Common genetic variants and modifiable risk factors underpin hypertrophic cardiomyopathy susceptibility and expressivity

Abstract

Hypertrophic cardiomyopathy (HCM) is a common, serious, genetic heart disorder. Rare pathogenic variants in sarcomere genes cause HCM, but with unexplained phenotypic heterogeneity. Moreover, most patients do not carry such variants. We report a genome-wide association study of 2,780 cases and 47,486 controls that identified 12 genome-wide-significant susceptibility loci for HCM. Single-nucleotide polymorphism heritability indicated a strong polygenic influence, especially for sarcomere-negative HCM (64% of cases; h²_g = 0.34 ± 0.02). A genetic risk score showed substantial influence on the odds of HCM in a validation study, halving the odds in the lowest quintile and doubling them in the highest quintile, and also influenced phenotypic severity in sarcomere variant carriers. Mendelian randomization identified diastolic blood pressure (DBP) as a key modifiable risk factor for sarcomere-negative HCM, with a one standard deviation increase in DBP increasing the HCM risk fourfold. Common variants and modifiable risk factors have important roles in HCM that we suggest will be clinically actionable.

Fig. 1 |. Study design for the HCM genome-wide association analysis.

Two independent HCM GWASs were performed before fixed-effects inverse-variance meta-analysis was conducted. Genetic risk scores (GRSs) were generated and stratified by sarcomere variant status. The findings were validated using three independent cohorts (GeL, RBH and the Netherlands cohort (Amsterdam, Rotterdam and Groningen)). Two-sample Mendelian randomization was performed, stratified by sarcomere variant status, to provide insight into heritable risk factors for HCM. SNP heritability (h²_g) estimates were compared between component GWASs using GReML-LDMS and stratified by sarcomere variant status. Standard errors for h²_g estimates are presented in parentheses.

Fig. 2 |. Validation of an HCM GRS.

A GRS was generated from 27 SNPs with <5% FDR and weighted by the β estimate from the multi-ancestry meta-analysis joint model GCTA results. The GRS was evaluated in all-comers, sarcomere-positive and sarcomere-negative HCM cases, in three validation cohorts. a, A quintile-based analysis demonstrates the protective effects of the GRS in the lowest 20% of the population compared with the middle 60%. Similarly, the upper 20% show increased susceptibility towards a risk of developing HCM compared with the middle 60%. b, To facilitate comparison between other GRSs, a per-standard deviation estimate is reported. In a and b, ORs (x axis) are reported, with error bars denoting 95% CIs. The validation cohorts included GeL (n = 435 HCM cases versus n = 36,500 controls), RBH (n = 359 HCM cases versus n = 1,211 controls) and the Netherlands cohort (n = 975 HCM cases versus n = 2,117 controls).

Fig. 3 |. Relationship between standardized GRS and maximum left ventricular wall thickness.

Linear regression was performed to assess the most frequently observed HCM variant classes: truncating variants in MYBPC3; MYH7 missense variants; and the most frequently observed pathogenic variant (MYPBC3^R502W). a, Carriers of pathogenic or likely pathogenic MYBPC3 truncating variants (n = 232; β = 0.71 ± 0.35; P = 0.048). b, Carriers of pathogenic or likely pathogenic MYH7 missense variants (n = 186; β = 0.73 ± 0.35; P = 0.036). c, Carriers of the most frequently observed pathogenic variant in HCM, MYBPC3^R502W (n = 48; β = 1.61 ± 0.80; P = 0.051) evaluated in HCMR cases (n = 36) and participants from the UKBB (n = 12). Linear regression lines are denoted in blue, with 95% Cls in gray. P values are uncorrected for multiple testing.

Fig. 4 |. Two-sample inverse-variance-weighted Mendelian randomization identifies modifiable risk factors for HCM.

a, Effect of presumed risk phenotypes, based on previous observational evidence, on sarcomere-positive (n = 871) and sarcomere-negative (n = 1,635) HCM. ORs are represented per standard deviation for systolic blood pressure (SBP), DBP, body mass index (BMI) and waist-to-hip ratio adjusted for BMI (WHRadjBMI). The error bars represent 95% CIs. As type 2 diabetes is a binary phenotype, risk is represented as the per log-odds unit of type 2 diabetes. b, Relative impact of DBP on sarcomere-positive and sarcomere-negative HCM susceptibility in relation to other established hypertension-associated phenotypes. The OR was measured per standard deviation of DBP (11.3 mmHg). The error bars represent 95% CIs. Ischemic stroke reflects all TOAST subtypes. Numbers of cases and controls, respectively, were as follows: 47,309 and 930,014 for heart failure; 65,446 and 522,744 for atrial fibrillation; 9,006 and 454,450 for cardioembolic stroke; 60,341 and 454,450 for ischemic stroke; 74,124 and 824,006 for type 2 diabetes; 64,164 and 625,219 for chronic kidney disease; and 122,733 and 424,528 for coronary artery disease.