Evaluation of plasma p-tau217 for detecting amyloid pathology in a heterogeneous community-based cohort

Abstract

Introduction: Studies suggest excellent performance of plasma phosphorylated tau 217 (p-tau217) for detecting amyloid pathology, though studies in more representative populations are needed to validate previously determined cutpoints.

Methods: Plasma p-tau217 utility for detecting amyloid pathology (Aβ) via amyloid positron emission tomography (PET) was assessed in a heterogeneous, community-based cohort in the Wake Forest Alzheimer's Disease Research Center (WFADRC). Participants with baseline plasma data (n = 598) were 21% Black; 313 cognitive unimpaired (CU), 214 mild cognitive impairment (MCI), and 64 dementia (DEM); 49% prediabetic, 44% hypertensive, 29% overweight/obese; and 64% had mild-to-moderate kidney disease. Gaussian-mixture models, logistic regression, and receiver operating curve analyses were performed.

Results: Plasma p-tau217 was associated with elevated Aβ deposition and accurately classified Aβ-positive participants (PET [n = 307]: area under the curve [AUC] = 94%-97%, cutpoint ≥ 0.338 pg/mL).

Discussion: Plasma p-tau217 is an accurate indicator of amyloid pathology in a heterogeneous cohort and is superior to other plasma biomarkers assessed.

Highlights: The Wake Forest Alzheimer's Disease Research Center (WFADRC) is a heterogeneous cohort. p-tau217 levels were lower, on average, in cognitively unimpaired participants, females, and Black participants. Plasma p-tau217 classified amyloid positron emission tomography (PET)-positive individuals with high precision and performed better than p-tau181. Cutpoints and reference ranges of plasma p-tau217 were lower compared to recently published thresholds. Combining cutpoint approaches, a four-tier system captured cohort heterogeneity.

Keywords: PET; biomarkers; comorbidities; dementia; plasma; risk.

FIGURE 1

Associations between amyloid PET deposition (Centiloids) and plasma p‐tau181 (A) and p‐tau217(B). Scatterplots highlight a stronger correlation between p‐tau217 and amyloid PET deposition overall and across cognitive status groups compared to p‐tau181 and amyloid PET. Dotted lines represent optimal (binary) and two‐point detection cutpoints for each p‐tau variable observed when classifying amyloid PET positivity. Data points are color‐coded by diagnosis and of different shapes based on eGFR status (Stage 1 [normal high] eGFR ≥ 90; Stage 2 [mild] > 60 eGFR < 90; Stage 3 [mild‐moderate] > 45 eGFR < 60; Stage 4 [moderate severe] eGFR < 45). CL, Centiloids; CU, cognitively unimpaired; DEM, dementia; eGFRg, estimated glomerular filtration rate; MCI, mild cognitive impairment; PET, positron emission tomography; p‐tau, phosphorylated tau; SENS, sensitivity; SPEC, specificity

FIGURE 2

Performance of p‐tau181 and p‐tau217 for classifying amyloid PET positivity defined as ≥ 24 Centiloids. Cutpoints derived using receiver operating curve analyses and Gaussian‐mixture models for both plasma (A) p‐tau181 and (B) p‐tau217. Optimal (e.g., binary; Youden index = 0.338 pg/mL) and two‐point ROC detection thresholds (Max‐SENS = 0.253 pg/mL; MAX‐SPEC = 0.472 pg/mL) were combined, resulting in a 4‐tier system: (1) Negative [< 0.253; N∼39%]; (2) Intermediate‐low [0.253–0.338; N∼20%]; (3) Intermediate‐high [0.338–0.472; N~13%]; and (4) Positive [> 0.472; N~28%]. Aβ, amyloid b eta; CL, Centiloids; GMM, Gaussian‐mixture models; NA, missing; PET, positron emission tomography; p‐tau, phosphorylated tau; ROC, receiver operating characteristic curve; SENS, sensitivity; SPEC, specificity (e.g., plasma p‐tau217 datapoints without a corresponding amyloid PET scan)