Homologous recombination repair germline variants and subsequent neoplasm risk among childhood cancer survivors

Abstract

Childhood cancer radiation therapy (RT) increases subsequent neoplasm risk. Radiation dose may modulate DNA damage responses, but the small sample sizes of prior human studies of homologous recombination repair hampered dose-specific investigations. We pooled data for 12 180 survivors (8339 from the Childhood Cancer Survivor Study and 3841 from the St Jude Lifetime Cohort) to estimate associations between deleterious homologous recombination repair variants and RT-related subsequent neoplasms (most commonly breast cancer, meningioma, thyroid cancer, and sarcoma) using conditional logistic regression with matched controls. In all, 1253 (10.3%) survivors were homologous recombination repair variant carriers, and 1301 (10.7%) developed at least 1 RT-related subsequent neoplasms. Variants increased the risk of out-of-field RT-related subsequent neoplasms (cases, 40/190 [21.1%]; control individuals, 9.7%; odds ratio [OR] = 2.5, 95% confidence interval [CI] = 1.7 to 3.6; P = 4.80 ×10⁻⁶), with consistent results across cohorts (Childhood Cancer Survivor Study, OR = 2.5, 95% CI = 1.6 to 3.7, P = 3.77 ×10⁻⁵; St Jude Lifetime Cohort, OR = 2.5, 95% CI = 1.0 to 6.4, P = 3.07 ×10⁻²). No association was observed for in-field or near-field subsequent neoplasms or individuals not undergoing RT. Findings emphasize homologous recombination repair variant-conferred susceptibility to RT-related subsequent neoplasms and dose-dependent DNA damage repair.

Radiation therapy (RT) for childhood cancer substantially increases the risk of subsequent neoplasms.¹ Treatment with ionizing radiation can induce several types of DNA aberrations, including base damage and single-base and double-strand breaks.² DNA double-strand breaks repaired by the high-fidelity homologous recombination repair pathway should not hamper cellular functions, whereas the error-prone nonhomologous end-joining pathway increases the likelihood of under-repaired or misrepaired DNA double-strand breaks,³ accumulation of which is highly cytotoxic and contributes to genome instability.⁴ Factors such as cell cycle stage and the extent of DNA damage determine the choice between the activation of homologous recombination repair and nonhomologous end-joining pathways, which may be further modulated by radiation dose, as suggested by radiobiological data.⁵ The contribution of germline genetic variation to this balance and risk of subsequent neoplasm development remains understudied, however, in part because of the difficulty of collecting biospecimens, comprehensive treatment information, and long-term follow-up data.

The Childhood Cancer Survivor Study (CCSS) and St Jude Lifetime Cohort Study (SJLIFE), 2 of the largest cohorts of childhood cancer survivors, have recently reported that rare, potentially protein-damaging variants in the homologous recombination repair pathway may increase subsequent neoplasm risk among childhood cancer survivors.^6,7 The CCSS analysis, however, found that risk was associated only with out-of-field subsequent neoplasm development, whereas the SJLIFE study did not investigate risks by radiation dose. We therefore aimed to replicate and extend the original CCSS report by comprehensively investigating of subsequent neoplasm risks by RT exposure among childhood cancer survivors in CCSS and SJLIFE to advance the understanding of RT-related carcinogenesis and further inform long-term follow-up guidelines for this population.

We pooled data for 12 180 survivors (CCSS = 8339, SJLIFE = 3841), including extended follow-up data for the individuals in the original CCSS report with whole exome sequencing data (ncbi.nlm.nih.gov/gap, phs001327.v2.p1; survivors diagnosed between 1970 and 1986) and newly harmonized treatment, follow-up, and whole genome sequencing data from CCSS (stjude.cloud, SJC-DS-1002; survivors diagnosed between 1987 and 1999) and SJLIFE (stjude.cloud, SJC-DS-1005; survivors diagnosed since 1962). Sequencing, alignment, and quality control were performed as previously described.^6,8

In the pooled dataset, we identified carriers of rare, potentially protein-damaging germline variants in 87 homologous recombination repair pathway genes,⁶ annotated using ClinVar (National Library of Medicine) and SnpEff (deleterious homologous recombination repair variants). Radiation therapy exposure has previously been described⁹ and was reconstructed to specific body regions (eg, head and neck, chest, abdomen, pelvis) based on medical record data for treatments occurring within 5 years of the first primary childhood cancer. Subsequent neoplasms were assigned to the same body regions. Conditional logistic regression estimated the association between deleterious homologous recombination repair variants and RT-related subsequent neoplasm types, which were defined a priori as those types previously associated with radiation-related risk in other exposed populations. Breast cancer, meningioma, thyroid cancer, and bone/soft-tissue sarcoma were the most prevalent RT-related subsequent neoplasms in this analysis. Basal cell carcinoma was excluded. Up to 100 subsequent neoplasm–free controls matched to cases by age (±3 years), sex, primary childhood cancer type, RT dose to the subsequent neoplasm location (no RT, out-of-field [ie, occurring in a body region that was not irradiated; estimated doses >0 to <10 Gy], and in-field or near-field [ie, occurring in an irradiated body region; estimated doses ≥10 Gy]), chemotherapy, study, and follow-up time (ie, free of the subsequent neoplasm of interest for a period at least equal to that of their matched case). The chemotherapy variable was coded to indicate receipt (yes, no, or unknown) of any alkylating agents, platinum-based chemotherapy, or anthracyclines. We examined associations between homologous recombination repair status and risk of any radiation-related subsequent neoplasm (ie, considering the first RT-related subsequent neoplasm), and between homologous recombination repair status and risk of selected subsequent neoplasm types (thyroid cancer, breast cancer, bone or soft-tissue sarcoma, colorectal cancer, other gastrointestinal cancers, glioma, or meningioma), following survivors until the first subsequent neoplasm of that cancer type and allowing for intervening of other subsequent neoplasm types. Genetic ancestry was inferred using SNPweights, version 2.1, software¹⁰ (Harvard T.H. Chan School of Public Health) and used in sensitivity analyses, including only survivors with 80% or greater inferred European genetic ancestry. We used the Integrative Genomics Viewer,¹¹ version 2.19.1 (University of California San Diego), to manually inspect and validate individual variants with a 0.2% or greater carrier prevalence.

Among all survivors, 50.6% were female, with a median (interquartile range) age at childhood cancer diagnosis of 7.3 (3.2-13.4) years and at last follow-up of 29.7 (22.5-36.3) years. A total of 1253 (10.3%) individuals carried deleterious homologous recombination repair variants (Figure 1), and 1301 (10.7%) developed at least 1 RT-related subsequent neoplasms. Deleterious homologous recombination repair variants were associated with increased risk of out-of-field RT-related subsequent neoplasms (cases, 40/190 [21.1%]; control individuals, 9.7%; odds ratio [OR] = 2.5, 95% confidence interval [CI] = 1.7 to 3.6; P = 4.80 ×10⁻⁶), with similar study-specific results (CCSS, OR = 2.5, 95% CI = 1.6 to 3.7, P = 3.77 ×10⁻⁵ and SJLIFE, OR = 2.5, 95% CI = 1.0 to 6.4, P = 3.07 ×10⁻²) (Figure 2, A). In contrast, no association was observed for in-field or near-field RT-related subsequent neoplasms (cases, 87/832 [10.5%]; control individuals, 11.7%; P = .54) or for RT-related subsequent neoplasms occurring among survivors who did not receive RT (cases, 22/220 [10.0%]; control individuals, 10.2%; P = .88). In analyses of specific RT-related subsequent neoplasm types after out-of-field RT exposure, deleterious homologous recombination repair variants were associated with increased risk of thyroid cancer (OR = 2.4, 95% CI = 1.3 to 4.6) and breast cancer (OR = 2.4, 95% CI = 1.5 to 4.9), while associations with other subsequent neoplasm types were not statistically significant (Figure 2, B). The association between deleterious homologous recombination repair variants and risk of out-of-field RT-related subsequent neoplasms did not appear to be driven by any individual gene, though case numbers were limited per gene. Results remained robust in a sensitivity analysis that included only survivors with at least 80% inferred European genetic ancestry (n = 10 430).

Figure 1.

Prevalence of childhood cancer survivors carrying deleterious variants in homologous repair combination pathway genes, restricted to genes with a variant carrier prevalence of at least 0.2% in the study population (n = 12 180). The frequency and proportion of variant carriers with 1 or more radiation therapy (RT)–related subsequent neoplasms (breast cancer, thyroid cancer, or other RT-related cancers, excluding basal cell carcinoma) is denoted at the top of the bar for each gene. Among BLM, HFM1, and RAD50 survivors four incident subsequent breast cancers were removed because these survivors developed an additional unique subsequent neoplasm; thus, the frequency bars only count one RT-related subsequent neoplasm per survivor.

Figure 2.

Associations of deleterious variants in homologous repair combination genes with (A) risk of any RT-related subsequent neoplasms (ie, considering the first RT-related subsequent neoplasm) and (B) risk of selected specific subsequent neoplasms (considering the first subsequent neoplasm of that cancer type) with out-of-field RT exposure in the combined cohort. CCSS = Childhood Cancer Survivor Study; OR = odds ratio; RT = radiotherapy; SJLIFE = St Jude Lifetime Cohort.

In this pooled cohort of childhood cancer survivors, we observed that carriers of germline variants in the homologous recombination repair pathway were at increased risk of out-of-field subsequent neoplasms after RT exposure, extending and replicating previous findings to provide further evidence that homologous recombination repair pathway germline genetic variation can modify the risk of treatment-related subsequent neoplasms after childhood cancer. Specifically, the association between homologous recombination repair variants and increased RT-related subsequent neoplasm risk replicates findings from a smaller CCSS study (n = 1108) and represents a novel finding in the SJLIFE cohort.

Although investigation of individual homologous recombination repair pathway genes was not possible due to the low prevalence of variant carriers for most genes, the pathway-level results emphasized that genetic susceptibility patterns may depend on RT dose because the observed associations were not equivalent across the full RT dose spectrum. In addition, the low genomic diversity of the current cohort methodologically limited our ability to match by genetic ancestry. Although the frequency of homologous recombination repair variant carriers was similarly distributed between survivors with European and non-European genetic ancestries (10.2% and 11.1%, respectively), these results should be interpreted with caution for survivors with non-European genetic ancestry, and future efforts are needed to enhance diversity in genomic studies.¹²

Our results support preclinical data suggesting that DNA double-strand break repair is influenced by the level of radiation exposure.^13,14 Specifically, homologous recombination repair may be the predominant repair pathway at lower RT doses, whereas more error-prone DNA double-strand break repair mechanisms (eg, nonhomologous end-joining) are more active at higher doses due to the increased DNA double-strand break burden. Therefore, the adverse consequences of radiation-induced DNA double-strand break may be most evident among survivors with germline variant–impaired homologous recombination repair function who were exposed to lower RT doses. Further functional characterization of germline variants in the homologous recombination repair pathway and their impact on DNA double-strand break repair are needed to further understand the dose-related associations observed in this study. Future studies also should consider potential impacts of cytotoxic chemotherapy in addition to radiation exposure. These findings support consideration of homologous recombination repair germline genetic variant testing in long-term clinical follow-up guidelines by raising the potential for clinical surveillance for out-of-field subsequent neoplasms among childhood cancer survivors.

Funding

This work was supported by the Intramural Research Program of the National Cancer Institute (NCI), National Institutes of Health (NIH), and the US Department of Health and Human Services. Support for CCSS was provided by the NCI (CA55727) and the Leukemia & Lymphoma Society. The SJLIFE study was funded by the NCI (CA195547), Cancer Center Support Grants (CA21765), and the American Lebanese Syrian Associated Charities. The contributions of the NIH author(s) were made as part of their official duties as NIH employees, are in compliance with agency policy requirements, and are considered works of the US government. The findings and conclusions presented in this paper, however, are those of the author(s) and do not necessarily reflect the views of the NCI, NIH, or the Department of Health and Human Services.

Conflicts of interest

The authors have no conflicts of interest to disclose.

Data availability

The data included in this study are available via dbGaP (ncbi.nlm.nih.gov/gap; accession No. phs001327.v2.p1) and St Jude Cloud (stjude.cloud; accession No. SJC-DS-1002, SJC-DS-1005).