Background: Known genetic causes of congenital heart disease (CHD) explain <40% of CHD cases, and interpreting the clinical significance of variants with uncertain functional impact remains challenging. We aim to improve diagnostic classification of variants in patients with CHD by assessing the impact of noncanonical splice region variants on RNA splicing.
Methods: We tested de novo variants from trio studies of 2649 CHD probands and their parents, as well as rare (allele frequency, <2×10-6) variants from 4472 CHD probands in the Pediatric Cardiac Genetics Consortium through a combined computational and in vitro approach.
Results: We identified 53 de novo and 74 rare variants in CHD cases that alter splicing and thus are loss of function. Of these, 77 variants are in known dominant, recessive, and candidate CHD genes, including KMT2D and RBFOX2. In 1 case, we confirmed the variant's predicted impact on RNA splicing in RNA transcripts from the proband's cardiac tissue. Two probands were found to have 2 loss-of-function variants for recessive CHD genes HECTD1 and DYNC2H1. In addition, SpliceAI-a predictive algorithm for altered RNA splicing-has a positive predictive value of ≈93% in our cohort.
Conclusions: Through assessment of RNA splicing, we identified a new loss-of-function variant within a CHD gene in 78 probands, of whom 69 (1.5%; n=4472) did not have a previously established genetic explanation for CHD. Identification of splice-altering variants improves diagnostic classification and genetic diagnoses for CHD.
Registration: URL: https://clinicaltrials.gov; Unique identifier: NCT01196182.
Keywords: RNA splicing; algorithms; alleles; child; humans.