Utility of genetics for risk stratification in pediatric hypertrophic cardiomyopathy

Children with hypertrophic cardiomyopathy (HCM) experience sudden cardiac death (SCD) and other life‐threatening events. We assessed if affected gene and variant burden predict outcomes. Patients <18 years old with primary HCM with a pathogenic variant or variant of uncertain significance in cardiomyopathy genes were included. Association of gene and variant number and type with freedom from major adverse cardiac events (MACE), that is, ICD insertion, myectomy, aborted SCD, transplantation or death, was assessed by Cox regression. A total of 98 of 155 gene‐tested patients carried a non‐benign variant. The primary affected gene was MYH7 in 35% (MYH7+) and MYBPC3 in 49% (MYBPC3+). MYH7+ patients had earlier disease onset and higher risk of MACE (hazard ratio 2.7, 95% CI 1.3‐5.7). Risk of MACE was also higher in patients with multiple variants (n = 16) (HR 2.5, CI: 1.1‐5.9) compared to a propensity score‐matched single variant subset, after adjustment for primary gene, and in patients with de novo (n = 18) vs inherited variants (HR 5.7, CI: 2.6‐12.7). Affected gene (eg, MYH7), higher variant burden and de novo variant status are independently associated with earlier onset and higher frequency of adverse outcomes in pediatric HCM, highlighting the importance of genetic risk stratification in HCM.


| INTRODUCTION
The description of familial hypertrophic cardiomyopathy (HCM) due to a missense mutation in the β-myosin heavy chain in 1990 marked the initial association of sarcomeric mutations with HCM. 1 Everbroadening gene panels can now identify a genetic cause in up to 63% of familial HCM cases [2][3][4] and 50% to 60% of non-familial cases. 5,6 At least 18 genes have been associated with HCM, with varying levels of evidence for pathogenicity. 7 Hitherto, such genetic data have been used primarily for cascade or predictive screening of at-risk relatives after the causative lesion is identified in a HCM proband, rather than to inform prognosis or guide management in the affected individual. 8,9 There is a high incidence of sudden cardiac death (SCD) and other complications in HCM. Decision making regarding prevention of SCD is driven entirely by clinical and echocardiographic risk factors rather than genetic factors, as encapsulated in the recent HCM SCD risk calculator in adults published by the European Society of Cardiology in 2014. 10 There are scant data describing the association between genetic etiology and outcomes in HCM, especially in children. Adult studies have reported a more severe phenotype and earlier onset of ventricular hypertrophy in patients with MYH7 mutations compared to those with other mutations albeit the data are inconsistent. 3,4,8,[11][12][13][14] Adult studies have also reported earlier presentation, more severe hypertrophy and higher rates of myectomy and ICD implantation in those with multiple mutations. 9,15,16 Pediatric studies have been limited to case reports 17,18 with no systematic evaluation of influence of gene type and mutation number on outcomes. Due to perceived greater risk of adverse outcomes in children with early onset HCM, knowledge of genetic predictors is critical to facilitate timely interventions before the onset of these adverse events.
This study aimed to define the association of (1) the gene involved, and (2) the number of variants with clinical outcomes in a longitudinally followed pediatric HCM cohort. or end-diastolic IVSD:LVPWD ratio greater than 1. 5  guidelines. 20 In applying these guidelines, the following definitions and tools were used. Prior report of variants was determined by review of institutional experience, testing company report, the literature, ClinVar and Human Gene Mutation databases. [21][22][23] Rare variants were those with a prevalence of ≤0.1% in all reference populations (Exome Variant Server, Exome Aggregation Consortium).
Variants were considered previously reported if reported ≥2 times and were considered to co-segregate with HCM if they did so with ≥3 phenotype-positive individuals in ≥2 affected kindreds in 1 or more of these sources.

| Statistical analysis
The composite primary endpoint was freedom from major adverse cardiac events (MACE) defined as implantable defibrillator-cardioverter (ICD) implantation, surgical myectomy, resuscitated cardiac arrest, transplantation or death. Secondary endpoints included freedom from each of the constituent MACE endpoints, freedom from severe (IVSD or LVPWD z-score ≥ 5) or massive hypertrophy (IVSD or LVPWD zscore ≥ 10), freedom from occurrence of any LV outflow tract obstruction (LVOTO; LVOT peak gradient ≥50 mmHg), as well as freedom from severe LVOTO (LVOT peak gradient ≥100 mmHg).
Data were reported as medians with interquartile ranges (IQR) or frequencies with percentages as appropriate. Intergroup differences were assessed with the Kruskal-Wallis test and Fisher exact tests.
Timed endpoints were modeled using the Kaplan-Meier method using univariable Cox regression to compare cumulative event rates between groups. The proportional hazards assumption was verified by applying a test based on Schoenfeld-residuals and by visual inspection of Schoenfeld-residual plots. Patients with multiple variants were propensity score-matched to those with single variants by family history, maximal wall-dimension z-score at presentation and use of panel testing (to adjust for otherwise unmeasured confounders), in 2:1 ratio without replacement, using an optimal matching algorithm (minimizing the sum of pair-wise distances).
Multivariable analysis was performed on this matched subset by Cox proportional hazards modeling with use of a robust variance estimator to account for matching. 29 Statistical tests were 2-tailed and P values of less than .05 were considered statistically significant.

| Patient characteristics
During the study period, 507 patients were seen in our institutional cardiomyopathy clinic for screening due to a positive family history of HCM (n = 423; 83%), or symptoms or findings in the patient leading to a diagnosis of HCM (n = 84, 17%). The breakdown of this cohort is depicted in Figure S1 (Supporting information). Genetic testing was performed in 155 patients, yielding 98 patients from 76 kindreds who had 1 or more non-benign variants, and who constitute the study cohort. Patient characteristics are shown in Table S1. The following gene panels were used: HCM 5 gene (n = 22), HCM 11 gene (n = 32), HCM 18 gene (n = 8) and expanded panels with 46 to 62 genes (n = 8). Panel-tested patients were less likely to have a family history of HCM and were more frequently phenotype positive at initial encounter, with greater LV wall and LA dimensions than those who underwent variant-specific testing.

| Association of genotype with phenotype
Patients in the overall cohort became phenotype positive at a median age of 12.4 years (IQR 5.7; 15.2). On subgroup analysis, MYH7+ patients were phenotype positive earlier at a median age of 9.0 years (IQR 5.2; 14.1) compared to 13 years (IQR 7.0; 15.8) for those with variants in other genes (see Table 1), and were more likely to have a missense variant than those with primary variants in other genes. Table 1 describes baseline clinical characteristics by the primary gene involved and by single vs multiple non-benign variants. The latter were more likely to be phenotype positive at presentation, to have greater wall thickness z-scores at presentation and to have undergone panel genetic testing. Patients with multiple variants were phenotype positive at a younger median age of 8.9 years (IQR 3.1; 11.2) compared to 13.1 years (IQR 6.9; 15.8) years in single variant patients. Patients with multiple variants were less likely to have presented for screening due to a family history (38% vs 70%, P = .009). There was no difference in age of presentation in those with de novo vs inherited variants. Median (IQR) age at presentation in those with de novo variants was 5.9 (1.4-13.1) and 6.3 (3.0-10.5) in those with inherited variants (P = .75).

| Myectomy
Eight patients underwent myectomy at a median age of 8.     Figure 4). This association remained significant when analysis was limited to the 48 propensity score-matched subset described above (Table 3).

| Genotype-phenotype association in HCM
Attempts at genotype-phenotype correlation in HCM have suffered from tremendous genetic heterogeneity, with most kindreds harboring private variants 34 and with few reported founder variants. 35,36 Uncontrolled early series emphasized individual variants and noted earlier presentation and a high incidence of early mortality in kindreds with lesions such as R719W, 37 G716R, 38 R403Q, 39   Chest pain Another large report found multiple variants in 2.6% and were associated with a significantly earlier age at presentation, greater wall thickness during cross-sectional assessment and greater incidence of myectomy. 15 Our observation that the presence of multiple variants influences outcomes, despite most secondary variants being considered VUS, would support that variant burden matters in disease outcomes.
We believe our findings are clinically significant since they not only identify a genetically vulnerable subset within the larger cohort of childhood onset HCM but also highlight the young age at which adverse events start occurring in this high-risk subset. About 15% patients in the overall cohort had a MACE by 10 years of age. These findings have the potential to change when and how clinical and genetic surveillance is conducted in at risk individuals. Current American Heart Association clinical guidelines for HCM recommend that clinical and genetic testing be initiated at the age of 12 years in children who are deemed at risk due to family history of HCM. 46 Our findings suggest that surveillance should not be delayed till 12 years of age since many patients, especially those with a high risk genetic profile, suffer adverse events or require major cardiac interventions before the age of 12 years. This subset may therefore benefit from more frequent surveillance starting at an earlier age which may improve timeliness of interventions like myectomy or ICD insertion prior to onset of lethal or life-threatening cardiac events. It also argues for studies to specifically evaluate how genetic etiology can be incorporated into SCD risk prediction in early onset HCM.

| Study Limitations
Genetic testing panels changed during the study period which can influence genetic yield. However, with the exception of TTN, the most frequently represented genes were included on testing panels throughout the study period. Though we adjusted for non-random utilization of panel genetic testing using propensity score-matching, there may be unmeasured confounders that associate with outcomes. Finally, based on our sample size, we were not powered to compare the association of individual genes but compared MYH7 with non-MYH7 positive cases. Further studies are needed to explore the association of individual genes with phenotype and outcomes in HCM.

| CONCLUSION
This study finds an important contribution of the primary gene, of variant burden and of variant type (de novo) to risk of adverse events in pediatric HCM. This highlights the importance of incorporating genetic findings into decision making regarding age and frequency of clinical surveillance for HCM and timing of interventions like ICD or myectomy in genetically high risk patients.