Mastermind of the Alzheimer’s blood‐based biomarkers: development of cutoffs and a visualization tool for use in clinical practice

Abstract

AbstractBackgroundPhosphorylated (P‐)tau, amyloid‐beta (Abeta)1‐42/1‐40, glial fibrillary acidic protein (GFAP) and neurofilament light (NfL) are the current key Alzheimer’s disease (AD) blood‐based biomarkers. Interpreting results of the four biomarkers simultaneously in a clinical context is challenging. A decision algorithm could aid in the interpretation, tailored to clinically relevant questions: 1) identify positive amyloid status among preclinical and prodromal AD stages, 2) discriminate AD‐dementia from frontotemporal dementia (FTD), 3) discriminate AD‐dementia from dementia with Lewy bodies (DLB). We aimed to develop such an interpretation tool.MethodWe included 1199 participants from the Amsterdam Dementia Cohort (table 1) with subjective cognitive decline (SCD), mild cognitive impairment (MCI), AD‐dementia, FTD (mixed pathologies) or DLB. Plasma P‐tau181, Abeta1‐42/1‐40, GFAP and NfL were measured with Simoa. NfL levels were adjusted for age (method: https://doi.org/10.1002/acn3.51676). We used LASSO regression with bootstrap (1000 iterations) to identify which plasma markers robustly contribute to the prediction of the outcomes of our clinically relevant questions. When a marker was selected in >95% of the iterations in at least one of the clinical scenarios, we qualified it as a contributing marker. Subsequently, we constructed UpSet plots applying Youden’s cutoffs as interpretation tool for the plasma results.ResultP‐tau181, GFAP and NfL were selected as contributing biomarkers. Abeta1‐42/1‐40 was not selected. The panel of P‐tau181, GFAP and NfL had an AUC = 83% (95%CI: 76‐90%) to identify amyloid positivity in SCD+MCI, AUC = 87% (95%CI: 79‐94%) to differentiate AD‐dementia from FTD, and AUC = 73% (95%CI: 62‐84%) to differentiate AD‐dementia from DLB. The UpSet plots show the simultaneous interpretation of P‐tau181, GFAP and NfL results for each clinically relevant question (figure 1): e.g., when all three biomarkers are abnormal in participants with SCD+MCI, there is 83% agreement with an abnormal PET or CSF amyloid status, and when all three plasma markers are normal, there is 88% agreement with a normal amyloid status.ConclusionThe combination of P‐tau181, GFAP and NfL in plasma could support decision making in different clinical contexts. UpSet plots enable easy interpretation of a combination of the markers, and could serve as a tool to support both detection and differential diagnosis of Alzheimer’s disease.