Abstract 467: Identification of Novel Pathogenic Mutations in Non-Canonical RNA Splice Sites in Congenital Heart Disease

Rationale: Congenital heart disease (CHD) occurs in about 1 in 100 live births, yet known genetic causes explain less than 20% of CHD cases. While variants that cause frameshift, nonsense, start/stop site gain or loss, and canonical splice site alterations are readily categorized as being pathogenic or loss-of-function (LOF), interpreting the clinical significance of variants without obvious functional consequences remains challenging. Here, we aim to improve classification of variants of unknown significance (VUS) in non-canonical RNA splice sites that may be pathogenic for CHD. Methods: We tested candidate LOF de novo (DNV) and rare (p < 2E-5) inherited variants from whole exome sequencing of 4474 CHD probands and their parents in the NHLBI Pediatric Cardiac Genetics Consortium. Briefly, variants underwent computational selection to prioritize VUS in splice regions that are likely to alter splicing (“high-likelihood VUS”). These variants then underwent in vitro analysis including Minigene construction, transfection, RNA isolation, and sequencing to confirm splicing outcomes. Results: Preliminary results limited to DNV variants showed that 163 of 2678 (6.08%) were high-likelihood VUS. Subsequent analysis in vitro assay of high-likelihood VUS yielded 53 as splice-altering (p < 0.05) and thus LOF. Combined with previously identified 366 DNV LOF variants, the addition of these splice-altering variants represents a 15.3% increase in total LOF DNV variant calls. This includes new pathogenic mutations in known CHD genes such as KMT2D . In one case, a CHD proband with features of Kabuki Syndrome without a definitive diagnosis was found to have a splice-altering variant in KMT2D . We have extended this assay to 34518 rare, inherited variants in the same cohort, of which 868 (2.54%) are in genes previously associated with CHD. Conclusion: Consideration of non-canonical RNA splice sites in this assay increased the yield of LOF mutations from traditional sequencing methods by 15.3% in the CHD cohort. Further analysis of splice-altering variants in both known and unknown pathogenic genes will improve diagnostic classification of VUS and gene-based diagnosis of CHD.