Uncovering the Genetic Etiology of the (Posttherapy) Broken Heart

The success of childhood cancer treatments is reflected in current 5-year overall survival rates of 80%-85% in the United States (1). However, treatment-related complications can be associated with lifelong chronic disease and cancer risks that contribute to substantial morbidity and mortality in these survivors. Cardiovascular diseases (CVD) are now the most common noncancer-related causes of death in adult survivors of childhood cancers (2) with an estimated cumulative incidence of 4.8% of any CVD 30 years after diagnosis of primary childhood cancer (3).

Treatment-related CVD and associated conditions include cardiomyopathy, ischemic heart disease, valvular disease, conduction disorders, hypertension, dyslipidemia, and diabetes, among others (4,5). These conditions can contribute alone or in combination to the functional, structural, and clinical features defining the syndrome of heart failure (HF) (5). Acute and chronic CVD risks associated with anthracyclines and/or radiotherapy to the chest with mean cardiac doses of 15 Gray or more have been known for decades (4,6,7). Among children exposed to both anthracyclines and chest-directed radiotherapy, 12% develop congestive heart failure (CHF) (8). However, increased CVD risks have also been reported following exposure to alkylators, antimetabolites, vinca alkaloids, platinum agents (4,9), and more recently, targeted and immunomodulatory agents (4). Other factors associated with adult CVD following childhood cancer treatments include cumulative drug dose, younger age at exposure, female sex, African American ancestry, hypertension, dyslipidemia, and diabetes (10,11). In addition, individual variation in normal tissue response to radiotherapy suggests a role of germline genetics and possibly epigenetics in the risk of CVD after cancer therapy (12). Studies of certain inherited cancer predisposition syndromes, such as those with germline mutations in DNA repair genes, are consistent with this hypothesis (13). However, the putative role of germline genetics in anthracycline-induced cardiotoxicity has been based mostly on candidate gene analysis (13).

In this issue of the Journal, Sapkota and colleagues (14) used whole genome sequencing data from 1870 St Jude Life (SJLIFE) childhood cancer survivors of European ancestry to identify common genetic variants associated with cancer therapy–induced cardiac dysfunction (CCD) diagnosed in 227 survivors. Validation cohorts were 301 SJLIFE survivors of African ancestry (43 with CCD) and 4020 survivors of European ancestry from the Childhood Cancer Survivor Study (CCSS) (230 with self-reported CHF). A single nucleotide polymorphism (SNP) in the 3′ untranslated region of the potassium 2-pore domain channel subfamily K member 17 (KCNK17) gene (rs2815063-A) showed a genome-wide significant association with ejection fraction (EF) in SJLIFE European ancestry survivors and was replicated in both validation cohorts. KCNK17 encodes the TASK-4 protein, a member of the 2-pore domain superfamily of background K(+) channels, located in a cluster of related genes on chromosome 6 (15,16). These K(+) channels are open at all membrane potentials, contribute to cellular resting membrane potential, are expressed in human heart atrial tissue (17), and have been suggested as therapeutic targets to treat atrial and ventricular arrhythmias (18). Sapkota et al. (14) found that the variant rs2815063-A was specifically associated with treatment with doxorubicin only but not with daunarubicin only or heart radiotherapy only. Evaluation of whole blood DNA methylation at 2 sites near rs2815063-A (34 kb downstream of the gene and in KCNK17 intron 1) in the SJLIFE European ancestry survivors showed a correlation with dysregulation of KCNK17 enhancers, which replicated in the SJLIFE African ancestry survivors. As a K(+) channel, it is biologically plausible for KCNK17 to affect cardiac function, and germline missense mutations in KCNK17 have been reported in a very small number of patients with cardiac conduction defects (19,20). The data from Sapkota et al. (14) suggest that rs2815063-A is associated with changes in gene expression and methylation of nearby sites, but these are primarily correlative studies not directly assessing the actual SNP.

Strengths of the Sapkota study include the size of the SJLIFE, a median of 22- to 25-year follow-up period, and replication of findings in 2 populations (CCSS European ancestry and SJLIFE African ancestry populations) (14). Furthermore, echocardiography was used to evaluate cardiac dysfunction in all SJLIFE survivors (8), and high-quality detailed childhood cancer treatment data were evaluated in the primary and validation populations. The whole genome sequencing effort in SJLIFE (and the CCSS validation cohort) is the largest gene discovery to date for CVD. The correlation of the rs2815063-A SNP with DNA methylation and KCNK17 expression suggests a functional role of this variant as well.

Although echocardiography is an important tool in surveillance of CVD in childhood cancer survivors, inter- and intraobserver variability in EF measurements preclude precise quantification of EF differences less than 10% (5,8,21). Therefore, the EF reductions ranging from 1.1% to 3.3% reported by Sapkota et al. (14), although statistically significant, are of questionable clinical importance. Furthermore, it is unclear whether the single EF measurement studied per survivor represents an early asymptomatic or preintervention time point in the HF trajectory or a later symptomatic and/or postintervention time point. Analysis using the lowest EF (SJLIFE) (or highest National Cancer Institute’s Common Terminology Criteria for Adverse Events grade for CCD in the CCSS) (22) when multiple measurements were available may have biased results away from the null if the lowest EF (or highest Common Terminology Criteria for Adverse Events grade) was not reflective of the true baseline. HF diagnoses in the CCSS were limited by the lack of linkage to medication history, and as noted by the authors, self-report likely resulted in misclassification of CHF. Finally, although reduced EF measurements (EF <50% and/or ≥10% absolute decline in EF) were incorporated into the CCD severity classification in the SJLIFE cohort, the authors did not account for HF occurring among survivors with preserved EF (≥50%), potentially underestimating the diagnosis of HF and obscuring genetic effects.

The study power is limited to accurately estimate risks for the potential effect modifiers of age at diagnosis and/or treatment  of the primary childhood cancers, sex, types of childhood cancer, treatment with agents other than anthracyclines and/or radiotherapy and other known CVD risk factors. This also complicates interpretation of the small number of CVD outcomes by treatment subgroups although the lack of an association of rs2815063-A with radiotherapy may not be surprising if anthracyclines and radiotherapy cause cardiomyopathy by different mechanisms (23). Future genetic and gene-environment interaction studies of treatment-related CVD (and other adverse health outcomes) require much larger sample sizes and comprehensive clinical assessments that could be implemented through establishment of international consortia of childhood cancer survivors. Genetic predisposition to cancer treatment–related CVD is also likely to occur in individuals of all ages diagnosed with cancer. As adults with cancer live longer, incorporating their outcomes with that of younger childhood cancer survivors could also contribute to the additional statistical power required to understand the contribution of germline genetic variation related to cancer treatment and CVD risk.

Despite notable decline in mortality from CVD among childhood cancer survivors during the past few decades (24), substantial morbidity remains. Although the armamentarium for treating childhood cancer has been expanding, the notable benefits of anthracyclines and radiotherapy are likely to lead to ongoing use with some modifications. Other systemic therapy (including other chemotherapy, immunotherapy, and targeted therapies) also requires further evaluation given their potential cardiotoxicity (4). Findings from Sapkota et al. (14) offer promising clues into common genetic variants associated with CCD, but these results represent early-stage genomic discovery and require further replication. The findings and limitations point the way to the value of international collaboration and consortia that would include such efforts as those begun by the Childhood Cancer Data Initiative (25). Increasing data access for not only genomics data but also more precise and longitudinal measures of HF and incorporating information on CVD risk factors, effect modifiers and other behavioral, lifestyle, and health outcomes would expedite discovery. Use of recently proposed definitions for HF (5,26) may improve standardization of diagnostic criteria for consortia and the Childhood Cancer Data Initiative. In the meantime, promising frameworks have been established to develop standardized protocols based on international consensus (27) involving international multidisciplinary teams (eg, pediatric oncologists, cardiologists, pediatric cancer and cardiovascular epidemiologists, statisticians, pharmacologists, and other specialists). These approaches could incorporate a baseline pretreatment, detailed cardiovascular system assessment and improved questionnaires for the ongoing longitudinal follow-up of the CCSS cohort with greater sensitivity and specificity for assessing a broad range of CVD and precursor outcomes posttreatment. As part of the long-term follow-up in large consortial cohort studies, implementation of improved strategies should address the monitoring of acute CVD treatment–related toxicity during administration of therapy and early identification of subsequent CVD treatment–related toxicity. Better methods are also needed to assess the effect of therapeutic interventions on cardiac function. In addition, sophisticated statistical analysis strategies should be used to consider potential continuous effects of cardiotoxic treatments at all treatment levels, survivor well-being, and primary cancer and CVD survival endpoints. Ongoing follow-up and collaboration are essential given the young, attained age in the SJLIFE, CCSS, and other national and international cohorts of childhood cancer survivors.

Funding

This work was supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics of the National Cancer Institute, National Institutes of Health, Bethesda, MD, USA.

Notes

Role of the funder: The funder had no role in the writing of this editorial or the decision to submit it for publication.

Disclosures: The authors have no disclosures.

Author contributions: Conceptualization: MSL, GMD, SAS; writing—original draft: MSL; writing—review & editing: MSL, GMD, SAS.

Data Availability

No data were generated or analyzed for this editorial.