Lantero‑Rodriguez et al. Alzheimer’s Alzheimer’s Research & Therapy (2023) 15:48 https://doi.org/10.1186/s13195-023-01201-0 Research & Therapy RESEARCH Open Access Clinical performance and head‑to‑head comparison of CSF p‑tau235 with p‑tau181, p‑tau217 and p‑tau231 in two memory clinic cohorts Juan Lantero‑Rodriguez1*†, Agathe Vrillon2,3†, Aida Fernández‑Lebrero4,5,6†, Paula Ortiz‑Romero4,5, Anniina Snellman1,7, Laia Montoliu‑Gaya1, Wagner S. Brum1,8, Emmanuel Cognat2,3, Julien Dumurgier3, Albert Puig‑Pijoan5,6,9, Irene Navalpotro‑Gómez4,5,6, Greta García‑Escobar6, Thomas K. Karikari1,10, Eugeen Vanmechelen11, Nicholas J. Ashton1,12,13,14, Henrik Zetterberg1,15,16,17,18,19, Marc Suárez‑Calvet4,5,6,20, Claire Paquet2,3 and Kaj Blennow1,15 Abstract Background Cerebrospinal fluid (CSF) p-tau235 is a novel biomarker highly specific of Alzheimer’s disease (AD). How‑ ever, CSF p-tau235 has only been studied in well-characterized research cohorts, which do not fully reflect the patient landscape found in clinical settings. Therefore, in this multicentre study, we investigated the performance of CSF p-tau235 to detect symptomatic AD in clinical settings and compared it with CSF p-tau181, p-tau217 and p-tau231. Methods CSF p-tau235 was measured using an in-house single molecule array (Simoa) assay in two independent memory clinic cohorts: Paris cohort (Lariboisière Fernand-Widal University Hospital Paris, France; n=212) and BIODEG‑ MAR cohort (Hospital del Mar, Barcelona, Spain; n=175). Patients were classified by the syndromic diagnosis (cogni‑ tively unimpaired [CU], mild cognitive impairment [MCI] or dementia) and their biological diagnosis (amyloid-beta [Aβ]+ or Aβ −). Both cohorts included detailed cognitive assessments and CSF biomarker measurements (clinically validated core AD biomarkers [Lumipulse CSF Aβ1–42/40 ratio, p-tau181 and t-tau] and in-house developed Simoa CSF p-tau181, p-tau217 and p-tau231). Results High CSF p-tau235 levels were strongly associated with CSF amyloidosis regardless of the clinical diagno‑ sis, being significantly increased in MCI Aβ+ and dementia Aβ+ when compared with all other Aβ− groups (Paris cohort: P ˂0.0001 for all; BIODEGMAR cohort: P ˂0.05 for all). CSF p-tau235 was pronouncedly increased in the A+T+ profile group compared with A−T− and A+T− groups (P ˂0.0001 for all). Moreover, CSF p-tau235 demonstrated high diagnostic accuracies identifying CSF amyloidosis in symptomatic cases (AUCs=0.86 to 0.96) and discriminating AT groups (AUCs=0.79 to 0.98). Overall, CSF p-tau235 showed similar performances to CSF p-tau181 and CSF p-tau231 † Juan Lantero-Rodriguez, Agathe Vrillon and Aida Fernández-Lebrero contributed equally as first authors. *Correspondence: Juan Lantero‑Rodriguez juan.rodriguez.2@gu.se Full list of author information is available at the end of the article © The Author(s) 2023. 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://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 2 of 15 when discriminating CSF amyloidosis in various scenarios, but lower than CSF p-tau217. Finally, CSF p-tau235 associ‑ ated with global cognition and memory domain in both cohorts. Conclusions CSF p-tau235 was increased with the presence of CSF amyloidosis in two independent memory clinic cohorts. CSF p-tau235 accurately identified AD in both MCI and dementia patients. Overall, the diagnostic perfor‑ mance of CSF p-tau235 was comparable to that of other CSF p-tau measurements, indicating its suitability to support a biomarker-based AD diagnosis in clinical settings. Keywords Alzheimer’s disease, CSF, Biomarkers, p-tau235, p-tau181, p-tau217, p-tau231, Memory clinic Background with control cases [15]. These findings align with previ- Alzheimer’s disease (AD) is characterized by the accu- ous studies, indicating that phosphorylation at serine 235 mulation of aggregated amyloid-beta (Aβ) and hyper- hampers tubulin assembly, compromising normal micro- phosphorylated tau protein into extracellular Aβ plaques tubules dynamics [20, 21]. We developed a single mole- and intracellular neurofibrillary tangles (NFTs), respec- cule array (Simoa) assay capable of measuring p-tau235 tively [1, 2]. These two immunohistochemical findings, in CSF, demonstrating that this novel biomarker is highly together with gross atrophy and synaptic and neuronal specific for AD and that it increases during preclinical loss, represent the hallmarks of AD and post-mortem stages, when only subtle abnormalities in CSF Aβ1–42/40 examination demonstrating the presence of these pro- can be detected [15]. Interestingly, when examining post- teinopathies is required for definitive diagnosis [3]. Vali- mortem brain samples, phosphorylation at serine 235 dated imaging and fluid biomarkers increasingly support appears to occur only when preceded by nearby phos- the clinical assessment and diagnosis of AD during life. phorylation at threonine 231, in what appears to be a Imaging biomarkers include positron emission tomog- sequential phosphorylation event. Due to the translat- raphy (PET) using radio-ligands capable of specifically ability of this sequential phosphorylation event from the binding Aβ and tau aggregates, whereas widely validated brain into CSF, p-tau235 has been previously proposed as fluid biomarkers currently include Aβ1–42 (or Aβ1–42/40 a potentially useful staging biomarker of preclinical AD ratio), total tau (t-tau) and phosphorylated tau at threo- (CSF p-tau235 positivity being indicative of late asymp- nine 181 (p-tau181), all measured in cerebrospinal fluid tomatic AD stage) [15]. With clinical trials increasingly (CSF) [4–7] often by full-automated instruments. focusing their intervention at the earliest stages of AD, Recently, the field of fluid biomarkers in neurodegener- CSF p-tau235 may be useful to discriminate early from ation has undergone a remarkable expansion, particularly late preclinical AD cases and to evaluate if novel com- with regard to biomarkers measuring different variants pounds effectively prevent disease progression. of tau protein. This has resulted in a wide range of novel In order to further characterize this novel p-tau p-tau biomarkers, measured with different methods and biomarker and contextualize its performance with platforms (fundamentally immunoassays and mass spec- other CSF p-tau species, we previously compared CSF trometry) [8–11]. Different p-tau residues may change p-tau235 with CSF p-tau217 and p-tau231 in a well- in different stages of the Alzheimer’s continuum. For characterized research cohort including the AD con- example, p-tau231 is the first p-tau residue increasing tinuum and a preclinical AD cohort. We demonstrated in preclinical AD, confirming the earliest underlying AD that CSF p-tau235 displays a statistically equal per- processes [11, 12] whereas p-tau217, with its pronounced formance to that of CSF p-tau217 and p-tau231 when fold changes and strong association with AD pathologi- discriminating cognitively impaired CSF Aβ-positive cal hallmarks, appears to be the most suitable p-tau bio- cases from cognitively impaired CSF Aβ-negative cases. marker for AD diagnosis and patient monitoring [13, 14]. However, the cases enrolled in that study belonged to Moreover, different p-tau epitopes may provide different a well-characterized research cohort, which does not information about the disease and some of them be more fully reflect the patients from daily clinical practice suited as a state, stage, or prognostic biomarkers. The including atypical AD, other dementia and comor- most recent p-tau biomarker to demonstrate potential bidities. Therefore, evaluating the performance of CSF utility in relation to AD is p-tau235 [15], a phosphoryla- p-tau235 in the routine practice of a memory clinic tion site that has been found to be a prominent feature in is much needed, since in these settings patients pre- paired helical filaments (PHFs) [16–19]. Recently, using sent with a higher heterogeneity in demographics, a targeted mass spectrometry method, our group has comorbidities and disease presentations [22–25]. As a demonstrated that p-tau235 is highly increased in neu- secondary aim, we tested whether CSF p-tau235 was ropathologically confirmed AD brain when compared associated with cognitive performance. Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 3 of 15 To this end, we measured this novel p-tau biomarker blood sampling. Clinical evaluation was performed by a in two independent memory clinic cohorts, compared neurologist, including anamnesis, physical examination its performance with N-terminal directed Simoa immu- and clinical diagnosis. Neuropsychological evaluation noassays targeting CSF p-tau181, p-tau217 and p-tau231, was performed by a neuropsychologist and consisted of and assessed its relation with cognitive status. a series of standardized cognitive tests and functional scales. A comprehensive description of the BIODEG- Methods MAR cohort has been previously published, with fur- Study participants ther details on the inclusion and exclusion criteria and Paris cohort on core AD CSF biomarker procedures, including assay Paris cohort enrolled a total of 212 patients who had details and cutoff determination [31, 32]. In this study, undergone CSF analysis at the Centre of Cognitive Neu- 175 patients of the BIODEGMAR cohort were included rology at Lariboisière Fernand-Widal University Hospi- from 27 April 2017 to 24 July 2020. The clinical diag- tal between March 2014 and December 2019, including noses included SCD (n=18, similar diagnostic criteria participants with subjective cognitive decline (SCD, as for Paris cohort), MCI (n=74) and, for subjects with n=21), non-AD mild cognitive impairment (non-AD moderate to severe cognitive impairment, possible AD MCI, n=45), AD-MCI (n=40), AD dementia (n=75) and dementia (n=11), probable AD dementia (n=43), LBD non-AD dementia (n=31). Non-AD dementia patients (n=2), extrapyramidal syndrome (n=2), vascular cogni- encompassed patients with dementia with Lewy bodies tive impairment and dementia (VCID; n=4), progressive (DLB, n=12), frontotemporal dementia (FTD, n=15), supranuclear palsy (PSP; n=2), corticobasal syndrome vascular cognitive impairment and dementia (VCID, (n=4), behavioral variant of FTD (bvFTD; n=4), primary n=3) and Creutzfeldt Jakob disease (n=1). Patients progressive aphasia (PPA; n=8) and cerebral amyloid underwent a thorough clinical examination involving angiopathy (CAA; n=3). personal medical and family histories, treatment, neu- rological examination, extensive neuropsychological Cohort stratification assessment, APOE genotyping, brain magnetic resonance Patient data from the Paris cohort and the BIODEGMAR imaging (MRI) extensive neuropsychological evaluation, cohort was examined using two stratification criteria: (i) MRI, APOE genotyping, blood and CSF analysis and fluid according to clinical syndrome (cognitively unimpaired sampling for collection (blood and CSF). The diagnosis [CU], MCI, or dementia) and CSF amyloid status (Aβ−/ for each patient was made during multidisciplinary con- Aβ+, as defined by Lumipulse CSF Aβ1-42/40, Supplemen- sensus meetings (including neurologists, neuropsycholo- tary Table 3), resulting in six groups: CU Aβ−, CU Aβ+, gists, gerontologists, neuroradiologist and biochemists) MCI Aβ−, MCI Aβ+, dementia Aβ− and dementia Aβ+ considering results of validated CSF biomarkers and (clinical diagnosis included in each group are available according to clinical diagnostic criteria for AD demen- in Supplementary Tables 1 and 2); (ii) based on the Aβ p-tau181, respectively (Lumipulse®) into A−T−, A+T− tia [3], MCI due to AD (AD-MCI) [26], DLB [27] and (A) and tau (T) status defined using CSF Aβ1–42/40 and FTD [28]. AD patients displayed CSF biomarkers on the AD continuum [3]. MCI of other causes (non-AD MCI) and A+T+ (the A−T+ group considered suspected non- included patients with psychiatric disorder, sleep apnea, AD pathology [SNAP] was not included in the statistical or systemic disease. Non-AD MCI presented with nor- analysis, but is depicted in the boxplots). The clinical mal CSF biomarkers or suspected non-Alzheimer patho- diagnoses included in each group are available in Sup- physiology (normal Aβ1-42/40, high p-tau and/or high plementary Tables 5 and 6). Additionally, CSF p-tau235 t-tau). Included SCD participants were individuals with levels across clinical diagnostic groups for both cohorts several years of clinical follow-up for a clinical complaint, are available in Supplementary Figure 1. presenting with normal cognitive testing and no abnor- malities at imaging and CSF examinations [29, 30]. Neuropsychological assessment All included participants were evaluated using the Mini- BIODEGMAR cohort Mental State Examination (MMSE). Detailed cognitive The BIODEGMAR cohort is an observational longitu- assessment was available for a subsample of participants dinal study that enrolls patients with neurodegenerative in both the Paris cohort (total n=135: CU [n=14], MCI diseases visiting the Cognitive Decline and Movement patients [n=64], dementia patients [n=57]) and the Disorders Unit of Hospital del Mar (Barcelona, Spain). BIODEGMAR cohort (total n=139: CU [n=18], MCI The procedures of the BIODEGMAR study include [n=59], dementia [n=62]). In the Paris cohort, memory extensive neuropsychological evaluation, MRI, APOE domain scores were evaluated using total immediate and ε4 genotyping, lumbar puncture for CSF collection and delayed recall of Free and Cued Selective Reminding Test Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 4 of 15 (FCSRT), and the delayed matching-to-samples 48 test Further details on assay specifications and validation can (DMS 48) for visual memory testing. Executive func- be found elsewhere [15]. tion was assessed using forward and backward digit span, frontal assessment battery and letter and animal fluencies. The language domain was evaluated using the Statistical analysis Dénomination Orale 80 test, a naming test. In BIODEG- Statistics were performed using SPSS (v26, IBM, Armonk, MAR cohort, memory domain was evaluated using total NY, USA) and Scistat (Ostend, Belgium). Graphs were immediate and delayed recall of FCSRT and the Memory generated using GraphPad PRISM (v7.03, San Diego, CA, Impairment Screen; executive functions, with backward USA) and R version 4.1 (https://​www.r-​proje​ct.​org/). Box- digit span, TMT A and B; language domain, with the plots show the median and interquartile range (IQR), with Boston naming test. Z-scores were computed from the individual data points for all participants always shown. control group scores as a reference. Domain scores were Statistical analyses were performed using parametric or obtained by averaging z-scores of the individual tests non-parametric methods when appropriate, based on the available results within that domain and the global cogni- normality or not of the data. Group comparisons between tion score by averaging the 3 domains’ z-scores. two categories were performed using a t-test and Mann- Whitney U test. All comparisons of multiple groups were tested with a one-way ANOVA adjusted for age and sex Biomarker measurements followed by Bonferroni-corrected post-hoc pairwise com- In the Paris cohort, core AD CSF biomarkers (Aβ1–42/40 parisons. Adjusted P-values for covariates are indicated as cally validated Lumipulse® G1200 assay (Fujirebio) [33]. ratio, p-tau181 and t-tau) were measured using the clini- PADJ. For each of the CSF p-tau biomarkers, accuracy for binary outcomes was determined using receiver operating Biomarker measurements for CSF p-tau181, p-tau217 characteristics (ROC) analysis and area under the curve and p-tau231 were performed using in-house Simoa (AUC). AUC values are accompanied by the 95% confi- assays. CSF p-tau181 and p-tau217 measurements were dence interval, abbreviated as C ­ I95%. Performance com- previously reported [34] and were included here in order parison between different p-tau biomarkers was analysed to allow the comparison with novel CSF p-tau235. In the using DeLong test (MedCalc). Spearman’s rank correla- BIODEGMAR cohort, core AD CSF biomarkers (Aβ1– tion ­(rS) was used to determine associations between bio- 42/40 ratio, p-tau181 and t-tau) were measured at Labo- markers. Association of CSF p-tau biomarkers with global ratori de Referència de Catalunya (LRC) with Lumipulse and domain-specific cognitive z-scores was studied with G600II (Fujirebio). Cutoff values for biomarkers and their linear regression models adjusted for age, sex and level of ratios (Aβ1–42/40, p-tau181, t-tau) were previously defined education. in the CORCOBIA study [31]. Biomarker measurements for CSF p-tau181 and p-tau231 were performed using in- house Simoa assays and were previously reported [32]. Results Cohort characteristics CSF p‑tau235 measurements A total number of 387 participants were included in CSF p-tau235 measurements were performed at the this study, which belong to two independent clinical Clinical Neurochemistry Laboratory at the University cohorts (Paris cohort [n=212]; BIODEGMAR cohort of Gothenburg using a Simoa HD-X instrument (Quan- [n=175]). Demographics and stratification in syn- terix), and all samples were blinded. Prior to the analysis, dromic groups (CU Aβ−/Aβ+, MCI Aβ−/Aβ+ and samples were allowed to thaw at room temperature for dementia Aβ−/Aβ+) are presented in Table 1 for both 45 min. Thawed samples were subsequently vortexed for cohorts. In the Paris cohort, MCI Aβ+ and dementia 15 s, after which they were plated and diluted 1:2 using Aβ+ were older than CU Aβ− group (P<0.0001). There Tau2.0 sample assay diluent (Quanterix). Due to sample was a weak yet significant association with age in both volume availability, CSF samples from the Paris cohort cohorts (Paris cohort: rS=0.32, P <0.0001; BIODE- were run in singlicates, while BIODEGMAR cohort GMAR cohort: rS =0.17, P <0.05). There were no sex samples were run in duplicates. An eight-point calibra- differences between groups (P ≥0.05; χ2 test). In both tion curve was generated using recombinant full-length cohorts, CSF p-tau levels were significantly higher GSK-3β-phosphorylated tau-441 (SignalChem) and run in APOE ε4 carriers (overall P <0.0001). As expected, in duplicates. Two internal quality controls (iQC, low and there were significant differences in MMSE scores high) were run at the beginning and end of each plate, across groups in both cohorts (P <0.0001), which, as also in duplicates. Repeatability (%CVr) and intermedi- compared with CU, gradually decreased in MCI and ate precision (%CVRw) values for both cohorts were ˂15%. further in patients with dementia. Table 1 Demographics and biomarkers levels Paris cohort (n = 212) BIODEGMAR cohort (n =175 ) CU MCI Dementia MCI Dementia P-value CU MCI Dementia CU MCI Dementia P-value Aβ− Aβ− Aβ− Aβ+ Aβ+ Aβ− Aβ− Aβ− Aβ+ Aβ+ Aβ+ n = 21 n = 47 n = 27 n = 40 n = 77 n=11 n=28 n=19 n=7 n=46 n=64 Age, years 64.6 66.30 (9.58) 66.04 (7.73) 72.40 (7.96)a 71.90 (8.61)a < 0.0001 70.4 71.4 70.0 70.1 74.4 73.5 0.009 (9.42) (5.8) (6.1) (6.1) (4.5) (4.8) (4.5) Male (% 8 16 15 16 30 0.338 3 16 6 2 19 25 0.141 male) (38%) (34%) (56%) (40%) (39%) (27%) (57%) (32%) (29%) (41%) (39%) APOEε4 car‑ 7/21 5/47 7/26 22/40 51/76 < 0.001 1/10 4/24 3/18 3/7 24/38 (63%)a 39/62 (61%)a < 0.02 riers (33%) (11%)a (27%) (55%) (67%)a (10%) (17%) (17%) (43%) Level of 11.5 (2.4) 10.7 (2.2) 11.0 (2.1) 10.0 (2.5) 10.6 (2.4) 0.299 10.4 (4.9) 8.5 (4.6) 7.75 (4.7) 11.4 (4.0) 8.4 (4.6) 7.7 (3.6) 0.0267 Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy education, years MMSE score 27.14 (2.45) 24.00 (4.01)a 23.85 (4.87)a 23.82 (3.84)a 19.09 (5.74)a < 0.0001 28.4 25.6 22.2 28.7 23.3 20.4 < 0.0001 (2.3) (2.7) (5.8)a (1.5) (3.9)a (4.8)a Core biomarkers Lumipulse CSF 1060.95 1090.66 1027.78 553.00 497.88 < 0.0001 1186 1164 986 634 556 522 < 0.0001 (2023) 15:48 Aβ1–42 (245.23) (468.75) (354.02) (209.36)a (155.27)a (357) (549) (449) (206)a (178)a (153)a CSF 0.093 0.088 (0.012) 0.09 0.044 0.042 < 0.0001 0.091 (0.017) 0.0879 0.091 (0.017) 0.049 0.043 0.043 < 0.0001 Aβ1–42/40 (0.010) (0.01) (0.009)a (0.009)a (0.012) (0.008)a (0.009)a (0.008)a CSF pTau, 32.25 40.97 (20.81) 33.59 94.80 113.53 < 0.0001 56.8 50.2 41.0 59.1 112 121.4 (57.9)a < 0.0001 pg/mL (8.16) (10.50) (47.31)a (60.02)a (31.4) (30.9) (24.4) (14.3) (59.6)a CSF t-tau, 236.62 333,085 362.26 594.83 722.01 < 0.0001 307 357 311 407 669 759 < 0.0001 pg/mL (63.69) (172.630) (341.90) (285.45)a (393.87)a (141) (230) (166) (107) (323)a (342)a Biomarker profile Normal 21 (100%) 39 (83%) 25 (92%) 0 (0%) 0 (0%) 9 (82%%) 24 (86%) 17 (89%) 0 (0%) 0 (0%) 0 (0%) profile AD con‑ 0 (0%) 0 (0%) 0 (0%) 40 (100%) 77 (100%) 0 (0%) 0 (0%) 0 (0%) 7 (100%) 46 (100%) 64 (100%) tinuum A+T− / / / 7 (18%) 10 (13%) / / / 6 (86%) 11 3 (24%) (5%) A+T+ / / / 33 (82%) 67 (87%) / / / 1 35 61 (14%) (76%) (95%) SNAP 0 8 (17%) 2 (8%) 0 0 2 (18%) 4 (14%) 2 (11%) 0 (0%) 0 (0%) 0 (0%) (0%) (0%) (0%) CSF p-tau biomarkers CSF 224.72 300.708 242.23 965.91 1223.94 < 0.0001 427 410 426 494 771 871 < 0.0001 N-term (66.33) (245.52) (110.58) (628.08) a (802.66)a (110) (164) (168) (129) (257)a (343)a p-tau181, pg/mL Page 5 of 15 Table 1 (continued) Paris cohort (n = 212) BIODEGMAR cohort (n =175 ) CU MCI Dementia MCI Dementia P-value CU MCI Dementia CU MCI Dementia P-value Aβ− Aβ− Aβ− Aβ+ Aβ+ Aβ− Aβ− Aβ− Aβ+ Aβ+ Aβ+ n = 21 n = 47 n = 27 n = 40 n = 77 n=11 n=28 n=19 n=7 n=46 n=64 Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy CSF 1.84 2.90 2.14 13.88 (9.48)a 16.62 (9.74)a < 0.0001 N/A N/A N/A N/A N/A N/A N/A N-term (0.94) (2.50) (1.67) p-tau217, pg/mL CSF 271.93 325.05 286.62 657.64 800.58 < 0.0001 313 348 331 541 788 893 < 0.0001 (2023) 15:48 N-term (64.74) (148.44) (114.72) (205.53)a (640.95)a (211) (201) (236) (144) (399)a (469)a p-tau231, pg/mL CSF 11.19 12.74 (5.67) 11.56 (4.30) 24.75 (8.11)a 28.03 < 0.0001 17.6 18.5 17.4 23.1 34.8 41.6 < 0.0001 N-term (2.94) (11.52)a (8.10) (11.3) (7.23) (3.65) (12.9)a (20.0)a p-tau235, pg/mL Data is shown as mean (SD) or n (%), as appropriate. We used Kruskal-Wallis test to compare age between groups and Pearson’s chi-square to compare sex and APOE ε4 frequencies between groups. Fluid biomarkers levels and MMSE were compared with a one-way ANOVA adjusted by age and sex. aP < 0.0001 versus controls. aP < 0.05 versus CU Aβ− Aβ1-42 β-amyloid 42, Aβ1-40 β-amyloid 40, CSF Cerebrospinal fluid, CU Cognitively unimpaired, MCI Mild cognitive impairment, MMSE Mini-Mental State Examination, N-term N-terminal, p-tau181 Tau phosphorylated at threonine 181, p-tau217 Tau phosphorylated at threonine 217, p-tau231 Tau phosphorylated at threonine 231, p-tau235 Tau phosphorylated at threonine 181, t-tau Total tau Page 6 of 15 Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 7 of 15 CSF p‑tau235 across syndromic and CSF Aβ‑positive In the Paris cohort, CSF p-tau235 displayed high accura- and CSF Aβ‑negative groups cies when discriminating dementia Aβ− from demen- All comparisons of multiple groups were adjusted for tia Aβ+ (AUC​Dementia Aβ− vs Aβ+=0.96, ­CI95%=0.90–0.99) age and sex. In the Paris cohort, CSF p-tau235 was sig- and MCI Aβ+ from MCI Aβ− (AUC​MCI Aβ− vs Aβ+=0.90, nificantly higher in MCI Aβ+ and dementia Aβ+ when ­CI95%=0.82–0.96) (Fig. 2a). In these two scenarios, CSF compared with CU Aβ−, MCI Aβ− and dementia p-tau235 performance showed no statistical differences Aβ− groups (all PADJ <0.0001) (Fig. 1a). Similarly, in the when compared with CSF p-tau181 (AUC​DementiaAβ− vs BIODEGMAR cohort, CSF p-tau235 was significantly Aβ+=0.98, ­CI95%=0.93–1.00; AUC​ MCI Aβ− vs Aβ+=0.94; increased in MCI Aβ+ and dementia Aβ+ when com- ­CI95%=0.86–0.98) and CSF p-tau231 (AUC​DementiaAβ− vs pared with CU Aβ−, MCI Aβ− and dementia Aβ− (all Aβ+=0.97, ­CI95%=0.92–0.99, AUC​ MCI Aβ− vs Aβ+=0.92, PADJ <0.001) (Fig. 1b). CSF p-tau235 was also significantly ­CI95%=0.85–0.97). In contrast, lower accuracies were higher in dementia Aβ+ participants when compared observed when compared with CSF p-tau217 (AUC​ with CU Aβ+ (despite this group being only comprised DementiaAβ− vs Aβ+=0.99, ­CI95%=0.94–1.00; AUC​ MCI Aβ− vs by seven subjects, PADJ =0.038), but no statistical differ- Aβ+=0.95, ­CI95%=0.88–0.99; both P˂0.05) (Fig. 2a). In the ences were found between MCI Aβ+ and CU Aβ+. No BIODEGMAR cohort, CSF p-tau235 showed high per- significant differences were found between CU Aβ− and formance discriminating dementia Aβ+ from demen- CU Aβ+, likely due to small sample sizes (eleven and tia Aβ− (AUC​Dementia Aβ− vs Aβ+=0.93, ­CI95%=0.86–0.98) seven cases, respectively), although higher CSF p-tau235 and MCI Aβ+ from MCI Aβ− (AUC​MCI Aβ− vs Aβ+=0.86, levels in CU Aβ+ were noticeable (Fig. 1b). ­CI95%=0.76–0.93) (Fig. 2b). When compared with other p-tau species, CSF p-tau235 performance discriminat- CSF p‑tau235 discriminatory accuracy identifying CSF ing dementia Aβ+ from dementia Aβ− was equal to CSF amyloidosis in syndromic groups p-tau181 (AUC​DementiaAβ− vs Aβ+=0.92, ­CI95%=0.84–0.97) We then investigated the discriminatory accuracies of and CSF p-tau231 (AUC​ DementiaAβ− vs Aβ+=0.88, CSF p-tau235 to detect CSF amyloidosis between each ­CI95%=0.79–0.94). On the other hand, the accuracy of of the syndromic groups and compared its performance CSF p-tau235 distinguishing MCI Aβ+ from MCI Aβ− with other novel N-terminal directed CSF p-tau biomark- approached significance compared with CSF p-tau181 ers, specifically p-tau181, p-tau217 and p-tau231 (Fig. 2). (AUC​MCI Aβ− vs Aβ+=0.91, ­CI95%=0.82–0.96; P=0.0513), Fig. 1 CSF levels of p-tau235 across syndromic groups stratified by CSF amyloidosis. a In the Paris cohort, CSF p-tau235 was significantly increased in both dementia Aβ+ and MCI Aβ+ when compared with CU Aβ−, MCI Aβ− and dementia Aβ−. b In the BIODEGMAR cohort, CSF p-tau235 was significantly higher in both dementia Aβ+ and MCI Aβ+ when compared with CU Aβ−, MCI Aβ− and dementia Aβ−. Additionally, CSF p-tau235 was significantly increased in dementia Aβ+ when compared with CU Aβ+. Data information: Boxplots show the median, IQR, and all the participants colour-coded based on the presence (red) or absence (blue) of CSF amyloidosis measured with Lumipulse CSF Aβ Aβ1–42/40. P-values were determined using one-way ANOVA adjusted by age and sex, followed by Bonferroni-corrected post hoc comparison (*P <0.05, **P <0.01, ***P <0.001, ****P <0.0001) Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 8 of 15 Fig. 2 Diagnostic performance of CSF p-tau235 when discriminating syndromic groups stratified by CSF amyloidosis. Both in a the Paris cohort and b the BIODEGMAR cohort CSF p-tau235 displayed high accuracies differentiating dementia Aβ+ from dementia Aβ− and MCI Aβ+ from MCI Aβ−, similar to CSF p-tau181 and p-tau231, and lower than CSF p-tau217. Data information: forest plots showing AUC values from ROC analysis (­ CI95% available in Supplementary Table 4). Comparisons of AUC values between CSF p-tau235 and other available CSF p-tau biomarkers were determined using DeLong test (significance is indicated in bold) while matched CSF p-tau231 (AUC​MCI Aβ− vs Aβ+=0.88, higher in A+T+ when compared with A−T− (PADJ ­CI95%=0.78–0.94) (Fig. 2b). <0.0001) (Fig. 3a). In the BIODEGMAR cohort, CSF p-tau235 also followed an increasing trajectory across the AD continuum, although no significant differences CSF p‑tau235 across AT groups were found between A−T− and A+T− subjects when All comparisons of multiple groups were adjusted for performing multiple comparisons (Fig. 3b). On the age and sex. In the Paris cohort, CSF p-tau235 lev- other hand, the levels of CSF p-tau235 were signifi- els progressively increased across the AD continuum, cantly higher in A+T+ when compared with A−T− that is, from A−T− to A+T− and from A+T− to and A+T− (PADJ <0.0001 for both) (Fig. 3b). Finally, A+T+ (Fig. 3a). A borderline significant increase in we stratified patients based on exclusively CSF Aβ, CSF p-tau235 levels was observed between A−T− and which resulted in CSF p-tau235 being highly increased A+T− (PADJ =0.054). This was followed by a prominent in A+ cases when compared with A− in both cohorts increase between A+T− and A+T+ (PADJ <0.0001). (P<0.0001 for all) (Supplementary Figure 2). CSF p-tau235 concentration was also significantly Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 9 of 15 Fig. 3 CSF levels of p-tau235 across AT groups. In both a the Paris cohort and b the BIODEGMAR cohort, CSF p-tau235 followed an increasing trajectory across the AT continuum. CSF p-tau235 was significantly increased in A+T+ when compared with A−T− and A+T−, and approached significance between A−T− and A+T− in the Paris cohort. Data information: Boxplots show the median, IQR, and all the participants colour-coded based on the presence (red) or absence (blue) of CSF amyloidosis measured with Lumipulse CSF Aβ1–42/40. P-values were determined using one-way ANOVA adjusted by age and sex, followed by Bonferroni-corrected post hoc comparison (*P <0.05, **P <0.01, ***P <0.001, ****P <0.0001) CSF p‑tau235 discriminatory accuracy identifying CSF of CSF p-tau235 when discriminating A+T− and amyloidosis in AT groups A+T+ (AUC​A+T− vs A+T+=0.95, ­C I95%=0.88–0.98) sta- We then examined the performance of CSF p-tau235 for tistically matched that of CSF p-tau181 (AUC​A+T- vs discriminating the AT groups which comprise the AD A+T+=0.94, ­C I95%=0.88–0.98) and slightly outper- continuum, that is A−T−, A+T− and A+T+ (Fig. 4). formed that of CSF p-tau231 (AUC​A+T− vs A+T+=0.86, In the Paris cohort, CSF p-tau235 displayed high accu- ­C I95%=0.79–0.92, P=0.0448). Finally, CSF p-tau235 racy discriminating A−T− from A+T− (AUC​ A−T− vs showed a nearly perfect accuracy discriminating A−T− A+T−=0.88, ­CI95%=0.80−0.93), similar to CSF p-tau231 and A+T+ (AUC​A−T− vs A+T+=0.97, ­C I95%=0.93–0.99), (AUC​A−T− vs A+T−=0.93, ­CI95%=0.87–0.97), but lower same as that observed for CSF p-tau181 (AUC​A−T− vs than both CSF p-tau181 (AUC​ A−T− vs A+T−=0.95, A+T+=0.98, ­C I95%=0.94–1.00) and CSF p-tau231 (AUC​ ­CI95%=0.89–0.98, P˂0.05) and CSF p-tau217 (AUC​A−T− A−T− vs A+T+=0.95, ­C I95%=0.90–0.98) (Fig. 4b). vs A+T−=0.95, ­CI95%=0.89–0.98, P˂0.05) (Fig. 4a). CSF Finally, we evaluated the performance of CSF p-tau235 showed high accuracy discriminating A+T− p-tau235 when discriminating Aβ+ from Aβ− indi- from A+T+ participants (AUC​ A+T− vs A+T+=0.89, viduals (Supplementary Figure 3, Supplementary ­CI95%=0.82–0.94) and A−T− from A+T+ (AUC​A−T− vs Table 8). In the Paris cohort, CSF p-tau235 showed A+T+=0.98, ­CI95%=0.95–1.00). In both scenarios, CSF high accuracy distinguishing these two groups (AUC​ p-tau235 performance was statistically equal to that of Aβ− vs Aβ+= 0.94, C­ I95%=0.90–0.97). When compared CSF p-tau181 (AUC​A+T− vs A+T+=0.93, ­CI95%=0.87–0.97; with other p-tau species (Supplementary Table 8), CSF AUC​A−T− vs A+T+=0.99, ­CI95%=0.97–1.00), CSF p-tau217 p-tau235 performance was significantly lower than (AUC​A+T− vs A+T+=0.91, ­CI95%=0.84–0.96; AUC​ A− CSF p-tau181 (AUC​Aβ− vs Aβ +=0.97, ­C I95%=0.93–0.99, T− vs A+T+=0.99, ­CI95%=0.97–1.00) and CSF p-tau231 P˂0.05) and CSF p-tau 217 (AUC​ Aβ− vs Aβ +=0.98, (AUC​A+T− vs A+T+=0.88, ­CI95%=0.81–0.93; AUC​A−T− vs ­C I95%=0.95–0.99, P˂0.01), but equal to CSF p-tau231 A+T+=0.98, ­CI95%=0.95–1.00) (Fig. 4a). (AUC​Aβ− vs Aβ+=0.96, ­C I95%=0.92–0.98) (Supplemen- In the BIODEGMAR cohort (Fig. 4b), CSF p-tau235 tary Figure 3a). In the BIODEGMAR cohort, CSF displayed equal accuracies when discriminat- p-tau235 (AUC​ Aβ− vs Aβ+ =0.89, ­C I95%=0.83–0.93) ing A−T− from A+T− (AUC​ A−T− vs A+T−=0.79, matched the performance of both CSF p-tau181 (AUC​ ­C I95%=0.68–0.87) as CSF p-tau181 (AUC​ A−T− vs Aβ− vs Aβ+=0.91, ­C I95%=0.85–0.94) and CSF p-tau231 A+T−=0.80, ­C I95%=0.69–0.88) and CSF p-tau231 (AUC​ (AUC​Aβ− vs Aβ+=0.88, ­C I95%=0.82–0.92) (Supplemen- A−T− vs A+T−=0.80, ­C I95%=0.70–0.88). The accuracy tary Figure 3b). Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 10 of 15 Fig. 4 Diagnostic performance of CSF p-tau235 when discriminating AT groups. a In the Paris cohort, CSF p-tau235 showed high accuracies in all three scenarios. CSF p-tau235 performance discriminating A−T− from A+T− was lower than CSF p-tau181 and p-tau217, but matched that of CSF p-tau231. When differentiating A+T− from A+T+ and A−T− from A+T+, CSF p-tau235 matched the performance of CSF p-tau181, p-tau217 and p-tau231. b In the BIODEGMAR cohort, CSF p-tau235 displayed high accuracies in all three scenarios, equal to those of CSF p-tau181 and p-tau231, slightly outperforming the latter when discriminating A+T− from A+T+. Data information: Forest plots showing AUC values from ROC analysis ­(CI95% available in Supplementary Table 7). Comparisons of AUC values between CSF p-tau235 and other available CSF p-tau biomarkers were determined using DeLong test (significance is indicated in bold) CSF p‑tau235 association with cognition with global and domain-specific cognition are shown in In both cohorts, MMSE score was associated with CSF Fig. 5, after adjustment on age, sex and level of education. p-tau235 after adjustment on age, sex and level of edu- CSF p-tau 235 associated with global cognition in both cation (Paris: β=−0.304, PADJ <0.0001; BIODEGMAR: cohorts (Paris β=−0.202, PADJ =0.046; BIODEGMAR β=−0.247, PADJ =0.0009). The association remained sig- β=−0.363, PADJ <0.0001), similar to all CSF p-tau meas- nificant after further adjustment on Aβ1–42/40 ratio (Paris urements but CSF p-tau217. Regarding specific cognitive Cohort: β=−0.243, PADJ =0.0012; BIODEGMAR cohort: domains, the strongest association was found with mem- β=−0.225, PADJ =0.004). ory impairment, both for CSF p-tau235 and other p-tau Neuropsychological assessment was available for a species (CSF p-235: Paris β= −0.255, PADJ =0.0031; BIO- subset of patients in the Paris (n=136) and BIODEG- DEGMAR: β= −0.454, PADJ <0.0001). A weak association MAR cohort (n=139). The cross-sectional associations of was also found with executive domain function but only CSF p-tau235 levels and other CSF p-tau measurements in BIODEGMAR (β= −0.229, PADJ =0.0051). Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 11 of 15 Fig. 5 CSF p-tau 235 association with global cognition and specific cognitive domains. Comparison of CSF p-tau235 and other p-tau markers (p-tau181, p-tau217 and p-tau231) for global and domain-specific cognition in a the Paris and b the BIODEGMAR cohort. Data information: Linear regressions were adjusted for age, sex and level of education. Estimates are presented standardized to allow for comparison in between the different p-tau biomarkers. (*P <0.05, **P <0.01, ***P <0.001, ****P <0.0001) Discussion performance to other CSF p-tau species [15]. However, In this study, we investigated the novel CSF biomarker these findings cannot be directly translated into the p-tau235 in two independent memory clinic cohorts real-world of memory clinics, which is characterized and compared its diagnostic performance with other rel- by higher heterogeneity in population (both in social- evant p-tau species, specifically CSF p-tau181, p-tau217 economic and ethnic terms), clinical presentations and and p-tau231. We found that (i) CSF p-tau235 is strongly co-pathologies, vastly contrasting with the stringent associated with CSF amyloidosis, regardless of clinical inclusion/exclusion criteria and detailed pathophysi- presentation; (ii) high levels of CSF p-tau235 in cogni- ological characterizations in research cohorts. Thus, tively impaired patients are highly indicative of MCI or the main goal of the present study was to investigate the dementia of the AD type; (iii) the main increase in CSF translatability of previous CSF p-tau235 results into the p-tau235 levels occurs between A+T− and A+T+, irre- routine practice of memory clinic. Consistent with our spective of the presence of symptoms or not; (iv) CSF previous findings in selected research cohorts (simi- p-tau235 displayed overall the same diagnostic perfor- larly stratified using Aβ1–42/40) [15], CSF p-tau235 was mance as CSF p-tau181 and p-tau231, but a slightly lower elevated across CSF Aβ-positive groups in both clinical diagnostic performance compared with CSF p-tau217; cohorts. CSF p-tau235 was increased in MCI Aβ+ and and (v) CSF p-tau235 levels associate with global and dementia Aβ+ showing similar levels between the two specific-domain cognitive decline. groups. This aligns well with previous results in a pre- In a previous study, we described CSF p-tau235 across clinical AD cohort in which CSF p-tau235 measurements the AD continuum using two research cohorts, demon- seemingly plateau between MCI Aβ+ and AD Aβ+ [15]. strating its high specificity for AD and its comparable Moreover, in the BIODEGMAR cohort, CSF p-tau235 Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 12 of 15 was noticeably higher in CU Aβ+ when compared with diagnostic accuracy discriminating Aβ− and Aβ+, how- CU Aβ−, and despite lacking sufficient statistical power ever, slightly lower than CSF p-tau181 and clearly lower (eleven and seven cases, respectively), it approached sig- than CSF p-tau217. Despite this, CSF p-tau235 exhibited nificance. Hence, these results from clinical practice are the same performance as CSF p-tau231 in both the Paris concordant with our previous results in research cohorts, and BIODEGMAR cohorts. In our previous publication, where CSF p-tau235 was increased in Aβ+ subjects both CSF p-tau235 started to increase in A+T− cases; how- in preclinical and symptomatic AD [15]. Furthermore, ever, the magnitude or extent of this increase was smaller these results indicate that high levels of CSF p-tau235 than for other CSF p-tau species, whereas the increase are strongly associated with CSF amyloidosis and highly was more pronounced between A+T− and A+T+ [15]. indicative of MCI or dementia of an AD type. In terms of The same was observed here in this study, and the result- diagnostic accuracy identifying Aβ+ in MCI and demen- ing overlap between the levels of CSF p-tau235 in A−T− tia, CSF p-tau235 showed equal performance to that of and A+T−groups rendered the lower performance of CSF p-tau181 and p-tau231, but slightly lower than CSF this novel biomarker species identifying CSF amyloido- p-tau217. Because these differences were very marginal, sis. Thus, despite its high performance, other p-tau bio- with almost fully overlapping 95% confidence intervals, markers emerge earlier than CSF p-tau235 in response to and together with the lack of CSF p-tau217 measure- CSF amyloidosis and, therefore, provide a superior per- ments in the BIODEGMAR cohort to corroborate these formance discriminating A−T− from A+T− and Aβ− results, this should be cautiously interpreted. from Aβ+. CSF p-tau235 followed an increasing trajectory across Higher CSF p-tau235 levels were associated with AT groups, which was characterized by a minor incre- lower global cognition and memory function in both ment between A−T− and A+T−, followed by a promi- cohorts, which had not been previously investigated. CSF nent increase from A+T− to A+T+. These differences p-tau181 and CSF p-tau217 have been reported to associ- across AT groups are similar to those previously reported ate with global cognition and memory impairment cross- in preclinical AD [15], despite the fact that the cohorts sectional and longitudinally, with p-tau217 showing analysed here are memory clinic-based, and thus included stronger associations with cognitive decline [36]. In our proportionally more symptomatic cases. Therefore, study, CSF p-tau235 performed overall similarly to other this suggests that the increase in CSF p-tau235 is more p-tau species. We observed modest differences between prominent when abnormalities in CSF Aβ and p-tau are assays; however, we cannot determine if this reflects dif- established (i.e. A+T+ cases), regardless of these cases ferences in the analytical performances of the assays or being symptomatic, as shown here, or asymptomatic, differences in the biology underlying tau phosphorylation like previously reported [15]. CSF p-tau235 showed high association with cognition. performance discriminating all three AT groups and was, for the most part, comparable to other CSF p-tau Limitations/strengths biomarkers. When compared with CSF p-tau181, CSF The strengths of this study include the use of two inde- p-tau235 was outperformed when discriminating A-T- pendent memory clinic-based cohorts, allowing us to from A+T-, but only in the Paris cohort. This may sug- validate and confirm our findings in two different clini- gest that N-terminal p-tau181 species emerge earlier in cal settings. Secondly, both cohorts had clinically vali- response to Aβ pathology than N-terminal p-tau235 dated biomarker measurements available, which enabled species (which in turn start increasing earlier than mid- the detailed stratification of participants and the subse- region p-tau181) [15, 35]. CSF p-tau217 also outper- quent investigation of CSF p-tau235 in syndromic or AT formed CSF p-tau235 in the same scenario (A−T− vs groups. Moreover, the studied cohorts included measure- A+T−) in the Paris cohort, which aligns well with pre- ments of other novel p-tau species, specifically N-termi- vious results indicating that CSF p-tau217 emerges nal directed CSF p-tau181, p-tau217 and p-tau231 (all prior to CSF p-tau235 [15]. On the other hand, no dif- measured in the same analytical platform), making pos- ferences in performance could be found between CSF sible a head-to-head comparison between different p-tau p-tau235 and CSF p-tau231, except when discriminating residues. However, this study does not go without limita- A+T− and A+T+ in the BIODEGMAR cohort, where tions. CSF p-tau217 measurements were not available in CSF p-tau235 slightly outperformed CSF p-tau231. The the BIODEGMAR cohort, and therefore, the biomarker marginal difference observed is likely due to the fact that comparison with CSF p-tau235 could only be explored CSF p-tau231 emerges very early in preclinical stages, in the Paris cohort. Another limitation is that CU Aβ+ and tends to plateau later on the continuum AD, thus cases were limited to the BIODEGMAR cohort, and rendering a lower discrimination between A+T− and their sample size was rather small. Considering that CSF A+T+ groups [12]. Finally, CSF p-tau235 showed a high p-tau235 increases late during preclinical AD [15], it is Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 13 of 15 expected that in cohorts richer in CU cases (especially ROC Receiver operating characteristic curve SCD Subjective cognitive decline A+T−), other CSF p-tau biomarkers that abnormally rS Spearman’s rank correlation coefficient emerge earlier in the AD continuum (such as p-tau231 Simoa Single molecule array and p-tau217) [15] would provide a superior performance t-tau Total tau ULOQ Upper limit of quantification identifying CSF amyloidosis in asymptomatic individuals. VCID Vascular contributions to cognitive impairments and dementia This would subsequently affect the A−T− vs A+T- anal- ysis, where CSF p-tau235 discriminatory accuracy would Supplementary Information likely be lower than that of other CSF p-tau biomarkers. The online version contains supplementary material available at https://​doi.​ A limited number of patients in non-AD dementia group org/​10.​1186/​s13195-​023-​01201-0. also warranted further studies. Additional file 1: Supplementary Figure 1. CSF levels of p-tau235 across clinical diagnosis. Supplementary Figure 2. CSF levels of p-tau235 in Conclusions Aβ+ and Aβ- cases. Supplementary Figure 3. CSF p-tau235 diagnos‑ In conclusion, our memory-clinic-based study including tic performance discriminating Aβ+ from Aβ- cases. Supplementary two well-characterized cohorts brings further evidence Table 1. Clinical diagnosis included in each syndrome group in Paris cohort. Supplementary Table 2. Clinical diagnosis included in each that p-tau235 is a novel and specific CSF biomarker for syndrome group in BIODEGMAR cohort. Supplementary Table 3. CSF AD AD diagnosis, both at MCI and dementia stages. Com- biomarkers cut-offs for BIODEGMAR and Paris Cohort. Supplementary parison with other CSF p-tau species, including p-tau181, Table 4. Accuracies of CSF p-tau181, p-tau217, p-tau231 and p-tau235 when identifying CSF amyloidosis in dementia and MCI cases in Paris and p-tau217 and p-tau231, supports that CSF p-tau235 BIODEGMAR cohort. Supplementary Table 5. Clinical diagnosis included should be suitable for use in clinical settings. In addi- in each AT group in Paris cohort. Supplementary Table 6. Clinical diagno‑ tion, significant association with cognitive decline adds sis included in each AT group in BIODEGMAR cohort. Supplementary Table 7. Accuracies of CSF p-tau181, p-tau217, p-tau231 and p-tau235 an argument to its relevance as a clinical tool to iden- when discriminating AT groups in Paris and BIODEGMAR cohort. Sup‑ tify and monitor CSF amyloidosis along the whole AD plementary Table 8. Accuracies of CSF p-tau181, p-tau217, p-tau231 and continuum. Future studies will aim to elucidate whether p-tau235 when discriminating Aβ- from Aβ+ in Paris and BIODEGMAR cohort. CSF p-tau235 can predict cognitive decline in longitudi- nal samples and attempt to measure p-tau235 in blood Authors’ contributions samples. JLR, AV, AFL, HZ, MSC, CP, and KB created the concept and design. Data acqui‑ sition was performed by JLR, AFL, AV, EC and JD. JLR, AV and AFL performed data analysis. JLR, AV, AFL, POR, AES, LGM, WSB, EC, JD, APJ, ING, GGE, TK, EV, Abbreviations NJA, HZ, MSC, CP, KB contributed to sample selection and/or interpretation AD Alzheimer’s disease of data. JLR, AV and AFL drafted the manuscript and all authors revised. All AD-MCI Mild cognitive impairment due to AD authors read and approved the final manuscript. Aβ Amyloid beta ANOVA Analysis of variance Funding AUC​ Area under the curve Open access funding provided by University of Gothenburg. This study was Aβ1–40 Aβ peptide 1–40 funded by the Swedish Research Council (#2017-00915 and #2022-00732), Aβ1–42 Aβ peptide 1–42 the Alzheimer Drug Discovery Foundation (ADDF), USA (#RDAPB-201809- Aβ1–42/40 Ratio Aβ1–42/Aβ1–40 2016615), the Swedish Alzheimer Foundation (#AF-742881), Hjärnfonden, bvFTD Behavioural variant frontotemporal dementia Sweden (#FO2017-0243), the Swedish state under the agreement between CAA​ Cerebral amyloid angiopathy the Swedish government and the County Councils, the ALF-agreement CBS Corticobasal syndrome (#ALFGBG-715986), the European Union Joint Program for Neurodegenerative CJD Creutzfeldt-Jakob disease Disorders (JPND2019-466-236) and the National Institute of Health (NIH), USA, CSF Cerebrospinal fluid (grant #1R01AG068398-01). AV is supported by Fondation Ophtalmologique CU Cognitively unimpaired Adolphe de Rothschild and Fondation Vaincre Alzheimer. AES was supported CU Aβ− Cognitively unimpaired Aβ negative by the Paulo Foundation, the Orion Research Foundation sr, and currently CU Aβ− Cognitively unimpaired Aβ positive holds a postdoctoral fellowship from the Academy of Finland (#341059). TKK CI95% 95% confidence interval was funded by the Swedish Research Council (Vetenskapsrådet #2021-03244), CV Coefficient of variation the Alzheimer’s Association Research Fellowship (#AARF-21-850325), the DLB Dementia with Lewy bodies Aina (Ann) Wallströms and Mary-Ann Sjöbloms stiftelsen and the Emil och FTD Frontotemporal dementia Wera Cornells stiftelsen. HZ is a Wallenberg Scholar supported by grants from IP-MS Immunoprecipitation-mass spectrometry the Swedish Research Council (#2018-02532), the European Union’s Horizon MCI Mild cognitive impairment Europe research and innovation programme under grant agreement No MCI Aβ− Mild cognitive impairment Aβ negative 101053962, Swedish State Support for Clinical Research (#ALFGBG-71320), MCI Aβ+ Mild cognitive impairment Aβ positive the Alzheimer Drug Discovery Foundation (ADDF), USA (#201809-2016862), MMSE Mini-Mental State Examination the AD Strategic Fund and the Alzheimer’s Association (#ADSF-21-831376-C, MS Mass spectrometry #ADSF-21-831381-C and #ADSF-21-831377-C), the Bluefield Project, the NFT Neurofibrillary tangles Olav Thon Foundation, the Erling-Persson Family Foundation, Stiftelsen för PPA Primary progressive aphasia Gamla Tjänarinnor, Hjärnfonden, Sweden (#FO2022-0270), the European PSP Progressive supranuclear palsy Union’s Horizon 2020 research and innovation programme under the Marie p-tau181 Tau phosphorylated at threonine 181 Skłodowska-Curie grant agreement No 860197 (MIRIADE), the European Union p-tau217 Tau phosphorylated at threonine 217 Joint Programme – Neurodegenerative Disease Research (JPND2021-00694) p-tau231 Tau phosphorylated at threonine 231 and the UK Dementia Research Institute at UCL (UKDRI-1003). MSC receives p-tau235 Tau phosphorylated at serine 235 Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 14 of 15 funding from the European Research Council (ERC) under the European London, London, UK. 17 UK Dementia Research Institute, University College Union’s Horizon 2020 research and innovation programme (Grant agreement London, London, UK. 18 Hong Kong Center for Neurodegenerative Diseases, No. 948677), Project “PI19/00155”, funded by Instituto de Salud Carlos III (ISCIII) Hong Kong, China. 19 Wisconsin Alzheimer’s Disease Research Center, Univer‑ and co-funded by the European Union, and from a fellowship from “la Caixa” sity of Wisconsin School of Medicine and Public Health, University of Wiscon‑ Foundation (ID 100010434) and from the European Union’s Horizon 2020 sin-Madison, Madison, WI, USA. 20 Centro de Investigación Biomédica en Red research and innovation programme under the Marie Skłodowska-Curie grant de Fragilidad y Envejecimiento Saludable (CIBERFES), Madrid, Spain. agreement No 847648 (LCF/BQ/PR21/11840004). Received: 7 December 2022 Accepted: 28 February 2023 Availability of data and materials Bulk anonymized data can be shared by request from qualified investigators, providing data transfer is in agreement with EU legislation and decisions by the institutional review board of each participating center. References 1. Braak H, Braak E. Neuropathological stageing of Alzheimer-related Declarations changes. Acta Neuropathol. 1991;82(4):239–59. 2. Thal DR, et al. Phases of A beta-deposition in the human brain and its Ethics approval and consent to participate relevance for the development of AD. Neurology. 2002;58(12):1791–800. All patients or their legal relatives in case of severe dementia gave written 3. Jack CR Jr, et al. NIA-AA research framework: toward a biological defini‑ informed consent to their participation in this study. Collection and analysis tion of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535–62. of samples were approved by the ethic committee of Bichat University, Paris, 4. Blennow K, Vanmechelen E, Hampel H. CSF total tau, Abeta42 and France (CEERB GHU Nord n°10-037) for Paris cohort and by the local ethics phosphorylated tau protein as biomarkers for Alzheimer’s disease. Mol committee (CEIC PSMAR, project code 2018/7805I) for the BIODEGMAR Neurobiol. 2001;24(1–3):87–97. cohort. 5. Ashton NJ, et al. Update on biomarkers for amyloid pathology in Alzhei‑ mer’s disease. Biomark Med. 2018;12(7):799–812. Consent for publication 6. Molinuevo JL, et al. Current state of Alzheimer’s fluid biomarkers. Acta Not applicable Neuropathol. 2018;136(6):821–53. 7. Scholl M, et al. Biomarkers for tau pathology. Mol Cell Neurosci. Competing interests 2019;97:18–33. KB has served as a consultant, at advisory boards, or at data monitoring 8. Barthelemy NR, et al. Tau phosphorylation rates measured by mass committees for Abcam, Axon, Biogen, JOMDD/Shimadzu. Julius Clinical, Lilly, spectrometry differ in the intracellular brain vs. extracellular cerebrospinal MagQu, Novartis, Prothena, Roche Diagnostics, and Siemens Healthineers, and fluid compartments and are differentially affected by Alzheimer’s disease. is a co-founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which Front Aging Neurosci. 2019;11:121. is a part of the GU Ventures Incubator Program. HZ has served at scientific 9. Karikari TK, et al. Blood phosphorylated tau 181 as a biomarker for advisory boards and/or as a consultant for Abbvie, Acumen, Alector, Alzinova, Alzheimer’s disease: a diagnostic performance and prediction model‑ ALZPath, Annexon, Apellis, Artery Therapeutics, AZTherapies, CogRx, Denali, ling study using data from four prospective cohorts. Lancet Neurol. Eisai, Nervgen, Novo Nordisk, Passage Bio, Pinteon Therapeutics, Prothena, Red 2020;19(5):422–33. Abbey Labs, reMYND, Roche, Samumed, Siemens Healthineers, Triplet Thera‑ 10. Palmqvist S, et al. Discriminative accuracy of plasma phospho-tau217 peutics, and Wave, has given lectures in symposia sponsored by Cellectricon, for Alzheimer disease vs other neurodegenerative disorders. JAMA. Fujirebio, Alzecure, Biogen, and Roche, and is a co-founder of Brain Biomarker 2020;324(8):772–81. Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incuba‑ 11. Ashton NJ, et al. Plasma p-tau231: a new biomarker for incipient Alzhei‑ tor Program (outside submitted work). MSC has served as a consultant and at mer’s disease pathology. Acta Neuropathol. 2021;141(5):709–24. advisory boards for Roche Diagnostics International Ltd and has given lectures 12. Ashton NJ, et al. Cerebrospinal fluid p-tau231 as an early indica‑ in symposia sponsored by Roche Diagnostics, S.L.U and Roche Farma, S.A. tor of emerging pathology in Alzheimer’s disease. EBioMedicine. MSC was granted with a project funded by Roche Diagnostics International 2022;76:103836. Ltd; payments were made to the institution (BBRC). EVM is cofounder of ADx 13. Janelidze S, et al. Cerebrospinal fluid p-tau217 performs better NeuroSciences. than p-tau181 as a biomarker of Alzheimer’s disease. Nat Commun. 2020;11(1):1683. Author details 1 14. Ashton NJ, et al. Differential roles of Abeta42/40, p-tau231 and p-tau217 Department of Psychiatry & Neurochemistry, Institute of Neuroscience for Alzheimer’s trial selection and disease monitoring. Nat Med. and Physiology, the Sahlgrenska Academy at the University of Gothenburg, 2022;28(12):2555–62. Gothenburg, Sweden. 2 Institut national de la santé et de la recherche médi‑ 15. Lantero-Rodriguez J, et al. P-tau235: a novel biomarker for staging pre‑ cale U1144 Optimisation Thérapeutique en Neuropsychopharmacologie, clinical Alzheimer’s disease. EMBO Mol Med. 2021;13(12):e15098. Université de Paris, Paris, France. 3 Centre de Neurologie Cognitive, Groupe 16. Hasegawa M, et al. Protein sequence and mass spectrometric analyses of Hospitalo Universitaire Assistance Publique Hôpitaux de Paris Nord Hôpital tau in the Alzheimer’s disease brain. J Biol Chem. 1992;267(24):17047–54. Lariboisière Fernand-Widal, Paris, France. 4 Barcelonaβeta Brain Research Center 17. Hoffmann R, et al. Unique Alzheimer’s disease paired helical filament (BBRC), Pasqual Maragall Foundation, Barcelona, Spain. 5 IMIM (Hospital del Mar specific epitopes involve double phosphorylation at specific sites. Bio‑ Medical Research Institute), Barcelona, Spain. 6 Cognitive Decline and Move‑ chemistry. 1997;36(26):8114–24. ment Disorders Unit, Neurology Department, Hospital del Mar, Barcelona, 18. Singer D, et al. Neighbored phosphorylation sites as PHF-tau spe‑ Spain. 7 Turku PET Centre, University of Turku, Turku University Hospital, Turku, cific markers in Alzheimer’s disease. Biochem Biophys Res Commun. Finland. 8 Graduate Program in Biological Sciences: Biochemistry, Universidade 2006;346(3):819–28. Federal do Rio Grande do Sul (UFRGS), Porto Alegre, Brazil. 9 Department 19. Hanger DP, et al. Novel phosphorylation sites in tau from Alzheimer brain of Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain. 10 Depart‑ support a role for casein kinase 1 in disease pathogenesis. J Biol Chem. ment of Psychiatry, School of Medicine, University of Pittsburgh, Pittsburgh, 2007;282(32):23645–54. PA, USA. 11 ADx NeuroSciences, Technologiepark 94, Ghent, Belgium. 12 Centre 20. Sengupta A, et al. Phosphorylation of tau at both Thr 231 and Ser 262 for Age‑Related Medicine, Stavanger University Hospital, Stavanger, Norway. 13 is required for maximal inhibition of its binding to microtubules. Arch Department of Old Age Psychiatry, Maurice Wohl Clinical Neuroscience Insti‑ Biochem Biophys. 1998;357(2):299–309. tute, King’s College London, London, UK. 14 NIHR Biomedical Research Centre 21. Cho JH, Johnson GV. Primed phosphorylation of tau at Thr231 by for Mental Health & Biomedical Research Unit for Dementia at South London & glycogen synthase kinase 3beta (GSK3beta) plays a critical role in Maudsley NHS Foundation, London, UK. 15 Clinical Neurochemistry Laboratory, regulating tau’s ability to bind and stabilize microtubules. J Neurochem. Sahlgrenska University Hospital, Mölndal, Sweden. 16 Department of Neurode‑ 2004;88(2):349–58. generative Disease, Queen Square Institute of Neurology, University College Lantero‑Rodriguez et al. Alzheimer’s Research & Therapy (2023) 15:48 Page 15 of 15 22. Van der Mussele S, et al. Behavioral syndromes in mild cognitive impair‑ ment and Alzheimer’s disease. J Alzheimers Dis. 2014;38(2):319–29. 23. Valero S, et al. Interaction of neuropsychiatric symptoms with APOE epsilon4 and conversion to dementia in MCI patients in a Memory Clinic. Sci Rep. 2020;10(1):20058. 24. Xiong C, et al. Complex interactions underlie racial disparity in the risk of developing Alzheimer’s disease dementia. Alzheimers Dement. 2020;16(4):589–97. 25. Santiago JA, Potashkin JA. The impact of disease comorbidities in Alzhei‑ mer’s disease. Front Aging Neurosci. 2021;13:631770. 26. Petersen RC. Mild cognitive impairment criteria in Alzheimer’s disease neuroimaging initiative: meeting biological expectations. Neurology. 2021;97(12):597–9. 27. McKeith IG, et al. Diagnosis and management of dementia with Lewy bodies: Fourth consensus report of the DLB Consortium. Neurology. 2017;89(1):88–100. 28. Rascovsky K, et al. Sensitivity of revised diagnostic criteria for the behav‑ ioural variant of frontotemporal dementia. Brain. 2011;134(Pt 9):2456–77. 29. Molinuevo JL, et al. Implementation of subjective cognitive decline criteria in research studies. Alzheimers Dement. 2017;13(3):296–311. 30. Jessen F, et al. The characterisation of subjective cognitive decline. Lancet Neurol. 2020;19(3):271–8. 31. Puig-Pijoan A, et al. The CORCOBIA study: cut-off points of Alzheimer’s disease CSF biomarkers in a clinical cohort. Neurologia (Engl Ed). 2022. 32. Ashton NJ, et al. Plasma and CSF biomarkers in a memory clinic: head- to-head comparison of phosphorylated tau immunoassays. Alzheimers Dement. 2022. 33. Leitao MJ, et al. Clinical validation of the Lumipulse G cerebrospinal fluid assays for routine diagnosis of Alzheimer’s disease. Alzheimers Res Ther. 2019;11(1):91. 34. Karikari TK, et al. Head-to-head comparison of clinical performance of CSF phospho-tau T181 and T217 biomarkers for Alzheimer’s disease diagno‑ sis. Alzheimers Dement. 2021;17(5):755–67. 35. Snellman A, et al. N-terminal and mid-region tau fragments as fluid biomarkers in neurological diseases. Brain. 2022;145(8):2834–48. 36. Mielke MM, et al. Comparison of CSF phosphorylated tau 181 and 217 for cognitive decline. Alzheimers Dement. 2022;18(4):602–11. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions