Re-evaluatingpathogenicity of variants associated with the long QT syndrome.

Abstract

Genetic testing for congenital long QT syndrome (LQTS) has become common. Recent studies have shown that some variants labelled as pathogenic might be misclassified due to sparse case reports and relatively common allele frequencies (AF) in the general population. This study aims to evaluate the presence of LQTS-associated variants in the Genome Aggregation Database (gnomAD) population, and assess the functional impact of these variants. Variants associated with LQTS from the Human Gene Mutation Database were extracted and matched to the gnomAD to evaluate population-based AF. We used MetaSVM to predict the function of LQTS variants. Allele distribution by protein topology in KCNQ1, KCNH2, and SCN5A was compared between gnomAD (n=123,136) and a cohort of LQTS patients aggregated from eight published studies (n=2,683). Among the 1,415 LQTS-associated single nucleotide variants in 30 genes, 347 (25%) are present in gnomAD; 24% of the 347 variants were predicted as functionally tolerated compared with 4% of variants not present in gnomAD (P<0.001). Of the 347 pathogenic variants in gnomAD, seven (2%) had an AF of ≥ 0.001 and 65 (19%) variants had an AF of ≥ 0.0001. In KCNQ1, KCNH2, and SCN5A, allele distribution by protein functional region was significantly different with gnomAD alleles appearing less frequently in highly pathogenic domains than case alleles. A significant number of LQTS variants have insufficient evidence for pathogenicity and relatively common AF in the general population. Caution should be used when ascribing pathogenicity to these variants.