Use of plasma p‐tau217 as a pre‐screening method for detecting amyloid‐PET positivity in cognitively unimpaired participants: A multicenter study

Abstract Background Recent results from clinical trials in Alzheimer’s disease (AD) emphasize the importance of treating early‐stage disease. However, recruitment of preclinical AD participants is difficult due to the lack of symptoms, and the costs and/or invasiveness of established CSF and PET tests. We aimed to investigate whether plasma p‐tau217 could be used to pre‐screen cognitively unimpaired (CU) potential participants for amyloid‐β (Aβ) pathology to improve the efficiency of clinical trial recruitment. Method We included 1,471 CU participants from eight cohorts (Table 1) with available plasma p‐tau217, Aβ CSF biomarkers and Aβ‐PET status (served as standard‐of‐truth). Plasma p‐tau217 concentrations were z‐scored based on Aβ‐negative participants and harmonized across cohorts using neuroCombat. Cut‐offs for plasma p‐tau217 were derived in the BioFINDER‐1 cohort (n=104) based on different specificity levels (90%, 95% and 97.5%) to maximize positive predictive values (PPV). These cut‐offs were used in the other cohorts to assess the accuracy of plasma p‐tau217 for detecting Aβ‐PET positivity. Next, within plasma positive participants only, we evaluated the value of dichotomized Aβ CSF (based on established clinical thresholds) on assessing Aβ‐PET positivity. All models included age and APOE‐ε4 carriership as covariates. Result 334 (24.4%) of participants were Aβ‐PET positive. Using the a priori defined cutoffs, plasma p‐tau217 categorization resulted in high PPVs (72.9%‐81.2%), negative predictive values (NPVs, 82.5%‐86.2%), and accuracy (82.4%‐83.8%), with an overall rate of positivity between 10.9%‐18.1% (Figure 1). When applying CSF biomarkers to the plasma positive participants in a second step, the PPVs increased up to 90.8%‐95.3%, with NPVs ranging between 82.8%‐86.7% and accuracies between 84.0%‐87.3%, with a slight decrease in the proportion of overall positive cases (9.3%‐14.3% from the original sample) given that the CSF positivity in the plasma positive participants ranged between 79.4%‐85.2% (Figure 2). Conclusion Plasma p‐tau217 can identify Aβ‐PET positive CU individuals with PPVs reaching 81%, which can be further improved to PPVs of up to 95% with a subsequent CSF measurement. Plasma p‐tau217 could be used, either as stand‐alone biomarker, or as an initial step before CSF biomarkers (reducing their need by ∼80‐90%), for pre‐screening in clinical trials of preclinical AD depending on the certainty needed for Aβ‐PET positivity.