Utilization of the genome aggregation database, in silico tools, and heterologous expression patch-clamp studies to identify and demote previously published type 2 long QT syndrome: Causative variants from pathogenic to likely benign

Background: Loss-of-function variants in the KCNH2-encoded Kv11.1 potassium channel cause long QT syndrome (LQTS) type 2 (LQT2). Presently, hundreds of KCNH2 missense variants (MVs) have been published as "disease-causative." However, an estimated 10% of rare published LQTS MVs may be "false positives."

Objective: The purpose of this study was to determine which published KCNH2 MVs are likely false positives and warrant demotion to "likely benign" status.

Methods: A list of 337 LQT2-associated MVs from 6 large compendia was compiled. MV frequency within the Genome Aggregation Database (gnomAD) (n = 141,352 individuals) was assessed, and MVs were analyzed with 8 in silico tools. Variants with minor allele frequency (MAF) >7*10E-6, the calculated maximum credible frequency of LQT2, and predicted "benign" by all tools were demoted to "likely benign." Ultra-rare variants (n = 8) absent in gnomAD but predicted "benign" by all tools were considered as potential false positives and were characterized functionally using whole-cell patch clamp.

Results: Overall, 14 of 337 published KCNH2 MVs (4%) were observed at MAF >7*10E-6, whereas 252 of 337 (75%) were absent in gnomAD. Among the latter, 8 variants (I96V, G187S, A203T, P241L, H254Q, G314S, P935S, P963T) were predicted benign by 8 tools and lacked characterization. Patch clamp showed no functional perturbation for these 8 MVs.

Conclusion: This study offers compelling evidence for the demotion of 22 of 337 KCNH2 variants (6.5%) in the literature. Meticulous "pruning" of compendia using exome/genome databases, in silico tools, and in vitro functional studies must be conducted not only for putatively pathogenic LQTS MVs but for the entire field of genetic heart disease.