https://orcid.org/0000-0001-5192-3962
Abstract
More than one-third of type 2 long QT syndrome (LQT2) patients carry KCNH2 non-missense variants that can result in haploinsufficiency (HI), leading to mechanistic loss-of-function. However, their clinical phenotypes have not been fully investigated. The remaining two-thirds of patients harbour missense variants, and past studies uncovered that most of these variants cause trafficking deficiency, resulting in different functional changes: either HI or dominant-negative (DN) effects. In this study, we examined the impact of altered molecular mechanisms on clinical outcomes in LQT2 patients.
We included 429 LQT2 patients (234 probands) carrying a rare KCNH2 variant from our patient cohort undergoing genetic testing. Non-missense variants showed shorter corrected QT (QTc) and less arrhythmic events (AEs) than missense variants. We found that 40% of missense variants in this study were previously reported as HI or DN. Non-missense and HI-groups had similar phenotypes, while both exhibited shorter QTc and less AEs than the DN-group. Based on previous work, we predicted the functional change of the unreported variants—whether they cause HI or DN via altered functional domains—and stratified them as predicted HI (pHI)- or pDN-group. The pHI-group including non-missense variants exhibited milder phenotypes compared to the pDN-group. Multivariable Cox model showed that the functional change was an independent risk of AEs (P = 0.005).
Stratification based on molecular biological studies enables us to better predict clinical outcomes in the patients with LQT2.
AE, arrhythmia event; CI, confidence interval; CNBHD, cyclic nucleotide-binding homology domain; DN, dominant-negative; EAGD, ether-à-go-go domain; ER, endoplasmic reticulum; ERAD, endoplasmic reticulum-associated degradation; HR, hazard ratio; HI, haploinsufficiency; LOF, loss-of-function; LQT2, long QT syndrome type 2; mRNA, messenger RNA; NMD, nonsense mediated mRNA decay; PD, pore domain; pDN, predicted DN; pHI, predicted HI; PTC, premature termination codon; QTc, corrected QT interval; VSD, voltage sensor domain; WT, wild-type.
More than 30% of type 2 long QT syndrome (LQT2)-related KCNH2 variants are non-missense variants, presumably causing haploinsufficiency that are not amenable to conventional functional assays; thus, the impact of functional consequences of non-missense variants on clinical outcomes has been unclear.
We found that patients with a non-missense variant showed better arrhythmia outcomes than those with a rare LQT2-causing missense variant. Furthermore, we found that missense variants in certain functional domains caused similar phenotypes and outcomes as non-missense variants, suggesting that these missense variants may cause haploinsufficiency.
Our results and approach will advance our knowledge about interpretation of in vitro studies and biological consequences of non-missense variants in LQT2.
Introduction
Congenital long QT syndrome (LQTS) is an inherited arrhythmia disease characterized by prolonged QT intervals on the electrocardiogram (ECG), leading to a bizarre form of ventricular tachycardia, Torsade de Pointes (TdP), and sudden death.1,2 The human ether-à-go-go-related gene potassium channel (Kv11.1), encoded by KCNH2, assembles as functional tetramers and generates repolarizing current IKr. Loss-of-function variants of KCNH2 cause type 2 LQTS (LQT2).3,4
More than one-third of patients with LQT2 carry a non-missense variant such as nonsense, frameshift, or splicing variants3,5,6 that may undergo nonsense-mediated messenger RNA decay (NMD)7,8 leading to haploinsufficiency (HI). However, as of yet, only a few studies have focused on the clinical outcome of the patients with non-missense variants.9,10,11 The remaining two-thirds of LQT2-related variants are missense variants, scattering over known functional domains in Kv11.1 such as ether-à-go-go domain (EAGD), voltage sensor domain (VSD), pore domain (PD), and cyclic nucleotide-binding homology domain (CNBHD).12–14 Among them, variants in the PD have been shown to exert dominant-negative (DN) effects and are thereby associated with worst arrhythmia outcomes.10,15
Recent in vitro studies have revealed that more than 80% of KCNH2 missense variants cause mis-trafficking of Kv11.1 onto the cell surface, resulting in the reduction of IKr16,17 and affecting channel abundance at a different stage during Kv11.1 translation via endoplasmic reticulum-associated degradation.18,19 Most missense variants in the PD are degraded after assembling tetramers, thus resulting in the DN trait by degrading the wild-type (WT) channel. On the other hand, variants in the EAGD, C-linker, or CNBHD likely are degraded before channel assembly, therefore not affecting the multi-merization of healthy subunits. This may lead to trafficking of fewer functional WT channels, thereby producing HI effects.17
LQT2-related gene variants thus result in different levels of loss-of-function. In 429 patients with LQT2, we aimed to test our hypothesis that potentially HI-causing non-missense and missense variants will exhibit better clinical outcomes than variants that can cause DN. This study will significantly help bridge an insight of molecular evidence with the interpretation of LQT2 patient clinical manifestations.
Methods
Study population
This study was comprised of 429 patients from 243 families with a single rare variant of KCNH2, who were referred to Shiga University of Medical Science or Kyoto University Graduate School of Medicine between 1996 and 2021 for examination of primary arrhythmia syndrome. The study was approved by the institutional review committees, and the patients or their guardians gave written informed consent.
Declaration of Helsinki
This study complies with the Declaration of Helsinki. We obtained written informed consent from all the patients and their guardians per the guidelines approved by the institutional review board of Shiga University of Medical Science and Kyoto University hospital.
Genotyping and variant interpretation
Genomic DNA was extracted from peripheral lymphocytes. Denaturing high-performance liquid chromatography (WAVE system Model 3500, Transgenomic, NE, USA) (between 1996 and 2014 for KCNQ1, KCNH2, SCN5A, KCNE1, and KCNE2) or targeted panel sequencing (between 2015 and 2021) was used for initial screening. All identified variants were verified by Sanger sequencing.20KCNH2 variants with minor allele frequency (MAF) greater than 0.00025 were excluded in consideration of the prevalence of LQTS in 1/2000.21 Cases with compound KCNH2 variants were also excluded; if a patient carried another variant in LQTS-related genes in addition to a KCNH2 variant and the MAF of the concurrent variant were < 0.00025, it was excluded from the analysis. MAF of each variant was assessed using both Genome Aggregation Database (gnomAD: https://gnomad.broadinstitute.org) and the National Bioscience Database Center's integrated database of Japanese genomic variation (Togovar: https://togovar.biosciencedbc.jp).
Classification of KCNH2 variants
KCNH2 variants were divided into two types; ‘missense variant’ included all missense variants regardless of variant domains and ‘non-missense variant’ included nonsense, frameshift, and canonical (± 2) splice site variants.22 Kv11.1’s domains were defined as follows: EAGD (p. 1–135), N-linker (p. 136–398), VSD (p. 399–545), PD (p. 546–670), C-linker (p. 671–736), CNBHD (p. 737–863), and C-tail domain (p. 864–1159).23
To assess the functional changes of each KCNH2 missense variant in our cohort, we made an extensive survey of previous experimental reports. According to the functional assay, there were reports on 41 variants in our cohort, and we classified them as either HI or DN variant (see Supplementary material online, Table S1). Furthermore, to predict the functional changes of the remaining 61 variants which were not previously investigated, we examined functional changes of each domain by reviewing past literature and then predicted their loss-of-function mechanisms as HI or DN (see Supplementary material online, Table S2). Online packages (SIFT, https://sift.bii.a-star.edu.sg; PROVEAN, https://www.jcvi.org/research/provean; polyphen-2, http://genetics.bwh.harvard.edu/pph2) were used to predict functional changes of the missense variants.
For non-missense variants, we considered them all as HI due to the lack of evidence that conventional functional assays using exogenous complementary DNA are capable of incorporating NMD mechanism in the cell model. In this study, we did not consider whether they escape NMD depending on the locations where the premature termination codons occurred.7,8
Assessment of the clinical characteristics and outcomes
The clinical characteristics of patients were evaluated before initiation of drug treatment. QT intervals were measured in II or V5 lead by the threshold method24 using Image J software (NIH) and were corrected for heart rate by the Bazett’s formula [corrected QT (QTc)]. To plot QTc across domains, QTc was plotted at each amino acid residue for the corresponding variant. When a variant was identified in multiple patients, averaged QTc was plotted. Bradycardia was diagnosed using normal limits in pediatric patients25 or below 50 bpm for patients at 4-year-old or older. When available, a Holter-monitoring was subject for diagnosis of bradycardia. Notched T wave was considered to be positive if a notch on the T wave was found in three or more leads.2 LQTS risk score (or Schwartz score) was calculated based on the most recent updates.26 Arrhythmia events (AEs) were assessed up to 40 years of age, combining syncope including unknown cause or suspected cardiac syncope, and lethal AEs including documented TdP or ventricular tachycardia/fibrillation, cardiopulmonary arrest, and sudden death with or without documentation of fatal arrhythmias.
Statistics
Data are shown as mean ± SD for continuous variables and number (percentage) for categorical variables. For continuous variables, unpaired Student’s t-test or analysis of variance (ANOVA) was used followed by Tukey’s test if an ANOVA found differences among groups. For categorical variables, the χ2 test followed by the Bonferroni corrections was used. The probability of first AEs was estimated using Kaplan–Meier method with log-rank test up to 40 years. Patients were censored when drug treatment was started. The Cox proportional hazard model was used to predict the association between AEs and predictor variables. At first, we analysed the effect of each variable on probability of AEs over time and identified sex as a time-varying covariate, a confirmatory result consistent with previous studies.10,27 Given this result, we used time-dependent Cox models, using an age of 12 as an optimal cut-point, as indicated by likelihood-based exhaustive search method, to create a step function of time to evaluate the time-dependent effect of sex.28 Two-tailed P < 0.05 was considered as significant. All statistical analyses were performed using SPSS 28.0 (IBM, Armonk, NY, USA).
Results
Study population and identified KCNH2 variants
We included 429 Japanese patients (59% female) with a single rare KCNH2 variant for analysis. Among them, 147 patients (34%) experienced arrhythmia events (syncope and/or lethal AE). There were no sudden death cases. The trigger of arrhythmia events was drug or hypokalaemia in 12 patients (8% of patients with AEs) and sudden arousal or emotional stress in 33 (22%). The situation of arrhythmia events was at rest in 97 patients (66% of patients with AEs), exercise in 10 (7%), and both (rest and exercise) in 24 (16%).
We identified 178 unique variants (see Supplementary material online, Table S1): 102 missense variants (n = 239 patients, 56% of study cohort) and 76 non-missense variants (n = 190, 44%). Missense variants included 28 variants in EAGD (n = 56), 5 in n-linker (n = 7), 11 in VSD (n = 15), 35 in PD (n = 114), 3 in C-linker (n = 5), 11 in CNBHD (n = 24), and 9 in C-tail domains (n = 18). Figure 1 shows the distribution of QTc at each corresponding amino acid residue with a missense (Figure 1A) and a non-missense (Figure 1B) variant, respectively, across functional domains of Kv11.1.
Qtc values across functional domains of Kv11.1. The topology of Kv11.1 and the mean QTc of respective variant carriers were plotted on the Y-axis, at each amino acid residue on the X-axis for all missense variants (A) and non-missense variants (B). CNBHD, cyclic nucleotide-binding homology domain; EAGD, ether-à-go-go domain; PD, pore domain; QTc, corrected QT interval; VSD, voltage sensor domain.
Comparison of variant types: non-missense vs. missense variants
Table 1 shows clinical characteristics and outcomes between non-missense and missense variants. QTc intervals in non-missense variant carriers were significantly shorter than those with missense variants (484 ± 40 vs. 495 ± 55 ms, P = 0.019). The incidence of AEs was lower in non-missense than missense variant carriers (29% vs. 39%, P = 0.023), and the Kaplan–Meier survival analysis revealed that the probability of AEs was lower in non-missense variant carriers than those of missense variant (log-rank P = 0.006, Figure 2A).
Kaplan–Meier survival analyses for the probability of AEs. (A) Probability of arrhythmia events (AEs) among the non-missense variant carriers and the missense variant carriers. (B) Probability of AEs among carriers with non-missense variant, HI-group, and DN-group. (C) Probability of AEs among carriers with pHI-group, and pDN-group. AEs, arrhythmia events; DN, dominant-negative; HI, haploinsufficiency; pDN, predicted DN; pHI, predicted HI.
Clinical characteristics of two variant types: non-missense vs. missense
| Non-missense variant | Missense variant | P-value | |
|---|---|---|---|
| n | 190 | 239 | |
| Proband, n (%) | 93 (49) | 141 (59) | 0.038 |
| Female, n (%) | 111 (58) | 144 (60) | 0.70 |
| Age at genotyping, years | 28 ± 19 | 27 ± 20 | 0.59 |
| QT, ms | 459 ± 62 | 462 ± 76 | 0.73 |
| QTc (Bazett), ms | 484 ± 40 | 495 ± 55 | 0.019 |
| FH of SD (≤30 years), n (%) | 16 (9) | 33 (14) | 0.12 |
| Bradycardia, n (%) | 34 (18) | 47 (20) | 0.64 |
| Notched T wave, n (%) | 106 (56) | 139 (60) | 0.46 |
| LQTS score | 4.1 ± 1.9 | 4.2 ± 1.9 | 0.48 |
| AEs, n (%) | 54 (29) | 93 (39) | 0.023 |
| Age at onset, years | 16 ± 7 | 16 ± 10 | 0.63 |
| ȃSyncope, n (%) | 45 (24) | 77 (32) | 0.052 |
| ȃAge at syncope, years | 17 ± 7 | 16 ± 9 | 0.84 |
| ȃLethal AE, n (%) | 25 (13) | 45 (19) | 0.11 |
| ȃAge at lethal AE, years | 21 ± 8 | 20 ± 12 | 0.77 |
| Non-missense variant | Missense variant | P-value | |
|---|---|---|---|
| n | 190 | 239 | |
| Proband, n (%) | 93 (49) | 141 (59) | 0.038 |
| Female, n (%) | 111 (58) | 144 (60) | 0.70 |
| Age at genotyping, years | 28 ± 19 | 27 ± 20 | 0.59 |
| QT, ms | 459 ± 62 | 462 ± 76 | 0.73 |
| QTc (Bazett), ms | 484 ± 40 | 495 ± 55 | 0.019 |
| FH of SD (≤30 years), n (%) | 16 (9) | 33 (14) | 0.12 |
| Bradycardia, n (%) | 34 (18) | 47 (20) | 0.64 |
| Notched T wave, n (%) | 106 (56) | 139 (60) | 0.46 |
| LQTS score | 4.1 ± 1.9 | 4.2 ± 1.9 | 0.48 |
| AEs, n (%) | 54 (29) | 93 (39) | 0.023 |
| Age at onset, years | 16 ± 7 | 16 ± 10 | 0.63 |
| ȃSyncope, n (%) | 45 (24) | 77 (32) | 0.052 |
| ȃAge at syncope, years | 17 ± 7 | 16 ± 9 | 0.84 |
| ȃLethal AE, n (%) | 25 (13) | 45 (19) | 0.11 |
| ȃAge at lethal AE, years | 21 ± 8 | 20 ± 12 | 0.77 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; FH, family history; LQTS, long QT syndrome; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
Clinical characteristics of two variant types: non-missense vs. missense
| Non-missense variant | Missense variant | P-value | |
|---|---|---|---|
| n | 190 | 239 | |
| Proband, n (%) | 93 (49) | 141 (59) | 0.038 |
| Female, n (%) | 111 (58) | 144 (60) | 0.70 |
| Age at genotyping, years | 28 ± 19 | 27 ± 20 | 0.59 |
| QT, ms | 459 ± 62 | 462 ± 76 | 0.73 |
| QTc (Bazett), ms | 484 ± 40 | 495 ± 55 | 0.019 |
| FH of SD (≤30 years), n (%) | 16 (9) | 33 (14) | 0.12 |
| Bradycardia, n (%) | 34 (18) | 47 (20) | 0.64 |
| Notched T wave, n (%) | 106 (56) | 139 (60) | 0.46 |
| LQTS score | 4.1 ± 1.9 | 4.2 ± 1.9 | 0.48 |
| AEs, n (%) | 54 (29) | 93 (39) | 0.023 |
| Age at onset, years | 16 ± 7 | 16 ± 10 | 0.63 |
| ȃSyncope, n (%) | 45 (24) | 77 (32) | 0.052 |
| ȃAge at syncope, years | 17 ± 7 | 16 ± 9 | 0.84 |
| ȃLethal AE, n (%) | 25 (13) | 45 (19) | 0.11 |
| ȃAge at lethal AE, years | 21 ± 8 | 20 ± 12 | 0.77 |
| Non-missense variant | Missense variant | P-value | |
|---|---|---|---|
| n | 190 | 239 | |
| Proband, n (%) | 93 (49) | 141 (59) | 0.038 |
| Female, n (%) | 111 (58) | 144 (60) | 0.70 |
| Age at genotyping, years | 28 ± 19 | 27 ± 20 | 0.59 |
| QT, ms | 459 ± 62 | 462 ± 76 | 0.73 |
| QTc (Bazett), ms | 484 ± 40 | 495 ± 55 | 0.019 |
| FH of SD (≤30 years), n (%) | 16 (9) | 33 (14) | 0.12 |
| Bradycardia, n (%) | 34 (18) | 47 (20) | 0.64 |
| Notched T wave, n (%) | 106 (56) | 139 (60) | 0.46 |
| LQTS score | 4.1 ± 1.9 | 4.2 ± 1.9 | 0.48 |
| AEs, n (%) | 54 (29) | 93 (39) | 0.023 |
| Age at onset, years | 16 ± 7 | 16 ± 10 | 0.63 |
| ȃSyncope, n (%) | 45 (24) | 77 (32) | 0.052 |
| ȃAge at syncope, years | 17 ± 7 | 16 ± 9 | 0.84 |
| ȃLethal AE, n (%) | 25 (13) | 45 (19) | 0.11 |
| ȃAge at lethal AE, years | 21 ± 8 | 20 ± 12 | 0.77 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; FH, family history; LQTS, long QT syndrome; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
Missense variants: haploinsufficient vs. dominant-negative effects
We then focused on the functional outcome of missense variants. Many experimental studies have examined the functional change induced by previously reported variants. Among the missense variants in our cohort, 41 (40%) variants had been previously studied: 22 HI and 19 DN variants (see Supplementary material online, Table S1).
As shown in Table 2, non-missense variants showed nearly identical phenotypes with missense variants that had been reported as HI (HI-group), including QTc and incidence of AEs. In contrast, reported DN missense variants (DN-group) showed longer QTc and higher incidence of AEs compared to the non-missense variants or the HI-group. In Kaplan–Meier analysis (Figure 2B), the DN-group showed significantly higher probability of AEs than the other variant carriers (log-rank P = 0.001). HI-group showed an intermediate prognosis between non-missense and DN-groups.
Clinical characteristics of non-missense, HI and DN variants
| Non-missense variant | HI-group | DN-group | P-value | |
|---|---|---|---|---|
| n | 190 | 60 | 71 | |
| Proband, n (%) | 93 (49) | 32 (53) | 45 (63) | 0.12 |
| Female, n (%) | 111 (58) | 36 (60) | 47 (66) | 0.52 |
| Age at genotyping, years | 28 ± 19 | 25 ± 19 | 27 ± 21 | 0.64 |
| QT, ms | 459 ± 62 | 444 ± 77 † | 476 ± 76 | 0.039 |
| QTc (Bazett), ms | 484 ± 40 ‡ | 484 ± 40 ‡ | 509 ± 52 | < 0.001 |
| FH of SD (≤30 years), n (%) | 16 (9) * | 4 (7) | 14 (21) | 0.017 |
| Bradycardia, n (%) | 34 (18) | 12 (20) | 11 (15) | 0.80 |
| Notched T wave, n (%) | 106 (56) | 32 (56) | 46 (67) | 0.29 |
| LQTS score | 4.1 ± 1.9 | 4.0 ± 1.8 | 4.6 ± 1.9 | 0.096 |
| AEs, n (%) | 54 (29) ** | 19 (32) | 36 (51) | 0.002 |
| Age at onset, years | 16 ± 7 † | 19 ± 8 ‡ | 12 ± 9 | 0.004 |
| ȃSyncope, n (%) | 45 (24) * | 18 (30) | 29 (41) | 0.021 |
| ȃAge at syncope, years | 17 ± 7 | 19 ± 8 † | 13 ± 7 | 0.023 |
| ȃLethal AE, n (%) | 25 (13) | 9 (14) | 14 (22) | 0.26 |
| ȃAge at lethal AE, years | 21 ± 8 † | 23 ± 6 † | 13 ± 12 | 0.018 |
| Non-missense variant | HI-group | DN-group | P-value | |
|---|---|---|---|---|
| n | 190 | 60 | 71 | |
| Proband, n (%) | 93 (49) | 32 (53) | 45 (63) | 0.12 |
| Female, n (%) | 111 (58) | 36 (60) | 47 (66) | 0.52 |
| Age at genotyping, years | 28 ± 19 | 25 ± 19 | 27 ± 21 | 0.64 |
| QT, ms | 459 ± 62 | 444 ± 77 † | 476 ± 76 | 0.039 |
| QTc (Bazett), ms | 484 ± 40 ‡ | 484 ± 40 ‡ | 509 ± 52 | < 0.001 |
| FH of SD (≤30 years), n (%) | 16 (9) * | 4 (7) | 14 (21) | 0.017 |
| Bradycardia, n (%) | 34 (18) | 12 (20) | 11 (15) | 0.80 |
| Notched T wave, n (%) | 106 (56) | 32 (56) | 46 (67) | 0.29 |
| LQTS score | 4.1 ± 1.9 | 4.0 ± 1.8 | 4.6 ± 1.9 | 0.096 |
| AEs, n (%) | 54 (29) ** | 19 (32) | 36 (51) | 0.002 |
| Age at onset, years | 16 ± 7 † | 19 ± 8 ‡ | 12 ± 9 | 0.004 |
| ȃSyncope, n (%) | 45 (24) * | 18 (30) | 29 (41) | 0.021 |
| ȃAge at syncope, years | 17 ± 7 | 19 ± 8 † | 13 ± 7 | 0.023 |
| ȃLethal AE, n (%) | 25 (13) | 9 (14) | 14 (22) | 0.26 |
| ȃAge at lethal AE, years | 21 ± 8 † | 23 ± 6 † | 13 ± 12 | 0.018 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; DN, dominant-negative; FH, family history; HI, haploinsufficiency; LQTS, long QT syndrome; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
†P < 0.05, ‡P < 0.01 vs. DN-group by analysis of variance followed by Tukey post-hoc test.
*P < 0.05, **P < 0.01 vs. DN-group by the χ2 test followed by the Bonferroni corrections.
Clinical characteristics of non-missense, HI and DN variants
| Non-missense variant | HI-group | DN-group | P-value | |
|---|---|---|---|---|
| n | 190 | 60 | 71 | |
| Proband, n (%) | 93 (49) | 32 (53) | 45 (63) | 0.12 |
| Female, n (%) | 111 (58) | 36 (60) | 47 (66) | 0.52 |
| Age at genotyping, years | 28 ± 19 | 25 ± 19 | 27 ± 21 | 0.64 |
| QT, ms | 459 ± 62 | 444 ± 77 † | 476 ± 76 | 0.039 |
| QTc (Bazett), ms | 484 ± 40 ‡ | 484 ± 40 ‡ | 509 ± 52 | < 0.001 |
| FH of SD (≤30 years), n (%) | 16 (9) * | 4 (7) | 14 (21) | 0.017 |
| Bradycardia, n (%) | 34 (18) | 12 (20) | 11 (15) | 0.80 |
| Notched T wave, n (%) | 106 (56) | 32 (56) | 46 (67) | 0.29 |
| LQTS score | 4.1 ± 1.9 | 4.0 ± 1.8 | 4.6 ± 1.9 | 0.096 |
| AEs, n (%) | 54 (29) ** | 19 (32) | 36 (51) | 0.002 |
| Age at onset, years | 16 ± 7 † | 19 ± 8 ‡ | 12 ± 9 | 0.004 |
| ȃSyncope, n (%) | 45 (24) * | 18 (30) | 29 (41) | 0.021 |
| ȃAge at syncope, years | 17 ± 7 | 19 ± 8 † | 13 ± 7 | 0.023 |
| ȃLethal AE, n (%) | 25 (13) | 9 (14) | 14 (22) | 0.26 |
| ȃAge at lethal AE, years | 21 ± 8 † | 23 ± 6 † | 13 ± 12 | 0.018 |
| Non-missense variant | HI-group | DN-group | P-value | |
|---|---|---|---|---|
| n | 190 | 60 | 71 | |
| Proband, n (%) | 93 (49) | 32 (53) | 45 (63) | 0.12 |
| Female, n (%) | 111 (58) | 36 (60) | 47 (66) | 0.52 |
| Age at genotyping, years | 28 ± 19 | 25 ± 19 | 27 ± 21 | 0.64 |
| QT, ms | 459 ± 62 | 444 ± 77 † | 476 ± 76 | 0.039 |
| QTc (Bazett), ms | 484 ± 40 ‡ | 484 ± 40 ‡ | 509 ± 52 | < 0.001 |
| FH of SD (≤30 years), n (%) | 16 (9) * | 4 (7) | 14 (21) | 0.017 |
| Bradycardia, n (%) | 34 (18) | 12 (20) | 11 (15) | 0.80 |
| Notched T wave, n (%) | 106 (56) | 32 (56) | 46 (67) | 0.29 |
| LQTS score | 4.1 ± 1.9 | 4.0 ± 1.8 | 4.6 ± 1.9 | 0.096 |
| AEs, n (%) | 54 (29) ** | 19 (32) | 36 (51) | 0.002 |
| Age at onset, years | 16 ± 7 † | 19 ± 8 ‡ | 12 ± 9 | 0.004 |
| ȃSyncope, n (%) | 45 (24) * | 18 (30) | 29 (41) | 0.021 |
| ȃAge at syncope, years | 17 ± 7 | 19 ± 8 † | 13 ± 7 | 0.023 |
| ȃLethal AE, n (%) | 25 (13) | 9 (14) | 14 (22) | 0.26 |
| ȃAge at lethal AE, years | 21 ± 8 † | 23 ± 6 † | 13 ± 12 | 0.018 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; DN, dominant-negative; FH, family history; HI, haploinsufficiency; LQTS, long QT syndrome; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
†P < 0.05, ‡P < 0.01 vs. DN-group by analysis of variance followed by Tukey post-hoc test.
*P < 0.05, **P < 0.01 vs. DN-group by the χ2 test followed by the Bonferroni corrections.
Estimation of functional changes and arrhythmia outcome with HI and DN variants
The remaining 61 (60%) of missense variants in our cohort were not biophysically examined previously, and so we attempted to predict their functional outcomes based on their locations. We surveyed previous reports and classified a total of 224 missense variants as DN or not (see Supplementary material online, Table S2). Figure 3 shows the details of known functional changes according to the variant domain among variants in our dataset (n = 41 previously studied) (Figure 3A). These functional changes correlate with previously reported functional changes of the variants in each domain (n = 224 previously studied) (Figure 3B). We thus determined the dominant functional changes in each domain. Following this finding, 61 functionally unknown variants were classified into either predicted DN (pDN)-group (VSD and PD) or predicted HI (pHI)-group (other domains) (see Supplementary material online, Table S1). Then, above mentioned functionally known DN and HI variants were included to pDN and pHI-group, respectively. Finally, we classified all non-missense variants as pHI-group since they likely exert HI effects through the NMD phenomenon.7,8
Functional changes in each domain. Proportion of known functional changes in each domain in our cohort (A) and in the literature (B). Numbers in the pie chart indicate the percentage of each functional change among variants previously studied in vitro. References of the literature are in Supplementary material online, Table S2. CNBHD, cyclic nucleotide-binding homology domain; DN, dominant-negative; EAGD, ether-à-go-go domain; HI, haploinsufficiency; PD, pore domain; VSD, voltage sensor domain; WT, wild-type.
We then compared the clinical outcomes between pHI- and pDN-groups. Table 3 summarizes the clinical characteristics in two groups. PHI-group showed mild phenotypes compared with pDN-group: shorter QTc, lower LQTS scores, lower presence of notched T wave, and lower incidence of AEs at older age. The incidence of both syncope and lethal AEs was lower in pHI-group than pDN-group. The Kaplan–Meier analysis (Figure 2C) revealed that the pHI-group was at a significantly lower risk of AEs compared to the pDN-group (log-rank P < 0.001).
Two groups of variant carriers based on functional prediction
| pHI-group | pDN-group | P-value | |
|---|---|---|---|
| n | 314 | 115 | |
| Proband, n (%) | 162 (52) | 72 (63) | 0.042 |
| Female, n (%) | 182 (58) | 73 (64) | 0.30 |
| Age at genotyping, years | 27 ± 19 | 27 ± 21 | 0.99 |
| QT, ms | 455 ± 67 | 475 ± 76 | 0.013 |
| QTc (Bazett), ms | 483 ± 45 | 511 ± 54 | < 0.001 |
| FH of SD (≤30 years), n (%) | 28 (9) | 21 (19) | 0.008 |
| Bradycardia, n (%) | 60 (19) | 21 (18) | 0.84 |
| Notched T wave, n (%) | 168 (54) | 77 (68) | 0.011 |
| LQTS score | 4.0 ± 1.9 | 4.7 ± 1.9 | < 0.001 |
| AEs, n (%) | 91 (29) | 56 (49) | < 0.001 |
| Age at onset, years | 18 ± 8 | 12 ± 9 | < 0.001 |
| ȃSyncope, n (%) | 77 (25) | 45 (39) | 0.003 |
| ȃAge at syncope, years | 18 ± 8 | 13 ± 7 | 0.001 |
| ȃLethal AE, n (%) | 43 (14) | 27 (23) | 0.015 |
| ȃAge at lethal AE, years | 22 ± 8 | 16 ± 13 | 0.024 |
| pHI-group | pDN-group | P-value | |
|---|---|---|---|
| n | 314 | 115 | |
| Proband, n (%) | 162 (52) | 72 (63) | 0.042 |
| Female, n (%) | 182 (58) | 73 (64) | 0.30 |
| Age at genotyping, years | 27 ± 19 | 27 ± 21 | 0.99 |
| QT, ms | 455 ± 67 | 475 ± 76 | 0.013 |
| QTc (Bazett), ms | 483 ± 45 | 511 ± 54 | < 0.001 |
| FH of SD (≤30 years), n (%) | 28 (9) | 21 (19) | 0.008 |
| Bradycardia, n (%) | 60 (19) | 21 (18) | 0.84 |
| Notched T wave, n (%) | 168 (54) | 77 (68) | 0.011 |
| LQTS score | 4.0 ± 1.9 | 4.7 ± 1.9 | < 0.001 |
| AEs, n (%) | 91 (29) | 56 (49) | < 0.001 |
| Age at onset, years | 18 ± 8 | 12 ± 9 | < 0.001 |
| ȃSyncope, n (%) | 77 (25) | 45 (39) | 0.003 |
| ȃAge at syncope, years | 18 ± 8 | 13 ± 7 | 0.001 |
| ȃLethal AE, n (%) | 43 (14) | 27 (23) | 0.015 |
| ȃAge at lethal AE, years | 22 ± 8 | 16 ± 13 | 0.024 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; FH, family history; LQTS, long QT syndrome; pDN, predicted dominant-negative; pHI, predicted haploinsufficiency; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
Two groups of variant carriers based on functional prediction
| pHI-group | pDN-group | P-value | |
|---|---|---|---|
| n | 314 | 115 | |
| Proband, n (%) | 162 (52) | 72 (63) | 0.042 |
| Female, n (%) | 182 (58) | 73 (64) | 0.30 |
| Age at genotyping, years | 27 ± 19 | 27 ± 21 | 0.99 |
| QT, ms | 455 ± 67 | 475 ± 76 | 0.013 |
| QTc (Bazett), ms | 483 ± 45 | 511 ± 54 | < 0.001 |
| FH of SD (≤30 years), n (%) | 28 (9) | 21 (19) | 0.008 |
| Bradycardia, n (%) | 60 (19) | 21 (18) | 0.84 |
| Notched T wave, n (%) | 168 (54) | 77 (68) | 0.011 |
| LQTS score | 4.0 ± 1.9 | 4.7 ± 1.9 | < 0.001 |
| AEs, n (%) | 91 (29) | 56 (49) | < 0.001 |
| Age at onset, years | 18 ± 8 | 12 ± 9 | < 0.001 |
| ȃSyncope, n (%) | 77 (25) | 45 (39) | 0.003 |
| ȃAge at syncope, years | 18 ± 8 | 13 ± 7 | 0.001 |
| ȃLethal AE, n (%) | 43 (14) | 27 (23) | 0.015 |
| ȃAge at lethal AE, years | 22 ± 8 | 16 ± 13 | 0.024 |
| pHI-group | pDN-group | P-value | |
|---|---|---|---|
| n | 314 | 115 | |
| Proband, n (%) | 162 (52) | 72 (63) | 0.042 |
| Female, n (%) | 182 (58) | 73 (64) | 0.30 |
| Age at genotyping, years | 27 ± 19 | 27 ± 21 | 0.99 |
| QT, ms | 455 ± 67 | 475 ± 76 | 0.013 |
| QTc (Bazett), ms | 483 ± 45 | 511 ± 54 | < 0.001 |
| FH of SD (≤30 years), n (%) | 28 (9) | 21 (19) | 0.008 |
| Bradycardia, n (%) | 60 (19) | 21 (18) | 0.84 |
| Notched T wave, n (%) | 168 (54) | 77 (68) | 0.011 |
| LQTS score | 4.0 ± 1.9 | 4.7 ± 1.9 | < 0.001 |
| AEs, n (%) | 91 (29) | 56 (49) | < 0.001 |
| Age at onset, years | 18 ± 8 | 12 ± 9 | < 0.001 |
| ȃSyncope, n (%) | 77 (25) | 45 (39) | 0.003 |
| ȃAge at syncope, years | 18 ± 8 | 13 ± 7 | 0.001 |
| ȃLethal AE, n (%) | 43 (14) | 27 (23) | 0.015 |
| ȃAge at lethal AE, years | 22 ± 8 | 16 ± 13 | 0.024 |
Data are expressed as mean ± SD or number (%) unless stated otherwise.
AE, arrhythmia event; FH, family history; LQTS, long QT syndrome; pDN, predicted dominant-negative; pHI, predicted haploinsufficiency; QT, QT interval; QTc, corrected QT interval; SD, sudden death.
Among patients with AEs (91 patients in pHI-group and 56 in pDN-group), there was no difference in medical therapy between groups. Recurrent arrhythmia event rate under medication (on-drug) was similar between groups (11% and 18%, respectively; P = 0.24) (see Supplementary material online, Table S3).
Risk stratification based on predicted functional changes
Finally, we evaluated variables by the Cox regression hazard model analysis. As previously reported,10,27,29 sex effect on AEs was not observed before age 12 but was clearly discernible after age 12 in our dataset (see Supplementary material online, Figure S1). We conducted the time-dependent Cox regression analysis by setting sex as a time-dependent covariate.
In Table 4, pDN-group showed significantly higher lifetime risk of AEs than pHI-group [hazard ratio (HR): 2.05, 95% confidence interval (CI): 1.25–3.38, P = 0.005], whereas the risk in variant types (missense vs. non-missense variants) was not significantly different (HR: 0.92, 95% CI: 0.55–1.55, P = 0.75). Proband status (HR: 7.02, 95% CI: 3.63–13.6, P < 0.001), females older than 12-year-old (HR: 2.33, 95% CI: 1.23–4.40, P = 0.009), severely prolonged QTc (≥480 ms) (HR: 1.68, 95% CI: 1.03–2.73, P = 0.039) and positive notched T wave (HR: 1.56, 95% CI: 1.02–2.41, P = 0.043) were also risk variables for lifetime AEs.
Univariable and multivariable cox regression analysis for AEs
| Univariable | Multivariable | |||
|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Status | ||||
| ȃProband vs. family | 11.2 (6.47–19.3) | < 0.001 | 7.02 (3.63–13.6) | < 0.001 |
| Sex (< 12 years) | ||||
| ȃMale vs. female | 1.69 (0.88–3.23) | 0.11 | 1.93 (0.95–3.92) | 0.069 |
| Sex (≥ 12 years) | ||||
| ȃFemale vs. male | 3.84 (2.17–6.78) | < 0.001 | 2.33 (1.23–4.40) | 0.009 |
| QTc | ||||
| ȃ≥ 480 vs. < 480 ms | 4.16 (2.66–6.52) | < 0.001 | 1.68 (1.03–2.73) | 0.039 |
| FH of SD | ||||
| ȃYes vs. no | 1.15 (0.70–1.89) | 0.59 | 0.73 (0.40–1.31) | 0.29 |
| Bradycardia | ||||
| ȃYes vs. no | 1.49 (1.01–2.20) | 0.044 | 1.19 (0.75–1.88) | 0.46 |
| Notched T wave | ||||
| ȃYes vs. no | 1.79 (1.24–2.58) | 0.002 | 1.56 (1.02–2.41) | 0.043 |
| Variant type | ||||
| ȃMissense vs. non-missense | 1.65 (1.15–2.37) | 0.007 | 0.92 (0.55–1.55) | 0.75 |
| Predicted functional change | ||||
| ȃpDN-group vs. pHI-group | 2.09 (1.46–2.98) | < 0.001 | 2.05 (1.25–3.38) | 0.005 |
| Univariable | Multivariable | |||
|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Status | ||||
| ȃProband vs. family | 11.2 (6.47–19.3) | < 0.001 | 7.02 (3.63–13.6) | < 0.001 |
| Sex (< 12 years) | ||||
| ȃMale vs. female | 1.69 (0.88–3.23) | 0.11 | 1.93 (0.95–3.92) | 0.069 |
| Sex (≥ 12 years) | ||||
| ȃFemale vs. male | 3.84 (2.17–6.78) | < 0.001 | 2.33 (1.23–4.40) | 0.009 |
| QTc | ||||
| ȃ≥ 480 vs. < 480 ms | 4.16 (2.66–6.52) | < 0.001 | 1.68 (1.03–2.73) | 0.039 |
| FH of SD | ||||
| ȃYes vs. no | 1.15 (0.70–1.89) | 0.59 | 0.73 (0.40–1.31) | 0.29 |
| Bradycardia | ||||
| ȃYes vs. no | 1.49 (1.01–2.20) | 0.044 | 1.19 (0.75–1.88) | 0.46 |
| Notched T wave | ||||
| ȃYes vs. no | 1.79 (1.24–2.58) | 0.002 | 1.56 (1.02–2.41) | 0.043 |
| Variant type | ||||
| ȃMissense vs. non-missense | 1.65 (1.15–2.37) | 0.007 | 0.92 (0.55–1.55) | 0.75 |
| Predicted functional change | ||||
| ȃpDN-group vs. pHI-group | 2.09 (1.46–2.98) | < 0.001 | 2.05 (1.25–3.38) | 0.005 |
Multivariable model includes all variables simultaneously.
AEs, arrhythmia events; CI, confidence interval; FH, family history; HR, hazard ratio; pDN, predicted dominant-negative; pHI, predicted haploinsufficiency; QTc, corrected QT interval; SD, sudden death.
Univariable and multivariable cox regression analysis for AEs
| Univariable | Multivariable | |||
|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Status | ||||
| ȃProband vs. family | 11.2 (6.47–19.3) | < 0.001 | 7.02 (3.63–13.6) | < 0.001 |
| Sex (< 12 years) | ||||
| ȃMale vs. female | 1.69 (0.88–3.23) | 0.11 | 1.93 (0.95–3.92) | 0.069 |
| Sex (≥ 12 years) | ||||
| ȃFemale vs. male | 3.84 (2.17–6.78) | < 0.001 | 2.33 (1.23–4.40) | 0.009 |
| QTc | ||||
| ȃ≥ 480 vs. < 480 ms | 4.16 (2.66–6.52) | < 0.001 | 1.68 (1.03–2.73) | 0.039 |
| FH of SD | ||||
| ȃYes vs. no | 1.15 (0.70–1.89) | 0.59 | 0.73 (0.40–1.31) | 0.29 |
| Bradycardia | ||||
| ȃYes vs. no | 1.49 (1.01–2.20) | 0.044 | 1.19 (0.75–1.88) | 0.46 |
| Notched T wave | ||||
| ȃYes vs. no | 1.79 (1.24–2.58) | 0.002 | 1.56 (1.02–2.41) | 0.043 |
| Variant type | ||||
| ȃMissense vs. non-missense | 1.65 (1.15–2.37) | 0.007 | 0.92 (0.55–1.55) | 0.75 |
| Predicted functional change | ||||
| ȃpDN-group vs. pHI-group | 2.09 (1.46–2.98) | < 0.001 | 2.05 (1.25–3.38) | 0.005 |
| Univariable | Multivariable | |||
|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Status | ||||
| ȃProband vs. family | 11.2 (6.47–19.3) | < 0.001 | 7.02 (3.63–13.6) | < 0.001 |
| Sex (< 12 years) | ||||
| ȃMale vs. female | 1.69 (0.88–3.23) | 0.11 | 1.93 (0.95–3.92) | 0.069 |
| Sex (≥ 12 years) | ||||
| ȃFemale vs. male | 3.84 (2.17–6.78) | < 0.001 | 2.33 (1.23–4.40) | 0.009 |
| QTc | ||||
| ȃ≥ 480 vs. < 480 ms | 4.16 (2.66–6.52) | < 0.001 | 1.68 (1.03–2.73) | 0.039 |
| FH of SD | ||||
| ȃYes vs. no | 1.15 (0.70–1.89) | 0.59 | 0.73 (0.40–1.31) | 0.29 |
| Bradycardia | ||||
| ȃYes vs. no | 1.49 (1.01–2.20) | 0.044 | 1.19 (0.75–1.88) | 0.46 |
| Notched T wave | ||||
| ȃYes vs. no | 1.79 (1.24–2.58) | 0.002 | 1.56 (1.02–2.41) | 0.043 |
| Variant type | ||||
| ȃMissense vs. non-missense | 1.65 (1.15–2.37) | 0.007 | 0.92 (0.55–1.55) | 0.75 |
| Predicted functional change | ||||
| ȃpDN-group vs. pHI-group | 2.09 (1.46–2.98) | < 0.001 | 2.05 (1.25–3.38) | 0.005 |
Multivariable model includes all variables simultaneously.
AEs, arrhythmia events; CI, confidence interval; FH, family history; HR, hazard ratio; pDN, predicted dominant-negative; pHI, predicted haploinsufficiency; QTc, corrected QT interval; SD, sudden death.
Discussion
In 429 patients with LQT2, the present study demonstrated that KCNH2 non-missense variant carriers showed shorter QTc and lower risk for cardiac events compared to those with missense variants which included various functional effects on Kv11.1 (Figure 2A). According to previously reported functional analyses, 40% of our missense variants could be classified either into HI- or DN-groups. Non-missense and HI variant groups showed a significantly better prognosis than the DN-group (Figure 2B). Finally, prediction of functional changes induced by remaining missense variants (60%) was made based on their location. Because of NMD, non-missense variants were classified as pHI-group. Such a prediction showed that pHI presented a significantly better prognosis than pDN (Figure 2C). The classification of variants according to the molecular mechanism, not variant types, contrasted the difference in phenotypes, thereby predicting the prognosis in LQT2 patients more clearly. Subsequent multivariable analyses revealed that significant risk factors are: proband status, female sex older than the age of 12 years, QTc ≥480 ms, positive notched T wave, and pDN-group.
Functional analyses are thus helpful for the understanding of LQTS mechanisms. However, functional assay of non-missense variants is technically difficult because they are postulated to undergo NMD.7,8 Although these variants are thought to be relatively benign compared to DN variants, there were no systematic analyses covering this. Indeed, using the international LQTS registry, Moss et al.15 (2002) examined 44 KCNH2 variants according to the location, irrespective of variants types. They did not analyse non-missense and missense variants separately. Using the same and expanded registry, Shimizu et al.10 (2009) showed that there was no significant difference in event rates between missense and non-missense variants, which contrasts with our finding (Figure 2A) that missense patients showed a worse prognosis. This difference may be driven by the different study protocol by which patients were censored at the time when pharmacologic therapy was initiated in our study whereas the previous study did not. This difference may partially result from the inclusion of more patients with milder phenotypes in our cohort. Specifically, our cohort included school-age children who took routine ECG check-ups in Japan.30 The school-based ECG check-up can identify various variants including non-missense variants (n = 190, 44%) and therefore made a sharp contrast with overall missense carriers that contained both HI and DN variants.
The loss-of-function type of KCNH2 variants may cause LQT2 by four different mechanisms: abnormal transcription or translation of Kv11.1 (Class 1); mis-trafficking (Class 2); disrupting channel gating or kinetics (Class 3); or altering ion selectivity or permeability (Class 4).7,31 PD variants reportedly induce DN effects by interfering with WT proteins through endoplasmic reticulum-associated degradation, leading to worse outcomes.16,32,33,34 An in-vitro study revealed that 88% of LQT2-associated missense variants showed loss-of-function by a Class 2 mechanism.17 In their study, among mis-trafficking variants, 76% of PD variants exerted DN effects, while all (100%) of EAGD, C-linker, and CNBHD variants produced HI effects, suggesting that they were degraded with or without WT subunits, revealing that the PD-mediated DN was caused through Class 2 rather than Class 3 or 4 mechanisms. In contrast, other missense variants may cause HI by Class 2 mechanism without affecting the transport of WT subunit.
In this study with many LQT2 patients, the prediction of loss-of-function mechanisms by functional domains (Figure 2C) was found to be helpful for assessing the prognosis. These dominant functional effects in each domain (Figure 3) showed accordance with their severity of clinical characteristics (see Supplementary material online, Table S4). Furthermore, our data support for the first time that trans-membrane domains including not only PD but VSD presented poorer clinical outcomes than others. In the literature, 12 of 18 previously studied variants in the VSD exerted a DN phenotype (Figure 3), and in our cohort, 15 carriers with variants in the VSD showed severe phenotypes (see Supplementary material online, Table S4).
Study limitations
The functional consequences of identified variants in this study were limited in the literature and were not fully evaluated, including non-missense variants that may escape NMD. This may result in discordance between predicted functional changes and in vitro findings. Our binary classification of HI vs. DN status may fail to fully reflect the underlying complex properties of genetic variants. Further validation of this new approach to verify our findings in different cohorts is necessary.
Conclusion
We identified, for the first time, that the clinical phenotypes of both non-missense variants and known HI missense variants of KCNH2 exhibited better outcomes than those with variants that are known to cause DN effects. Risk stratification supplemented by prediction of functional consequence revealed that HI-causing variants have better arrhythmia outcomes than DN-causing variants. Our study highlights the advantage of combining molecular biology and clinical phenotypes to bridge the knowledge gap between biological consequence of the non-missense variants and clinical outcomes in LQT2 patients.
Supplementary material
Supplementary material is available at Europace online.
Acknowledgements
We thank Ms. Arisa Ikeda, Ms. Kazu Toyooka, and Ms. Madoka Tanimoto for their contributions to genetic testing.
Funding
This work was supported by a grant from the Ministry of Health, Labor and Welfare of Japan for Clinical Research on Intractable Disease (H27-032 and H29-055 to M.H., T.M., S.O.), MEXT KAKENHI (grant numbers: 15H04818 to M.H., 19K08538 to T.M., 18K07875 to S.O.), Heart Rhythm Society (Clinical Research Award in Honor of Mark Josephson and Hein Wellens to Y.W.), and American Heart Association postdoctoral fellowship (grant no. 830951 to Y.W.).
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
References
Author notes
Conflict of interest: None declared.
