Abstract 15476: Genome-wide Transcriptome Study for Myocardial Infarction and Coronary Artery Calcification Using RNA-Seq

Abstract

Introduction: Coronary artery calcification (CAC) is a noninvasive measure of coronary atherosclerosis, which underlies most cases of myocardial infarction (MI). Genome-wide association (GWA) studies have identified common genetic variants associated with MI and CAC. However, the mechanisms underlying MI and CAC remain unclear. We sought to identify gene expression signatures of early MI and high CAC, and further investigate the underlying mechanisms between clinical MI and subclinical atherosclerosis in the Framingham Heart Study (FHS). Methods: We conducted a paired-end RNA sequencing on whole blood samples collected from 202 FHS participants (57 with early MI, 74 with high CAC without MI, and 71 control subjects free of elevated CAC levels and MI), with Illumina HiSeq 2000. After filtering non- expressed genes, for each of 12,063 protein-coding genes and 3,704 long intergenic non-coding RNAs (lincRNAs), we applied DESeq2 to identify genes differentially expressed in MI and high CAC, respectively, compared with the control group after adjusting for sex, batch and hidden confounders. Results: On average, 150 million paired-end reads were obtained for each sample. Among them, 71% of reads were mapped uniquely to the genome. At the false discovery rate (FDR) <0.1, we found 71 coding genes and 9 lincRNAs differentially expressed between early MI and control. Pathway and gene ontology analysis further indicate that immune response, lipid metabolic process and interferon regulatory factor were enriched in these gene signatures. By contrast, only 3 coding genes are expressed differently between high CAC and controls at FDR<0.1. APOD, encoding a component of high density lipoprotein, was expressed significantly lower in both early MI (FDR=0.007) and in high CAC (FDR=0.01) compared with controls, respectively, identifying a novel candidate for clinical and subclinical MI. Conclusion: We identified early MI expression signatures including protein-coding genes and lincRNAs, indicating the power of our extremely deep sequenced data to identify both coding genes and, in particular, lincRNA biomarkers expressed at levels much lower than coding genes. Our results suggest an important role for both protein-coding genes and lincRNAs in the pathogenesis of MI.