Correspondence between blood‐based amyloid‐β by immuno‐precipitation mass spectrometry and PIB‐PET imaging in a population cohort

Abstract

AbstractBackgroundRecently developed immuno‐precipitation mass spectrometry (IP‐MS) assays to measure blood‐based amyloid‐β with high accuracy holds exciting potential. Adoption of amyloid‐β measurement will be especially valuable in large population studies where not previously feasible. We evaluated the correspondence between a new IP‐MS assay developed on a commercially available platform with brain amyloid PET imaging as the gold standard.MethodN=99 older participants without dementia, active in the population‐based Monongahela‐Youghiogheny Healthy Aging Team (MYHAT) study, underwent blood draws at the time of brain amyloid imaging with PiB‐PET (50‐70 minutes post‐injection, SUVR with cerebellar reference). Adapting methods of Nakamura et al. (2018), immunoprecipitation from plasma was followed by time‐of‐flight MS, using a benchtop mass spectrometer, to measure relative amounts of endogenous peptides Aβ1‐40, Aβ1‐42, APP669‐711, and a stable‐isotope‐labeled internal standard. A composite plasma biomarker was calculated as the mean of the normalized ratios APP669711/Aβ142 and Aβ140/Aβ142.Resultn= 24 (24%) of participants were PiB‐positive by a global cutoff, 7 (7%) were CDR 0.5, and 13 (13%) were APOE*4 carriers. Higher global PiB SUVR was associated with older age, higher CDR (0.5 vs. 0) and APOE*4 genotype (p’s < .05), while higher plasma composite biomarker was similarly associated with CDR and APOE*4 but not age. ROC analysis yielded an AUC of 0.78 (SE 0.05) for the composite plasma biomarker predicting PiB‐PET status, with an optimized cut‐point corresponding to sensitivity of 0.75 and specificity of 0.69. Including age and APOE*4 into the model yielded an overall AUC of 0.89, and plasma biomarker sensitivity of 0.91 and specificity of 0.77.ConclusionThis IP‐MS assay was successfully implemented using a commercially‐available, benchtop MALDI‐TOF instrument that is easy to operate and already used in clinic settings. Concurrent prediction of amyloid status from plasma amyloid alone was significant, with moderately high accuracy. Combining plasma amyloid with age and APOE genotype improved prediction to high accuracy. Planned follow‐up of the sample with repeat PiB scans will yield yet further information about prospective prediction of brain amyloid status. This non‐invasive approach using a peripheral biomarker could be highly useful and feasible to apply in large population studies of AD risk.