Evaluation of plasma biomarkers for causal association with peripheral artery disease

Abstract

BackgroundHundreds of biomarkers for Peripheral artery disease (PAD) have been reported in the literature, however, the observational nature these studies limit robust causal inference due to the potential of reverse causality and confounding. We aimed to perform a systematic review of plasma biomarkers associated with PAD followed by Mendelian Randomization (MR) to systematically address residual confounding and better understand the causal pathophysiology of PAD. Combining a systematic review with MR facilitates cross-domain verification of observational and experimental results. MethodsWe performed a systematic literature review for terms related to PAD and/or biomarkers using Pubmed, Cochrane, and Embase, followed by manual review to extract biomarkers and their direction of effect. To evaluate evidence for causality, we employed Two-sample Mendelian randomization (MR). We developed genetic instruments for the biomarkers by mapping them to genome wide association studies (GWAS) of circulating biomolecules aggregated by the IEU Open GWAS and deCODE projects. We tested the association of the genetic instruments with PAD using summary statistics from a GWAS of 31,307 individuals with and 211,753 individuals without PAD in the VA Million Veteran Program. We employed Wald ratio or inverse-variance weighted MR; weighted median and weighted mode methods were applied as sensitivity analyses. ResultsWe identified a total of 1,993 unique papers related to PAD and biomarkers using extant genetic instruments, and MeSH terms across PubMed, Embase, and Cochrane. After filtering and manual review, 170 unique papers remained, mentioning 204 unique biomarkers. Genetic instruments based on publicly available data were developed for 175 biomarkers. After accounting for multiple testing by controlling the false discovery rate (q < 0.05), 19/175 (10.9%) biomarkers had significant associations with PAD. Of the 19 significant associations, only 13/19 (58.3%) had concordant directions of effects with published reports. These 19 biomarkers represented broad categories including plasma lipid regulation (HDL-C, LPA, Triglycerides, APOA1, EPA, APOB, APOA5, and SHBG), coagulation-inflammatory response (CD36, IL6-sRa, VWF, IL18BP, and CD163), and endothelial damage/dysfunction (HLA-G, NPPA, VCAM-1, CDH5, MMP1, and INS). ConclusionThis systematic review paired with Mendelian randomization elucidates biomarkers with genetic evidence for causality relevant to PAD, and highlights discrepancies between published reports and human genetic findings. Conventional studies have previously highlighted biomarkers that have correlation to PAD but have not emphasized the causal pathobiology of this disease. Expansion of genetic datasets to increase the power of these analyses will be crucial to further understand the causal role of plasma biomarkers and highlighting key biological pathways in PAD.