Integrating plasma p-tau217 and digital cognitive assessments for early detection in Alzheimer's disease

Abstract

Introduction: Plasma phosphorylated tau (p-tau)217 is an early Alzheimer's disease (AD) biomarker, but the timing of pathological changes and cognitive decline varies substantially. Digital cognitive assessments can detect subtle cognitive changes, suggesting they may complement p-tau217 for early detection. Here, we evaluate whether combining these tools improves the detection of individuals at risk for future decline.

Methods: We analyzed 954 amyloid-positive cognitively unimpaired individuals who completed a digital cognitive assessment and a blood test for p-tau217, assessing their ability to predict future decline on the Preclinical Alzheimer Cognitive Composite (PACC) and Mini-Mental State Examination (MMSE).

Results: Combining performance on a digital cognitive assessment with p-tau217 improved identification of individuals who declined on the PACC and MMSE in the next 5 years. The predictive value was stronger in apolipoprotein E ε4 noncarriers but did not differ by sex.

Discussion: This approach offers a sensitive method for identifying individuals at high risk for AD-related cognitive decline.

Highlights: Combining plasma phosphorylated tau 217 with baseline digital cognitive assessment improved the prediction of cognitive decline on gold-standard neuropsychological tests over the next 5 years, achieving greater accuracy than either measure alone. This combination also predicted a decline in a global cognitive screening test. Pairing a blood test with a digital cognitive assessment offers a scalable and feasible approach for Alzheimer's disease screening.

Keywords: Alzheimer's disease; digital cognitive assessments; early detection; low‐burden measures; memory; phosphorylated tau 217.

FIGURE 1

Integrating plasma p‐tau217 and the C3 enhances prediction of future decline on the Preclinical Alzheimer Cognitive Composite (PACC). Comparison of models using demographics alone (green), demographics plus p‐tau217 (blue), C3 composite (red), or both p‐tau217 and C3 (purple) for predicting low performers and future cognitive decline. Top, ROC curves illustrating the predictive value of each model. Middle, Violin plots depicting model sensitivity at a false positive rate of 20%. Bottom, Odds ratios for future decline based on low memory (red), high p‐tau217 (blue), or both (purple). A, At baseline, the C3 composite was the strongest predictor of high and low performers, showing the highest sensitivity and odds ratios. Over 2 (B) and 5 (C) years, the combined model (p‐tau217 + C3) was the most predictive of future decline, achieving the highest AUCs and sensitivity. Critically, individuals with both high baseline p‐tau217 and low baseline C3 performance had significantly higher odds of cognitive decline compared to those with only one of these risk factors. AUC, area under the curve; p‐tau, phosphorylated tau; C3, Computerized Cognitive Composite; ROC, receiver operating characteristic

FIGURE 2

Predictive capacity of the combined model (demographics, plasma p‐tau217, and C3) differs by APOE ε4 carrier status but not sex. Top, ROC curves illustrating the predictive value of the combined model in males (blue) and females (red). Bottom, ROC curves illustrate the predictive value of the combined model in APOE ε4 non‐carriers (blue) and APOE ε4 carriers (red). A, At baseline, the model performed similarly in males and females, with a slight advantage in females (top). Additionally, no significant difference was observed between APOE ε4 carriers and non‐carriers in baseline performance. B, C, When predicting future cognitive decline, the models performed equally well in males and females at both 2 (B) and 5 (C) years. However, the predictive capacity was stronger in APOE ε4 non‐carriers compared to carriers at both time points. APOE, apolipoprotein E; AUC, area under the curve; C3, Computerized Cognitive Composite; p‐tau, phosphorylated tau; ROC, receiver operating characteristic

FIGURE 3

Integrating plasma p‐tau217 and C3 enhances the prediction of individuals who progress to clinical impairment on the MMSE. Comparison of models using demographics alone (green), demographics plus p‐tau217 (blue), C3 composite (red), or both p‐tau217, and C3 (purple) for predicting low performers and future cognitive decline. Top, ROC curves illustrate the predictive value of each model. Bottom, Odds ratios for future impairment on the MMSE based on low memory (red), high p‐tau217 (blue), or both (purple). The combined model incorporating both measures was the strongest predictor of individuals who would progress to clinical impairment on the MMSE over the next (A) 3, (B) 4, or (C) 5 years, achieving the highest AUCs (top) and demonstrating that individuals with both high p‐tau217 and low C3 performance at baseline had the highest odds ratio for progressing to an impaired MMSE (bottom). AUC, area under the curve; C3, Computerized Cognitive Composite; MMSE, Mini‐Mental State Examination; p‐tau, phosphorylated tau; ROC, receiver operating characteristic

FIGURE 4

Longitudinal trajectories over 240 weeks on the Preclinical Alzheimer Cognitive Composite (PACC) and Mini‐Mental State Examination (MMSE). A, Individuals in the p‐tau217*Mem* group exhibited the most rapid decline in the PACC compared to the other three groups. p‐tau217*Mem_o showed the second fastest decline, followed by p‐tau217_oMem*, which declined more rapidly than p‐tau217_oMem_o. B, On the MMSE, p‐tau217*Mem* individuals again exhibited the steepest decline. p‐tau217*Mem_o and p‐tau217_oMem* declined at similar rates, with both declining faster than p‐tau217_oMem_o. p‐tau, phosphorylated tau