A Profile of Candidate Causal Genes of Spontaneous Coronary Artery Dissection Identified Through Machine Learning

Background: Spontaneous coronary artery dissection (SCAD), defined as the dissection of coronary arterial intima and media, is a genetic disorder but currently only two genes, PHACTR1 and TSR1, have been reported as it is not strongly familial. Given the limitations of association tests such as genome wide association test and gene based association test, we applied a new strategy of machine learning for identification of new candidate genes. Methods: Whole exome sequencing (WES) was performed on 85 SCAD cases and 285 controls, as well as 60 SCAD cases in replication cohort, recruited from Tongji Hospital, Wuhan, China. Accumulated gene burden scores were computed based on average scores of VEST3, MetaLR and M-CAP annotated by ANNOVAR. A machine learning model was constructed to identify candidate causal genes of SCAD by logistic regression with penalty. Findings: The machine learning model identified 116 genes associated with SCAD, whose mutational burden was elevated in cases relative to controls and yielded an area under curve area of 0.83 in cross validation. Compared to control group, patients with SCAD carried more rare deleterious variants of machine learning identified genes. Multiple variants in different machine learning identified genes were more prevalent in SCAD cohort. In a replication cohort, 73 genes were confirmed, with FBN3 ranking the top. Specifically, two deleterious variants of FBN3 were exactly the same as in discovery cohort. Interpretation: We built a machine learning model and identified a profile of candidate genes associated with SCAD where rare deleterious and multiple variants were enriched in patients with SCAD. Funding Statement: This work was supported in part by key grants from National Key R&D Program of China (No. 2017YFC0909400), National Nature Science Foundation of China (No. 91839302 and 81790624), Shanghai Municipal Science and Technology Major Project (No. 2017SHZDZX01) and Major Project of Technological Innovation of Hubei province (No. 2017ACA170). Declaration of Interests: All authors declares no conflict of interests Ethics Approval Statement: The study protocol was approved by the Tongji Hospital Review Board and all participants have signed written informed content.