Multiplexed Assays of Variant Effect and Automated Patch Clamping Improve KCNH2-LQTS Variant Classification and Cardiac Event Risk Stratification

Abstract

Background: Long QT syndrome is a lethal arrhythmia syndrome, frequently caused by rare loss-of-function variants in the potassium channel encoded by KCNH2. Variant classification is difficult, often because of lack of functional data. Moreover, variant-based risk stratification is also complicated by heterogenous clinical data and incomplete penetrance. Here we sought to test whether variant-specific information, primarily from high-throughput functional assays, could improve both classification and cardiac event risk stratification in a large, harmonized cohort of KCNH2 missense variant heterozygotes.

Methods: We quantified cell-surface trafficking of 18 796 variants in KCNH2 using a multiplexed assay of variant effect (MAVE). We recorded KCNH2 current density for 533 variants by automated patch clamping. We calibrated the strength of evidence of MAVE data according to ClinGen guidelines. We deeply phenotyped 1458 patients with KCNH2 missense variants, including QTc, cardiac event history, and mortality. We correlated variant functional data and Bayesian long QT syndrome penetrance estimates with cohort phenotypes and assessed hazard ratios for cardiac events.

Results: Variant MAVE trafficking scores and automated patch clamping peak tail currents were highly correlated (Spearman rank-order ρ=0.69; n=433). The MAVE data were found to provide up to pathogenic very strong evidence for severe loss-of-function variants. In the cohort, both functional assays and Bayesian long QT syndrome penetrance estimates were significantly predictive of cardiac events when independently modeled with patient sex and corrected QT interval (QTc); however, MAVE data became nonsignificant when peak tail current and penetrance estimates were also available. The area under the receiver operator characteristic curve for 20-year event outcomes based on patient-specific sex and QTc (area under the curve, 0.80 [0.76-0.83]) was improved with prospectively available penetrance scores conditioned on MAVE (area under the curve, 0.86 [0.83-0.89]) or attainable automated patch clamping peak tail current data (area under the curve, 0.84 [0.81-0.88]).

Conclusions: High-throughput KCNH2 variant MAVE data meaningfully contribute to variant classification at scale, whereas long QT syndrome penetrance estimates and automated patch clamping peak tail current measurements meaningfully contribute to risk stratification of cardiac events in patients with heterozygous KCNH2 missense variants.

Keywords: LQTS; arrhythmias; automated patch clamping; multiplexed assay of variant effect; risk stratification.

Figure 1.. Project Overview.

We integrate two high-throughput functional assays with clinical deep phenotyping, quantitative penetrance estimates, and prospective risk prediction models to improve classification and risk estimation of the KCNH2-LQT2 genotype-phenotype pair.

Figure 2.. Results of a KCNH2 Multiplexed Assay of Variant Effect.

A) Schematic of MAVE assay. We employed a barcode abundance-based MAVE of cell-surface KCNH2 variant expression to quantify variant trafficking, the primary mechanism of KCNH2 variant loss-of-function. B) Distribution of WT-normalized variant trafficking scores among missense (score 71.4 +/− 46.4, N=16562), synonymous (score 100.4 +/− 23.9, N=775), and nonsense variants (score 17.9 +/− 36.0, N=986). Each dot represents average of %WT across 6 biological replicates. p < 0.01, ANOVA with post-hoc Tukey test C) Heatmap depicting trafficking scores across the coding region of KCNH2. Dark orange indicated less than WT trafficking, white similar trafficking to WT, and blue increased trafficking. Missing data are depicted in gray.

Figure 3.. Classification and Diagnostic Value of MAVE Trafficking Assay.

A) Distribution of trafficking scores among control variants for assay calibration, each dot represents averaged, WT-normalized trafficking score of each individual variant. Controls selected from those described by Thomson et al., including 25 B/LB and 26 P/LP variants. The dotted lines form a boundary of ‘normal’ functional assay, as determined by z-score’s of assay performance on the 25 B/LB controls. Dots falling outside of this boundary are considered abnormal (quantified in B). No hypothesis testing intentionally performed for assay calibration. B) 2x2 table of assay result vs variant classification in control group. Scores for abnormal and normal function are calibrated to assay performance based on B/LB controls (24/25 concordant). The mean distribution was used to calculate z-score thresholds of abnormal function (see methods and thresholds in panel A). Using these thresholds, 23/26 variants were concordant for Pathogenic functional evidence. C) Receiver operator characteristic curve of MAVE data applied across all readily available ClinVar B/LB and P/LP annotations (N=152). D-E) Violin plots showing trafficking scores of VUS and Conflicting Interpretation variants in ClinVar. Each dot represents averaged, WT-normalized trafficking score of each individual variant. Dotted lines reflect bins of ACMG evidence criteria for variant classifications, derived using OddsPath thresholds (D), and Log-likelihood ratios (E).

Figure 4.. Results of a KCNH2 Automated Patch Clamp Assay.

A) Example APC peak tail currents recorded at −50 mV showing different levels of function. Y-axis is 500 pA and X-axis is 500ms. B) KCNH2 peak-tail current densities for 533 variants (n=38,772 recordings) across 6 domains of the protein observed in our clinical cohort, gnomAD, and previous literature reports. Benign variant controls from gnomAD are shown as white circles. Blue range depicts variants with ‘normal function’, as defined by a ±2 z-score window from the mean current density for B/LB variants. Positions were delineated as follows: EAG domain: 1-135; Proximal N: 136-399; VSD: 400-550; Pore: 551-667; C-linker/CNBHD: 668-870; Distal-C: 871-1159. C) Matrix of z-score determined normal and abnormal variants studied by both functional assays. D) Functional assay outcomes by datapoint sampled at each residue position across K_V11.1. Top panel shows peak tail current function at each position (N = 533), while the bottom panel represents the heatmap from Figure 2C aligned to protein residue number (N = 18,796). Broadly, we observe regional correlation between areas of mutation intolerance from the hemizygous MAVE (bottom panel) with decreased heterozygous functional scores from the APC (top panel) across the protein. For example, residue 876, there are infrequently observed MAVE loss-of-function variants and infrequent loss-of-function APC variants. Please see Figure S5 for direct MAVE vs APC outcome comparisons.

Figure 5.. Descriptive Correlations of Functional Data with Missense Heterozygote Cohort Clinical Features.

A-C) Correlations between participant QTc and functional scores for all available cohort members with each descriptor (Spearman rho). D-F) Stratification of cardiac event status across the cohort by variant-specific features including MAVE trafficking scores (D, no event n = 988, event n = 221), LQTS penetrance estimates (E, no event n = 1171, event n = 250), and APC densities (F, no event n = 910, event n = 231). p-values indicated in figure following Wilcox rank sum test.

Figure 6.. Clinical Risk Models and Applications of Variant-specific and Patient-Specific features.

A-C) Royston-Parmar Hazard Ratios (RP HR) for first 20-year cardiac event with baseline patient-specific features of sex and adjusted QTc, and variant-specific data of MAVE, LQTS penetrance, and APC. D) Royston-Parmar Hazard Ratios (RP HR) for first 20-year cardiac event cardiac event with all patient-specific and variant-specific data. E) ROCs/AUCs for three different models for all cardiac events through 20 years of age.