Plasma phosphorylated tau 217 and longitudinal trajectories of Aβ, tau, and cognition in cognitively unimpaired older adults - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Nat Commun . 2026 Apr 14;17:3188. doi: 10.1038/s41467-026-71269-3 Search in PMC Search in PubMed View in NLM Catalog Add to search Plasma phosphorylated tau 217 and longitudinal trajectories of Aβ, tau, and cognition in cognitively unimpaired older adults Hyun-Sik Yang Hyun-Sik Yang 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA 4 Broad Institute of MIT and Harvard, Cambridge, MA USA Find articles by Hyun-Sik Yang 1, 2, 3, 4, ✉ , Juliana A U Anzai Juliana A U Anzai 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA Find articles by Juliana A U Anzai 1, 2 , Wai-Ying Wendy Yau Wai-Ying Wendy Yau 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Wai-Ying Wendy Yau 1, 2, 3 , Brian C Healy Brian C Healy 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Brian C Healy 1, 3 , Andrea M Román Viera Andrea M Román Viera 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA Find articles by Andrea M Román Viera 1, 2 , Courtney Maa Courtney Maa 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA Find articles by Courtney Maa 1, 2 , Dylan Kirn Dylan Kirn 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA Find articles by Dylan Kirn 1, 2 , Michael J Properzi Michael J Properzi 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA Find articles by Michael J Properzi 2 , Jean-Pierre Bellier Jean-Pierre Bellier 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA Find articles by Jean-Pierre Bellier 1 , Aaron P Schultz Aaron P Schultz 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Aaron P Schultz 2, 3 , Michelle E Farrell Michelle E Farrell 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Michelle E Farrell 2, 3 , Heidi I L Jacobs Heidi I L Jacobs 3 Harvard Medical School, Boston, MA USA 5 Department of Radiology, Massachusetts General Hospital, Boston, MA USA Find articles by Heidi I L Jacobs 3, 5 , Rachel F Buckley Rachel F Buckley 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA 6 Melbourne School of Psychological Sciences, University of Melbourne, VIC, Australia Find articles by Rachel F Buckley 2, 3, 6 , Kathryn V Papp Kathryn V Papp 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Kathryn V Papp 1, 2, 3 , Gad A Marshall Gad A Marshall 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Gad A Marshall 1, 2, 3 , Rebecca E Amariglio Rebecca E Amariglio 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Rebecca E Amariglio 1, 2, 3 , Dorene M Rentz Dorene M Rentz 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Dorene M Rentz 1, 2, 3 , Lei Liu Lei Liu 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Lei Liu 1, 3 , Dennis J Selkoe Dennis J Selkoe 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Dennis J Selkoe 1, 3 , Philip B Verghese Philip B Verghese 7 C2N Diagnostics LLC, St. Louis, MO USA Find articles by Philip B Verghese 7 , Joel B Braunstein Joel B Braunstein 7 C2N Diagnostics LLC, St. Louis, MO USA Find articles by Joel B Braunstein 7 , Keith A Johnson Keith A Johnson 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA 5 Department of Radiology, Massachusetts General Hospital, Boston, MA USA Find articles by Keith A Johnson 1, 2, 3, 5 , Reisa A Sperling Reisa A Sperling 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Reisa A Sperling 1, 2, 3 , Jasmeer P Chhatwal Jasmeer P Chhatwal 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA Find articles by Jasmeer P Chhatwal 1, 2, 3 Author information Article notes Copyright and License information 1 Department of Neurology, Brigham and Women’s Hospital, Boston, MA USA 2 Department of Neurology, Massachusetts General Hospital, Boston, MA USA 3 Harvard Medical School, Boston, MA USA 4 Broad Institute of MIT and Harvard, Cambridge, MA USA 5 Department of Radiology, Massachusetts General Hospital, Boston, MA USA 6 Melbourne School of Psychological Sciences, University of Melbourne, VIC, Australia 7 C2N Diagnostics LLC, St. Louis, MO USA ✉ Corresponding author. Received 2025 Mar 28; Accepted 2026 Mar 16; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, 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 you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. 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-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13079864 PMID: 41980948 Abstract Plasma phosphorylated tau 217 (pTau217) is an excellent biomarker of Alzheimer’s disease (AD) pathology, but it remains uncertain whether pTau217 can predict amyloid-β (Aβ) and tau accumulation prior to Aβ positron emission tomography (PET) positivity. Here, we leverage data from a well-characterized prospective cohort of cognitively unimpaired older adults to examine mass spectrometry-based plasma %pTau217 (pTau217/non-phosphorylated-Tau217×100) relative to changes in Aβ/tau PET and cognition. A higher baseline %pTau217 was associated with faster Aβ and tau accumulation on PET, which then led to greater cognitive decline. Among individuals Aβ PET-negative at baseline, higher %pTau217 levels presaged increases in Aβ and tau PET signals. Together, our results suggest that very low %pTau217 in cognitively unimpaired older adults is associated with a minimal risk of AD pathology accumulation and cognitive decline. Subject terms: Neurodegeneration, Predictive markers Plasma phosphorylated tau 217 levels predict future accumulation of Alzheimer’s disease pathology in cognitively unimpaired older adults without elevated brain amyloid-β (“A-”) on positron emission tomography. Introduction Alzheimer’s disease (AD) is the most common cause of late-onset dementia and among the leading causes of death, affecting close to 7 million older adults in the United States and over 55 million worldwide 1 . Recently, two anti-amyloid-β (anti-Aβ) treatments that can partially slow down AD progression 2 , 3 have become available. As a result, biomarker confirmation of AD pathology—required for these treatments—has become increasingly important. Presently, the primary modalities for clinical detection of AD pathology depend on testing of cerebrospinal fluid (CSF) Aβ and tau markers and/or Aβ positron emission tomography (PET), tests that are not widely available beyond resource-rich settings. Recently, plasma phosphorylated tau (tau phosphorylated at the 217 th residue [threonine]; pTau217), particularly when expressed as a ratio against non-phosphorylated tau to normalize for inter-individual variability 4 , has emerged as a highly accurate, non-invasive, and less expensive AD pathology biomarker that has the potential to be widely used in primary care and specialty clinic settings 5 . Elevated plasma pTau217 is AD-specific 6 , 7 , and elevated plasma %pTau217 was highly correlated with Aβ and tau PET, showing equivalent to superior performance compared to the widely used CSF tests 8 . Thus, plasma pTau217 is accepted as a core biomarker that can be used for AD diagnosis 9 , and is increasingly being introduced into clinical practice to determine the presence of AD pathology in mild cognitive impairment (MCI) or dementia patients. AD includes up to two decades of asymptomatic or preclinical stage when the pathophysiology progresses before clinical symptom onset 9 – 11 . As a successful disease-modifying intervention in the preclinical stage of AD might prevent irreversible neurodegeneration and progression to symptomatic AD, prevention trials targeting AD pathophysiology in cognitively unimpaired (CU) participants have been completed or are ongoing 12 , 13 . However, identifying individuals on an AD trajectory during the earliest, presymptomatic phases of the disease remains challenging, both from diagnostic and feasibility perspectives. In this context, blood-based assays of AD pathology, such as plasma pTau217, may greatly benefit the execution of AD prevention trials and broader screening efforts. Indeed, prior work demonstrates that plasma pTau217 accurately reflects Aβ and tau burden measured by PET, even in the preclinical stage 14 , 15 , and is being used as a first-stage screening tool for prevention trials 16 . Moreover, plasma pTau217 level in CU older adults without elevated Aβ (“A-“) has been shown to predict future Aβ accumulation 17 , 18 , and a longitudinal increase in pTau217 has been shown to correlate with progressive brain atrophy and cognitive decline 14 , 15 , 19 , suggesting that plasma pTau217 may be used to track individual-level disease progression and treatment response in the earliest preclinical stages of AD. However, the predictive potential of plasma pTau217 levels in CU older adults in the accumulation of AD pathology and cognitive decline remains to be clarified, as previous studies 17 – 19 have not examined longitudinal tau PET changes and focused on changes of either Aβ or cognitive decline during a relatively short timeframe. Here, we analyze baseline and longitudinal plasma pTau217 from the Harvard Aging Brain Study (HABS) 20 , a prospective cohort study of cognitive aging and preclinical AD (mean follow-up 8.0 ± 3.9 years) that enrolled older CU at baseline. We use the pTau217/npTau217 ratio expressed as a percentage (%pTau217) from the C 2 N mass spectrometry-based platform and quantify its longitudinal association with Aβ PET, tau PET, and cognition. We then perform path analyses to connect baseline %pTau217 with longitudinal Aβ, tau, and cognition in CU older adults, clarifying the longitudinal implications of plasma pTau217 in CU older adults. Results Higher plasma %pTau217 is associated with longitudinal Aβ accumulation in early-stage AD We examined 317 HABS participants who were CU at baseline and had plasma pTau217 and Aβ ( 11 C-Pittsburgh compound B [PiB]) PET data at baseline (Table 1 , Supplementary Table 1 ). Table 1. Baseline Participant Characteristics All ( n = 317) A- ( N = 242) A+ ( N = 75) Age, years (SD) 71.7 (8.0) 70.5 (8.1) 75.3 (6.2) Females, n (%) 192 (61) 147 (61) 45 (60) Education, yr, mean (SD) 15.9 (2.9) 15.8 (2.9) 16.1 (3.0) White, n (%) 257 (81) 191 (79) 66 (88) APOE ε4 carriers, n (%) 89 (28) 44 (18) 45 (60) Mean length of follow-up, Years (SD) 8.0 (3.9) 8.2 (3.9) 7.5 (3.8) Median MMSE (IQR) 29 (1) 29 (1) 29 (2) Mean PiB cortical composite, Centiloid (SD) 21.0 (26.9) 8.2 (7.0) 62.5 (25.6) Mean entorhinal cortex FTP SUVR a (SD) 1.10 (0.12) 1.07 (0.088) 1.20 (0.17) Mean inferior temporal cortex FTP SUVR a (SD) 1.18 (0.097) 1.16 (0.066) 1.25 (0.15) %pTau217 (SD) 3.6 (2.8) 2.5 (1.3) 7.30 (3.28) Open in a new tab PiB Pittsburgh compound B (Aβ PET tracer), FTP flortaucipir (tau PET tracer), SUVR standardized uptake value ratio (cerebellar gray reference; partial volume corrected). a FTP PET was introduced mid-study, and the first available FTP PET data from each participant (available from n = 233, with an average of 2.6 ± 1.7 years after baseline) are summarized. MMSE the Mini-Mental State Examination. Baseline plasma %pTau217 was strongly associated with baseline cortical Aβ burden measured by Aβ PET (Fig. 1a ). Receiver operating characteristics (ROC) area under the curve (AUC) to classify Aβ status (A+ vs. A-, with Aβ CL cut point of 24.1 Centiloid [CL]) with %pTau217 was 0.94 (Fig. 1b, c ). Using the cut point of 4.2% recommended as a component of the current clinical C 2 N assay for use in individuals with cognitive impairment 21 , sensitivity and specificity to detect A+ were 0.89 and 0.90, respectively, demonstrating consistent performance of the plasma %pTau217 in detecting elevated cortical Aβ burden. Notably, three participants had high Aβ PET signals (Aβ CL > 50 CL) but with plasma %pTau217 levels below 4.2% at baseline (Fig. 1b ); two of these participants had subsequent %pTau217 measures, all of which were greater than 4.2% (Supplementary Table 2 ). Fig. 1. Plasma %pTau217 and cross-sectional and longitudinal cortical Aβ (PiB PET). Open in a new tab a Baseline PiB PET Centiloid [CL] and plasma %pTau217. The dashed vertical line shows the threshold for A+ (24.1 CL), and the dashed horizontal line shows the clinical %pTau217 cut point (4.2%). b Plasma %pTau217 by baseline PiB groups (“A + ”: cortical composite region PiB DVR > 24.1 centiloid [CL], n = 75; “A-“: cortical composite region PiB DVR ≤ 24.1 CL, n = 242). Center lines indicate median; box limits indicate upper and lower quartiles; whiskers indicate 1.5×interquartile range. c Receiver operating characteristic (ROC) curve of %pTau217 in discriminating between A- and A + . Area under the curve (AUC) = 0.94. The sensitivity (0.89) and specificity (0.90) for the clinical %pTau217 cut point (4.2%) are shown as horizontal and vertical dashed lines, respectively. d Estimated longitudinal cortical Aβ burden (PiB PET CL) trajectory (with 95% confidence interval [CI]) among all participants with longitudinal PiB PET. Mean values of %pTau217 among A- (2.45) and A+ (7.56) were used for visualization. e , f Estimated longitudinal cortical Aβ burden (PiB PET CL) trajectory (with 95% confidence interval) in the A- subgroup ( < 24.1 CL; e ) and the < 10 CL subgroup ( f ). Mean values of each %pTau217 tertile within each subgroup (1st tertile [blue]: 1.34, 2nd tertile [orange]: 2.13, 3rd tertile [red]: 3.91) were used for visualization. The dashed horizontal line indicates the threshold for A+ (24.1 CL). In ( d –f ), we showed p-values for the [baseline %pTau217]×[time] term in predicting longitudinal PiB PET CL in linear mixed effect models are shown (see main text for the effect size and 95% CI). g The differential risk of A- to A+ progression among the CU A- subgroup ( n = 188 after excluding three participants who progress to A+ but later return to A-) across the baseline %pTau217 tertile was shown with Kaplan-Meier curves. The log-rank test Chi-square statistic is shown. The number of at-risk (upper table) and cumulative events per group (lower table) at years 3, 6, and 9 are noted below the plot. All p -values shown are two-sided and are not adjusted for multiple comparisons. Source data are provided as a Source Data file. Higher baseline plasma %pTau217 was associated with a longitudinal increase in Aβ measured by PET. In n = 245 participants with longitudinal Aβ PET data, 92, 66, and 87 participants had 2, 3, and 4 scans, respectively; the mean duration between baseline and last Aβ PET scan was 6.0 ± 2.3 years. Baseline plasma %pTau217 predicted a longitudinal increase in cortical Aβ accumulation after adjusting for APOE ε4 carrier state, age, and sex (Fig. 1d , b = 0.35 [CL/Year per one percent increase in %pTau217], 95% CI 0.27–0.43, p = 8.9 × 10 −16 ). Associations between higher baseline %pTau217 levels and greater longitudinal increases in cortical Aβ accumulation were observed even in the subgroup with sub-threshold Aβ at baseline (Fig. 1e , baseline A- subgroup, n = 191, b = 0.26, 95% CI 0.13–0.40, p = 1.6 × 10 −4 ) and in a subgroup with even lower levels of baseline Aβ (CL < 10; Fig. 1f , n = 123, b = 0.22, 95% CI 0.035–0.40, p = 0.020). Thus, higher plasma %pTau217 may predict future cortical Aβ accumulation even when the baseline Aβ burden is well below the conventional PET thresholds for “amyloid positivity” (A + ). Furthermore, in the baseline A- subgroup, baseline plasma %pTau217 predicted the likelihood of reaching the threshold for elevated Aβ (“A + ”; >24.1 CL) during the study period (Fig. 1g ). We conducted survival analyses across three groups categorized by baseline %pTau217 tertile (within the baseline A- subgroup only) and found that the survival probably (i.e., remaining A-) at 6 years from the baseline was 0.98 (95% CI 0.95 to 1.0) for participants in the first (lowest) %pTau217 tertile and 0.72 (95% CI 0.60 to 0.87) for those in the third (highest) %pTau217 tertile. We additionally used a Cox regression model adjusting for APOE ε4, age, and sex, and observed a consistent result: higher baseline %pTau217 was associated with higher risk of progression to A+ ( n = 188 participants, n = 28 A+ progression events, hazard ratio [HR] per 1% increase in %pTau217 = 1.40, 95% CI 1.13 to 1.72, p = 1.7 × 10 −3 ), while APOE ε4, age, and sex were not associated with the risk of progression to A+ (all p > 0.05). Notably, after additionally adjusting for baseline Aβ CL, baseline %pTau217 was no longer associated with Aβ progression ( p = 0.31), while baseline Aβ was a stronger predictor of Aβ progression in this model (HR per 1 CL increase in Aβ PET = 1.22, 95% CI 1.13–1.32, p = 3.9 × 10 −7 ). This observation was likely driven by individuals with their baseline Aβ burden closer to the A+ threshold: after excluding 10 participants whose baseline Aβ was greater than 20 CL but less than cut point (24.1 CL), higher baseline %pTau217 predicted progression to A+ ( n = 178, HR = 1.52, 95% CI 1.06 to 2.18, p = 0.023), while the association between baseline Aβ CL and progression to A+ became numerically weaker (HR = 1.17, 95% CI 1.06–1.28, p = 1.4 × 10 −3 ). Moreover, greater baseline %pTau217 was associated with a higher risk of progression from baseline Aβ PET of less than 10 CL to Aβ positivity ( n = 123, HR = 2.67, 95% CI 1.17–6.11, p = 0.020); however, only 6 participants from this subgroup with very low initial Aβ PET progressed to amyloid PET positivity within the study period, and this result should be interpreted cautiously. Together, our results support the utility of plasma %pTau217 not only in diagnosing cross-sectional A+ status but also in the early identification of individuals at risk of faster Aβ accumulation and eventual progression to A + . Longitudinal increase in plasma %pTau217 appears to precede A- to A+ progression We utilized paired measures of longitudinal plasma %pTau217 and Aβ CL (from the same study years; n = 223) to evaluate whether changes in these biomarkers over time were correlated. The overall trajectory of %pTau217 (x-axis) and Aβ CL over time followed a sigmoid curve, suggesting that plasma %pTau217 increases before substantial cortical Aβ accumulation (Fig. 2a ). To examine this possibility, we first defined subgroups based on their baseline Aβ status (A+ vs. A-) and %pTau217 level (low vs. intermediate-high %pTau217). Fig. 2. Longitudinal relationship of %pTau217 and Aβ (PiB) PET. Open in a new tab Each node represents %pTau217 (x-axis) and Aβ PET Centiloid (CL) (y-axis) from the same study year of a participant. All data points from each individual are connected with edges in temporal order, with an arrowhead indicating the final available data point (average time interval between nodes: 3.0 ± 0.9 years). The red curved line shows the average trajectory estimated using a general additive model calculated for visualization. The blue dashed horizontal line shows the threshold for A+ (24.1 CL), and the blue dashed vertical line shows the %pTau217 cut point (2.6%) with maximum Youden’s index to distinguish Aβ progressors from Aβ non-progressors. a All study participants with matched %pTau217 and Aβ CL from at least two time points ( n = 223). b Baseline A+ subgroup ( n = 49). c Baseline A-/low %pTau217 subgroup ( n = 106). d Baseline A-/high %pTau217 subgroup ( n = 68). b – d The number of data points from the final visit is shown for each quadrant to demonstrate coordinated temporal changes of %pTau217 and Aβ PET CL. The source data for this figure are individual-level human participant data available upon request through the procedures outlined in the Data Availability statement. To define the subgroup with low %pTau217, we derived a cut point for %pTau217 that optimally predicts Aβ progression in our study. The ROC AUC for baseline %pTau217 in predicting Aβ progression among the baseline A- subgroup was 0.73 (95% CI 0.62 to 0.84), and a %pTau217 cut point of 2.6% maximized Youden’s index (sensitivity 0.75, specificity 0.66) (Supplementary Fig. 1 ). The ROC AUC we observed is lower than previously reported values ranging from 0.87 to 0.95 17 : this discrepancy might be due to different Aβ CL used to define A+ as well as study participant characteristics, as our participants were older, less likely to be APOE ε4 carriers, and had longer follow-up. It is important to note that the 2.6% cut point should be interpreted as a cohort-specific threshold optimized for our study, rather than a definite and generalizable value: this value was chosen to illustrate the longitudinal %pTau217 – Aβ trajectory of the A- participants with %pTau217 much lower than the clinical A+ cut point (4.2% 21 ). All but one baseline A+ participant had intermediate-high %pTau217 at all time points, and A + /low %pTau217 state was rare (Fig. 2b , n = 49). Further, one A+ participant with low %pTau217 at baseline moved to the A + /intermediate-high %pTau217 state in their next visit, suggesting that the baseline %pTau217 measure might have been an erroneously low value. None of the participants A+ at baseline moved to A− in their follow-up PET scans, suggesting that the A+ status strongly supports the presence of preclinical AD (brain β-amyloidosis) and could be used to define a target population in secondary AD prevention trials. Participants in the A-/low %pTau217 subgroup ( n = 106) rarely progressed to A+ (5%, n = 5) during the study period, while 32% ( n = 34) of this subgroup moved to A-/intermediate-high %pTau217 category. Critically, all of the A+ progressors also had their %pTau217 in the intermediate-high range when they progressed to A+ (Fig. 2c ). Therefore, the A-/low %pTau217 subgroup might represent a low-risk group who does not require Aβ PET until %pTau217 is elevated. The A−/intermediate-high %pTau217 subgroup ( n = 68) represented a more heterogeneous category (Fig. 2d ). By the time of their last visit, 25% ( n = 17) progressed to A + /high %pTau217, and 25% ( n = 17) had decreased %pTau217 and moved to the A-/low %pTau217 state. One participant moved to the A + /low-%pTau217 category, but their last Aβ CL was only borderline positive (24.11 CL). Notably, the A-/intermediate-high %pTau217 state can still revert to the A-/low-%pTau217 state. While it is unclear whether this is driven by true biological fluctuations or noise in measuring lower pTau217 levels, there is an important practical implication: the cohort-specific 2.6% threshold used in this study is not a definite indicator of irreversible brain pathology progression. Lastly, we examined the predictors of A+ progression in the intermediate-high %pTau217 subgroup, given heterogeneous outcomes associated with this state. In a Cox regression model in this subgroup of participants who are A-/intermediate-high-%pTau217 at baseline ( n = 68), longitudinal (time-varying) %pTau217 predicted the risk of A+ progression (HR = 1.53, 95% CI 1.20 to 1.96, p = 7.4 × 10 −4 ); by contrast, neither baseline %pTau217, APOE ε4, age, nor sex could predict A+ progression (all p > 0.05). Thus, in this subgroup, serial %pTau217 measurement may help identify those at the highest risk of A+ progression. Higher plasma %pTau217 is associated with longitudinal mesial temporal and neocortical tau accumulation We examined the association of baseline %pTau217 with longitudinal fibrillar tau pathology burden measured with flortaucipir (FTP) PET standardized uptake value ratio (SUVR). We first focused on the entorhinal cortex (EC) to capture the earliest regions of tau accumulation. Higher baseline %pTau217 was associated with greater increases in longitudinal EC tau (Fig. 3a ; n = 182, b = 4.6 × 10 −3 [SUVR/Year per one unit of %pTau217], 95% CI 2.2 × 10 −3 to 7.0 × 10 −3 , p = 1.6 × 10 −4 ), even after adjusting for baseline Aβ PET CL (Table 2 ). In the same model, the association of baseline Aβ with longitudinal EC tau was no longer significant once %pTau217 was considered (Table 2 ). Further, greater baseline %pTau217 was associated with greater longitudinal EC tau accumulation in the A- subgroup (Fig. 3b ; n = 145, b = 6.4 × 10 −3 , 95% CI 2.0 × 10 −3 to 0.011, p = 4.9 × 10 −3 ) and the Aβ CL < 10 subgroup (Fig. 3c ; n = 100, b = 7.6 × 10 −3 , 95% CI 9.1 × 10 −4 to 0.014, p = 0.026). By contrast, the association between baseline Aβ CL and longitudinal EC tau was absent in the Aβ CL < 10 subgroup ( p = 0.60). Together, these results demonstrate that greater plasma %pTau217 is modestly associated with fibrillar tau accumulation, even in individuals with sub-threshold levels of Aβ PET. Fig. 3. Baseline plasma %pTau217 and longitudinal tau (FTP PET). Open in a new tab a –c The estimated longitudinal trajectory of entorhinal cortex (EC) FTP standardized uptake value ratio (SUVR) (with 95% confidence interval [CI]), in all participants ( a ), the A- subgroup ( b ), and the baseline PiB CL < 10 subgroup ( c ). Mean %pTau217 values of A- and A+ subgroups ( a ) and each %pTau217 tertile ( b , c ) were shown. d The estimated longitudinal trajectory of inferior temporal cortex (ITC) FTP SUVR (with 95% CI). Mean %pTau217 values of A- and A+ subgroups were shown. In ( a –d ), we showed p-values for the [baseline %pTau217]×[time] term in predicting longitudinal FTP PET SUVR ( a – c : EC, d : ITC) in each linear mixed effect model (see main text for the effect size and 95% CI). e The baseline %pTau217 – longitudinal tau association in each bilateral cortical and subcortical FTP PET region of interest (ROI). Regions with FDR < 0.05 were colored with their association t-values. This image was created with an R package, “ggseg.” All p -values shown are two-sided and are not adjusted for multiple comparisons, except for the FDR values shown in e. Source data are provided as a Source Data file. Table 2. Association of baseline %pTau217 and Aβ with longitudinal temporal tau Outcome Independent variable b (95% CI), p Longitudinal EC tau Baseline %pTau217 × time b = 5.0 × 10 −3 (1.3×10 −3 to 8.7 × 10 −3 ), p = 7.5 × 10 −3 Baseline Aβ (CL) × time b = −6.5 × 10 −5 (−4.0×10 −4 to 2.7 × 10 −4 ), p = 0.70 Longitudinal ITC tau Baseline %pTau217 × time b = 5.0 × 10 −3 (2.5×10 −3 to 7.5 × 10 −3 ), p = 1.3 × 10 −4 Baseline Aβ (CL) × time b = −2.8 × 10 −5 (−2.6 × 10 −4 to 2.0 × 10 −4 ), p = 0.81 Open in a new tab Results from two separate linear mixed effect models were shown, one with longitudinal EC tau and another with longitudinal ITC tau as an outcome ( n = 182). Each model included (Baseline %pTau217)×(time), (Baseline Aβ)×(time), and their main terms as well as covariates ( APOE ε4 status, age, and sex) and their time interaction terms. All p -values shown are two-sided and are not adjusted for multiple comparisons. EC entorhinal cortex, ITC inferior temporal cortex. We next examined tau PET measures from the inferior temporal cortex, an early neocortical region of AD-related tau accumulation. Similar to results in the EC, greater baseline %pTau217 was associated with greater longitudinal increases in ITC tau (Fig. 3d ; n = 182, b = 4.8 × 10 −3 , 95% CI 3.2 × 10 −3 to 6.5 × 10 −3 , p = 1.7 × 10 −8 ). The association between baseline %pTau217 and longitudinal ITC tau accumulation remained significant even after adjusting for baseline Aβ. Conversely, the association of baseline Aβ with longitudinal ITC tau was no longer significant after adjusting for baseline %pTau217 (Table 2 ). The %pTau217 – ITC tau association was not observed in the A- subgroup and the baseline Aβ CL < 10 subgroups (both p > 0.05), consistent with the literature that neocortical tau accumulation usually occurs in the setting of elevated Aβ 22 , 23 . We explored the association of baseline %pTau217 with longitudinal tau across regions of interest (ROIs). We observed positive associations in the temporoparietal and amygdala regions, consistent with the well-described topology of early AD-related tau accumulation (Fig. 3e ). Together, these results suggest that higher plasma %pTau217 reflects pathophysiologic processes leading to fibrillar tau accumulation, including the link between baseline Aβ and future fibrillar tau accumulation. Further, in those with low fibrillar Aβ burden, %pTau217 may also capture Aβ-independent pathways leading to early mesial temporal tau accumulation, although the larger dynamic range of %pTau217 compared to Aβ PET in this subgroup might have driven the results. Higher plasma %pTau217 is associated with greater cognitive decline, an effect mediated via longitudinal Aβ and tau accumulation Baseline %pTau217 was associated with longitudinal change in cognitive composite (PACC5) in all participants ( n = 308, b = −0.019, 95% CI −0.024 to −0.013, p = 1.6 × 10 −10 ; Fig. 4a ). This association was not present when the analysis was limited to the baseline A- subgroup ( n = 235, b = −5.7 × 10 −3 , 95% CI −0.017 to 5.3 × 10 −3 , p = 0.31; Fig. 4b ). Thus, the association of higher plasma %pTau217 with faster cognitive decline in CU older adults is likely driven by those with elevated Aβ (A + ); %pTau217 did not predict cognitive decline in CU A- older adults even after mean follow-up of 8.2 ± 3.9 years (for PACC5). Fig. 4. Baseline plasma %pTau217 and longitudinal Δ Aβ, Δ Tau, and PACC5. Open in a new tab a , b The predicted longitudinal trajectory of PACC5 (with 95% confidence interval) in all participants ( a ) and the A- subgroup ( b ). Mean %pTau217 values in the A- (2.5) and A+ (7.3) subgroups were used in panel ( a ), and mean %pTau217 values for each tertile among the A- subgroup were used in ( b ) for visualization. In ( a , b ), we showed p-values for the [baseline %pTau217]×[time] term in predicting longitudinal PACC in each linear mixed effect model (see main text for the effect size and 95% confidence interval). c Path analysis shows that higher baseline %pTau217 is associated with longitudinal Aβ accumulation ( Δ Aβ; Aβ CL slope), which in turn leads to longitudinal neocortical tau accumulation in the neocortex ( Δ ITC-Tau) and cognitive decline (PACC5 slope). Δ Aβ and Δ ITC-Tau explained most of the association between %pTau217 with cognitive decline, and direct %pTau217 – Δ PACC was not significant in this model. Solid arrows indicate significant associations ( p < 0.05); the associations with p > 0.05 were shown in dashed arrows. Numbers adjacent to each arrow show the relative strength of each association (completely standardized solution). Figure ( c ) was partially created with BioRender. Yang, H. (2026) https://BioRender.com/yts4cwp . All p -values shown are two-sided and are not adjusted for multiple comparisons. PACC5: Preclinical Alzheimer’s Cognitive Composite 5. Source data are provided as a Source Data file. Finally, by combining the observed association of %pTau217 with longitudinal cortical Aβ ( Δ Aβ), neocortical tau ( Δ ITC-Tau), and cognition, we conducted a path analysis to examine a potential sequence of biomarker changes connecting baseline %pTau217 to longitudinal cognition (Fig. 4c ). Our model was based on the previously described associations between Aβ, neocortical tau, and cognition 22 – 24 , including those in the present report. In the path analysis, the association between baseline %pTau217 and Δ ITC-Tau was largely mediated through Δ Aβ, while the association between Δ Aβ and PACC5 decline was mediated through Δ ITC-Tau. The direct associations of baseline %pTau217 or Δ Aβ with PACC5 decline were not statistically significant once Δ ITC-Tau was considered (Fig. 4c ). Of note, Δ EC-tau was highly collinear with Δ ITC-tau ( r = 0.63), and Δ EC-tau did not exhibit a direct association with Δ PACC5 ( p > 0.05) once Δ ITC-tau was taken into consideration; thus, we excluded Δ EC-tau from this model. It is important to emphasize that the path analysis does not establish causality nor suggest that elevated plasma %pTau217 directly leads to increasing Aβ, tau, or cognitive decline. Instead, the primary aim of this analysis was to determine whether cognitively unimpaired older adults with higher %pTau217 are at increased risk of cognitive decline due to the accelerated progression of AD pathology. Discussion Leveraging a deeply-phenotyped cohort of CU older adults with biofluid, neuroimaging, and cognitive follow-up, we comprehensively examined the extent to which plasma %pTau217 levels predicted AD pathology progression and cognitive change. Our study utilized longitudinal Aβ (PiB PET), tau (FTP PET), and cognitive data spanning up to 13.8 years (average 8.0 ± 3.9 years) from a prospective cohort of unimpaired older adults that resembles an older adult population that may be targeted for AD prevention trials and may soon be considered for clinical screening using newly available blood tests for AD. Consistent with prior reports, we observe that elevated %pTau217 in CU older adults is strongly associated with AD pathology burden. Moreover, the length of available follow-up in the examined cohort enabled us to observe that greater plasma %pTau217 levels were associated with a greater accumulation of both Aβ and tau pathologies, even in individuals who entered the study with low Aβ burden. These results suggest that changes in plasma %pTau217 likely precede changes in Aβ PET, and that very low levels of plasma %pTau217 indicate a low risk of substantial Aβ and tau accumulation in the following several years. Our findings are consonant with a recent report that %pTau217 can predict longitudinal Aβ PET change even in those with low baseline Aβ 17 , 18 , and expand these findings by incorporating longitudinal measures of %pTau217 and tau PET. Specifically, using paired, longitudinal %pTau217 and Aβ PET data, we observed that an increase in %pTau217 temporally preceded the emergence of supra-threshold levels of Aβ PET signal (i.e., progression to A + ) in almost all cases examined. This observation suggests that screening strategies based on longitudinal plasma %pTau217 levels may efficiently identify older adults entering the preclinical stages of AD. Concordant results were observed in examining tau PET, where %pTau217 correlated with PET-based measures of tau accumulation both in mesial temporal and neocortical areas. %pTau217 predicted future EC tau accumulation even when Aβ PET was below the threshold, suggesting that elevated %pTau217 may also reflect early phase mesial temporal tau aggregation that precedes clear elevations in fibrillar Aβ burden. Thus, elevated %pTau217 might identify CU individuals at a higher risk of progressive Aβ and tau accumulation, and in turn, faster accumulation of these AD pathologies may underlie the link between higher %pTau217 and faster cognitive decline. Importantly, our results suggest that individuals with very low levels of plasma pTau217 may not need frequent blood-based screening or Aβ PET. In contrast, participants with modestly elevated %pTau217 (intermediate range, still below the clinical cut point of 4.2%) represent a group with an intermediate risk of future AD, and may require more frequent screening, as increasing %pTau217 on serial measures was found in most cases that progressed to A + . Finally, those with a high %pTau217 (> 4.2%) are very likely to have preclinical AD, suggesting that confirmatory Aβ PET may be appropriate in certain clinical circumstances. Notably, since %pTau217 appears to change very early in the disease process, not everyone with elevated %pTau217 may be at an imminent risk of cognitive decline. A second biomarker (e.g., tau PET) is likely necessary to identify individuals at risk for cognitive decline in the next few years. Several limitations of our study should be considered when interpreting our data. First, we did not analyze plasma Aβ measures. A previous study has shown that, when combined with %pTau217, plasma Aβ42/40 ratio can explain additional variance in Aβ PET change in A- CU older adults 17 . Analyzing Aβ42/40 ratio and additional promising biomarkers of tau pathology such as microtubule binding region (MTBR)-tau243 25 in the future could further clarify the relationship among fluid and neuroimaging biomarkers in early AD progression. Second, the cut points we used are specific to the C 2 N v2 %pTau217 assay (which is a component of the PrecivityAD2 TM test with intended use in individuals with cognitive impairment). The manufacturer-established clinical cut point of 4.2% 21 is applicable to the assay used in this study, but cannot be directly applied to other assay platforms 6 or to earlier versions of the C 2 N assay 16 . The lower cut points we used here to examine the A+ progression—%pTau217 tertiles and the 2.6% cut point—are assay- and cohort-specific cut points selected primarily for illustration purposes, and may not represent biologically meaningful thresholds. While cut points are useful tools for operationalizing biomarker interpretation and application, they do not always capture the continuous nature of disease biology, especially when the trait of interest varies continuously without a clear transition point. Third, our findings do not yet support the widespread use of plasma pTau217 screening in CU older adults, as clinical and observational studies are needed to fully understand the prognostic utility of pTau217 alone, and how it may be combined with further confirmatory testing (e.g., amyloid PET) to best identify individuals at high-risk of progression. Fourth, our study participants are comprised largely of self-identified white older adults with a relatively high level of education, characteristics which might limit the generalizability of the results presented. Notably, previous large-scale studies reported that race or education did not impact the relationship between biomarkers 26 or pathologies 27 , but our study participant characteristics might have impacted the %pTau217 – cognition association. Despite these limitations, our study leverages the unique strengths of the HABS cohort and thoroughly delineates the long-term implications and temporal dynamics of changes in pTau217 in CU older adults with and without established preclinical AD. Future studies combining multiple plasma biomarkers of Aβ and tau might expand our understanding of AD pathophysiology and further inform clinical screening strategies and AD prevention trial design. Methods Participants This study included 317 HABS 20 participants with baseline plasma %pTau217, PiB PET data and APOE ε2/ε3/ε4 genotype. HABS participants were recruited from the community (volunteers, convenience sample). At baseline, the HABS participants were CU and between 50 and 90 years old. All participants had a global Clinical Dementia Rating (CDR) 28 of 0, an education-adjusted Mini-Mental State Examination (MMSE) 29 score of 27 or greater, and education-adjusted Logical Memory IIa Delayed Recall performance within the normal range at baseline 30 . Individuals who had a modified Hachinski ischemic score greater than 4 or a history of stroke with persistent neurological deficits 20 were excluded from the study. The participants were assessed with longitudinal neuroimaging (in Years 1 [baseline], 4, 6, 9, and 12) and annual cognitive/clinical assessments. Data was collected from April 2010 through August 2024, and quantitative data were used in this study. The protocols and procedures were approved by the Mass General Brigham Institutional Review Board (IRB), and all participants signed a written informed consent form before undergoing any study procedures. The HABS participants receive remuneration for their participation in visits and procedures. Plasma phospho-tau217 measures Fasting plasma samples from Year 1 ( n = 318), Year 4 ( n = 210), Year 6 ( n = 127), Year 9 ( n = 70), and Year 12 ( n = 2) were analyzed. Plasma phospho-tau217 (tau phosphorylated at the 217 th residue [threonine]; pTau217) and non-phosphorylated tau-217 (npTau217) were quantified by C 2 N through immunoprecipitation (IP) followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) 21 . Following clinical reporting practices at C 2 N 21 , pTau217 measures below the limit of detection (LOD, 1.3 pg/mL) were assigned half of the LOD value (0.65 pg/mL). In subsequent analyses, we used the pTau217/npTau217 percent ratio (%pTau217). Of note, among different %pTau217 cut points to detect AD pathology 5 , 16 , 21 , we chose to test the 4.2% cut point in our dataset because it is a manufacturer-established cut point for clinical use in individuals with cognitive impairment, and was derived using the same version of the assay (C 2 N v2 assay) and gold standard (Aβ PET) as our study. PET Imaging 11 C-Pittsburgh Compound-B (PiB) positron emission tomography (PET) was performed at Massachusetts General Hospital (MGH) using the ECAT EXACT HR+ scanner (Siemens) to measure brain Aβ burden. PiB PET scans were done in Years 1, 4, 6, 9, and 12. We used the distribution volume ratio (DVR) of a cortical composite region of interest (ROI), combining frontal, lateral temporal, parietal, and retrosplenial regions (FLR), using cerebellar gray matter as a reference. The DVR values were transformed to the Centiloid (CL) scale using the following formula 31 : [PiB CL] = 143.06× [PiB FLR DVR]−145.60. The subgroup with elevated Aβ (“A + ”) was defined with Gaussian mixture modeling with the threshold of 24.1 CL 32 . FTP PET was introduced into HABS mid-study, and participants had their first FTP PET at 2.6 ± 1.7 years after their baseline visits. Subsequent scans were performed in Years 4, 6, 9, and 12, and n = 183 participants had longitudinal FTP PET data. Standardized uptake value ratios (SUVR) with cerebellar gray reference and partial volume correction (PVC; geometric transfer matrix method 33 ) were computed for each ROI (based on FreeSurfer 34 v6.0). Two ROIs were selected for main analyses: the entorhinal cortex (EC) to represent early mesial temporal changes and the inferior temporal cortex (ITC) to capture early neocortical tau deposition. Longitudinal FTP SUVR in other ROIs were explored for their association with baseline %pTau217. Cognitive Assessment Participants completed the Preclinical Alzheimer’s Cognitive Composite (PACC5) 35 , which was derived by combining the z scores (baseline reference) of MMSE 29 , Weschler Adult Intelligence Scale-Revised Digit Symbol Coding 36 , Wechsler Memory Scale-Revised Logical Memory delayed recall 30 , Free and Cued Selective Reminding Test (free recall plus total recall) 37 , and Category Fluency Test 38 done each year. Statistical Analyses The HABS data were downloaded on August 29, 2024, and R version 4.3.2 was used for all analyses. All p-values shown are two-sided and are not adjusted for multiple comparisons. Pearson’s correlation and Wilcoxon rank sum test were used for cross-sectional associations of %pTau217 with Aβ (PiB) CL and Aβ status (A + /A−), respectively. The associations of baseline %pTau217 and longitudinal Aβ, tau, and cognition were assessed using the following linear mixed effect models (R “nlme” package): Aβ: [PiB CL] ~ [%pTau217]*[time] + [baseline age, sex, APOE ε4 status]*[time] + [main terms] Tau: [FTP SUVR] ~ [%pTau217]*[time] + [baseline age, sex, APOE ε4 status]*[time] + [main terms] Cognition: [PACC5] ~ [%pTau217]*[time] + [baseline age, sex, APOE ε4 status, years of education]*[time] + [main terms] We used random intercepts for Aβ and tau models (given that most participants had ≤ 3 longitudinal data points) and used random intercepts and slopes for the PACC models. The baseline %pTau217 vs. regional longitudinal FTP analyses were done using FTP SUVR from each cortical and subcortical region of interest using the above linear mixed effect model (total 41 regions; multiplicity corrected with false discovery rate [FDR]). We did not assume laterality and used the combined bilateral ROIs in these analyses. The results were visualized using R “ggseg” package. In the baseline A- subgroup, we examined the likelihood of progression to A+ using a log-rank test across baseline %pTau217 tertiles among A- participants, and the Kaplan-Meier curve was presented (R “survival” package). We excluded 3 participants who converted to A+ but later reverted to A-. We performed a log-rank test to assess survival probability differences across the tertiles. To evaluate the continuous dose-response relationship between baseline %pTau217 and the likelihood of A+ progression while also considering potential confounders, we used the following Cox regression models (R “survival” package): [Progression to A+ (event)] ~ [baseline %pTau217] + [baseline age, sex, APOE ε4 status] [Progression to A+ (event)] ~ [baseline %pTau217] + [baseline Aβ CL] + [baseline age, sex, APOE ε4 status] [Progression to A+ (event)] ~ [time-varying %pTau217] + [baseline %pTau217] + [baseline Aβ CL] + [baseline age, sex, APOE ε4 status] (for the A-/intermediate-high %pTau217 subgroup, as defined below) To examine whether longitudinal changes in plasma %pTau217 and Aβ PET CL are correlated, we leveraged the paired measurements of %pTau217 and Aβ CL from the same study years. First, we assigned subgroups based on their baseline measures (low/intermediate-high %pTau217 by A−/A + ; 4 subgroups) and assessed how the participants moved across the subgroups by the time of their final measures. Here, we used a cohort-specific %pTau217 cut point of 2.6% that best differentiates A+ progressors from non-progressors in our dataset (maximum Youden’s index) to define low versus intermediate-high %pTau217 and illustrate coordinated temporal changes of %pTau217 and Aβ CL. For the path analysis, we first used separate linear mixed-effect models to derive random slopes of PiB, ITC-FTP, and PACC5, adjusting for baseline age, sex, APOE ε4 status, years of education (only for PACC5), and time interaction terms. Here, we focused on the ITC-tau rather than the EC-tau, as ITC-tau has been shown to have a more proximate association with longitudinal cognition 24 . Baseline %pTau217 and these random slopes were modeled together in a just-identified (saturated) model using non-parametric bootstrapping ( n = 10,000 simulations), and significant ( p < 0.05) pathways were highlighted (R “lavaan” package). Since the model we specified is a just-identified (saturated) model, and the overall model fit indices are not informative, our focus was on the path coefficients. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information (128.4KB, pdf) Reporting Summary (93.8KB, pdf) Transparent Peer Review file (532.6KB, pdf) Source data Source Data (68.7KB, xlsx) Acknowledgements We thank the study participants of the Harvard Aging Brain Study. We also thank Dr. Tim West, who has provided helpful comments related to the various cut points of C 2 N %pTau217 assays. This work was supported by grants K23 AG062750 (Dr. Yang), K23 AG084868 (Dr. Yau), P01 AG036694 (Harvard Aging Brain Study; Drs. Johnson and Sperling), and R01 AG071865 (Dr. Chhatwal) from the National Institute on Aging, as well as the Shelby Cullom Davis Charitable Fund’s philanthropic gift (Drs. Selkoe and Sperling). This research was conducted in part at the Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, utilizing resources provided by the Center for Functional Neuroimaging Technologies (CFNT), P41EB015896, a P41 Biotechnology Resource Grant supported by the National Institute of Biomedical Imaging and Bioengineering (NIBIB), NIH. This work also involved the use of instrumentation supported by the NIH Shared Instrumentation Grant Program and/or High-End Instrumentation Grant Program; specifically, grant numbers S10RR021110, S10RR023401, and S10RR023043. Funding sources were not involved in study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication. Author contributions H.-S.Y., R.A.S., and J.P.C. conceptualized and designed the study. H.S.Y. performed statistical analysis, in consultation with B.C.H. H.-S.Y. drafted the manuscript and generated figures. All authors (H.-S.Y., J.A.U.A, W.W.Y., B.C.H., A.M.R., C.M., D.K., M.J.P., J.-P.B., A.P.S., M.E.F., H.I.L.J., R.F.B., K.V.P., G.A.M., R.E.A., D.M.R., L.L., D.J.S., P.B.V., J.B.B., K.A.J., R.A.S., and J.P.C.) contributed to data generation, interpreted the results, revised the manuscript for important intellectual content, and approved the manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Data availability All published Harvard Aging Brain Study (HABS) data have been deposited in the Synapse data sharing platform under accession code syn53910452 ( www.synapse.org/habs ). The HABS data, including all data used in this study, are individual-level human data available under restricted access. Access can be obtained by submitting a data request through the Synapse platform and signing a data use agreement. Source data, excluding the individual-level human data, are provided as a Source Data file. Source data are provided with this paper. Code availability R codes used in this study are available at: https://github.com/YangLabADRD/pTau217 . Competing interests Philip B. Verghese and Joel B. Braunstein are full-time employees of C2N Diagnostics LLC. Other authors declare no directly relevant conflict of interest. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-71269-3. References 1. Alzheimer's Association. 2024 Alzheimer’s disease facts and figures. Alzheimers Dement 20 , 3708–3821 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. van Dyck, C. H. et al. Lecanemab in Early Alzheimer’s Disease. N. Engl. J. Med 388 , 9–21 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 3. Sims, J. R. et al. Donanemab in Early Symptomatic Alzheimer Disease: The TRAILBLAZER-ALZ 2 Randomized Clinical Trial. JAMA 330 , 512–527 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Janelidze, S., Barthelemy, N. R., He, Y., Bateman, R. J. & Hansson, O. Mitigating the Associations of Kidney Dysfunction With Blood Biomarkers of Alzheimer Disease by Using Phosphorylated Tau to Total Tau Ratios. JAMA Neurol. 80 , 516–522 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Palmqvist, S. et al. Blood Biomarkers to Detect Alzheimer Disease in Primary Care and Secondary Care. JAMA 332 , 1245–1257 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Palmqvist, S. et al. Discriminative Accuracy of Plasma Phospho-tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders. JAMA 324 , 772–781 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Yu, L. et al. Plasma p-tau181 and p-tau217 in discriminating PART, AD and other key neuropathologies in older adults. Acta Neuropathol. 146 , 1–11 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Barthelemy, N. R. et al. Highly accurate blood test for Alzheimer’s disease is similar or superior to clinical cerebrospinal fluid tests. Nat. Med 30 , 1085–1095 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Jack, C. R. Jr. et al. Revised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup. Alzheimers Dement 10.1002/alz.13859 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Sperling, R., Mormino, E. & Johnson, K. The evolution of preclinical Alzheimer’s disease: implications for prevention trials. Neuron 84 , 608–622 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Sperling, R. A. et al. Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7 , 280–292 (2011). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Sperling, R. A. et al. Trial of Solanezumab in Preclinical Alzheimer’s Disease. N. Engl. J. Med 389 , 1096–1107 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Rafii, M. S. et al. The AHEAD 3-45 Study: Design of a prevention trial for Alzheimer’s disease. Alzheimers Dement 19 , 1227–1233 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Ashton, N. J. et al. Differential roles of Abeta42/40, p-tau231 and p-tau217 for Alzheimer’s trial selection and disease monitoring. Nat. Med 28 , 2555–2562 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Mattsson-Carlgren, N. et al. Longitudinal plasma p-tau217 is increased in early stages of Alzheimer’s disease. Brain 143 , 3234–3241 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Rissman, R. A. et al. Plasma Abeta42/Abeta40 and phospho-tau217 concentration ratios increase the accuracy of amyloid PET classification in preclinical Alzheimer’s disease. Alzheimers Dement 20 , 1214–1224 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Janelidze, S. et al. Plasma Phosphorylated Tau 217 and Abeta42/40 to Predict Early Brain Abeta Accumulation in People Without Cognitive Impairment. JAMA Neurol. 81 , 947–957 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Cogswell, P. M. et al. Plasma Alzheimer’s disease biomarker relationships with incident abnormal amyloid PET. Alzheimers Dement 21 , e14629 (2025). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Jonaitis, E. M. et al. Plasma phosphorylated tau 217 in preclinical Alzheimer’s disease. Brain Commun. 5 , fcad057 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Dagley, A. et al. Harvard Aging Brain Study: Dataset and accessibility. Neuroimage 144 , 255–258 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Meyer, M. R. et al. Clinical validation of the PrecivityAD2 blood test: A mass spectrometry-based test with algorithm combining %p-tau217 and Abeta42/40 ratio to identify presence of brain amyloid. Alzheimers Dement 20 , 3179–3192 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Johnson, K. A. et al. Tau positron emission tomographic imaging in aging and early Alzheimer disease. Ann. Neurol. 79 , 110–119 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Smith, V. D. et al. Overlapping but distinct TDP-43 and tau pathologic patterns in aged hippocampi. Brain Pathol. 28 , 264–273 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Hanseeuw, B. J. et al. Association of Amyloid and Tau With Cognition in Preclinical Alzheimer Disease: A Longitudinal Study. JAMA Neurol. 76 , 915–924 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Horie, K. et al. CSF MTBR-tau243 is a specific biomarker of tau tangle pathology in Alzheimer’s disease. Nat. Med 29 , 1954–1963 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Molina-Henry, D. et al. Relationship between Plasma P-Tau217 and Amyloid PET in Racial and Ethnic Underrepresented Groups (RE-URG) Compared with Non RE-URG in LEARN and A4. J. Prev. Alzheimers Dis. 11 , 831–837 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Bennett, D. A. et al. Religious Orders Study and Rush Memory and Aging Project. J. Alzheimers Dis. 64 , S161–S189 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Morris, J. C. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 43 , 2412–2414 (1993). [ DOI ] [ PubMed ] [ Google Scholar ] 29. Folstein, M. F., Folstein, S. E. & McHugh, P. R. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res 12 , 189–198 (1975). [ DOI ] [ PubMed ] [ Google Scholar ] 30. Wechsler, D. WMS-R: Wechsler Memory Scale–Revised: manual . (Psychological Corp.: Harcourt Brace Jovanovich, 1987). 31. Farrell, M. E. et al. Defining the Lowest Threshold for Amyloid-PET to Predict Future Cognitive Decline and Amyloid Accumulation. Neurology 96 , e619–e631 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Mormino, E. C. et al. Amyloid and APOE epsilon4 interact to influence short-term decline in preclinical Alzheimer disease. Neurology 82 , 1760–1767 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Rousset, O. G., Ma, Y. & Evans, A. C. Correction for partial volume effects in PET: principle and validation. J. Nucl. Med 39 , 904–911 (1998). [ PubMed ] [ Google Scholar ] 34. Fischl, B. et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron 33 , 341–355 (2002). [ DOI ] [ PubMed ] [ Google Scholar ] 35. Papp, K. V., Rentz, D. M., Orlovsky, I., Sperling, R. A. & Mormino, E. C. Optimizing the preclinical Alzheimer’s cognitive composite with semantic processing: The PACC5. Alzheimers Dement (N.Y.) 3 , 668–677 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Wechsler, D. Wechsler adult intelligence scale: WAIS-R manual . Rev. edn, (Harcourt Brace Jovanovich for The Psychological Corporation, 1981). 37. Grober, E., Lipton, R. B., Hall, C. & Crystal, H. Memory impairment on free and cued selective reminding predicts dementia. Neurology 54 , 827–832 (2000). [ DOI ] [ PubMed ] [ Google Scholar ] 38. Monsch, A. U. et al. Comparisons of verbal fluency tasks in the detection of dementia of the Alzheimer type. Arch. Neurol. 49 , 1253–1258 (1992). [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Information (128.4KB, pdf) Reporting Summary (93.8KB, pdf) Transparent Peer Review file (532.6KB, pdf) Source Data (68.7KB, xlsx) Data Availability Statement All published Harvard Aging Brain Study (HABS) data have been deposited in the Synapse data sharing platform under accession code syn53910452 ( www.synapse.org/habs ). The HABS data, including all data used in this study, are individual-level human data available under restricted access. Access can be obtained by submitting a data request through the Synapse platform and signing a data use agreement. Source data, excluding the individual-level human data, are provided as a Source Data file. Source data are provided with this paper. R codes used in this study are available at: https://github.com/YangLabADRD/pTau217 . Articles from Nature Communications are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (7.1 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top