PLASMA BIOMARKER FOR ALZHEIMER’S DISEASE: ARE WE READY NOW FOR CLINICAL PRACTICE AND DRUG TRIALS?

Abstract

To facilitate disease-modifying clinical trials for Alzheimer’s Disease (AD), a blood-based amyloid-β (Aβ) biomarker, which can accurately detect an early pathological signature of AD at prodromal or preclinical stages, has been strongly desired, because it is simpler, less invasive and less costly compared to PET or lumbar puncture. Despite plasma Aβ biomarkers having been extensively investigated, most studies failed to demonstrate clinical utility (1, 2), and at the end of 2016, there was a rather pessimistic mood that this objective might be impossible to realize (3). However, since the latter half of 2017, the situation appears to have changed dramatically, in that several groups have reported potential clinical utility of plasma Aβ biomarkers using different methodologies (4-7). Especially, immunoprecipitation followed by mass spectrometry (IP-MS) assays have shown promising converging evidence. In 2014, we, the National Center for Geriatrics and Gerontology (NCGG) and Koichi Tanaka Mass Spectrometry Research Laboratory at Shimadzu Corporation (Shimadzu), reported that the plasma ratio of Aβ1-42 to a novel APP669-711 fragment (APP669–711/Aβ 1–42) as determined by IP-MS could discriminate high Aβ (Aβ+) individuals from low Aβ (Aβ-) individuals (classified using PiB-PET) with more than 90% accuracy (n=62) (8). In 2017, the Washington University group analyzed detailed kinetics of plasma Aβs, and reported that Aβ42/Aβ40 as measured by IP-MS could distinguish Aβ+ and Aβ- individuals with 88.7% areas under the curve value (n=41) (5). Then very recently, we, in collaboration with the Australian Imaging, Biomarker and Lifestyle Study of Aging (AIBL), have demonstrated that plasma biomarkers, APP669-711/Aβ1-42, Aβ1-40/Aβ1-42, and their composites (composite biomarker), as generated by improved IP-MS methodology performs very well in larger independent datasets: a discovery dataset (NCGG, n=121) and a validation dataset (AIBL, n=252 which includes n=111 PiB-PET and 141 with other ligands) both of which included individuals with normal cognition, MCI and AD. Particularly, the composite biomarker showed very high AUCs in both datasets (discovery 96.7%, n=121, and validation 94.1%, n=111) with accuracy c.a. 90% when using PiB-PET as standard of truth. The findings of the study were considered to be robust, reproducible and reliable because biomarker performance was validated in a blinded manner using independent data sets (Japan and Australia) and involved an established large-scale multicenter cohort (AIBL).