Venous thromboembolism (VTE) is a heritable, life-threatening condition with known genetic predictors. The most recent agnostic search for genetic association focused on common variation (PMID 25772935). Here we studied rare variation using whole-genome sequence (WGS). In TOPMed, we performed a pooled case-control analysis using data from 3 cohort studies (nested case-control) and 3 case-only collections. Case-control matching was performed within sex-ancestry-study groups, based on age at diagnosis for cases and age at last follow-up for controls. Excess controls from one of the cohort studies were matched with cases from the case-only collections. The matched sample set consisted of 8,911 participants with an average ratio of 1.9 controls per case (5,816 controls and 3,095 cases) with 88% European-American (EA), 10% African-American (AA) and 2% other ancestries. Mixed-model logistic regression was used for single-variant and aggregate tests. Covariates were sex, sample set, age, and ln(case/control ratio per matched set).
Single-variant tests (minor allele count filter > 5) identified 6 loci known to be associated with VTE (PMID 25772935, PMID 20212171) and one novel locus. Conditioning on the known associations did not identify any new associations in previously known loci and the new locus remained significant (p = 3e-8). Significant variant rs10963226 at the novel locus had a MAF of 0.0001 in EA and 0.04 in AA participants. The high quality variant rs10963226 lies in the first intron of one of the SH3GL2 transcripts. Chromatin capture data indicates that the region overlapping rs10963226 interacts with the promoter of SH3GL2 suggesting rs10963226 may have a regulatory role. Furthermore, SH3GL2 has been reported to interact with coagulation factor XIII A subunit [PMID: 21900206]. These observations suggest that the rare variant association in the SH3GL2 locus is biologically plausible.

Burden and SKAT tests were performed using variants with MAF < 0.01, aggregating by sliding-window (50 kb and 5kb) and also by genes after applying various genome annotation filters. Only the analysis that aggregated loss of function and missense variants in each gene had a significant result (< Bonferroni p-value 2e-6). The unit identified in this analysis was the gene PROC, a locus well known to be associated with VTE based on family studies.

In summary, the WGS data confirmed previously known loci containing common and rare variants and identified a new locus.