Implementation of the Alzheimer’s blood‐based biomarker panel in clinical practice: the development of cutoffs for plasma Abeta_{1‐42/1‐40}, P‐tau181, GFAP and NfL across the clinical continuum

Abstract

BackgroundSeveral blood‐based biomarkers for Alzheimer’s disease (AD) can now be reliably measured. It is not yet clear how those markers should be interpreted in a real‐world clinical setting, where a heterogeneous group of patients presents in different syndromal stages and with different symptoms. We aim to establish a ready‐to‐implement blood‐based biomarker panel including Abeta1‐42/1‐40, P‐tau181, GFAP and NfL with cutoffs for clinical application among subjective cognitive decline (SCD), mild cognitive impairment (MCI), AD‐dementia, frontotemporal dementia (FTD) and dementia with Lewy bodies (DLB).MethodWe included n=1132 individuals of the Amsterdam Dementia Cohort (table 1) with known baseline amyloid‐beta (Aβ) status, who had SCD (n=237 Aβ‐, n=66 Aβ+), MCI (n=108 Aβ‐, n=151 Aβ+), AD‐dementia (n=297, all Aβ+), FTD (n=162 of whom n=121 Aβ) or DLB (n=111 of whom n=53 Aβ‐). Plasma markers were measured with Simoa (neurology 4‐plex E, P‐tau181 V2). We applied ROC analysis and subsequently LASSO regression to identify panels among the plasma markers.ResultROC analysis (table 2, figure 1) showed that among SCD and MCI, AUCs of the individual plasma markers to predict Aβ status were 0.738–0.799 (lowest for NfL, highest for P‐tau181). LASSO regression (table 2, figure 1) selected Abeta1‐42/1‐40, P‐tau181 and GFAP, but not NfL as the optimal panel (AUC=0.846, 95%CI: 0.812–0.880). Stratifying for SCD or MCI stage resulted in largely similar findings (not shown). When discriminating AD from FTD, AUCs of the individual plasma markers were 0.643–0.820 (lowest for Abeta1‐42/1‐40, highest for P‐tau). LASSO regression selected P‐tau181, GFAP and NfL as the optimal panel (AUC=0.889, 95%CI: 0.850–0.927). For AD versus DLB, AUCs were 0.514–0.741 (lowest for NfL, highest for P‐tau181). LASSO regression selected only P‐tau181 (AUC=0.741, 95%CI: 0.683–0.800). Plasma marker cutoffs at Youden’s indices differed depending on the diagnostic groups, with higher cutoffs for P‐tau181, GFAP and NfL in AD versus FTD or DLB analyses, compared to the SCD and MCI analyses (table 2).ConclusionWe showed that depending on the clinical question at hand, the four biomarkers add valuable information with their own cutoffs. Decision trees might lead to ready‐to‐apply cutoffs for daily clinical practice.

Investigating Compensatory Brain Activity in Older Adults with Subjective Cognitive Decline

“…Furthermore, we analyzed Aβ 42/40 ratio (mean = 0.068, SD = 0.014), pTau181 (mean = 2.34 pg/ml, SD = 1.22), glial fibrillary acidic protein (mean = 126.86 pg/ml, SD = 60.21), and neurofilament light chain (mean = 22.9 pg/ml, SD = 11.21). Based on currently recommended cut-off scores, ratios lower than 0.06 in the Aβ 42/40 ratio indicate amyloid positivity [ 57 ]. In our sample, this included eight participants.…”

“…In our sample, this included eight participants. For pTau181, a cut-off of 1.8 pg/ml has been suggested for amyloid positivity [ 57 ], which indicated 18 participants with amyloid positivity. This pTau181 cut-off score is in-line with previous research [ 58 ].…”

“…Seven participants were carriers of at least one APOE ε4 allele. Furthermore, we analyzed A␤ 42/40 ratio (mean = 0.068, SD = 0.014), pTau181 (mean = 2.34 pg/ml, SD = 1.22), glial fibrillary acidic protein (mean = 126.86 pg/ml, SD = 60.21), and neurofilament light chain (mean = 22.9 pg/ml, [57]. In our sample, this included eight participants.…”