Genetic Architecture of Dilated Cardiomyopathy in Individuals of African and European Ancestry

Abstract

Importance: Black patients with dilated cardiomyopathy (DCM) have increased familial risk and worse outcomes than White patients, but most DCM genetic data are from White patients.

Objective: To compare the rare variant genetic architecture of DCM by genomic ancestry within a diverse population of patients with DCM.

Design: Cross-sectional study enrolling patients with DCM who self-identified as non-Hispanic Black, Hispanic, or non-Hispanic White from June 7, 2016, to March 15, 2020, at 25 US advanced heart failure programs. Variants in 36 DCM genes were adjudicated as pathogenic, likely pathogenic, or of uncertain significance.

Exposure: Presence of DCM.

Main outcomes and measures: Variants in DCM genes classified as pathogenic/likely pathogenic/uncertain significance and clinically actionable (pathogenic/likely pathogenic).

Results: A total of 505, 667, and 26 patients with DCM of predominantly African, European, or Native American genomic ancestry, respectively, were included. Compared with patients of European ancestry, a lower percentage of patients of African ancestry had clinically actionable variants (8.2% [95% CI, 5.2%-11.1%] vs 25.5% [95% CI, 21.3%-29.6%]), reflecting the lower odds of a clinically actionable variant for those with any pathogenic variant/likely pathogenic variant/variant of uncertain significance (odds ratio, 0.25 [95% CI, 0.17-0.37]). On average, patients of African ancestry had fewer clinically actionable variants in TTN (difference, -0.09 [95% CI, -0.14 to -0.05]) and other genes with predicted loss of function as a disease-causing mechanism (difference, -0.06 [95% CI, -0.11 to -0.02]). However, the number of pathogenic variants/likely pathogenic variants/variants of uncertain significance was more comparable between ancestry groups (difference, -0.07 [95% CI, -0.22 to 0.09]) due to a larger number of non-TTN non-predicted loss of function variants of uncertain significance, mostly missense, in patients of African ancestry (difference, 0.15 [95% CI, 0.00-0.30]). Published clinical case-based evidence supporting pathogenicity was less available for variants found only in patients of African ancestry (P < .001).

Conclusion and relevance: Patients of African ancestry with DCM were less likely to have clinically actionable variants in DCM genes than those of European ancestry due to differences in genetic architecture and a lack of representation of African ancestry in clinical data sets.

Figure.. Characteristics of Pathogenic Variants, Likely Pathogenic Variants, and Variants of Uncertain Significance Observed in DCM Probands of a Given Ancestry

Variants denoted as being from African (n = 345 variants), European (n = 521), or Native American (n = 22) ancestry were observed only in study probands from that ancestry group and not in probands from other ancestry groups. For all panels, the y-axis is the proportion of variants from a particular ancestry with a given characteristic; underlying data are shown in eTables 4 and 5 in Supplement 1. For panels A-C, bars are mutually exclusive categories; proportions in each bar sum to 1 within each ancestry. For panels D-E, each bar is the proportion of variants from an ancestry meeting at least 1 of the variant interpretation criteria that provide evidence of pathogenicity in a particular domain, as denoted by American College of Medical Genetics acronyms listed in parentheses in the key (see definitions in eTable 4 in Supplement 1). As these domains are independent, space between bars indicates that the bars do not sum to 1 within each ancestry. Variant-based criteria include those in the population frequency, null variant in gene/region, and in silico prediction domains that relate to potential deleteriousness of the variant in general. Case-based criteria include those in the segregation and case counts or reported association domains, which relate to the availability of external data showing that the variant is associated specifically with dilated cardiomyopathy (DCM). A lower proportion of variants found only in probands of African ancestry were in definitive and strong evidence DCM genes (panel A) based on ClinGen classification or had predicted loss-of-function (pLOF) consequences (panel B) relative to those found only in probands of European ancestry (post hoc P < .001 for both). As a result, a lower proportion of variants found only in probands of African ancestry were classified as pathogenic/likely pathogenic compared with those found only in probands of European ancestry (panel C; post hoc P < .001). A reduced ability to apply variant-based (panel D) or case-based (panel E) criteria in favor of pathogenicity to variants in probands of African ancestry contributed to this outcome. Compared with variants found only in European ancestry, smaller proportions of variants found in African ancestry were absent or rare in nonfounder gnomAD populations (post hoc P < .001) or had evidence for pathogenicity from case counts or reported associations (post hoc P < .001). Inferences for the predicted impact distribution in panel B were based on the predicted impact multinomial variable, which did not distinguish TTN and non-TTN pLOF variants to avoid confounding gene class with variant impact. For variant-based criteria in panel D, inferences were based on the population frequency domain, null variant in gene/region, and in silico prediction domain multinomial variables. Inferences for case-based criteria in panel E were based on the segregation domain and case counts or reported association domain multinomial variables.