CORRELATION OF PLASMA P-TAU217 AND CSF P-TAU181 WITH CLINICAL DIAGNOSIS OF ALZHEIMER’S DISEASE: A CROSS-SECTIONAL STUDY OF 1,766 PARTICIPANTS IN THE ALZHEIMER’S DISEASE NEUROIMAGING INITIATIVE

Introduction

Two anti-amyloid monoclonal antibodies received accelerated approval, which increased clinical and public interest in blood biomarkers [1]. On May 16, 2025, the FDA authorized an assay that measures the plasma p-tau217 to β-amyloid1-42 ratio. Despite the rapid expansion of p-tau testing in outpatient labs and growing enthusiasm among clinicians and patients, most studies benchmark p-tau against CSF or PET [2] rather than bedside cognition. Therefore, we evaluated whether plasma p-tau217 and CSF p-tau181 relate to standard cognitive and functional measures that remain central to Alzheimer’s disease diagnosis.

Methods

This study is a cross-sectional analysis using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), a multi-phase, non-randomized, non-treatment study aimed to develop and validate biomarkers. results from ADNI are shared through the University of Southern California (USC) Laboratory of Neuro Imaging (LONI)’s Image and Data Archive (IDA) (https://ida.loni.usc.edu/login.jsp). We conducted a cross-sectional, complete-case analysis of baseline data from all ADNI phases, accessed July 7, 2025. Outcomes were MMSE, CDR Global, and ADAS-Cog13. Predictors were plasma p-tau217 and CSF ptau181, analyzed as continuous variables. Linear regressions were unadjusted given the exploratory aim. To reflect real clinical triage, participants were grouped as cognitively normal or cognitively impaired, where the latter included SMC, EMCI, LMCI, MCI, and AD per ADNI adjudication. For plasma p-tau217 we retained the first visit with a measured value for each participant; a similar baseline selection was applied for CSF p-tau181. No imputation or transformation was performed. Standardized ADNI biomarker and cognitive protocols were used to minimize measurement bias.

Results

Among 4,926 ADNI participants, 116 had baseline plasma p-tau217, 1,650 had baseline CSF p-tau181, and 50 had both. In cognitively normal participants, associations between plasma p-tau217 and ADAS-Cog13 or MMSE were not significant, and CSF ptau181 likewise showed negligible relations with cognition, with all R² values near zero. In cognitively impaired participants, higher plasma p-tau217 related to worse cognition on ADAS-Cog13 (β ≈ 0.0025, R² ≈ 0.32, p < 0.001) and to lower MMSE (β ≈ −0.0067, R² ≈ 0.25, p < 0.001). CSF p-tau181 showed concordant but weaker associations, including a positive association with ADAS-Cog13 (β ≈ 0.49, R² ≈ 0.13, p < 0.001) and a negative association with MMSE (β ≈ −0.91, R² ≈ 0.06, p < 0.001). Across biomarkers there were no clinically actionable thresholds that separated cognitively normal from MCI or AD, and distributions overlapped across diagnostic strata.

Conclusions

While plasma p-tau217 and CSF p-tau181 demonstrate high diagnostic accuracy for distinguishing AD from other neurodegenerative disorders, their clinical utility is primarily limited to populations with established cognitive impairment. These biomarkers showed significant correlations with cognitive assessments only in cognitively impaired individuals, not in cognitively normal populations, and lack established threshold values for reliable AD differentiation. The findings suggest that p-tau biomarkers should serve as complementary tools to support treatment decisions and disease monitoring rather than standalone diagnostic markers, with AD diagnosis remaining clinician-led through comprehensive cognitive and functional evaluation.