Blood-based biomarker prescreening with different testing combinations and cutoffs: A simulation study examining...

Blood-based biomarkers (BBBMs), including plasma amyloid beta (Aβ) or phosphorylated tau (p-tau), combined with apolipoprotein E (APOE) testing, are anticipated to serve as prescreening tools before amyloid positron emission tomography (PET) for recruiting participants for Alzheimer's disease (AD) prevention studies. The predictive efficacy and cost-effectiveness of prescreening may vary with different testing combinations, sequences, and cutoff levels.

We conducted a simulation study utilizing data from our ongoing Japanese Trial-Ready Cohort (J-TRC) onsite study (n = 202) recruited online. We included cognitively unimpaired individuals who had undergone amyloid PET, APOE genotyping, and evaluation of BBBMs (i.e., plasma Aβ42/Aβ40 ratio, plasma p-tau217, and plasma p-tau217/Aβ42 ratio). We examined 14 different prescreening models incorporating APOE genotype and/or BBBMs with varied combinations and cutoff levels. Models were evaluated for predictive performance (sensitivity, specificity, and positive predictive value [PPV]) and cost-effectiveness (cost per identified amyloid-positive case) across varied testing costs and the prevalence of amyloid positivity.

Applying BBBM prescreening significantly decreased sensitivity and increased specificity and PPV compared to the no-prescreening scenario. Although no single model was superior in all performance metrics, a trade-off between sensitivity and specificity was observed. Generalized linear models (GLMs) simultaneously incorporating plasma Aβ42/Aβ40 ratio and p-tau217 showed a balanced efficacy (the best level of improvement in number needed to screen (NNS) but modest worsening in sensitivity) and the best level of cost-effectiveness compared to other models, although there were substantial overlaps in their 95% confidence intervals (CIs). The minimum-required PET/BBBM cost ratio to achieve improved cost-effectiveness by employing the prescreening process was negatively associated with the background prevalence of amyloid positivity.

The choice of prescreening strategy in AD prevention studies/trials should be tailored to specific trial requirements, considering the relative importance of sensitivity versus cost-effectiveness, local testing cost environments, and background population characteristics.

We investigated different models of blood-based biomarker (BBBM) prescreening.Data from Japanese Trial-Ready Cohort Study were used for simulation.BBBM reduced the number needed to screen with positron emission tomography (PET) but decreased the sensitivity.The cost-effectiveness improved with a higher PET/BBBM cost ratio.No single model was optimal for all scenarios.