Individualized diagnostic and prognostic blood‐based models for use in clinical practice and trials

Abstract

AbstractBackgroundBlood‐based biomarker’s for Alzheimer’s disease are likely to revolutionize the diagnostic work‐up in both specialized clinics and in primary care. Further, they might be used to identify suitable participants for preclinical and prodromal AD trials, i.e., individuals with AD pathology who will exhibit cognitive decline in the coming 2‐4 years.MethodWe evaluated the diagnostic and prognostic performance of plasma P‐tau181, P‐tau217, Ab42/40, and NfL in the prospective and longitudinal BioFINDER I and II studies. We compared the accuracy of plasma biomarkers with CSF AD biomarkers, amyloid‐PET and tau‐PET. Diagnostic and prognostic algorithms were typically validated in an independent cohort.ResultPlasma P‐tau alone could with very high accuracy differentiate AD dementia from other neurodegenerative diseases, and a plasma P‐tau217 assay was shown to be non‐inferior to tau‐PET and CSF biomarkers in this setting. A combination of plasma P‐tau and NfL could be used for individualized prediction of cognitive decline and conversion to AD dementia within 2‐ 6 years in patients with mild cognitive impairment (MCI), and this combination performed equal to CSF AD biomarkers. Further, in a population including both patients with subjective cognitive decline (SCD) and MCI, as often seen in primary care settings, an optimal prognostic algorithm for prediction of AD dementia included plasma P‐tau and brief cognitive tests of memory and executive function. We have developed an online tool for individualized prediction of AD dementia in cases with SCD or MCI (https://brainapps.shinyapps.io/PredictionADdementia/), which was shown to outperform the clinical prediction by memory clinic physicians. In cognitively unimpaired (CU) individuals we showed that plasma Ab42/Ab40 and P‐tau together could detect cerebral Ab pathology, but a combination of plasma Ab42/Ab40, P‐tau and NfL was needed for accurate prediction of subsequent cognitive decline and to enrich pre‐clinical trials with suitable research participants. Finally, results will be presented on the use of fully automated assays for quantification of relevant plasma AD biomarkers.ConclusionDifferent combinations of easily accessible tests are optimal during different disease stages of AD, and they are likely to be useful in both clinical practice and trials after validation in multiple large‐scale, relevant and diverse cohorts.