Longitudinal measurements of a blood-based tau biomarker were associated with the timing of symptom onset in cognitively unimpaired individuals at risk for Alzheimer disease, according to a study published in Nature Medicine.
In the study, researchers used repeated measurements of plasma phosphorylated tau at position 217 (p-tau217) to construct individualized “biomarker clock” models that estimate the age at which individuals cross a biomarker positivity threshold and how long that point precedes the onset of clinical symptoms. The analysis drew on data from two independent cohorts: the Knight Alzheimer Disease Research Center and the Alzheimer’s Disease Neuroimaging Initiative (ADNI).
Using longitudinal plasma %p-tau217 values – the ratio of phosphorylated to non-phosphorylated tau217 – the team estimated the age at which individuals became biomarker positive, defined relative to an amyloid PET Centiloid threshold. That estimated age was then examined in relation to the timing of symptomatic Alzheimer disease onset among participants who were cognitively unimpaired at baseline.
Across cohorts and modeling approaches, the estimated age at plasma %p-tau217 positivity was moderately correlated with the age at symptom onset. Models predicting symptom onset showed median absolute errors of approximately 3 to 4 years. The interval between biomarker positivity and clinical symptoms also varied by age: individuals who became p-tau217 positive later in life progressed to symptoms more quickly than those who crossed the threshold at younger ages.
The models incorporated repeated within-person measurements rather than single time-point values, allowing estimation of individualized trajectories of biomarker change over time.
The authors emphasize that the approach is not intended for individual-level prognostication. Prediction error remained too large to support clinical decision-making, and biomarker testing in cognitively unimpaired individuals is not currently recommended outside research or trial settings. Instead, the models may be most useful for clinical trial design, where estimating time to symptom onset could help enrich enrollment with participants more likely to progress within a defined study period.
To assess generalizability, the team applied similar clock models to several commercially available plasma p-tau217 assays, including a p-tau217/Aβ42 ratio test. While model performance varied, the overall pattern linking the timing of biomarker positivity to subsequent symptom onset was preserved.
The authors suggest that translating plasma p-tau217 trajectories into time-based estimates of disease progression may be most useful for clinical trial design, particularly when identifying individuals likely to develop symptoms within a specific timeframe.
Source: Nature Medicine
