 4.88E 1.21E 1.97E 4.22E 1.49E 4.21E−01 −08 −04 −07 −04 −03 −02 Bold font indicates significant group difference Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 23 of 25 differences in tau PET uptake. Previous studies compar- than disease severity per se. This is supported by the ob- ing CSF tau measures with tau PET have been inconsist- servation that higher tau subgroups showed increasing ent [25–27]. Together with our results, it remains proportions of APOE e4 carriers, the strongest genetic unclear whether CSF t-tau and p-tau reflect similar or risk factor for AD [35]. Previous studies have also re- different aspects of neuronal injury. Higher levels of t- ported higher levels of tau in APOE-e4 carriers, also in and p-tau might result from passive release into extra- predementia stages [36]; however, also see [37] where cellular space due to neuronal death which increases tau levels were similar between carriers and non- with worse disease severity. However, tau is also actively carriers. Other genetic risk factors may contribute to dif- secreted by neurons as part of normal physiology [28] ferences in tau levels as well, as another study reported and can increase in the presence of amyloid pathology that a polygenic risk score, including SNPs with moder- [29]. The majority of individuals remained in their t-tau ate strength to detect AD, was strongly related to t-tau subgroup over time, suggesting that at least part of their and p-tau levels, also after correcting for APOE [38]. levels do not depend on disease stage, but perhaps re- This suggests that multiple genetic risk factors may ex- flect other biological aspects. The relative lack of change plain variability between individual tau levels. More over time in tau levels within individuals seems at odds studies with large sample sizes are needed to further in- with the idea that tau increases with worsening cogni- vestigate these biological factors associated with tau tion. Previous longitudinal CSF studies have reported levels in CSF. Also, future studies should further investi- conflicting results, observing increases in middle-age in- gate the longitudinal relationship of these tau subtypes dividuals with normal cognition during a follow-up with concurrent other biological measures that deterior- period of 6 years [30], but also a lack of change in indi- ate during the AD process, such as synaptic markers in viduals with normal cognition, MCI and AD over a me- CSF or on PET, and cognitive data, to better understand dian follow-up of 2 years [31, 32]. This literature differences in clinical progression amongst tau subtypes. together with our observations suggests that increases over time in t-tau levels in CSF are slow, and follow-up Limitations times longer than 2–3 years might be necessary for par- A potential limitation of our study is that although we ticipants to change subgroups. used large clinical cohorts, the number of subjects in One of the challenges in biomarkers research is how some subgroups and subanalyses was small: this was es- to define the cut-point between normal and abnormal pecially the case in the highest tau subgroup, as well as levels. Pathology is the gold standard, but is also the end in tau PET analyses. The small size of the highest tau stage of the disease and difficult to obtain for large sam- subgroup means that there is more uncertainty in the as- ple sizes. The cut-point for Luminex p-tau in ADNI was sociation of this subgroup with clinical characteristics. originally based on pathology [21], and we observed the Therefore, the results regarding the highest tau sub- same cut-off for the lowest p-tau subgroup (23 pg/ml). group and the tau PET analyses should be interpreted However, for t-tau, we observed a lower cut-off that was with caution, and if possible repeated in future studies in still related to increased risk for disease progression. A even larger cohorts. Furthermore, we used Gaussian recent study defined cut-points for t- and p-tau mea- mixtures as a data-driven approach to study potential sured with Elecsys (t-tau 300 pg/ml and p-tau 27 pg/ml) subgroups in tau levels as a first step, it is possible that in ADNI based on their association with clinical progres- more complex models may improve the fit of tau levels sion in MCI patients [23]. We expand upon previous distributions, which should be addressed in future stud- studies [6, 11, 33, 34] by identifying additional cut- ies. Also, we determined cut-points here as the intersec- points that may have practical use for more specific tions of the probability distributions of the normal prognoses to individual patients or in trial design: we mixtures, which may not be ideal in all settings. For ex- identified lower cut-points than defined in the literature ample, in studies where minimizing misclassification (resp. 193 and 22 pg/ml for t- and p-tau, respectively) costs is desired, e.g. in clinical trial design, it may be use- that were already associated with increased risk for clin- ful to choose cut-points so that misclassification is mini- ical progression, and also showed for the higher cut- mized of individuals with high tau as falling in the points, that the corresponding subgroups were associ- lowest tau group, to ensure that as many individuals ated with gradually increasing hazard ratios and steeper with potentially fast progression are included in the trial. decline on the MMSE. Future studies could test the efficacy of the data-driven The notion that higher tau subgroups also included cut-points in those settings. Strengths of the study are non-demented individuals, and that higher tau levels that we used a large cohort, and we validated the mix- were associated with faster cognitive decline, regardless ture modelling results in another independent cohort of disease stage, suggests that tau subgroups may reflect with two different analysis platforms for CSF tau, and differences in underlying biological processes, rather both cohorts had detailed information of the Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 24 of 25 characteristics of the study populations, including cogni- related to PET imaging. PS and WF led the design and set-up of the ADC, from which data were used in this article. LE, EWo, EWi, WF, BT, CT and PV tive measures, follow-up data on clinical progression and made suggestions on improvement of the manuscript. All authors read and information on APOE genotype available. approved the manuscript for submission. Funding Conclusions This work has been supported by ZonMW Memorabel grant programme In conclusion, our studies suggest that abnormal levels of #733050824 (KW, BMT and PJV), Alzheimer Nederland grant #NL18003P (FD) and the Sigrid Juselius Foundation (LE). Funding was used in the analysis, CSF t-tau and p-tau may convey different biological aspects interpretation and writing of the manuscript. Statistical analyses were in AD, which might be in part driven by genetic factors performed at the VUmc Alzheimer Center that is part of the such as different APOE genotypes. The data-driven cut- neurodegeneration research programme of the Neuroscience Campus Amsterdam. The VUmc Alzheimer Center is supported by Stichting points we found may aid daily practice in prognosis of pa- Alzheimer Nederland and Stichting VUmc fonds. tients and may aid trial design by allowing stratification of individuals according to their risk of clinical progression. Availability of data and materials The ADNI dataset analysed during the current study is available in the ADNI repository, www.adni-info.org. For ADC, the data that support the findings of Supplementary information this study are available from the corresponding author upon reasonable Supplementary information accompanies this paper at https://doi.org/10. request. 1186/s13195-020-00713-3. Ethics approval and consent to participate Additional file 1. The institutional review boards of all institutions participating in ADNI approved the procedures that were part of the study. For ADC, all procedures were Additional file 2. approved by the local medical ethics committee. In both ADNI and ADC, written informed consent was obtained from all participants or surrogates. Abbreviations Aβ42: Amyloid-β 1-42; AD: Alzheimer’s disease; ADC: Amsterdam Dementia Consent for publication Cohort; ADNI: Alzheimer’s disease Neuroimaging Initiative; Not applicable. APOE: Apolipoprotein E; CI: Confidence interval; CSF: Cerebrospinal fluid; HR: Hazard ratio; NC: Normal cognition; MCI: Mild cognitive impairment; Competing interests MMSE: Mini-Mental State Examination; MRI: Magnetic resonance imaging; Prof. dr. Scheltens has acquired grants for the institution from GE Healthcare PET: Positron emission tomography; p-tau-181: Tau phosphorylated at and Piramal and received consultancy/speaker fees paid to the institution threonine 181; t-tau: Total tau from Novartis, Probiodrug, Biogen, Roche, and EIP Pharma, LLC in the past 2 years. Research programmes of Prof. dr. Wiesje van der Flier received funding Acknowledgements by ZonMW, NWO, EU-JPND, Alzheimer Nederland, CardioVascular Onderzoek Data was used for this project of which collection and sharing was funded Nederland, Health~Holland, Topsector Life Sciences & Health, stichting Dior- by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes aphte, Gieskes-Strijbis fonds, stichting Equilibrio, Biogen MA Inc., Life-MI, of Health Grant U01 AG024904) and DOD ADNI (Department of Defense AVID, Combinostics. WF holds the Pasman chair. WF has performed contract award number W81XWH-12-2-0012). ADNI is funded by the National Institute research for Biogen MA Inc. All funding is paid to her institution. Prof. dr. on Aging, the National Institute of Biomedical Imaging and Bioengineering, Teunissen received grants from the European Commission, the Dutch Re- and through generous contributions from the following: AbbVie, Alzheimer’s search Council (ZonMW), Association of Frontotemporal Dementia/Alzhei- Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioCli- mer’s Drug Discovery Foundation, The Weston Brain Institute, Alzheimer nica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Netherlands. Prof. Dr. Teunissen has functioned in advisory boards of Roche, Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. received non-financial support in the form of research consumables from Hoffmann-La Roche Ltd. and its affiliated company Genentech, Inc.; Fujirebio; ADxNeurosciences and Euroimmun, performed contract research or received GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & De- grants from Probiodrug, Biogen, Esai, Toyama, Janssen Prevention Center, velopment, LLC.; Johnson & Johnson Pharmaceutical Research & Develop- Boehringer, AxonNeurosciences, EIP farma, PeopleBio, Roche. The other au- ment LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, thors reported no conflicts of interest. LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Author details Company; and Transition Therapeutics. The Canadian Institutes of Health Re- Department of Neurology, Alzheimer Center Amsterdam, Amsterdam search is providing funds to support ADNI clinical sites in Canada. Private Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, sector contributions are facilitated by the Foundation for the National Insti- the Netherlands. Turku PET Centre, University of Turku and Turku University tutes of Health (www.fnih.org). The grantee organization is the Northern Cali- Hospital, Turku, Finland. Department of Radiology & Nuclear Medicine, fornia Institute for Research and Education, and the study is coordinated by Amsterdam Neuroscience, Amsterdam UMC, Amsterdam Neuroscience, the Alzheimer’s Therapeutic Research Institute at the University of Southern Amsterdam, Netherlands. Department of Clinical Chemistry, Neurochemistry California. ADNI data are disseminated by the Laboratory for Neuro Imaging Laboratory, Amsterdam UMC, Amsterdam Neuroscience, Amsterdam, at the University of Southern California. Netherlands. Department of Epidemiology and Biostatistics, Amsterdam Part of the data used in preparation of this article were obtained from the UMC, Amsterdam, The Netherlands. Alzheimer Center Limburg, Department Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc. of Psychiatry & Neuropsychology, School of Mental Health and Neuroscience, edu). As such, the investigators within the ADNI contributed to the design Maastricht University, Maastricht, The Netherlands. Division of and implementation of ADNI and/or provided data but did not participate in Neurogeriatrics, Department of Neurobiology, Care Sciences and Society, analysis or writing of this report. A complete listing of ADNI investigators can Karolinska Institutet, Stockholm, Sweden. be found at: http://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ ADNI_Acknowledgement_List.pdf Received: 16 June 2020 Accepted: 22 October 2020 Authors’ contributions FD wrote the first drafts of the manuscript; KW and BT contributed further References revisions. FD and BT performed the statistical analyses. LE contributed to the 1. Olsson B, Lautner R, Andreasson U, Öhrfelt A, Portelius E, Bjerke M, et al. CSF writing of the discussion. EWo contributed to methods and discussion and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 25 of 25 review and meta-analysis. Lancet Neurol. 2016;15:673–84 https://doi.org/10. linked immunosorbent assay and multiplex platforms in a longitudinal 1016/S1474-4422(16)00070-3. Alzheimer’s disease study. Alzheimers Dement. 2013;9:276–83 https://doi. 2. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH, et org/10.1016/j.jalz.2012.01.004. al. The diagnosis of dementia due to Alzheimer’s disease: recommendations 21. Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen from the National Institute on Aging-Alzheimer’s Association workgroups RC, et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011; neuroimaging initiative subjects. Ann Neurol. 2009;65:403–13 https://doi. 7:263–9 https://doi.org/10.1016/J.JALZ.2011.03.005. org/10.1002/ana.21610. 22. Verhage F. Intelligentie en Leeftijd: Onderzoek bij Nederlanders van Twaalf 3. Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo JL, Blennow K, et al. tot Zevenenzeventig Jaar [Intelligence and Age: Study with Dutch People Advancing research diagnostic criteria for Alzheimer’sdisease: the IWG-2 from Age 12 to 77]. Assen: Van Gorcum; 1964. criteria. Lancet Neurol. 2014;13:614–29 https://doi.org/10.1016/S1474- 23. Blennow K, Shaw LM, Stomrud E, Mattsson N, Toledo JB, Buck K, et al. 4422(14)70090-0. Predicting clinical decline and conversion to Alzheimer’s disease or 4. Mattsson N, Zetterberg H, Hansson O, Andreasen N, Parnetti L, Jonsson M, dementia using novel Elecsys Aβ (1–42), pTau and tTau CSF immunoassays. et al. CSF biomarkers and incipient Alzheimer disease in patients with mild Sci Rep. 2019;9:19024 https://doi.org/10.1038/s41598-019-54204-z. cognitive impairment. JAMA. 2009;302:385 https://doi.org/10.1001/jama. 24. Iqbal K, Liu F, Gong C-X. Tau and neurodegenerative disease: the story so 2009.1064. far. Nat Rev Neurol. 2016;12:15–27 https://doi.org/10.1038/nrneurol.2015.225. 5. Mulder C, Verwey NA, van der Flier WM, Bouwman FH, Kok A, van Elk EJ, et 25. Chhatwal JP, Schultz AP, Marshall GA, Boot B, Gomez-Isla T, Dumurgier J, et al. Amyloid- (1-42), total tau, and phosphorylated tau as cerebrospinal fluid al. Temporal T807 binding correlates with CSF tau and phospho-tau in biomarkers for the diagnosis of Alzheimer disease. Clin Chem. 2010;56:248– normal elderly. Neurology. 2016;87:920–6. 53 https://doi.org/10.1373/clinchem.2009.130518. 26. Mattsson N, Schöll M, Strandberg O, Smith R, Palmqvist S, Insel PS, et al. 18 6. Duits FH, Teunissen CE, Bouwman FH, Visser P-J, Mattsson N, Zetterberg H, F-AV-1451 and CSF T-tau and P-tau as biomarkers in Alzheimer’s disease. et al. The cerebrospinal fluid “Alzheimer profile”: easily said, but what does it EMBO Mol Med. 2017;9:1212–23. mean? Alzheimer’s Dement. 2014;10:713–23.e2 https://doi.org/10.1016/J. 27. Gordon BA, Friedrichsen K, Brier M, Blazey T, Su Y, Christensen J, et al. The JALZ.2013.12.023. relationship between cerebrospinal fluid markers of Alzheimer pathology 7. Toledo JB, Zetterberg H, van Harten AC, Glodzik L, Martinez-Lage P, and positron emission tomography tau imaging. Brain. 2016;139:2249–60. Bocchio-Chiavetto L, et al. Alzheimer’s disease cerebrospinal fluid biomarker 28. Pooler AM, Phillips EC, Lau DHW, Noble W, Hanger DP. EMBO Reports. 2013; in cognitively normal subjects. Brain. 2015;138:2701–15 https://doi.org/10. 14(4):389–94. https://doi.org/10.1038/embor.2013.15. 1093/brain/awv199. 29. Sato C, Barthélemy NR, Mawuenyega KG, Patterson BW, Gordon BA, Jockel- 8. Mirra SS. The CERAD neuropathology protocol and consensus Balsarotti J, et al. Tau kinetics in neurons and the human central nervous recommendations for the postmortem diagnosis of Alzheimer’s disease: a system. Neuron. 2018;97:1284–98.e7 https://doi.org/10.1016/j.neuron.2018. commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih. 02.015. gov/pubmed/9330994. Accessed 9 Sept 2019. 30. Sutphen CL, Jasielec MS, Shah AR, Macy EM, Xiong C, Vlassenko AG, et al. 9. Knopman DS, DeKosky ST, Cummings JL, Chui H, Corey-Bloom J, Relkin N, et al. Longitudinal cerebrospinal fluid biomarker changes in preclinical Alzheimer Practice parameter: diagnosis of dementia (an evidence-based review): report disease during middle age. JAMA Neurol. 2015;72:1029 https://doi.org/10. of the Quality Standards Subcommittee of the American Academy of 1001/jamaneurol.2015.1285. Neurology. Neurology. 2001;56:1143–53 https://doi.org/10.1212/WNL.56.9.1143. 31. Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. 10. Beach TG, Monsell SE, Phillips LE, Kukull W. Accuracy of the clinical diagnosis Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s of Alzheimer disease at National Institute on Aging Alzheimer Disease disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; Centers, 2005–2010. J Neuropathol Exp Neurol. 2012;71:266–73 https://doi. https://doi.org/10.1016/j.jalz.2019.01.015. org/10.1097/NEN.0b013e31824b211b. 32. Wildsmith KR, Schauer SP, Smith AM, Arnott D, Zhu Y, Haznedar J, et al. 11. Degerman Gunnarsson M, Ingelsson M, Blennow K, Basun H, Lannfelt L, Identification of longitudinally dynamic biomarkers in Alzheimer’s disease Kilander L. High tau levels in cerebrospinal fluid predict nursing home cerebrospinal fluid by targeted proteomics. Mol Neurodegener. 2014;9:22 placement and rapid progression in Alzheimer’s disease. Alzheimers Res https://doi.org/10.1186/1750-1326-9-22. Ther. 2016;8:22 https://doi.org/10.1186/s13195-016-0191-0. 33. van Rossum IA, Vos SJB, Burns L, Knol DL, Scheltens P, Soininen H, et al. 12. De Meyer G, Shapiro F, Vanderstichele H, Vanmechelen E, Engelborghs S, De Injury markers predict time to dementia in subjects with MCI and amyloid Deyn PP, et al. Diagnosis-independent Alzheimer disease biomarker pathology. Neurology. 2012;79:1809–16 https://doi.org/10.1212/WNL. signature in cognitively normal elderly people. Arch Neurol. 2010;67:949 0b013e3182704056. https://doi.org/10.1001/archneurol.2010.179. 34. Kester MI, van der Vlies AE, Blankenstein MA, Pijnenburg YAL, van Elk EJ, 13. Bertens D, Tijms BM, Scheltens P, Teunissen CE, Visser PJ. Unbiased Scheltens P, et al. CSF biomarkers predict rate of cognitive decline in estimates of cerebrospinal fluid β-amyloid 1-42 cutoffs in a large memory Alzheimer disease. Neurology. 2009;73:1353–8 https://doi.org/10.1212/WNL. clinic population. Alzheimers Res Ther. 2017;9:8 https://doi.org/10.1186/ 0B013E3181BD8271. s13195-016-0233-7. 35. Jansen WJ, Ossenkoppele R, Knol DL, Tijms BM, Scheltens P, Verhey FRJ, et 14. Zwan M, van Harten A, Ossenkoppele R, Bouwman F, Teunissen C, al. Prevalence of cerebral amyloid pathology in persons without dementia. Adriaanse S, et al. Concordance between cerebrospinal fluid biomarkers and JAMA. 2015;313:1924 https://doi.org/10.1001/jama.2015.4668. [11C] PIB PET in a memory clinic cohort. J Alzheimers Dis. 2014;41:801–7 36. Slot RER, Kester MI, Van Harten AC, Jongbloed W, Bouwman FH, Teunissen https://doi.org/10.3233/JAD-132561. CE, et al. ApoE and clusterin CSF levels influence associations between 15. Zwan MD, Rinne JO, Hasselbalch SG, Nordberg A, Lleó A, Herukka S-K, et al. APOE genotype and changes in CSF tau, but not CSF Aβ42, levels in non- Use of amyloid-PET to determine cutpoints for CSF markers. Neurology. demented elderly. Neurobiol Aging. 2019;79:101–9. 2016;86:50–8 https://doi.org/10.1212/WNL.0000000000002081. 37. Konijnenberg E, Tijms BM, Gobom J, Dobricic V, Bos I, Vos S, et al. APOE ϵ4 16. Palmqvist S, Zetterberg H, Blennow K, Vestberg S, Andreasson U, Brooks DJ, genotype-dependent cerebrospinal fluid proteomic signatures in et al. Accuracy of brain amyloid detection in clinical practice using Alzheimer’s disease. Alzheimers Res Ther. 2020;12:65 https://doi.org/10.1186/ cerebrospinal fluid β-amyloid 42. JAMA Neurol. 2014;71:1282 https://doi.org/ s13195-020-00628-z. 10.1001/jamaneurol.2014.1358. 38. Reus LM, Stringer S, Posthuma D, Teunissen CE, Scheltens P, Pijnenburg 17. Blennow K, Hampel H. CSF markers for incipient Alzheimer’s disease. Lancet YAL, et al. Degree of genetic liability for Alzheimer’s disease associated with Neurol. 2003;2:605–13. specific proteomic profiles in cerebrospinal fluid. Neurobiol Aging. 2020;93: 18. Blennow K, Hampel H, Weiner M, Zetterberg H. Cerebrospinal fluid and 144 e1–144.e15. plasma biomarkers in Alzheimer disease. Nat Rev Neurol. 2010;6:131–44. 19. van der Flier WM, Scheltens P. Amsterdam dementia cohort: performing research to optimize care. J Alzheimers Dis. 2018;62:1091–111 https://doi. Publisher’sNote org/10.3233/JAD-170850. Springer Nature remains neutral with regard to jurisdictional claims in 20. Jongbloed W, Kester MI, van der Flier WM, Veerhuis R, Scheltens P, published maps and institutional affiliations. Blankenstein MA, et al. Discriminatory and predictive capabilities of enzyme- http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Alzheimer s Research & Therapy Springer Journals http://www.deepdyve.com/lp/springer-journals/four-subgroups-based-on-tau-levels-in-alzheimer-s-disease-observed-in-qpkECnqdq4

Loading next page...

References (44)

(Mirra SS. The CERAD neuropathology protocol and consensus recommendations for the postmortem diagnosis of Alzheimer’s disease: a commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih.gov/pubmed/9330994. Accessed 9 Sept 2019.)
Mirra SS. The CERAD neuropathology protocol and consensus recommendations for the postmortem diagnosis of Alzheimer’s disease: a commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih.gov/pubmed/9330994. Accessed 9 Sept 2019.
Mirra SS. The CERAD neuropathology protocol and consensus recommendations for the postmortem diagnosis of Alzheimer’s disease: a commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih.gov/pubmed/9330994. Accessed 9 Sept 2019., Mirra SS. The CERAD neuropathology protocol and consensus recommendations for the postmortem diagnosis of Alzheimer’s disease: a commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih.gov/pubmed/9330994. Accessed 9 Sept 2019.
S. Mirra (1997)
The CERAD Neuropathology Protocol and Consensus Recommendations for the Postmortem Diagnosis of Alzheimer’s Disease: A Commentary
Neurobiology of Aging, 18
(2020)
Degree of genetic liability for Alzheimer’s disease associated with specific proteomic profiles in cerebrospinal fluid
Neurobiol Aging, 93
F. Verhage (1964)
Intelligentie en leeftijd : onderzoek bij Nederlanders van twaalf tot zevenenzeventig jaar
(2016)
Temporal T807 binding correlates with CSF tau and phospho-tau in normal elderly
Neurology., 87
(2009)
CSF biomarkers predict rate of cognitive decline in Alzheimer disease
Neurology., 73
(2014)
Accuracy of brain amyloid detection in clinical practice using cerebrospinal fluid β-amyloid 42
JAMA Neurol., 71
(2010)
Cerebrospinal fluid and plasma biomarkers in Alzheimer disease
Nat Rev Neurol., 6
C Mulder (2010)
248
Clin Chem., 56
(2015)
Prevalence of cerebral amyloid pathology in persons without dementia
JAMA., 313
(2014)
Concordance between cerebrospinal fluid biomarkers and [11C] PIB PET in a memory clinic cohort
J Alzheimers Dis., 41
(2017)
Unbiased estimates of cerebrospinal fluid β-amyloid 1-42 cutoffs in a large memory clinic population
Alzheimers Res Ther., 9
JB Toledo (2015)
2701
Brain., 138
(2016)
High tau levels in cerebrospinal fluid predict nursing home placement and rapid progression in Alzheimer’s disease
Alzheimers Res Ther., 8
(2019)
Predicting clinical decline and conversion to Alzheimer’s disease or dementia using novel Elecsys Aβ (1–42), pTau and tTau CSF immunoassays
Sci Rep., 9
(2016)
Tau and neurodegenerative disease: the story so far
Nat Rev Neurol., 12
A. Lleó, D. Alcolea, P. Martínez-Lage, P. Scheltens, L. Parnetti, J. Poirier, A. Simonsen, M. Verbeek, P. Rosa-Neto, R. Slot, M. Tainta, Andrea Izaguirre, Babette Reijs, L. Farotti, M. Tsolaki, Rik Vandenbergue, Y. Freund-Levi, F. Verhey, J. Clarimón, J. Fortea, L. Frolich, I. Santana, J. Molinuevo, S. Lehmann, P. Visser, C. Teunissen, H. Zetterberg, K. Blennow (2019)
Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer's disease continuum in the BIOMARKAPD study
Alzheimer's & Dementia, 15
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations
(2017)
18 F-AV-1451 and CSF T-tau and P-tau as biomarkers in Alzheimer’s disease
EMBO Mol Med., 9
(2016)
The relationship between cerebrospinal fluid markers of Alzheimer pathology and positron emission tomography tau imaging
Brain., 139
(2016)
Use of amyloid-PET to determine cutpoints for CSF markers
Neurology., 86
(Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; 10.1016/j.jalz.2019.01.015.)
Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; 10.1016/j.jalz.2019.01.015.
Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; 10.1016/j.jalz.2019.01.015., Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; 10.1016/j.jalz.2019.01.015.
(2018)
Tau kinetics in neurons and the human central nervous system
Neuron, 97
(2015)
Longitudinal cerebrospinal fluid biomarker changes in preclinical Alzheimer disease during middle age
JAMA Neurol., 72
(2012)
Accuracy of the clinical diagnosis of Alzheimer disease at National Institute on Aging Alzheimer Disease Centers, 2005–2010
J Neuropathol Exp Neurol., 71
(2003)
CSF markers for incipient Alzheimer’s disease
Lancet Neurol., 2
B Dubois (2014)
614
Lancet Neurol., 13
FH Duits (2014)
713
Alzheimer’s Dement, 10
(2012)
Injury markers predict time to dementia in subjects with MCI and amyloid pathology
Neurology., 79
(2013)
10.1038/embor.2013.15
EMBO Reports, 14
(2014)
Alzheimer profile": easily said, but what does it mean? Alzheimer's Dement
(2001)
Practice parameter: diagnosis of dementia (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology
Neurology., 56
(2009)
Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects
Ann Neurol., 65
(2018)
Amsterdam dementia cohort: performing research to optimize care
J Alzheimers Dis., 62
(2010)
Diagnosis-independent Alzheimer disease biomarker signature in cognitively normal elderly people
Arch Neurol., 67
GM McKhann (2011)
263
Alzheimers Dement., 7
B Olsson (2016)
673
Lancet Neurol., 15
(2013)
Discriminatory and predictive capabilities of enzyme-linked immunosorbent assay and multiplex platforms in a longitudinal Alzheimer’s disease study
Alzheimers Dement., 9
(2014)
Identification of longitudinally dynamic biomarkers in Alzheimer’s disease cerebrospinal fluid by targeted proteomics
Mol Neurodegener., 9
(2020)
APOE ϵ4 genotype-dependent cerebrospinal fluid proteomic signatures in Alzheimer’s disease
Alzheimers Res Ther., 12
H. Togt (2003)
Publisher's Note
J. Netw. Comput. Appl., 26
(2019)
ApoE and clusterin CSF levels influence associations between APOE genotype and changes in CSF tau, but not CSF Aβ42, levels in non-demented elderly
Neurobiol Aging., 79
(2021)
a systematic Duits et al
N Mattsson (2009)
385
JAMA., 302

Publisher: Springer Journals
Copyright: Copyright © The Author(s) 2020
eISSN: 1758-9193
DOI: 10.1186/s13195-020-00713-3
Publisher site: See Article on Publisher Site

Abstract

Background: As Alzheimer’s disease (AD) pathology presents decades before dementia manifests, unbiased biomarker cut-points may more closely reflect presence of pathology than clinically defined cut-points. Currently, unbiased cerebrospinal fluid (CSF) tau cut-points are lacking. Methods: We investigated CSF t-tau and p-tau cut-points across the clinical spectrum using Gaussian mixture modelling, in two independent cohorts (Amsterdam Dementia Cohort and ADNI). Results: Individuals with normal cognition (NC) (total n = 1111), mild cognitive impairment (MCI) (total n = 1213) and Alzheimer’s disease dementia (AD) (total n = 1524) were included. In both cohorts, four CSF t- and p-tau distributions and three corresponding cut-points were identified. Increasingly high tau subgroups were characterized by steeper MMSE decline and higher progression risk to AD (cohort/platform-dependent HR, t-tau 1.9–21.3; p-tau 2.2–9.5). Limitations: The number of subjects in some subgroups and subanalyses was small, especially in the highest tau subgroup and in tau PET analyses. Conclusions: In two independent cohorts, t-tau and p-tau levels showed four subgroups. Increasingly high tau subgroups were associated with faster clinical decline, suggesting our approach may aid in more precise prognoses. Keywords: Alzheimer’s disease, CSF tau, Gaussian mixture modelling, Prognosis * Correspondence: k.wesenhagen@amsterdamumc.nl Part of the data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: https://adni.loni.usc.edu/wpcontent/uploads/ how_to_apply/ADNI_Acknowledgement_List.pdf. Flora H. Duits and Kirsten E. J. Wesenhagen contributed equally to this work. Department of Neurology, Alzheimer Center Amsterdam, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, the Netherlands Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 2 of 25 Background distributions) was higher than clinically based cut- Abnormal levels of amyloid-β 1-42 (Aβ42), total tau (t- points, resulting in more sensitive detection of prede- tau) and tau phosphorylated at threonine 181 (p-tau-181) mentia AD [13–16]. As of yet, however, it remains un- are biomarkers for the presence of Alzheimer’s disease clear whether it is also possible to detect unbiased cut- (AD) pathology in the brain [1], and part of established re- points in t-tau and p-tau levels. search criteria for AD across the cognitive continuum [2, High t-tau levels in the cerebrospinal fluid (CSF) are 3]. Classification schemes based on biomarkers depend on thought to reflect neuronal degeneration or injury, and cut-points, and different approaches exist to determine elevated t-tau levels can be found in the CSF in various such cut-points. The most often used traditional approach conditions involving neuronal death, for example after determines cut-points by optimizing the sensitivity and an acute stroke. In contrast, p-tau-181 is presumed to specificity to detect clinical AD-type dementia compared reflect the formation of phosphorylated tau in the brain to controls [4–6]. However, approaches that use clinical and to represent more specifically the formation of labels as outcomes may not be optimal, because clinical la- neurofibrillary tangles, one of the neuropathological hall- bels do not optimally reflect the absence or presence of marks of AD [17, 18]. As tau pathology is a hallmark of AD pathology: For example, almost 30% of cognitively in- AD, it can be hypothesized that similarly to amyloid, t- tact individuals in their seventies have AD pathology [7], and p-tau levels may be a mixture of values belonging to and up to 20% of clinical AD dementia cases do not show normal and affected individuals, from which unbiased AD pathology at neuropathological examination [8–11]. cut-points might be determined. As such, cut-point based on clinical labels can be biased. The objective of this study was to investigate whether Gaussian mixture modelling provides an approach to subgroups can be identified in CSF t- and p-tau levels determine cut-points independent of clinical information using Gaussian mixture modelling and to determine cut- [12]. This approach is based on the notion that the dis- points. We characterized tau subgroups in terms of clin- tribution of biomarker values in a population is a mix- ical and biological characteristics and longitudinal trajec- ture of values belonging to subpopulations, i.e. normal tories of cognitive decline. We repeated analyses in the and affected individuals. Previous studies using this ap- independent ADNI cohort to determine the robustness proach have found a bimodal distribution of Aβ42 levels, of the identified subgroups and tested stability of group of which the cut-point (i.e. the intersection of these membership by studying longitudinal changes in t-tau Table 1 Participant characteristics of the Amsterdam Dementia Cohort (ADC) and ADNI cohorts ADC ADNI Characteristic NC MCI AD dementia NC MCI AD dementia N = 740 N = 591 N = 1296 N = 371 N = 622 N = 228 a c a c MMSE, mean ± SD 28.2 ± 1.8 26.5 ± 2.4 20.5 ± 5 29.1 ± 1.2 27.7 ± 1.8 23.3 ± 2 a a a b Age, mean ± SD 59.6 ± 8.9 66.4 ± 8.2 66.2 ± 8.1 73.8 ± 5.9 72.4 ± 7.5 74.9 ± 8.1 a c a c Female, n (%) 306 (41.4%) 217 (36.7%) 674 (52%) 195 (52.6%) 255 (41%) 95 (41.7%) a c a c APOE e4 carrier, n (%) NC 258 (36.3%) 287 (52.4%) 791 (65.3%) 103 (27.8%) 307 (49.4%) 154 (67.5%) a c Innotest: T-tau (pg/ml), mean ± SD 296.4 ± 200.8 466.4 ± 303.6 716.6 ± 417.1 n.a. n.a. n.a. a c Innotest: P-tau (pg/ml), mean ± SD 48.4 ± 22.7 66.8 ± 33.6 87.6 ± 39.5 n.a. n.a. n.a. a c Innotest: Aβ42 (pg/ml), mean ± SD 1071.2 ± 246.9 859.1 ± 288.1 648.4 ± 166.6 n.a. n.a. n.a. a c Innotest: Abnormal Aβ42 (< 813 pg/ml), n (%) 124 (16.8%) 326 (55.2%) 1173 (90.5%) n.a. n.a. n.a. a c Luminex T-tau (pg/ml), mean ± SD n.a. n.a. n.a. 67.4 ± 32.8 90.4 ± 54.8 126.6 ± 61.4 a c Luminex P-tau (pg/ml), mean ± SD n.a. n.a. n.a. 32.4 ± 18.8 39.2 ± 23.7 51.6 ± 30.7 a c Luminex Aβ42 (pg/ml), mean ± SD n.a. n.a. n.a. 201.7 ± 51.9 171.2 ± 52.5 139.6 ± 38.8 a c Luminex Abnormal Aβ42 (< 192 pg/ml), n(%) n.a. n.a. n.a. 156 (42%) 403 (64.8%) 210 (92.1%) a c Elecsys T-tau (pg/ml), mean ± SD n.a. n.a. n.a. 238.5 ± 90 284.9 ± 126.7 370.2 ± 144.4 a c Elecsys P-tau (pg/ml), mean ± SD n.a. n.a. n.a. 21.9 ± 9.2 27.6 ± 14.4 36.9 ± 15.7 a c Elecsys Aβ42 (pg/ml), mean ± SD n.a. n.a. n.a. 1337 ± 647.9 1020.7 ± 554.8 694.6 ± 420.7 a c Elecsys Abnormal Aβ42 (< 880 pg/ml), n (%) n.a. n.a. n.a. 104 (28%) 325 (53%) 189 (85%) n.a. not available, NC cognitively normal, MCI mild cognitive impairment, AD Alzheimer’s disease Differs from NC with p < .05 Differs from MCI with p < .001 Differs from MCI and NC with p < .001 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 3 of 25 and p-tau levels. Finally, we compared subgroups on tau www.adni-info.org) was used for validation of the results. PET uptake that was available for a subset of individuals ADNI started in 2003 as a public-private collaboration in ADNI. under the supervision of Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI is to study Methods whether serial magnetic resonance imaging (MRI), posi- We investigated the existence of CSF t- and p-tau sub- tron emission tomography (PET), other biological markers groups in data from two independent clinical cohorts. The and clinical and neuropsychological measures can be com- memory clinic-based Amsterdam Dementia Cohort bined to measure the progression of mild cognitive im- (ADC) was used for testing our hypothesis [19], and the pairment (MCI) and early Alzheimer’s disease (AD). Alzheimer’s Disease Neuroimaging Initiative (ADNI; Please see www.adni-info.org for the latest information. a) Amsterdam Dementia Cohort Amsterdam Dementia Cohort 0 0 0 1000 2000 3000 0 100 200 300 T−tau pg/ml P−tau pg/ml b) ADNI Luminex ADNI Luminex 0 0 0 100 200 300 400 0 50 100 150 200 T−tau pg/ml P−tau pg/ml c) ADNI Elecsys ADNI Elecsys 0 0 200 400 600 800 25 50 75 100 T−tau pg/ml P−tau pg/ml Normal amyloid Abnormal amyloid Subtype 1 Subtype 2 Subtype 3 Subtype 4 Fig. 1 Tetramodal distributions in t-tau and p-tau levels in ADC and ADNI. Levels in ADC are shown in (a); levels in ADNI are shown in (b) for Luminex and (c) for Elecsys assay. Grey colours in the distributions reflect for those tau levels the number of individuals with abnormal amyloid levels Count Count Count Count Count Count Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 4 of 25 The institutional review boards of all participating institu- further assessed whether subgroups showed differences tions approved the procedures for this study. Written in- in cognitive decline, stratifying subjects based on their formed consent was obtained from all participants or baseline cognitive state (defined as NC, MCI or demen- surrogates. Participant selection and CSF analyses are tia). First, decline in MMSE (outcome) was assessed with summarized below; for more detailed methods and infor- linear mixed models using the R package “lmer4”,includ- mation on Apolipoprotein E (APOE) genotyping and PET ing the main terms time and tau subgroup, and inter- imaging, see Additional file 1. action terms time*tau subgroup. For individuals without dementia at baseline, Cox proportional hazards models Participants were used to compare the rate of progression from NC to In short, patients from the ADC who visited our mem- MCI or AD dementia and from MCI to AD dementia be- ory clinic between November 2000 and December 2016 tween tau subgroups. We ran 5 models: (1) without covar- were selected (n = 2724) if they had baseline CSF tau iates; (2) including age, sex and educational level; (3) measurements available and had subjective cognitive de- model 2 + amyloid status; (4) model 3 + baseline cognitive cline (considered as normal cognition (NC)), mild cogni- state; and (5) model 4 + APOE-e4 carriership (dichotom- tive impairment (MCI) or AD dementia. Participants ous). For the Cox proportional hazards models, data from from ADNI who had baseline CSF biomarkers available were selected (n = 1221) for the replication analyses if Table 2 Consistency of subgroup labelling between t-tau and they met the study-specific criteria of NC, MCI or de- p-tau (ADC and ADNI), and across platforms (ADNI) mentia. A subset of 619 individuals in ADNI (51%; 183 Biomarker: Platform ADC p-tau Innotest NC, 345 MCI and 91 with AD dementia) with available ADC T-tau: Innotest Subgroup 1 2 3 4 follow-up CSF measures were selected for longitudinal analyses. 1 960 83 0 0 2 140 661 58 0 CSF biomarkers 3 6 209 399 18 In ADC, CSF biomarkers (β-amyloid , hTAU-Ag, (1-42) 40 0 33 60 and phospo-tau 181P) were assessed with INNOTEST P-tau: Luminex (Fujirebio, Ghent, Belgium) on a routine basis as de- T-tau: Luminex Subgroup 1 2 3 4 scribed before [20]. In ADNI, CSF biomarkers were ana- lysed using a multiplex xMAP Luminex platform 1 210 108 13 1 (Luminex Corp) with immunoassay kit-based reagents 2 116 261 89 2 (INNO-BIA Alzbio3; Innogenetics) [21](n = 1213 partic- 3 3 159 138 8 ipants), and on Elecsys (Roche, Basel, Switserland) [21] 4 0 10 76 19 (n = 1193 participants, overlap with Luminex 98%). P-tau: Elecsys T-tau: Elecsys Subgroup 1 2 3 4 Statistical analysis Gaussian mixture modelling was used to identify cut- 1 290 0 0 0 points in the distribution of t-tau and p-tau values. First, 2 186 309 0 0 the number of distributions that best described the data 3 0 145 185 2 was determined with the R boot.comp function. This 40 0 14 61 function sequentially tests increasing number of compo- T-tau: Elecsys nents in the data using parametric bootstrapping of the T-tau: Luminex Subgroup 1 2 3 4 likelihood ratio (i.e. likelihood of x components vs. likeli- hood of having one more component, i.e. x + 1), until 1 256 70 1 0 the null hypothesis cannot be rejected anymore (p > 2 35 381 43 1 0.05, i.e. no improvement of additional component for 3 0 44 255 4 model fit). Then, we identified data-driven cut-points as 40 0 33 70 the points where the lines of two fitted Gaussian distri- P-tau: Elecsys butions intersected. Using these cut-points, we labelled P-tau: Luminex Subgroup 1 2 3 4 subjects according to tau subgroups. Next, within each cohort, we compared subgroups based on demographi- 1 287 36 0 0 cal, clinical and biological characteristics with ANOVA 2 167 296 63 2 or chi-square tests, when appropriate. For a subset of in- 3 21 119 121 48 dividuals with available repeated mini-mental state 40 1 13 12 examination (MMSE) and/or clinical follow-up, we Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 5 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 T-tau: Innotest ADC N All 1043 859 632 93 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 349 350–671 672–1380 > 1380 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 576 (78%) 135 (18%) 24 (3%) 5 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 262 (44%) 207 (35%) 118 (20%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 205 (16%) 517 (40%) 490 (38%) 84 (6%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 57 (9.9%) 47 (34.8%) 16 (66.7%) 4 (80%) 2.71E−12 6.49E−15 7.84E−05 4.00E−02 6.73E−01 9.90E−01 n (%) MCI 74 (28.2%) 140 (67.6%) 108 (91.5%) 4 (100%) 2.45E−16 6.64E−29 6.01E−02 1.31E−05 9.90E−01 9.90E−01 AD dementia 146 (71.2%) 472 (91.3%) 474 (96.7%) 81 (96.4%) 5.86E−11 1.08E−21 2.76E−05 2.96E−03 9.83E−01 9.90E−01 MMSE, mean ± NC 28.2 ± 1.8 28.2 ± 1.6 27.4 ± 2.1 27.8 ± 1.9 1.00E+00 1.34E−01 9.65E−01 1.61E−01 9.62E−01 9.62E−01 SD MCI 26.7 ± 2.4 26.6 ± 2.2 25.9 ± 2.8 25.2 ± 2.6 9.85E−01 2.53E−02 6.41E−01 7.27E−02 6.84E−01 9.47E−01 AD dementia 21.3 ± 4.6 20.8 ± 5 20.2 ± 5.1 18.8 ± 5.1 5.53E−01 3.94E−02 7.13E−04 2.82E−01 5.16E−03 8.41E−02 Age, mean ± SD NC 58.2 ± 8.5 64.2 ± 8 66.1 ± 9.7 65.8 ± 11.7 0.00E+00 4.83E−05 1.86E−01 7.42E−01 9.75E−01 1.00E+00 MCI 64 ± 8.4 67.9 ± 7.6 69.1 ± 7.1 69.9 ± 12 1.24E−06 5.96E−08 4.54E−01 5.28E−01 9.58E−01 9.97E−01 AD dementia 66.5 ± 8 66.5 ± 7.9 65.7 ± 8.3 65.7 ± 8.1 1.00E+00 6.10E−01 8.52E−01 3.68E−01 8.11E−01 1.00E+00 Female, n (%) NC 234 (40.6%) 57 (42.2%) 11 (45.8%) 4 (80%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 65 (24.8%) 83 (40.1%) 67 (56.8%) 2 (50%) 3.53E−03 1.71E−08 9.90E−01 3.21E−02 9.90E−01 9.90E−01 AD dementia 81 (39.5%) 263 (50.9%) 277 (56.5%) 53 (63.1%) 4.51E−02 3.63E−04 2.58E−03 4.94E−01 2.98E−01 9.90E−01 APOE e4 carrier, NC 183 (33.1%) 58 (45%) 15 (62.5%) 2 (40%) 8.88E−02 3.56E−02 9.90E−01 1.00E+00 9.90E−01 9.90E−01 n (%) MCI 92 (36.9%) 115 (59.6%) 77 (75.5%) 3 (75%) 2.15E−05 6.98E−10 9.90E−01 5.64E−02 1.00E+00 9.90E−01 AD dementia 113 (58.9%) 313 (65.3%) 310 (67.5%) 55 (67.1%) 8.18E−01 2.55E−01 9.90E−01 1.00E+00 1.00E+00 9.90E−01 T-tau: Luminex ADNI N All 335 468 310 108 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 54 55–95 96–171 > 171 s n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 154 (42%) 151 (41%) 63 (17%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 168 (27%) 249 (40%) 146 (23%) 59 (9%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 13 (6%) 68 (30%) 101 (44%) 46 (20%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 54 (35.1%) 55 (36.4%) 45 (71.4%) 2 (66.7%) 9.90E−01 1.34E−05 9.90E−01 3.58E−05 9.90E−01 9.90E−01 n (%) MCI 53 (31.5%) 155 (62.2%) 138 (94.5%) 57 (96.6%) 8.68E−09 9.22E−29 1.73E−16 2.12E−11 4.08E−06 9.90E−01 AD dementia 9 (69.2%) 59 (86.8%) 97 (96%) 45 (97.8%) 9.90E−01 1.70E−02 4.10E−02 3.26E−01 5.23E−01 9.90E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 6 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 MMSE, mean ± NC 29.1 ± 1.2 29 ± 1 29.1 ± 1.4 29 ± 1 9.87E−01 1.00E+00 1.00E+00 9.97E−01 1.00E+00 1.00E+00 SD MCI 28.3 ± 1.6 27.8 ± 1.8 27.3 ± 1.8 26.8 ± 1.9 2.27E−02 1.41E−05 6.60E−07 7.24E−02 1.66E−03 2.86E−01 AD dementia 24.4 ± 1.3 23.6 ± 2 23 ± 2 23.3 ± 2 5.75E−01 1.02E−01 2.87E−01 2.60E−01 8.11E−01 9.11E−01 Age, mean ± SD NC 72.5 ± 5.6 74 ± 5.8 76.3 ± 6.2 74.1 ± 3.4 9.69E−02 5.53E−05 9.64E−01 3.49E−02 1.00E+00 9.10E−01 MCI 70.7 ± 7.5 72.6 ± 7.6 73.7 ± 7.4 73.3 ± 7 5.69E−02 2.46E−03 1.15E−01 4.89E−01 9.34E−01 9.78E−01 AD dementia 78.4 ± 6.6 75.6 ± 8.7 75.1 ± 7.5 72.2 ± 8.6 6.66E−01 4.98E−01 6.62E−02 9.72E−01 1.08E−01 1.73E−01 Female, n (%) NC 81 (52.6%) 78 (51.7%) 33 (52.4%) 3 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 68 (40.5%) 89 (35.7%) 64 (43.8%) 34 (57.6%) 1.00E+00 9.90E−01 2.01E−01 8.23E−01 1.98E−02 6.12E−01 AD dementia 3 (23.1%) 18 (26.5%) 45 (44.6%) 29 (63%) 1.00E+00 9.90E−01 1.51E−01 1.58E−01 1.30E−03 3.44E−01 APOE e4 carrier, NC 32 (20.8%) 44 (29.1%) 26 (41.3%) 1 (33.3%) 7.19E−01 2.05E−02 9.90E−01 7.07E−01 1.00E+00 9.90E−01 n (%) MCI 43 (25.6%) 117 (47%) 103 (70.5%) 44 (74.6%) 1.01E−04 2.45E−14 4.77E−10 5.27E−05 1.46E−03 9.90E−01 AD dementia 7 (53.8%) 43 (63.2%) 75 (74.3%) 29 (63%) 9.90E−01 9.90E−01 1.00E+00 9.90E−01 9.90E−01 9.90E−01 T-tau: Elecsys ADNI N All 291 495 332 75 n.t. n.t. n.t. n.t. n.t. n.t. T-tau pg/ml All ≤ 192 193–311 312–514 > 514 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 126 (35%) 164 (45%) 69 (19%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 152 (25%) 258 (42%) 157 (26%) 41 (7%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 13 (6%) 73 (33%) 106 (48%) 30 (14%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 40 (31.7%) 67 (40.9%) 40 (58%) 3 (75%) 8.49E−01 3.93E−03 9.90E−01 1.47E−01 9.90E−01 9.90E−01 n (%) MCI 52 (34.2%) 153 (59.3%) 146 (93%) 40 (97.6%) 9.25E−06 1.06E−25 1.23E−11 1.67E−12 2.77E−05 9.90E−01 AD dementia 8 (61.5%) 67 (91.8%) 100 (94.3%) 29 (96.7%) 6.33E−02 4.91E−03 6.03E−02 9.90E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 29.2 ± 1.1 29 ± 1.1 29 ± 1.4 29.2 ± 1 5.42E−01 8.40E−01 9.99E−01 9.95E−01 9.66E−01 9.79E−01 SD MCI 28.1 ± 1.7 27.9 ± 1.8 27.2 ± 1.8 27.1 ± 1.8 6.43E−01 1.70E−05 6.98E−03 2.46E−04 3.86E−02 9.98E−01 AD dementia 24.5 ± 1.5 23.5 ± 1.9 23 ± 2.1 23.5 ± 1.9 3.88E−01 7.21E−02 4.57E−01 3.92E−01 1.00E+00 6.83E−01 Age, mean ± SD NC 71.6 ± 5.5 74.4 ± 5.8 76.1 ± 6.2 76.7 ± 6 3.60E−04 1.86E−06 2.95E−01 1.56E−01 8.47E−01 9.96E−01 MCI 70.9 ± 7.9 72.4 ± 7.4 73.9 ± 7.4 71.7 ± 6.9 2.02E−01 3.72E−03 9.26E−01 2.48E−01 9.47E−01 3.81E−01 AD dementia 77.7 ± 7.3 75.6 ± 8.1 74.3 ± 8.1 73.5 ± 8.7 8.13E−01 4.81E−01 4.09E−01 7.40E−01 6.61E−01 9.69E−01 Female, n (%) NC 69 (54.8%) 80 (48.8%) 37 (53.6%) 4 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 7.74E−01 9.90E−01 MCI 55 (36.2%) 99 (38.4%) 72 (45.9%) 23 (56.1%) 9.90E−01 6.41E−01 2.01E−01 9.69E−01 2.90E−01 9.90E−01 AD dementia 3 (23.1%) 23 (31.5%) 51 (48.1%) 17 (56.7%) 9.90E−01 9.40E−01 5.40E−01 2.35E−01 1.86E−01 9.90E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 7 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 APOE e4 carrier, NC 26 (20.6%) 45 (27.4%) 29 (42%) 1 (25%) 9.90E−01 1.58E−02 1.00E+00 2.54E−01 9.90E−01 9.90E−01 n (%) MCI 41 (27%) 120 (46.5%) 105 (66.9%) 32 (78%) 8.40E−04 2.90E−11 3.90E−08 4.95E−04 2.03E−03 9.90E−01 AD dementia 5 (38.5%) 51 (69.9%) 75 (70.8%) 19 (63.3%) 3.67E−01 2.55E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Innotest ADC N All 1106 953 490 78 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 56 57–96 97–159 > 159 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 564 (76%) 153 (21%) 19 (3%) 4 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 272 (46%) 214 (36%) 98 (17%) 7 (1%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 270 (21%) 586 (45%) 373 (29%) 67 (5%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 61 (10.8%) 44 (28.8%) 16 (84.2%) 3 (75%) 3.22E−07 2.14E−18 6.88E−03 3.57E−05 8.98E−01 9.90E−01 n (%) MCI 78 (28.7%) 153 (71.5%) 89 (90.8%) 6 (85.7%) 9.08E−20 6.37E−25 2.78E−02 1.57E−03 9.90E−01 9.90E−01 AD dementia 204 (75.6%) 546 (93.2%) 358 (96%) 65 (97%) 4.83E−12 2.03E−13 1.07E−03 5.59E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 28.2 ± 1.8 28.1 ± 1.6 27.9 ± 2.1 27.5 ± 1.7 9.99E−01 9.53E−01 8.79E−01 9.71E−01 8.94E−01 9.68E−01 SD MCI 26.7 ± 2.3 26.6 ± 2.4 25.7 ± 2.7 27.1 ± 2.1 9.88E−01 1.99E−03 9.60E−01 7.13E−03 9.40E−01 3.93E−01 AD dementia 20.9 ± 5 20.7 ± 4.9 20.3 ± 5.2 19 ± 4.8 9.11E−01 4.53E−01 2.98E−02 7.26E−01 5.43E−02 2.07E−01 Age, mean ± SD NC 58.3 ± 8.6 63.1 ± 8.3 67.1 ± 6.5 68.3 ± 16.8 7.58E−09 7.08E−05 9.30E−02 2.07E−01 6.19E−01 9.94E−01 MCI 64.1 ± 8.2 68 ± 7.9 69.3 ± 6.8 70.2 ± 10.1 3.55E−07 1.67E−07 1.77E−01 5.48E−01 8.89E−01 9.91E−01 AD dementia 66.7 ± 7.8 66.1 ± 8.1 65.6 ± 8 67.4 ± 9.2 7.76E−01 3.62E−01 9.13E−01 8.01E−01 6.02E−01 3.47E−01 Female, n (%) NC 228 (40.4%) 67 (43.8%) 9 (47.4%) 2 (50%) 9.90E−01 1.00E+00 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 73 (26.8%) 85 (39.7%) 54 (55.1%) 5 (71.4%) 2.15E−02 4.97E−06 1.80E−01 9.45E−02 9.90E−01 9.90E−01 AD dementia 118 (43.7%) 302 (51.5%) 215 (57.6%) 39 (58.2%) 2.38E−01 3.89E−03 2.77E−01 4.48E−01 9.90E−01 9.90E−01 APOE e4 carrier, NC 181 (33.3%) 62 (43.1%) 14 (73.7%) 1 (25%) 2.22E−01 4.13E−03 9.90E−01 1.39E−01 9.90E−01 9.90E−01 n (%) MCI 90 (35%) 133 (66.2%) 60 (71.4%) 4 (66.7%) 4.10E−10 6.79E−08 9.90E−01 9.90E−01 9.90E−01 9.90E−01 AD dementia 148 (59.2%) 367 (66.6%) 228 (65.7%) 48 (75%) 3.09E−01 7.44E−01 1.74E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Luminex ADNI N All 329 538 316 30 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 23 24–46 47–99 > 99 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 138 (37%) 166 (45%) 62 (17%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 173 (28%) 266 (43%) 166 (27%) 13 (2%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 18 (8%) 106 (47%) 88 (39%) 14 (6%) n.t. n.t. n.t. n.t. n.t. n.t. Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 8 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 Abnormal ab42, NC 41 (29.7%) 66 (39.8%) 44 (71%) 3 (100%) 5.28E−01 6.79E−07 2.93E−01 3.11E−04 7.84E−01 9.90E−01 n (%) MCI 47 (27.2%) 184 (69.2%) 155 (93.4%) 13 (100%) 9.96E−17 4.99E−34 1.93E−06 3.28E−08 2.30E−01 9.90E−01 AD dementia 11 (61.1%) 98 (92.5%) 85 (96.6%) 14 (100%) 4.36E−03 1.28E−04 1.63E−01 9.90E−01 9.90E−01 9.90E−01 MMSE, mean ± NC 29.1 ± 1.2 29 ± 1.2 29.1 ± 1 30 ± 0 9.35E−01 1.00E+00 5.35E−01 9.83E−01 4.60E−01 5.32E−01 SD MCI 28.2 ± 1.6 27.5 ± 1.9 27.5 ± 1.8 27 ± 1.5 2.41E−04 1.71E−03 7.75E−02 1.00E+00 7.42E−01 7.33E−01 AD dementia 23.6 ± 1.9 23.3 ± 2 23.3 ± 2.1 23.6 ± 1.8 9.56E−01 9.15E−01 1.00E+00 9.94E−01 9.80E−01 9.55E−01 Age, mean ± SD NC 73 ± 5.3 73.7 ± 6.5 75.4 ± 5.6 72.8 ± 2.5 7.76E−01 4.23E−02 1.00E+00 1.98E−01 9.94E−01 8.74E−01 MCI 71.7 ± 7.7 73 ± 7.8 72.6 ± 6.9 69.2 ± 7.8 2.58E−01 6.72E−01 6.78E−01 9.41E−01 2.92E−01 4.12E−01 AD dementia 80.6 ± 7.8 76 ± 7.4 72.9 ± 8.5 72.4 ± 5.3 9.87E−02 9.25E−04 1.76E−02 2.83E−02 3.57E−01 9.96E−01 Female, n (%) NC 69 (50%) 88 (53%) 33 (53.2%) 3 (100%) 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 9.90E−01 MCI 65 (37.6%) 108 (40.6%) 73 (44%) 8 (61.5%) 9.90E−01 9.90E−01 9.47E−01 9.90E−01 9.90E−01 9.90E−01 AD dementia 4 (22.2%) 43 (40.6%) 36 (40.9%) 10 (71.4%) 9.90E−01 9.90E−01 9.20E−02 9.90E−01 3.45E−01 3.92E−01 APOE e4 carrier, NC 28 (20.3%) 43 (25.9%) 29 (46.8%) 2 (66.7%) 9.90E−01 1.47E−03 9.90E−01 2.57E−02 9.90E−01 9.90E−01 n (%) MCI 49 (28.3%) 132 (49.6%) 115 (69.3%) 10 (76.9%) 8.90E−05 6.32E−13 5.36E−03 5.41E−04 6.08E−01 9.90E−01 AD dementia 7 (38.9%) 73 (68.9%) 61 (69.3%) 12 (85.7%) 1.71E−01 1.74E−01 1.24E−01 9.90E−01 9.90E−01 9.90E−01 P-tau: Elecsys ADNI N All 467 454 199 63 n.t. n.t. n.t. n.t. n.t. n.t. P-tau pg/ml All ≤ 21 22–36 37–55 > 55 n.t. n.t. n.t. n.t. n.t. n.t. cutoffs N (%) NC 208 (57%) 127 (35%) 24 (7%) 3 (1%) n.t. n.t. n.t. n.t. n.t. n.t. MCI 241 (40%) 229 (38%) 105 (17%) 33 (5%) n.t. n.t. n.t. n.t. n.t. n.t. AD dementia 27 (12%) 98 (44%) 70 (32%) 27 (12%) n.t. n.t. n.t. n.t. n.t. n.t. Abnormal ab42, NC 63 (30.3%) 61 (48%) 23 (95.8%) 2 (66.7%) 9.91E−03 7.59E−09 1.00E+00 2.49E−04 1.00E+00 1.00E+00 n (%) MCI 83 (34.4%) 173 (75.5%) 102 (97.1%) 33 (100%) 5.17E−18 1.25E−25 2.03E−11 1.99E−05 1.75E−02 1.00E+00 AD dementia 18 (66.7%) 92 (93.9%) 68 (97.1%) 26 (96.3%) 2.61E−03 6.13E−04 8.52E−02 1.00E+00 1.00E+00 1.00E+00 MMSE, mean ± NC 29.1 ± 1.2 28.9 ± 1.2 29.5 ± 0.8 29 ± 1 9.35E−01 1.00E+00 5.35E−01 9.83E−01 4.60E−01 5.32E−01 SD MCI 28.1 ± 1.7 27.7 ± 1.8 27.2 ± 1.8 26.9 ± 1.9 2.41E−04 1.71E−03 7.75E−02 1.00E+00 7.42E−01 7.33E−01 AD dementia 24 ± 1.7 23.3 ± 2 23 ± 2.1 23.6 ± 1.8 9.56E−01 9.15E−01 1.00E+00 9.94E−01 9.80E−01 9.55E−01 Age, mean ± SD NC 72.5 ± 5.5 75.2 ± 6.3 77.3 ± 5.5 74.1 ± 3.4 7.76E−01 4.23E−02 1.00E+00 1.98E−01 9.94E−01 8.74E−01 MCI 70.9 ± 7.6 73.4 ± 7.6 73.4 ± 7.3 72.6 ± 7.1 2.58E−01 6.72E−01 6.78E−01 9.41E−01 2.92E−01 4.12E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 9 of 25 Table 3 T-tau and p-tau subgroup comparisons on baseline characteristics (Continued) Biomarker: Cohort Characteristic Cognitive Subgroup Subgroup Subgroup Subgroup p value 1 p value 1 p value 1 p value 2 p value 2 p value 3 platform state 1 2 3 4 vs 2 vs 3 vs 4 vs 3 vs 4 vs 4 AD dementia 78.1 ± 7.7 75.2 ± 7.7 73.8 ± 8.4 72.9 ± 8.7 9.87E−02 9.25E−04 1.76E−02 2.83E−02 3.57E−01 9.96E−01 Female, n (%) NC 101 (48.6%) 75 (59.1%) 11 (45.8%) 3 (100%) 4.77E−01 1.00E+00 1.00E+00 1.00E+00 1.00E+00 1.00E+00 MCI 94 (39%) 89 (38.9%) 46 (43.8%) 20 (60.6%) 1.00E+00 1.00E+00 1.79E−01 1.00E+00 1.76E−01 8.25E−01 AD dementia 7 (25.9%) 35 (35.7%) 37 (52.9%) 15 (55.6%) 1.00E+00 1.84E−01 3.15E−01 2.39E−01 6.04E−01 1.00E+00 APOE e4 carrier, NC 45 (21.6%) 41 (32.3%) 14 (58.3%) 1 (33.3%) 2.51E−01 1.50E−03 1.00E−03 1.66E−01 1.00E+00 1.00E+00 n (%) MCI 72 (29.9%) 122 (53.3%) 79 (75.2%) 25 (75.8%) 2.56E−06 7.89E−14 3.92E−06 1.36E−03 1.48E−01 1.00E+00 AD dementia 10 (37%) 72 (73.5%) 52 (74.3%) 16 (59.3%) 5.81E−03 8.60E−03 1.00E+00 1.00E−03 1.00E+00 1.00E+00 All pairwise comparisons are Tukey HSD adjusted for multiple testing n.t. not tested Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 10 of 25 945 subjects was available (357 normal and 588 MCI sub- dementia subjects had lowest baseline MMSE scores, jects). Subsets of individuals in ADNI also had repeated highest proportion of APOE e4 carriers, lowest levels of CSF tau measures available, for which we tested changes Aβ42, and highest levels of tau. MCI subjects had values over time in tau subgroups, or had TAU PET available, in between NC subjects and AD dementia patients. for which we compared tau uptake according to Braak stages between subgroups. All analyses except for Cox Gaussian mixture modelling reveals four subgroups proportional hazard analyses were stratified for baseline Mixture modeling showed that four distributions (i.e. a cognitive state, and adjusted for age and sex, and cognitive tetramodel distribution) best fitted the data for both t- outcomes additionally for level of education [22]. In Cox tau and p-tau levels, with an optimal fit for four distribu- proportional hazard analyses, no stratification for baseline tions (log-likelihood ratio for 3 vs. 4 distributions, for t- cognitive state was performed due to small size of the tau: 97.2, and p-tau: 28.3, both p < 0.001, no further im- resulting groups; instead, baseline cognitive state was provement for 5 distributions: log-likelihood ratio for 5 added as additional covariate. All statistical analyses were vs 4 distributions, for t-tau: 3.9, and p-tau 15, both p > performed in R version 3.6.1 “Action of the Toes”,mixture 0.05; see Additional file 2 for fit statistics of all fitted modelling was performed with the mixtools package (ver- models, and Fig. 1a for a visualisation of the four distri- sion 1.1.0), estimated marginal means and trends were butions). In the ADC (using Innotest), this yielded three computed with the R package “emmeans” v1.4, and sensi- cutoffs (95% confidence interval (CI)), for t-tau—349 tivity and specificity analyses with epiR v.1.0-15. (304–382), 671 (582–834) and 1380 (1260–1505) pg/mL, and for p-tau—56 (46–60), 96 (71–121) and 159 (138– Results 240) (for n per subgroup defined by cut-points, see Ta- Patient characteristics bles 2 and 3). The first cut-points for t-tau (349 pg/mL) Table 1 shows baseline characteristics of the ADC and and p-tau (56 pg/mL) were comparable to the t-tau and ADNI cohorts. Compared to the ADC, subjects in the p-tau cut-points of 375 pg/ml and 52 pg/ml we previ- ADNI cohort were approximately 10 years older and had ously reported [5], and showed similar sensitivity and a lower prevalence of AD dementia and a higher preva- specificity performance to distinguish between clinical lence of MCI. In ADC, subjects with NC were about AD dementia and controls (see Table 4 for sensitivity 7 years younger compared to MCI and AD patients, and and specificity comparisons). Sensitivity and specificity the NC and the MCI subjects were more often male for distinguishing NC vs MCI were also comparable to than AD dementia subjects. In ADNI, MCI subjects those resulting from the clinical cut-point (Table 4). were youngest, and MCI and AD dementia subjects were T-tau and p-tau strongly correlated across the total more often male than NC. In both cohorts, AD group (r = .92, p < .001); however, when comparing Table 4 Sensitivity and specificity for clinical comparisons NC vs AD-type dementia NC vs MCI First cut-point Cut-point (literature) First cut-point Cut-point (literature) Dataset First cut- Cut-point Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity point (literature) T-tau ADC: 349 375 0.84 0.78 0.81 0.82 0.56 0.78 0.52 0.82 Innotest (0.8–0.86) (0.74–0.82) (0.78–0.84) (0.78–0.86) (0.52–0.6) (0.75–0.81) (0.47–0.56) (0.79–0.84) ADNI: 54 93 0.95 0.4 0.64 0.81 0.73 0.42 0.35 0.81 Luminex (0.9–0.98) (0.33–0.48) (0.55–0.72) (0.74–0.86) (0.69–0.76) (0.36–0.47) (0.31–0.39) (0.76–0.84) ADNI: 192 300 0.94 0.33 0.64 0.78 0.76 0.34 0.37 0.77 Elecsys (0.88–0.97) (0.26–0.4) (0.55–0.72) (0.71–0.84) (0.72–0.79) (0.29–0.39) (0.33–0.4) (0.72–0.81) P-tau ADC: 56 52 0.78 0.77 0.83 0.72 0.54 0.76 0.59 0.7 Innotest (0.75–0.81) (0.72–0.81) (0.8–0.86) (0.68–0.77) (0.5–0.58) (0.73–0.79) (0.55–0.63) (0.67–0.73) ADNI: 23 23 0.9 0.4 0.9 0.4 0.72 0.37 0.72 0.37 Luminex (0.84–0.95) (0.33–0.47) (0.84–0.95) (0.33–0.48) (0.68–0.76) (0.32–0.42) (0.68–0.76) (0.32–0.43) ADNI: 21 24 0.84 0.56 0.8 0.67 0.61 0.56 0.5 0.68 Elecsys (0.76–0.9) (0.48–0.63) (0.72–0.87) (0.6–0.74) (0.57–0.65) (0.51–0.61) (0.46–0.54) (0.63–0.72) MCI mild cognitive impairment, NC normal cognition Source: [5] Source: [21] Source: [23] Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 11 of 25 Fig. 2 T-tau and p-tau subgroup comparisons within each cohort, stratified for cognitive state. Left, ADC; middle, ADNI Luminex; right, ADNI Elecsys. Comparisons for MMSE are shown in (a), for age in (b), for proportion female in (c) and for proportion of APOE-e4 carriers in (d). See Table 3 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 12 of 25 classification of individuals based on t-tau and p-tau, point of t-tau (54 pg/mL) with the cut-point of 93 pg/mL concordance was somewhat lower (79%; Table 2). Apply- previously determined for ADNI [21], our new cut-point ing mixture modelling in ADNI showed that similar to for t-tau resulted in higher sensitivity to detect clinical the ADC, a tetramodal distribution best fitted the CSF t- AD dementia versus controls, at the cost of lower speci- tau and p-tau data (log-likelihood ratio for 3 vs 4 distri- ficity (Table 4). The first p-tau cut-point (23 pg/mL) was butions, for t-tau: 25.2 and for p-tau: 54.3, both with p < identical to the cut-point reported in the literature [21]. 0.05, no further improvement for 5 distributions: log- In ADNI, we further repeated analyses on the novel likelihood ratio for 5 vs 4 distributions, for t-tau: 11.2, Elecsys data as an analytical validation, and again observed and p-tau: 20.3, with p = 0.08 and p = 0.05, respectively). a tetramodal distribution for t-tau (log-likelihood ratio for The tetramodal distribution yielded three different cut- 3 vs 4 distributions, for t-tau: 12.4, and p-tau: 19.6, both points for t-tau measured with Luminex (95%CI)—54 with p < 0.05, no further improvement for 5 distributions: (42–68), 95 (68–125) and 171 (146–263) pg/mL respect- log-likelihood ratio for 5 vs 4 distributions, for t-tau: 7.1, ively (Fig. 1b), and for p-tau levels (95%CI)—23 (20–28), and p-tau 10.6, both p > 0.05). The tetramodal distribution 46 (38–57) and 99 (74–124). Comparing the first cut- yielded for t-tau the cut-points (95%CI)—192 (129–235), Fig. 3 Comparison of annual MMSE decline for t-tau and p-tau subgroups, stratified for cognitive state. Left, t-tau; right, p-tau; top, ADC; middle, ADNI Luminex; bottom, ADNI Elecsys. See Table 5 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 13 of 25 312 (239–405) and 515 (417–679), and for p-tau (95%CI)—21 (17–23), 36 (28–43) and 55 (47–68) (Fig. 1c). For the first t- tau cut-point measured with Elecsys (192 pg/mL) compared to the cut- point of 300 pg/mL previ- ously reported for ADNI [23], our new cut-point also yielded higher sensi- tivity but lower specificity for Elecsys t-tau (Table 4). For Elecsys p-tau, the first cut-point of 21 pg/ mL was comparable to a previously reported cut- point of 24 pg/mL [23] and yielded similar sensi- tivity and lower specifi- city estimates (Table 4). In ADNI, 81% of the sub- jects were labelled identi- cally using t-tau labels for Luminex and Elecsys, whereas between- platform correspondence for p-tau labelling was 60% (Table 2). For Lumi- nex, the correlation be- tween t-tau and p-tau was .67 (p <.001) and the correspondence of t-tau and p-tau labelling was 52%. For Elecsys, the cor- relation between t-tau and p-tau was .98 (p <.001), and the corres- pondence of subgroup la- belling was 71%. Clinical and biological characteristics of tau subgroups Gradually higher t- and p-tau subgroups in ADC were characterized by increasingly high preva- lence of abnormal amyl- oid, with the highest two t-tau and p-tau groups consisting for more than 94% of amyloid Table 5 Tau subgroup comparisons on MMSE at first visit and annual change rates Subgroup Baseline estimated marginal means ± SE of p values of pairwise comparisons effect subgroup between subgroups Bio- Cognitive Cohort: N per p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 marker state platform subgroup 1/2/3/4 T-tau NC ADC: Innotest 240/73/13/4 0.428 28.4 ± 0.1 28 ± 0.2 27.3 ± 0.4 23.8 ± 0.9 6.45E−02 1.07E−02 2.00E−07 1.15E−01 2.20E−06 ADNI: Luminex 151/151/60/2 0.231 28.8 ± 0.1 28.8 ± 0.1 28.6 ± 0.1 29.3 ± 0.6 8.90E−01 1.07E−01 5.10E−01 7.92E−02 5.25E−01 ADNI: Elecsys 124/163/67/3 0.725 28.8 ± 0.1 28.8 ± 0.1 28.6 ± 0.1 28.7 ± 0.5 7.25E−01 1.58E−01 8.06E−01 2.20E−01 8.65E−01 MCI ADC: Innotest 199/170/92/4 0.270 26.6 ± 0.2 25 ± 0.2 24.2 ± 0.3 24.2 ± 1.7 7.98E−09 1.13E−11 1.47E−01 2.99E−02 6.39E−01 ADNI: Luminex 162/237/145/56 0.0002 27.9 ± 0.2 26.4 ± 0.2 25.1 ± 0.2 24 ± 0.4 1.36E−06 4.46E−16 4.12E−17 3.32E−05 3.80E−08 ADNI: Elecsys 145/247/154/41 0.0001 27.7 ± 0.2 26.7 ± 0.2 25.1 ± 0.2 23.7 ± 0.5 9.96E−04 2.04E−14 4.56E−14 1.12E−07 3.71E−09 AD dementia ADC: Innotest 122/303/280/38 0.172 20.4 ± 0.4 19.7 ± 0.3 18.9 ± 0.3 17.1 ± 0.8 1.58E−01 3.00E−03 1.58E−04 3.87E−02 1.32E−03 ADNI: Luminex 13/63/95/43 0.031 23.9 ± 0.9 22 ± 0.4 21.6 ± 0.3 20.8 ± 0.5 5.04E−02 1.41E−02 2.35E−03 4.16E−01 5.82E−02 ADNI: Elecsys 13/68/99/28 0.008 25 ± 0.9 21.8 ± 0.4 21.3 ± 0.3 21.3 ± 0.6 1.00E−03 1.33E−04 6.41E−04 3.69E−01 4.77E−01 P-tau NC ADC: Innotest 237/76/15/2 0.120 28.4 ± 0.1 28.2 ± 0.2 27.5 ± 0.4 22.2 ± 1 2.84E−01 2.48E−02 1.10E−08 1.03E−01 3.91E−08 ADNI: Luminex 137/164/59/3 0.883 28.9 ± 0.1 28.7 ± 0.1 28.7 ± 0.1 29.1 ± 0.8 6.64E−02 1.13E−01 7.68E−01 7.89E−01 5.74E−01 ADNI: Elecsys 206/124/24/2 0.367 28.8 ± 0.1 28.7 ± 0.1 28.5 ± 0.2 29.2 ± 0.7 2.81E−01 1.70E−01 5.56E−01 4.27E−01 4.39E−01 MCI ADC: Innotest 208/173/79/5 0.575 26.3 ± 0.2 25.1 ± 0.2 24.3 ± 0.3 23.8 ± 1.4 6.13E−05 7.46E−08 7.69E−02 2.19E−02 3.41E−01 ADNI: Luminex 164/259/161/12 0.009 27.7 ± 0.2 26.3 ± 0.2 25 ± 0.2 23.8 ± 0.8 1.03E−06 8.81E−17 8.02E−06 1.20E−05 4.47E−03 ADNI: Elecsys 228/224/103/32 0.011 27.6 ± 0.2 26.1 ± 0.2 24.6 ± 0.3 23.7 ± 0.5 6.25E−08 3.34E−18 1.74E−12 8.99E−06 1.14E−05 AD dementia ADC: Innotest 149/348/216/30 0.250 20.2 ± 0.4 19.4 ± 0.2 19 ± 0.3 16.3 ± 0.9 7.19E−02 1.25E−02 4.35E−05 2.97E−01 6.82E−04 ADNI: Luminex 17/98/83/14 0.483 23.7 ± 0.8 21.5 ± 0.3 21.5 ± 0.4 21.8 ± 0.9 1.21E−02 1.30E−02 1.20E−01 9.54E−01 7.47E−01 ADNI: Elecsys 26/91/66/25 0.118 23.4 ± 0.7 21.8 ± 0.3 21.2 ± 0.4 21.2 ± 0.7 3.21E−02 4.37E−03 1.91E−02 2.39E−01 4.10E−01 a b c All analyses were stratified for cognitive state and adjusted for age, sex and level of education. Slope that differs from 0 is indicated with when p < .05, when p < .01 or when p < .001 p values of pairwise comparisons Time Subgroup Estimated slopes ± SE of p value of pairwise comparisons of slope NC cognitively normal, MCI mild cognitive impairment, AD Alzheimer’s disease between subgroups effect x Time subgroup differences between subgroups Bio- 3vs4 p value p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 3 vs 4 marker Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 14 of 25 Table 5 Tau subgroup comparisons on MMSE at first visit and annual change rates (Continued) p values of pairwise comparisons Time Subgroup Estimated slopes ± SE of p value of pairwise comparisons of slope between subgroups effect x Time subgroup differences between subgroups Bio- 3vs4 p value p value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 2 vs 4 3 vs 4 marker a c T-tau 2.54E−04 1.14E−07 1.34E−05 0 ± 0.1 −0.2 ± 0.1 −0.3 ± 0.2 −2.2 ± 0.5 6.16E−02 1.51E−01 2.65E−06 7.06E−01 1.97E−05 8.86E−05 b c 3.06E−01 3.12E−01 3.23E−03 0±0 −0.1 ± 0 − 0.3 ± 0 0.1 ± 0.3 1.38E−01 3.36E−04 5.15E−01 1.21E−02 3.55E−01 1.36E−01 b c 8.82E−01 9.20E−02 1.15E−01 0 ± 0 −0.1 ± 0 −0.2 ± 0 0 ± 0.2 2.32E−01 1.60E−02 9.47E−01 1.20E−01 7.42E−01 4.69E−01 a c c 9.71E−01 2.05E−03 8.94E−13 −0.2 ± 0.1 −1.0 ± 0.1 − 1.3 ± 0.1 −0.6 ± 1 2.48E−10 3.78E−11 6.98E−01 1.35E−01 6.68E−01 5.13E−01 a c c c 1.39E−02 1.74E−47 7.39E−17 −0.2 ± 0.1 −0.7 ± 0.1 −1.2 ± 0.1 − 1.6 ± 0.2 4.68E−05 3.56E−13 1.82E−14 7.80E−05 1.39E−07 1.69E−02 a c c c 9.62E−03 7.02E−43 7.36E−16 −0.2 ± 0.1 − 0.6 ± 0.1 −1.1 ± 0.1 − 1.8 ± 0.2 5.03E−04 2.90E−11 3.83E−14 6.21E−05 5.28E−09 9.50E−04 c c c c 2.61E−02 1.72E−71 1.73E−04 −1.7 ± 0.2 −2.0 ± 0.1 −2.3 ± 0.1 − 3.5 ± 0.4 1.49E−01 1.55E−02 1.95E−05 2.00E−01 1.92E−04 1.69E−03 a c c c 1.77E−01 3.33E−16 1.29E−01 −1.5 ± 0.8 −2.0 ± 0.4 − 2.4 ± 0.3 − 3.2 ± 0.5 6.20E−01 3.06E−01 6.60E−02 3.67E−01 3.53E−02 1.34E−01 c c c 9.33E−01 8.53E−13 3.68E−02 −0.3 ± 0.8 −2.2 ± 0.4 − 2.4 ± 0.3 − 3.0 ± 0.6 2.48E−02 9.79E−03 4.88E−03 6.24E−01 2.33E−01 3.65E−01 P-tau 3.65E−06 2.80E−07 3.36E−05 −0.1 ± 0.1 −0.1 ± 0.1 −0.2 ± 0.2 −2.5 ± 0.5 3.79E−01 4.50E−01 2.05E−06 7.62E−01 5.29E−06 2.55E−05 5.43E−01 2.10E−01 1.35E−01 0 ± 0 −0.1 ± 0 −0.2 ± 0.1 −0.1 ± 0.4 5.87E−02 4.56E−02 8.32E−01 4.88E−01 9.82E−01 8.85E−01 a c c c 3.15E−01 1.24E−01 8.11E−03 −1.3 ± 0.6 −2.1 ± 0.3 −2.7 ± 0.4 −3.0 ± 0.6 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 c c c b 7.18E−01 2.94E−09 5.39E−08 −0.4 ± 0.1 −0.9 ± 0.1 −1.2 ± 0.1 −2.0 ± 0.7 1.32E−05 1.08E−07 2.23E−02 6.44E−02 1.32E−01 2.75E−01 a c c c 1.95E−01 2.19E−27 3.33E−16 −0.2 ± 0.1 −0.7 ± 0.1 −1.3 ± 0.1 −2.0 ± 0.3 1.04E−04 2.19E−16 3.76E−07 1.31E−07 1.50E−04 6.60E−02 b b c c 1.32E−01 1.76E−46 5.50E−19 −0.2 ± 0.1 −0.8 ± 0.1 −1.4 ± 0.1 −1.8 ± 0.2 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 c c c c 4.03E−03 2.75E−65 9.95E−05 −1.8 ± 0.2 −2 ± 0.1 −2.4 ± 0.1 −3.7 ± 0.4 3.14E−01 5.30E−03 7.50E−05 2.36E−02 2.48E−04 6.23E−03 c c a 7.26E−01 1.62E−10 9.00E−02 −0.7 ± 0.7 −2.5 ± 0.3 −2.6 ± 0.3 −2.1 ± 0.8 2.00E−02 1.33E−02 2.06E−01 7.23E−01 6.68E−01 5.53E−01 a c c c 9.88E−01 1.88E−18 9.79E−02 −1.3 ± 0.6 −2.1 ± 0.3 −2.7 ± 0.4 −3.0 ± 0.6 1.74E−01 3.60E−02 3.29E−02 2.52E−01 1.83E−01 5.96E−01 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 15 of 25 abnormal participants. This relationship also held for associations of higher t- and p-tau subgroups with amyl- lower t-tau values, with a higher prevalence of abnormal oid, cognitive state, and demographic factors were amyloid in the second subgroup than the lowest tau sub- mostly reproduced in ADNI. group. T- and p-tau subgroups were also associated with cognitive state, with lower subgroups containing the Rates of cognitive decline over time depend on tau highest proportion of cognitively normal participants, subgroups while highest subgroups contained more demented par- We further studied whether subjects across tau sub- ticipants (Table 3). Therefore, we stratified subsequent groups differed in rates of cognitive decline, as measured comparisons between tau subgroups for cognitive state. with the MMSE stratified for cognitive state. In ADC, Average MMSE was lower for higher tau subgroups, tau subgroups were not associated with cognitive decline with the strongest effects observed in AD-type dementia in MCI or NC; however, in the dementia phase, higher (Fig. 2; Table 3). Tau subgroups also differed in demo- tau subgroups were characterized by faster cognitive de- graphic factors, including age (on average lower in the cline on MMSE (Fig. 3; Table 4). In ADNI, faster MMSE lowest tau subgroup in NC and MCI), sex (higher pro- decline with higher tau subgroups in dementia was portion of women in higher t-tau and p-tau subgroups), reproduced. While in ADC no association between tau and APOE e4 carriership (higher prevalence in higher t- subgroups and MMSE decline was found for participants tau and p-tau subgroups) (Fig. 2; Table 3). The Fig. 4 Proportional Hazard curves for progression to dementia in initially non-demented individuals in different tau subgroups. T-tau subgroups are shown in (a) and p-tau subgroups in (b); left, ADC; middle, ADNI Luminex; right, ADNI Elecsys. See Table 6 for statistical descriptions Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 16 of 25 Table 6 Cox proportional hazard models for time to progression in individuals with NC or MCI Model 1: no Model 2: + sex, Model 3: model Model 4: model Model 5: model 4+ covariates age, education 2 + amyloid status 3 + diagnosis APOE-e4 carriership HR (95%CI) p HR (95%CI) sex, p HR (95%CI) sex, p HR (95%CI) sex, age, p HR (95%CI) sex, age, edu, p age, edu age, edu, abeta edu, abeta, diagnosis abeta, diagnosis, APOE-e4 T-tau subgroups Subgroup 1 All Reference Subgroup 2 ADC: Innotest 6.3 (4.3, 9.2) 5.79E 6 (3.9, 9.2) 1.32E 3.4 (2.2, 5.3) 5.96E 3.4 (2.1, 5.3) 1.08E 3.3 (2.1, 5.3) 4.31E −21 −16 −08 −07 −07 ADNI: Luminex 2.1 (1.4, 3) 8.76E 2 (1.4, 3) 1.58E 1.7 (1.2, 2.5) 5.84E 1.6 (1.1, 2.4) 1.24E 1.5 (1, 2.2) 2.98E −05 −04 −03 −02 −02 ADNI: Elecsys 1.9 (1.3, 2.8) 9.27E 1.8 (1.2, 2.7) 2.44E 1.8 (1.2, 2.7) 2.56E 1.8 (1.2, 2.6) 3.71E 1.7 (1.1, 2.5) 9.74E −04 −03 −03 −03 −03 Subgroup 3 ADC: Innotest 14.6 (9.8, 6.14E 13.2 (8.3, 20.8) 3.31E 6.2 (3.8, 10) 1.96E 6.0 (3.7, 9.8) 6.96E 6.1 (3.7, 10.1) 1.86E 21.8) −39 −28 −13 −13 −12 ADNI: Luminex 5.2 (3.6, 7.6) 1.31E 4.9 (3.4, 7.2) 3.53E 2.8 (1.9, 4.1) 2.86E 2.6 (1.8, 3.9) 1.57E 2.4 (1.6, 3.6) 1.06E −18 −17 −07 −06 −05 ADNI: Elecsys 4.4 (3, 6.4) 6.15E 4 (2.7, 5.9) 2.05E 3.1 (2.1, 4.5) 1.88E 2.8 (1.9, 4.2) 1.91E 2.7 (1.8, 4) 1.60E −14 −12 −08 −07 −06 Subgroup 4 ADC: Innotest 21.3 (7.5, 9.81E 15 (4.4, 50.9) 1.43E 6.5 (1.9, 22.3) 2.93E 6.6 (1.9, 22.7) 2.70E 6.4 (1.9, 22.2) 3.30E 60.6) −09 −05 −03 −03 −03 ADNI: Luminex 6.7 (4.2, 10.7) 1.01E 6.6 (4.1, 10.6) 3.89E 3.4 (2.1, 5.6) 8.31E 2.8 (1.7, 4.5) 6.50E 2.6 (1.5, 4.2) 2.63E −15 −15 −07 −05 −04 ADNI: Elecsys 7.1 (4.3, 11.7) 2.96E 7.2 (4.3, 12) 2.37E 4.6 (2.7, 7.7) 6.71E 3.9 (2.3, 6.6) 2.99E 3.5 (2.1, 6.1) 3.59E −14 −14 −09 −07 −06 P-tau subgroups Subgroup 1 All Reference Subgroup 2 ADC: Innotest 4.4 (3.1, 6.2) 4.13E 3.9 (2.7, 5.7) 2.35E 2.1 (1.4, 3.2) 1.71E 2.1 (1.4, 3.1) 2.28E 2.1 (1.4, 3.2) 4.88E −17 −12 −04 −04 −04 ADNI: Luminex 2.2 (1.5, 3) 7.78E 2.1 (1.5, 3) 1.38E 1.5 (1.0, 2.1) 3.30E 1.4 (1.0, 2.0) 5.85E 1.4 (1, 2) 7.19E −06 −05 −02 −02 −02 ADNI: Elecsys 2.6 (1.9, 3.6) 9.65E 2.6 (1.9, 3.5) 3.83E 2.2 (1.6, 3.0) 1.73E 2.0 (1.5, 2.8) 1.22E 2 (1.4, 2.7) 4.34E −10 −09 −06 −05 −05 Subgroup 3 ADC: Innotest 10.2 (7, 14.8) 1.48E 8.4 (5.6, 12.8) 1.10E 3.7 (2.4, 5.8) 6.58E 3.6 (2.3, 5.6) 2.03E 3.5 (2.2, 5.6) 1.02E −33 −23 −09 −08 −07 ADNI: Luminex 4.4 (3.1, 6.3) 1.81E 4.3 (3.0, 6.2) 4.37E 2.3 (1.6, 3.3) 1.80E 2.2 (1.5, 3.2) 7.21E 2 (1.4, 3) 2.88E −16 −16 −05 −05 −04 ADNI: Elecsys 5.0 (3.6, 7.1) 7.09E 4.7 (3.3, 6.7) 4.35E 3.0 (2.1, 4.3) 4.19E 2.7 (1.9, 3.9) 1.32E 2.6 (1.8, 3.7) 8.22E −20 −18 −09 −07 −07 Subgroup 4 ADC: Innotest 9.5 (3.4, 26.5) 1.69E 4.6 (1.4, 15.5) 1.28E 2.1 (0.6, 7.2) 2.19E 2.1 (0.6, 7.1) 2.29E 4 (1.2, 13.8) 2.72E −05 −02 −01 −01 −02 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 17 of 25 Table 6 Cox proportional hazard models for time to progression in individuals with NC or MCI (Continued) Model 1: no Model 2: + sex, Model 3: model Model 4: model Model 5: model 4+ covariates age, education 2 + amyloid status 3 + diagnosis APOE-e4 carriership HR (95%CI) p HR (95%CI) sex, p HR (95%CI) sex, p HR (95%CI) sex, age, p HR (95%CI) sex, age, edu, p age, edu age, edu, abeta edu, abeta, diagnosis abeta, diagnosis, APOE-e4 ADNI: Luminex 5.0 (2.4, 10.6) 2.83E 5.8 (2.7, 12.4) 5.32E 2.5 (1.2, 5.5) 1.76E 2.3 (1.0, 4.9) 3.81E 2.1 (0.9, 4.5) 6.85E −05 −06 −02 −02 −02 ADNI: Elecsys 7.7 (4.7, 12.6) 3.47E 8.2 (4.9, 13.4) 1.24E 4.7 (2.8, 7.7) 3.52E 3.9 (2.3, 6.6) 2.51E 3.6 (2.1, 6.1) 2.11E −16 −16 −09 −07 −06 Amyloid status Amyloid abnormal ADC: Innotest (< 10.8 (7.3, 8.71E 8.9 (5.9, 13.4) 6.65E n.t. 7.8 (5.0, 12.0) 1.88E 7.2 (4.5, 11.5) 5.27E 813 pg/ml) 15.9) −33 −25 −20 −17 ADNI: Luminex (< 5.2 (3.8, 7.0) 2.29E 5.0 (3.6, 6.8) 4.91E n.t. 4.4 (3.2, 6.0) 1.79E 3.8 (2.7, 5.3) 4.16E 192 pg/ml) −25 −24 −20 −15 ADNI: Elecsys (< 4.2 (3.2, 5.5) 1.60E 4.1 (3.1, 5.3) 8.27E n.t. 3.5 (2.7, 4.6) 1.30E 3 (2.2, 4) 2.51E 880 pg/ml) −26 −25 −19 −13 Continuous predictors Continuous ab1-42 (z ADC: Innotest 0.3 (0.2, 0.4) 3.31E 0.3 (0.3, 0.4) 8.16E 0.6 (0.4, 0.8) 2.84E 0.6 (0.4,0.9) 4.17E 0.6 (0.4, 0.8) 4.75E score; HR per SD) −36 −27 −03 −03 −03 ADNI: Luminex 0.5 (0.4, 0.5) 1.71E 0.5 (0.4, 0.5) 6.62E 0.6 (0.5, 0.8) 7.12E 0.7 (0.5, 0.9) 2.19E 0.7 (0.6, 0.9) 1.79E −30 −23 −05 −03 −02 ADNI: Elecsys 0.4 (0.3, 0.5) 2.24E 0.4 (0.3, 0.5) 7.05E 0.6 (0.4, 0.8) 1.14E 0.6 (0.5,0.8) 1.56E 0.6 (0.5, 0.8) 5.41E −23 −22 −04 −04 −04 Continuous t-tau (z score; ADC: Innotest 1.6 (1.5, 1.7) 7.67E 1.9 (1.7, 2.1) 3.60E 1.5 (1.3, 1.7) 2.42E 1.5 (1.3, 1.7) 4.57E 1.6 (1.4, 1.8) 3.46E HR per SD) −58 −32 −11 −11 −11 ADNI: Luminex 1.7 (1.6, 1.9) 7.95E 1.7 (1.6, 1.9) 5.61E 1.4 (1.3, 1.6) 1.80E 1.4 (1.2, 1.5) 1.15E 1.3 (1.2, 1.5) 1.69E −32 −31 −11 −08 −07 ADNI: Elecsys 1.6 (1.5, 1.8) 8.36E 1.7 (1.5, 1.8) 1.39E 1.4 (1.3, 1.6) 2.41E 1.4 (1.3, 1.5) 1.51E 1.4 (1.2, 1.5) 4.90E −27 −25 −13 −10 −09 Continuous p-tau (z ADC: Innotest 1.8 (1.7, 2.0) 2.92E 1.7 (1.5, 1.9) 3.31E 1.4 (1.2, 1.6) 1.11E 1.4 (1.2, 1.6) 3.56E 1.5 (1.3, 1.7) 3.03E score; HR per SD) −42 −22 −07 −07 −08 ADNI: Luminex 1.5 (1.4, 1.6) 5.58E 1.6 (1.4, 1.7) 1.75E 1.3 (1.2, 1.5) 1.67E 1.3 (1.2, 1.4) 2.46E 1.3 (1.1, 1.4) 4.28E −24 −25 −08 −07 −06 ADNI: Elecsys 1.7 (1.5, 1.8) 3.36E 1.7 (1.5, 1.9) 1.06E 1.5 (1.3, 1.6) 1.72E 1.4 (1.3, 1.5) 9.62E 1.4 (1.2, 1.5) 3.05E −31 −29 −13 −11 −09 N.t. not tested Source: Tijms BM et al., Clinical Chemistry. 2018;64(3):576–585 Source: [21] Source: Hansson O et al., Alzheimer’s & Dementia. 2018;14(11):1470–1481 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 18 of 25 Table 7 Frequencies of individuals remaining or changing and still showed higher HRs for progression to AD-type subgroup over time from baseline (rows) dementia compared to the first tau subgroup (HR Biomarker: Platform Subgroup at last (95%CI) = 2.1 (1.4, 3.0), p < .001). measurement T-tau: Luminex Baseline subgroup 1 2 3 4 Longitudinal changes in tau concentrations in ADNI 1 110 35 1 0 Examining transitions over time to higher tau groups in ADNI, we observed that the majority of individuals for 2 22 188 42 0 both Luminex and Elecsys t-tau subgroups remained in 3 0 18 139 15 the same subgroup as first measured (Luminex: 472 40 1 13 35 (76% of 619); Elecsys: 443 (76% of 586); Table 7; see T-tau: Elecsys Baseline subgroup 1 2 3 4 Table 8 and Fig. 5 for continuous results). Of individuals 192 25 1 0 who changed, the majority shifted to one tau group 2 11 206 36 0 higher (Table 8). 3 0 17 147 13 Comparison with tau PET in ADNI 40 1 4 33 Finally, we compared CSF tau subgroups on tau PET up- P-tau: Luminex Baseline subgroup 1 2 3 4 take values available for 345 individuals (235 NC; 93 1 103 63 11 0 MCI; 28 dementia; of note, these included n = 232 new 2 24 164 91 5 CSF observations not included in mixture analyses). Fig- 3 2 23 100 13 ure 6 shows that tau PET uptake increased with higher t-tau and p-tau subgroups. For all Braak regions, the up- 40 1 6 6 take of the highest two tau subgroups was significantly P-tau: Elecsys Baseline subgroup 1 2 3 4 higher than the lowest two (or three) subgroups 1 185 33 0 0 (Table 9). The second lowest t-tau subgroup also 2 9 197 25 0 showed higher average tau uptake in Braak I/II brain 3 0 14 75 12 areas compared to subgroup 1, and the second lowest p- 40 0 4 33 tau subgroup in addition also to Braak III/IV and V/VI compared to subgroup 1. with MCI, in ADNI, higher tau subgroups in MCI were associated with MMSE decline (Table 5). Discussion Next, we tested for individuals without dementia (i.e. In this study, we used Gaussian mixture modelling to NC and MCI) whether tau-subgroups differed in terms determine unbiased cut-points for CSF tau levels. We of progression to MCI or AD-type dementia. In the identified three cut-points resulting in four different dis- ADC, 46/381 (12%) of NC patients showed clinical pro- tributions, and the cut-point between the lowest two gression either to MCI (n = 39) within 2.3 ± 1.6 years or subgroups corresponded closely to an existing clinically to AD-type dementia (n = 16) in 4.5 ± 4.0 years, and 178/ defined cut-point [21]. Furthermore, two additional tau 591 (30%) of MCI patients progressed to AD-type de- groups with highest t- and p-tau levels were discovered mentia in 2.4 ± 1.6 years. Across the total group of non- in the data. We similarly observed four distributions in demented subjects, hazard ratios (HRs) increased with the independent ADNI cohort, and despite differences increasing tau or p-tau subgroups compared to the low- between ADC and ADNI in cohort composition, tau est tau or p-tau subgroups (Fig. 4; Table 6). Repeating subgroups showed similar clinical and biological charac- analyses including covariates sex, age and education level teristics in both study cohorts. These findings suggest (model 2), amyloid status (model 3), baseline cognitive that t-tau and p-tau levels may not necessarily reflect state (model 4) and APOE-e4 carriership (model 5) gen- disease stage, but possibly different biological subtypes erated largely similar results for t-tau subgroups, al- of AD. though HRs were somewhat attenuated. Results were Tau is an intracellular protein playing an important largely consistent for ADNI albeit with somewhat lower role in microtubule assembly and stabilization in axons HR values (Table 6), where 65/371 (17.5%) NC showed [24]. Hyperphosphorylation disturbs its function, result- clinical progression either to MCI (n = 47) within 3 ± 9 ing in the formation of aggregates or neurofibrillary tan- years or to AD-type dementia (n = 18) in 8 ± 3 years, and gles, which is one of the hallmarks of AD pathology. 212/622 (34%) MCI individuals to AD-type dementia in Still, the precise factors influencing t- and p-tau CSF 4.1 ± 2.3 years. Of note is that in ADNI, individuals in levels remain unclear. Measures correlated highly, and the second Luminex t-tau subgroup had levels below the even though subgroup labelling showed moderate con- official cut-point defined by ADNI (i.e. 93 pg/ml [21]) cordance, t-tau and p-tau subgroups showed similar Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 19 of 25 Table 8 T-tau and p-tau subgroup comparisons on annual change in CSF t-tau and p-tau values Effect Interaction Annual change (se) for each subgroup p values pairwise comparisons in slope time subgroup x time differences between subgroups Biomarker: Fp value Fp value Subgroup 1 Subgroup 2 Subgroup 3 Subgroup 4 1 vs 2 1 vs 3 1 vs 4 2 vs 3 platform a c b T-tau: 14 1.74E−04 0.2 8.96E−01 1.72 (0.75) 2.03 (0.57) 2.2 (0.71) 1.08 (1.44) 7.45E−01 6.43E−01 6.94E−01 8.50E−01 Luminex a c c a T-tau: 37 2.55E−09 2.5 5.98E−02 3.34 (1.68) 4.8 (1.16) 8.71 (1.46) 8.36 (3.31) 4.74E−01 1.62E−02 1.77E−01 3.67E−02 Elecsys c c b c P-tau: 2 1.63E−01 15 2.93E−09 3.17 (0.72) 4.67 (0.6) 2.89 (0.88) −15.23 (2.95) 1.11E−01 8.07E−01 2.58E−09 9.58E−02 Luminex a c a P-tau: 0.005 2.85E−02 1.5 2.13E−01 0.26 (0.13) 0.48 (0.13) 0.49 (0.21) −0.22 (0.37) 2.12E−01 3.28E−01 2.25E−01 9.68E−01 Elecsys a b c CSF t- and p-tau values are in pg/ml. Baseline effects are reported in the last columns. Bold font highlights significant effects. Slope that differs from 0 is indicated with when p < .05, when p < .01 or when p < .001 SE standard error Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 20 of 25 Table 8 T-tau and p-tau subgroup comparisons on annual change in CSF t-tau and p-tau values (Continued) p values pairwise comparisons in slope Baseline Baseline estimated marginal means (SE) for p values pairwise comparisons in baseline estimates between differences between subgroups effect tau group each subgroup subgroups Biomarker: 2vs4 3 vs4 Fp value Subgroup Subgroup Subgroup Subgroup 1vs2 1vs3 1vs4 2vs3 2vs4 3vs4 platform 1 2 3 4 T-tau: 5.41E−01 4.86E−01 1318 7.48E−273 44.6 (1.89) 76.4 (1.43) 129.3 (1.73) 221.9 (3.24) 1.65E−36 1.83E 6.43E 1.68E−87 3.96E 3.88E Luminex −137 −206 −177 −96 T-tau: 3.11E−01 9.23E−01 1330 1.30E−261 162.4 (4.49) 254.1 (3.05) 392.8 (3.64) 616.8 (7.86) 7.82E−53 1.58E 1.28E 7.20E 1.89E 4.47E Elecsys −168 −215 −117 −184 −99 P-tau: 8.98E−11 6.95E−09 808 6.11E−222 21.8 (1.06) 39.1 (0.84) 65.5 (1.22) 109.5 (3.99) 9.63E−33 2.98E 1.14E−73 5.97E−56 9.88E−54 7.98E Luminex −101 −24 P-tau: 7.35E−02 9.28E−02 1456 3.10E−270 16.8 (0.36) 28.4 (0.34) 43.4 (0.52) 66.0 (0.90) 1.96E 7.62E 7.62E 1.64E−90 1.76E 6.98E Elecsys −181 −221 −221 −167 −78 Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 21 of 25 ADNI Elecsys t-tau a) ADNI Luminex t-tau CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 300 700 0 0 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 250 600 0 0 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 250 600 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 250 600 01234567 01234567 01234567 01234567 01234567 01234567 Follow−up (years) Follow−up (years) b) ADNI Luminex p-tau ADNI Elecsys p-tau CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 CN Subtype 1 MCI Subtype 1 Dementia Subtype 1 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 CN Subtype 2 MCI Subtype 2 Dementia Subtype 2 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 CN Subtype 3 MCI Subtype 3 Dementia Subtype 3 100 100 80 80 60 60 40 40 20 20 0 0 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 CN Subtype 4 MCI Subtype 4 Dementia Subtype 4 100 100 80 80 60 60 40 40 20 20 0 0 01234567 01234567 01234567 01234567 01234567 01234567 Follow−up (years) Follow−up (years) Fig. 5 Changes over time in t-tau and p-tau levels, stratified for tau subgroup and cognitive state. Changes in t-tau levels are shown in (a) and in p-tau levels in (b). Left, ADNI Luminex; right, ADNI Elecsys. See Table 7 for statistical descriptions. NC, normal cognition; MCI, mild cognitive impairment; AD dementia, AD-type dementia T−tau pg/ml P−tau pg/ml P−tau pg/ml T−tau pg/ml Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 22 of 25 Fig. 6 Comparison of t-tau and p-tau subgroups in tau PET uptake according to Braak stages. Tau PET uptake for t-tau subgroups are shown in (a)and for p-tau subgroups in (b). Left, ADNI Luminex; right, ADNI Elecsys. See Table 9 for statistical descriptions. SUVr, standardized uptake value ratio Table 9 T-tau and p-tau subgroup comparison on tau PET uptake T-tau T-tau subgroup estimated marginal means (SE) T-tau subgroup pairwise comparisons p value subgroup effect Tau PET Subgroup Fp value Subgroup Subgroup Subgroup Subgroup 1vs2 1vs3 1vs4 2vs3 2vs4 3vs4 SUVr 1 2 3 4 Braak I/II T-tau 28 3.69E 1.18 (0.016) 1.22 (0.013) 1.36 (0.018) 1.51 (0.052) 2.91E 1.91E 1.65E 8.36E 1.03E 4.78E −16 −02 −12 −09 −09 −07 −03 P-tau 37.6 5.14E 1.18 (0.012) 1.27 (0.014) 1.43 (0.024) 1.47 (0.049) 1.02E 1.97E 1.95E 3.40E 1.22E 4.67E−01 −21 −06 −18 −08 −08 −04 Braak III/IV T-tau 30 2.64E 1.13 (0.019) 1.17 (0.016) 1.34 (0.022) 1.56 (0.061) 5.19E−02 2.98E 6.58E 3.73E 3.08E 7.24E −17 −12 −11 −09 −09 −04 P-tau 31 1.04E 1.13 (0.014) 1.23 (0.017) 1.38 (0.029) 1.51 (0.059) 1.79E 1.50E 7.23E 7.87E 3.88E 4.08E −17 −05 −13 −10 −06 −06 −02 Braak V/VI T-tau 16 7.49E 1.05 (0.018) 1.09 (0.015) 1.21 (0.021) 1.32 (0.058) 1.50E−01 3.80E 1.01E 2.69E 8.12E 5.80E−02 −10 −08 −05 −06 −05 P-tau 14.2 1.04E 1.06 (0.014) 1.13 (0.016) 1.22 (0.028) 1.27 (0.056) 4.88E 1.21E 1.97E 4.22E 1.49E 4.21E−01 −08 −04 −07 −04 −03 −02 Bold font indicates significant group difference Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 23 of 25 differences in tau PET uptake. Previous studies compar- than disease severity per se. This is supported by the ob- ing CSF tau measures with tau PET have been inconsist- servation that higher tau subgroups showed increasing ent [25–27]. Together with our results, it remains proportions of APOE e4 carriers, the strongest genetic unclear whether CSF t-tau and p-tau reflect similar or risk factor for AD [35]. Previous studies have also re- different aspects of neuronal injury. Higher levels of t- ported higher levels of tau in APOE-e4 carriers, also in and p-tau might result from passive release into extra- predementia stages [36]; however, also see [37] where cellular space due to neuronal death which increases tau levels were similar between carriers and non- with worse disease severity. However, tau is also actively carriers. Other genetic risk factors may contribute to dif- secreted by neurons as part of normal physiology [28] ferences in tau levels as well, as another study reported and can increase in the presence of amyloid pathology that a polygenic risk score, including SNPs with moder- [29]. The majority of individuals remained in their t-tau ate strength to detect AD, was strongly related to t-tau subgroup over time, suggesting that at least part of their and p-tau levels, also after correcting for APOE [38]. levels do not depend on disease stage, but perhaps re- This suggests that multiple genetic risk factors may ex- flect other biological aspects. The relative lack of change plain variability between individual tau levels. More over time in tau levels within individuals seems at odds studies with large sample sizes are needed to further in- with the idea that tau increases with worsening cogni- vestigate these biological factors associated with tau tion. Previous longitudinal CSF studies have reported levels in CSF. Also, future studies should further investi- conflicting results, observing increases in middle-age in- gate the longitudinal relationship of these tau subtypes dividuals with normal cognition during a follow-up with concurrent other biological measures that deterior- period of 6 years [30], but also a lack of change in indi- ate during the AD process, such as synaptic markers in viduals with normal cognition, MCI and AD over a me- CSF or on PET, and cognitive data, to better understand dian follow-up of 2 years [31, 32]. This literature differences in clinical progression amongst tau subtypes. together with our observations suggests that increases over time in t-tau levels in CSF are slow, and follow-up Limitations times longer than 2–3 years might be necessary for par- A potential limitation of our study is that although we ticipants to change subgroups. used large clinical cohorts, the number of subjects in One of the challenges in biomarkers research is how some subgroups and subanalyses was small: this was es- to define the cut-point between normal and abnormal pecially the case in the highest tau subgroup, as well as levels. Pathology is the gold standard, but is also the end in tau PET analyses. The small size of the highest tau stage of the disease and difficult to obtain for large sam- subgroup means that there is more uncertainty in the as- ple sizes. The cut-point for Luminex p-tau in ADNI was sociation of this subgroup with clinical characteristics. originally based on pathology [21], and we observed the Therefore, the results regarding the highest tau sub- same cut-off for the lowest p-tau subgroup (23 pg/ml). group and the tau PET analyses should be interpreted However, for t-tau, we observed a lower cut-off that was with caution, and if possible repeated in future studies in still related to increased risk for disease progression. A even larger cohorts. Furthermore, we used Gaussian recent study defined cut-points for t- and p-tau mea- mixtures as a data-driven approach to study potential sured with Elecsys (t-tau 300 pg/ml and p-tau 27 pg/ml) subgroups in tau levels as a first step, it is possible that in ADNI based on their association with clinical progres- more complex models may improve the fit of tau levels sion in MCI patients [23]. We expand upon previous distributions, which should be addressed in future stud- studies [6, 11, 33, 34] by identifying additional cut- ies. Also, we determined cut-points here as the intersec- points that may have practical use for more specific tions of the probability distributions of the normal prognoses to individual patients or in trial design: we mixtures, which may not be ideal in all settings. For ex- identified lower cut-points than defined in the literature ample, in studies where minimizing misclassification (resp. 193 and 22 pg/ml for t- and p-tau, respectively) costs is desired, e.g. in clinical trial design, it may be use- that were already associated with increased risk for clin- ful to choose cut-points so that misclassification is mini- ical progression, and also showed for the higher cut- mized of individuals with high tau as falling in the points, that the corresponding subgroups were associ- lowest tau group, to ensure that as many individuals ated with gradually increasing hazard ratios and steeper with potentially fast progression are included in the trial. decline on the MMSE. Future studies could test the efficacy of the data-driven The notion that higher tau subgroups also included cut-points in those settings. Strengths of the study are non-demented individuals, and that higher tau levels that we used a large cohort, and we validated the mix- were associated with faster cognitive decline, regardless ture modelling results in another independent cohort of disease stage, suggests that tau subgroups may reflect with two different analysis platforms for CSF tau, and differences in underlying biological processes, rather both cohorts had detailed information of the Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 24 of 25 characteristics of the study populations, including cogni- related to PET imaging. PS and WF led the design and set-up of the ADC, from which data were used in this article. LE, EWo, EWi, WF, BT, CT and PV tive measures, follow-up data on clinical progression and made suggestions on improvement of the manuscript. All authors read and information on APOE genotype available. approved the manuscript for submission. Funding Conclusions This work has been supported by ZonMW Memorabel grant programme In conclusion, our studies suggest that abnormal levels of #733050824 (KW, BMT and PJV), Alzheimer Nederland grant #NL18003P (FD) and the Sigrid Juselius Foundation (LE). Funding was used in the analysis, CSF t-tau and p-tau may convey different biological aspects interpretation and writing of the manuscript. Statistical analyses were in AD, which might be in part driven by genetic factors performed at the VUmc Alzheimer Center that is part of the such as different APOE genotypes. The data-driven cut- neurodegeneration research programme of the Neuroscience Campus Amsterdam. The VUmc Alzheimer Center is supported by Stichting points we found may aid daily practice in prognosis of pa- Alzheimer Nederland and Stichting VUmc fonds. tients and may aid trial design by allowing stratification of individuals according to their risk of clinical progression. Availability of data and materials The ADNI dataset analysed during the current study is available in the ADNI repository, www.adni-info.org. For ADC, the data that support the findings of Supplementary information this study are available from the corresponding author upon reasonable Supplementary information accompanies this paper at https://doi.org/10. request. 1186/s13195-020-00713-3. Ethics approval and consent to participate Additional file 1. The institutional review boards of all institutions participating in ADNI approved the procedures that were part of the study. For ADC, all procedures were Additional file 2. approved by the local medical ethics committee. In both ADNI and ADC, written informed consent was obtained from all participants or surrogates. Abbreviations Aβ42: Amyloid-β 1-42; AD: Alzheimer’s disease; ADC: Amsterdam Dementia Consent for publication Cohort; ADNI: Alzheimer’s disease Neuroimaging Initiative; Not applicable. APOE: Apolipoprotein E; CI: Confidence interval; CSF: Cerebrospinal fluid; HR: Hazard ratio; NC: Normal cognition; MCI: Mild cognitive impairment; Competing interests MMSE: Mini-Mental State Examination; MRI: Magnetic resonance imaging; Prof. dr. Scheltens has acquired grants for the institution from GE Healthcare PET: Positron emission tomography; p-tau-181: Tau phosphorylated at and Piramal and received consultancy/speaker fees paid to the institution threonine 181; t-tau: Total tau from Novartis, Probiodrug, Biogen, Roche, and EIP Pharma, LLC in the past 2 years. Research programmes of Prof. dr. Wiesje van der Flier received funding Acknowledgements by ZonMW, NWO, EU-JPND, Alzheimer Nederland, CardioVascular Onderzoek Data was used for this project of which collection and sharing was funded Nederland, Health~Holland, Topsector Life Sciences & Health, stichting Dior- by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes aphte, Gieskes-Strijbis fonds, stichting Equilibrio, Biogen MA Inc., Life-MI, of Health Grant U01 AG024904) and DOD ADNI (Department of Defense AVID, Combinostics. WF holds the Pasman chair. WF has performed contract award number W81XWH-12-2-0012). ADNI is funded by the National Institute research for Biogen MA Inc. All funding is paid to her institution. Prof. dr. on Aging, the National Institute of Biomedical Imaging and Bioengineering, Teunissen received grants from the European Commission, the Dutch Re- and through generous contributions from the following: AbbVie, Alzheimer’s search Council (ZonMW), Association of Frontotemporal Dementia/Alzhei- Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioCli- mer’s Drug Discovery Foundation, The Weston Brain Institute, Alzheimer nica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Netherlands. Prof. Dr. Teunissen has functioned in advisory boards of Roche, Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. received non-financial support in the form of research consumables from Hoffmann-La Roche Ltd. and its affiliated company Genentech, Inc.; Fujirebio; ADxNeurosciences and Euroimmun, performed contract research or received GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & De- grants from Probiodrug, Biogen, Esai, Toyama, Janssen Prevention Center, velopment, LLC.; Johnson & Johnson Pharmaceutical Research & Develop- Boehringer, AxonNeurosciences, EIP farma, PeopleBio, Roche. The other au- ment LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, thors reported no conflicts of interest. LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Author details Company; and Transition Therapeutics. The Canadian Institutes of Health Re- Department of Neurology, Alzheimer Center Amsterdam, Amsterdam search is providing funds to support ADNI clinical sites in Canada. Private Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, sector contributions are facilitated by the Foundation for the National Insti- the Netherlands. Turku PET Centre, University of Turku and Turku University tutes of Health (www.fnih.org). The grantee organization is the Northern Cali- Hospital, Turku, Finland. Department of Radiology & Nuclear Medicine, fornia Institute for Research and Education, and the study is coordinated by Amsterdam Neuroscience, Amsterdam UMC, Amsterdam Neuroscience, the Alzheimer’s Therapeutic Research Institute at the University of Southern Amsterdam, Netherlands. Department of Clinical Chemistry, Neurochemistry California. ADNI data are disseminated by the Laboratory for Neuro Imaging Laboratory, Amsterdam UMC, Amsterdam Neuroscience, Amsterdam, at the University of Southern California. Netherlands. Department of Epidemiology and Biostatistics, Amsterdam Part of the data used in preparation of this article were obtained from the UMC, Amsterdam, The Netherlands. Alzheimer Center Limburg, Department Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc. of Psychiatry & Neuropsychology, School of Mental Health and Neuroscience, edu). As such, the investigators within the ADNI contributed to the design Maastricht University, Maastricht, The Netherlands. Division of and implementation of ADNI and/or provided data but did not participate in Neurogeriatrics, Department of Neurobiology, Care Sciences and Society, analysis or writing of this report. A complete listing of ADNI investigators can Karolinska Institutet, Stockholm, Sweden. be found at: http://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ ADNI_Acknowledgement_List.pdf Received: 16 June 2020 Accepted: 22 October 2020 Authors’ contributions FD wrote the first drafts of the manuscript; KW and BT contributed further References revisions. FD and BT performed the statistical analyses. LE contributed to the 1. Olsson B, Lautner R, Andreasson U, Öhrfelt A, Portelius E, Bjerke M, et al. CSF writing of the discussion. EWo contributed to methods and discussion and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic Duits et al. Alzheimer's Research & Therapy (2021) 13:2 Page 25 of 25 review and meta-analysis. Lancet Neurol. 2016;15:673–84 https://doi.org/10. linked immunosorbent assay and multiplex platforms in a longitudinal 1016/S1474-4422(16)00070-3. Alzheimer’s disease study. Alzheimers Dement. 2013;9:276–83 https://doi. 2. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH, et org/10.1016/j.jalz.2012.01.004. al. The diagnosis of dementia due to Alzheimer’s disease: recommendations 21. Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen from the National Institute on Aging-Alzheimer’s Association workgroups RC, et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011; neuroimaging initiative subjects. Ann Neurol. 2009;65:403–13 https://doi. 7:263–9 https://doi.org/10.1016/J.JALZ.2011.03.005. org/10.1002/ana.21610. 22. Verhage F. Intelligentie en Leeftijd: Onderzoek bij Nederlanders van Twaalf 3. Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo JL, Blennow K, et al. tot Zevenenzeventig Jaar [Intelligence and Age: Study with Dutch People Advancing research diagnostic criteria for Alzheimer’sdisease: the IWG-2 from Age 12 to 77]. Assen: Van Gorcum; 1964. criteria. Lancet Neurol. 2014;13:614–29 https://doi.org/10.1016/S1474- 23. Blennow K, Shaw LM, Stomrud E, Mattsson N, Toledo JB, Buck K, et al. 4422(14)70090-0. Predicting clinical decline and conversion to Alzheimer’s disease or 4. Mattsson N, Zetterberg H, Hansson O, Andreasen N, Parnetti L, Jonsson M, dementia using novel Elecsys Aβ (1–42), pTau and tTau CSF immunoassays. et al. CSF biomarkers and incipient Alzheimer disease in patients with mild Sci Rep. 2019;9:19024 https://doi.org/10.1038/s41598-019-54204-z. cognitive impairment. JAMA. 2009;302:385 https://doi.org/10.1001/jama. 24. Iqbal K, Liu F, Gong C-X. Tau and neurodegenerative disease: the story so 2009.1064. far. Nat Rev Neurol. 2016;12:15–27 https://doi.org/10.1038/nrneurol.2015.225. 5. Mulder C, Verwey NA, van der Flier WM, Bouwman FH, Kok A, van Elk EJ, et 25. Chhatwal JP, Schultz AP, Marshall GA, Boot B, Gomez-Isla T, Dumurgier J, et al. Amyloid- (1-42), total tau, and phosphorylated tau as cerebrospinal fluid al. Temporal T807 binding correlates with CSF tau and phospho-tau in biomarkers for the diagnosis of Alzheimer disease. Clin Chem. 2010;56:248– normal elderly. Neurology. 2016;87:920–6. 53 https://doi.org/10.1373/clinchem.2009.130518. 26. Mattsson N, Schöll M, Strandberg O, Smith R, Palmqvist S, Insel PS, et al. 18 6. Duits FH, Teunissen CE, Bouwman FH, Visser P-J, Mattsson N, Zetterberg H, F-AV-1451 and CSF T-tau and P-tau as biomarkers in Alzheimer’s disease. et al. The cerebrospinal fluid “Alzheimer profile”: easily said, but what does it EMBO Mol Med. 2017;9:1212–23. mean? Alzheimer’s Dement. 2014;10:713–23.e2 https://doi.org/10.1016/J. 27. Gordon BA, Friedrichsen K, Brier M, Blazey T, Su Y, Christensen J, et al. The JALZ.2013.12.023. relationship between cerebrospinal fluid markers of Alzheimer pathology 7. Toledo JB, Zetterberg H, van Harten AC, Glodzik L, Martinez-Lage P, and positron emission tomography tau imaging. Brain. 2016;139:2249–60. Bocchio-Chiavetto L, et al. Alzheimer’s disease cerebrospinal fluid biomarker 28. Pooler AM, Phillips EC, Lau DHW, Noble W, Hanger DP. EMBO Reports. 2013; in cognitively normal subjects. Brain. 2015;138:2701–15 https://doi.org/10. 14(4):389–94. https://doi.org/10.1038/embor.2013.15. 1093/brain/awv199. 29. Sato C, Barthélemy NR, Mawuenyega KG, Patterson BW, Gordon BA, Jockel- 8. Mirra SS. The CERAD neuropathology protocol and consensus Balsarotti J, et al. Tau kinetics in neurons and the human central nervous recommendations for the postmortem diagnosis of Alzheimer’s disease: a system. Neuron. 2018;97:1284–98.e7 https://doi.org/10.1016/j.neuron.2018. commentary. Neurobiol Aging. 18 4 Suppl:S91–4. http://www.ncbi.nlm.nih. 02.015. gov/pubmed/9330994. Accessed 9 Sept 2019. 30. Sutphen CL, Jasielec MS, Shah AR, Macy EM, Xiong C, Vlassenko AG, et al. 9. Knopman DS, DeKosky ST, Cummings JL, Chui H, Corey-Bloom J, Relkin N, et al. Longitudinal cerebrospinal fluid biomarker changes in preclinical Alzheimer Practice parameter: diagnosis of dementia (an evidence-based review): report disease during middle age. JAMA Neurol. 2015;72:1029 https://doi.org/10. of the Quality Standards Subcommittee of the American Academy of 1001/jamaneurol.2015.1285. Neurology. Neurology. 2001;56:1143–53 https://doi.org/10.1212/WNL.56.9.1143. 31. Lleó A, Alcolea D, Martínez-Lage P, Scheltens P, Parnetti L, Poirier J, et al. 10. Beach TG, Monsell SE, Phillips LE, Kukull W. Accuracy of the clinical diagnosis Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s of Alzheimer disease at National Institute on Aging Alzheimer Disease disease continuum in the BIOMARKAPD study. Alzheimers Dement. 2019; Centers, 2005–2010. J Neuropathol Exp Neurol. 2012;71:266–73 https://doi. https://doi.org/10.1016/j.jalz.2019.01.015. org/10.1097/NEN.0b013e31824b211b. 32. Wildsmith KR, Schauer SP, Smith AM, Arnott D, Zhu Y, Haznedar J, et al. 11. Degerman Gunnarsson M, Ingelsson M, Blennow K, Basun H, Lannfelt L, Identification of longitudinally dynamic biomarkers in Alzheimer’s disease Kilander L. High tau levels in cerebrospinal fluid predict nursing home cerebrospinal fluid by targeted proteomics. Mol Neurodegener. 2014;9:22 placement and rapid progression in Alzheimer’s disease. Alzheimers Res https://doi.org/10.1186/1750-1326-9-22. Ther. 2016;8:22 https://doi.org/10.1186/s13195-016-0191-0. 33. van Rossum IA, Vos SJB, Burns L, Knol DL, Scheltens P, Soininen H, et al. 12. De Meyer G, Shapiro F, Vanderstichele H, Vanmechelen E, Engelborghs S, De Injury markers predict time to dementia in subjects with MCI and amyloid Deyn PP, et al. Diagnosis-independent Alzheimer disease biomarker pathology. Neurology. 2012;79:1809–16 https://doi.org/10.1212/WNL. signature in cognitively normal elderly people. Arch Neurol. 2010;67:949 0b013e3182704056. https://doi.org/10.1001/archneurol.2010.179. 34. Kester MI, van der Vlies AE, Blankenstein MA, Pijnenburg YAL, van Elk EJ, 13. Bertens D, Tijms BM, Scheltens P, Teunissen CE, Visser PJ. Unbiased Scheltens P, et al. CSF biomarkers predict rate of cognitive decline in estimates of cerebrospinal fluid β-amyloid 1-42 cutoffs in a large memory Alzheimer disease. Neurology. 2009;73:1353–8 https://doi.org/10.1212/WNL. clinic population. Alzheimers Res Ther. 2017;9:8 https://doi.org/10.1186/ 0B013E3181BD8271. s13195-016-0233-7. 35. Jansen WJ, Ossenkoppele R, Knol DL, Tijms BM, Scheltens P, Verhey FRJ, et 14. Zwan M, van Harten A, Ossenkoppele R, Bouwman F, Teunissen C, al. Prevalence of cerebral amyloid pathology in persons without dementia. Adriaanse S, et al. Concordance between cerebrospinal fluid biomarkers and JAMA. 2015;313:1924 https://doi.org/10.1001/jama.2015.4668. [11C] PIB PET in a memory clinic cohort. J Alzheimers Dis. 2014;41:801–7 36. Slot RER, Kester MI, Van Harten AC, Jongbloed W, Bouwman FH, Teunissen https://doi.org/10.3233/JAD-132561. CE, et al. ApoE and clusterin CSF levels influence associations between 15. Zwan MD, Rinne JO, Hasselbalch SG, Nordberg A, Lleó A, Herukka S-K, et al. APOE genotype and changes in CSF tau, but not CSF Aβ42, levels in non- Use of amyloid-PET to determine cutpoints for CSF markers. Neurology. demented elderly. Neurobiol Aging. 2019;79:101–9. 2016;86:50–8 https://doi.org/10.1212/WNL.0000000000002081. 37. Konijnenberg E, Tijms BM, Gobom J, Dobricic V, Bos I, Vos S, et al. APOE ϵ4 16. Palmqvist S, Zetterberg H, Blennow K, Vestberg S, Andreasson U, Brooks DJ, genotype-dependent cerebrospinal fluid proteomic signatures in et al. Accuracy of brain amyloid detection in clinical practice using Alzheimer’s disease. Alzheimers Res Ther. 2020;12:65 https://doi.org/10.1186/ cerebrospinal fluid β-amyloid 42. JAMA Neurol. 2014;71:1282 https://doi.org/ s13195-020-00628-z. 10.1001/jamaneurol.2014.1358. 38. Reus LM, Stringer S, Posthuma D, Teunissen CE, Scheltens P, Pijnenburg 17. Blennow K, Hampel H. CSF markers for incipient Alzheimer’s disease. Lancet YAL, et al. Degree of genetic liability for Alzheimer’s disease associated with Neurol. 2003;2:605–13. specific proteomic profiles in cerebrospinal fluid. Neurobiol Aging. 2020;93: 18. Blennow K, Hampel H, Weiner M, Zetterberg H. Cerebrospinal fluid and 144 e1–144.e15. plasma biomarkers in Alzheimer disease. Nat Rev Neurol. 2010;6:131–44. 19. van der Flier WM, Scheltens P. Amsterdam dementia cohort: performing research to optimize care. J Alzheimers Dis. 2018;62:1091–111 https://doi. Publisher’sNote org/10.3233/JAD-170850. Springer Nature remains neutral with regard to jurisdictional claims in 20. Jongbloed W, Kester MI, van der Flier WM, Veerhuis R, Scheltens P, published maps and institutional affiliations. Blankenstein MA, et al. Discriminatory and predictive capabilities of enzyme-

Journal

Alzheimer s Research & Therapy – Springer Journals

Published: Jan 4, 2021

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

Four subgroups based on tau levels in Alzheimer’s disease observed in two independent cohorts

References (44)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies