Blood‐based multi‐omic signatures of blood‐brain barrier impairment and CSF‐defined Alzheimer disease in older adults

Abstract

AbstractBackgroundBlood‐brain barrier (BBB) breakdown and accummulation of Alzheimer disease (AD) pathology are risk factors for age‐related dementia. However, both require either lumbar puncture or advanced neuroimaging for detection that are not widely available. Less invasive ways to detect these risk factors early on could permit more targeted therapies for prevention. Multiple‐level molecular biomarker approaches (e.g., ‘multiomics’) may identify candidates for diagnostic development and help initiate earlier risk factor modificaion. We deployed multi‐omics spanning plasma or serum of non‐coding RNAs, proteins, and lipids to comprehensive nutrient biomarkers to distinguish BBB impairment and CSF defined AD pathology in older adults.MethodCSF albumin index ≥ 9.0 indicated BBB impairment. CSF p‐tau181/abeta42 ratio > 0.0779 indicated AD pathology. miRNAs were measured with the HTG EdgeSeq technology; inflammatory biomarkers with MesoScale sandwich immunoassays; mass spectrometry was used for proteomics, lipidomics, metabolomics, measurements of amino acids, hydro‐and liposoluble vitamins, fatty acids, and minerals; mass spectrometry was coupled to liquid, supercritical fluid or gas chromatography where needed. LASSO logistic regression models selected multi‐omic signatures that best classified BBB impairment and AD pathology using formal diagnostic accuracy and area under the curve (AUC) statistical criteria.ResultCross‐sectional analysis was conducted in 120 subjects with a mean age of 70 years, 64% were female and the mean MMSE was 27. Prevalence of BBB impairment and AD pathology was recorded in 13.5% and 35%, respectively. A blood‐based multi‐omic signature, including 1 amino acid , 1 cholesterol ester, 1 mineral, 1 pro‐inflammatory cytokine, 3 proteins and 4 miRNAs produced a diagnostic accuracy of 93% for BBB impairment (AUC of 0.96 with 95% confidence interval (CI) of 0.91‐0.99; Figure 1). Another blood‐based multi‐omic signature, including 2 amino acids, ethanolamine and a phospholipid , 2 minerals, 3 inflammatory‐specific markers, 4 other proteins, 1 miRNA , APOE4 genotype and age diagnosed AD pathology with 86% accuracy and generated an AUC of 0.92 (95% CI of 0.86‐0.96; Figure 2)ConclusionWe identified distinct blood‐based signatures of BBB impairment and CSF biomarkers of AD pathology in a cohort of predominantly non‐demented older adults using pre‐specified outcome criteria. Multi‐omic approaches in blood can provide novel insights into mechanism and improve detection of underlying risk factors for age‐related dementia.