Retrospective Analysis of Clinical Genetic Testing in Pediatric Primary Dilated Cardiomyopathy: Testing Outcomes and the Effects of Variant Reclassification

Patients were retrospectively identified by querying records between January 1, 2008, and January 1, 2018, for the diagnostic Fyler code for DCM. In general, this code is assigned at our institution on the basis of the presence of structural characteristics of DCM, a dilated left ventricle with systolic dysfunction. Patients with secondary DCM or incomplete clinical documentation were excluded. Three patients with borderline inflammation on endomyocardial biopsy who were subsequently diagnosed with DCM were included (Figure S1). Positive family history was defined as the presence of cardiomyopathy in a first‐degree relative or multiple (≥2) second‐ and third‐degree relatives with cardiomyopathy or sudden cardiac death. All functional cardiac data were calculated from the patient's echocardiogram at the time of initial presentation. Ejection fraction was calculated using the 5/6 area‐length method.

Clinical cardiomyopathy genetic testing was performed at the Partners Healthcare Laboratory for Molecular Medicine (Cambridge, MA) as previously described.¹¹ The number of genes sequenced per patient and sequencing methodology differed over the course of the study period. There was an expansion in the number of genes included in the testing panel from 5 to 62, and 23 patients underwent testing with a next‐generation sequencing–based panel composed of 62 genes associated with cardiomyopathy (Table S1). Variants were classified at the time of initial testing as previously described.^{11, 22} Given significant changes in cardiomyopathy testing panel content, revisions in guidelines for variant classification, and an increase in the availability of reference population allele databases over the 10‐year study period²³ variants were reclassified in accordance with the American College of Medical Genetics and Genomics and the Association for Molecular Pathology Guidelines²⁴ and additional specifications for cardiovascular genes as established by the Clinical Genome inherited cardiomyopathy expert panel.²⁵ Gene‐specific testing was performed in a subset of parents and siblings of probands for segregation analysis.

Continuous variables are presented as medians with ranges, and a Wilcoxon rank sum test was used for comparison. Categorical variables are displayed as frequencies and percentages, and Fisher's exact test was used for statistical comparison. Survival analyses were performed with the Kaplan‐Meier method and were divided according to age (<1 year and >1 year) and the presence or absence of a DCM‐causing genetic variant. For transplant‐free survival analysis, end points of heart transplantation or death were used and subjects were censored at the time of last follow‐up. For transplant‐censored survival analysis, a single end point of death was used, and patients were censored at the time of heart transplantation or time of last follow‐up. Pairwise comparison of survival curves was performed with a log‐rank test. Multivariate Cox proportional hazard modeling was performed to determine the association between survival time and covariates of sex, age, and presence of a DCM‐causing variant. A P<0.05 was considered statistically significant. Statistical analysis was performed in R 3.5.1.

All data collection, storage, and analysis was approved by the Institutional Review Board at Boston Children's Hospital and Partners Healthcare. Requirement for informed consent was waived by the Institutional Review Board.

We identified 118 patients with DCM who were evaluated at our tertiary care center over a 10‐year period (Figure S1). The etiologies of DCM in this cohort were similar to those described in larger case series,⁴ with 80 patients having familial (12%) or idiopathic (56%) DCM and the remainder of cases caused by neuromuscular disease (6%), acute myocarditis (8%), metabolic disorder (10%), and genetic syndrome (8%) (Figure S1). Of the 80 patients who presented with primary DCM, 73 were probands with no prior genetic evaluation, and 63 (86%) of these patients underwent cardiomyopathy‐specific gene testing as part of their diagnostic evaluation. The median age at the time of presentation for probands that underwent genetic testing was 6.5 years (range, 0–22) and 37 of 63 (59%) were male (Table 1). There was a bimodal distribution of age at the time of presentation, with 35% of patients presenting before the age of 1 year, and a second group of patients presenting between the ages of 13 and 19 years. Eleven of 60 (18%) probands with available family history data had a positive family history of cardiomyopathy or sudden death, and all patients with a positive family history presented after 1 year of age (P=0.005). The median left ventricular ejection fraction at the time of presentation was 26% (range, 8–65) and tended to be lower in those patients presenting under 1 year of age (23% versus 29.5%; P=0.06). The median left ventricular end‐diastolic volume z score at the time of presentation was +5.85 (range, −0.1 to +21.1) and did not differ between age groups (Table 1). One individual presented with a decreased ejection fraction without ventricular dilation, and another individual with ventricular dilation with a normal ejection fraction, both of whom went on to have DCM.

Cardiomyopathy gene panel testing of 63 probands resulted in the identification of 116 variants that were classified at the initial time of testing as follows: 8 pathogenic, 11 likely pathogenic (LP), 90 variants of unknown significance (VUSs), 3 likely benign (LB), 2 benign, and 2 unclassified on the basis of their position in untranslated regions (Table 2). The burden of VUSs per patient correlated with the size of the testing panel as previously described (Table S2).^{11, 18} Following variant reevaluation for this study, the 116 variants were classified as 7 pathogenic, 8 LP, 68 VUSs (of which 6 are VUS‐favor pathogenic), 22 LB, and 11 benign (Table 2). The largest shift in variant classification following reevaluation was in the VUS category, where 26 of 90 (29%) of the original VUSs were reclassified as LB or benign (Table 2). The next largest change in classification occurred in variants previously designated as LP, with 4 being downgraded to VUS‐favor pathogenic and 1 being reclassified as pathogenic. No VUS, LB, or benign variants were advanced to a higher‐grade category upon reclassification. These changes in variant classification led to a decrease in the number of “inconclusive” test results with VUS (60%–52%), and an increase in the proportion of “negative” tests that identified only LB or benign variants (14%–24%) (Table S2).

Familial testing for 9 probands with a pathogenic or LP variant revealed that 3 variants were inherited and 6 variants were de novo mutations (Table 3). Additionally, 6 probands with a potentially disease‐causing VUS underwent familial testing to help further clarify the variant's clinical significance. Of these 6 VUSs, 3 occurred de novo, and 3 were inherited from individuals without a cardiac structural phenotype at the time of testing.

Pathogenic variants, LP, and VUS‐favoring pathogenic, were identified in 12 genes, including those integral to sarcomere function (MYH7, TTN, TNNT2, TNNI3, MYBPC3, TPM1, BAG3), cellular structure (LMNA, DES), cellular junctions (PKP2, DSP), and lysosomal function (LAMP2) (Table 3). Nearly all mutations were heterozygous and thus demonstrated an autosomal dominant pattern. One proband was found to be hemizygous for a truncating mutation on the X chromosome encoded LAMP2. Nine de novo variants were identified in the following genes: LAMP2, MYH7 (n=3), LMNA, TNNI3, TMP1, and TNNT2 (n=2). Three of these variants in MYH7, TNNT2, and TPM1 were classified as VUS‐favor pathogenic, and within the clinical context were interpreted as causal in their respective cases. No individuals had a LP or pathogenic variant or a VUS‐favor pathogenic variant in >1 gene. Collectively, 19 of 63 (30%) patients with idiopathic or familial DCM were determined to have a disease‐causing variant, with at least 9 of 19 (47%) variants occurring de novo (Table 3). The most frequently identified mutations were in the sarcomeric genes TNNT2 (n=5) and MYH7 (n=3). All patients with mutations in MYH7 presented in infancy, while those with TNNT2 mutations presented at 2 discrete time points, either in early infancy (n=2) or adolescence (n=3). Rates at which disease‐causing mutations were identified were similar between patients with a positive family history (3/11; 27%) and those without a positive family history (16/49; 33%). Additionally, the rate of detection of a causal variant did not differ by patient age (Table 1). The percentage of “positive” genetic tests that yielded a pathogenic or LP variant increased by 1.3‐fold with expansion of the gene testing panel to >50 genes (20% for panels <50 genes versus 27% for >50 genes tested) and was largely unaffected by variant reclassification (Tables S2 and S3).

Patients with DCM who underwent genetic testing had a median clinical follow‐up of 37 months (range, 0–118), and 29 of 63 (46%) patients underwent heart transplantation or died during follow‐up. Echocardiographic measurements of left ventricular size and systolic function did not differ on the basis of the presence or absence of a causal variant in age‐stratified patients (Table 1). Patients presenting after 1 year of age with a disease‐causing variant had decreased transplant‐free survival, with 10 of 11 (91%) patients having a composite outcome of death or transplant versus 13 of 29 (45%) patients >1 year of age without an identified variant (P=0.01) (Table 1 and Figure A).On proportional hazards modeling there was an increased risk of death or transplantation in patients presenting after the first year of life (hazard ratio [HR], 1.2; 95% CI, 1.05–1.3; P=0.007) (Figure B),consistent with previously reported age‐related risk of death or transplantation.⁴ The presence of a disease‐causing variant was not significantly associated with risk of transplantation or death in the overall cohort (HR, 1.9; 95% CI, 0.87–4.2; P=0.11) (Figure B). Transplant‐censored survival was not different between groups, and the observed differences were largely driven by rates of heart transplantation (Figure C).