Genome-Wide Association Study of Pregnancy in Childhood Cancer Survivors: A Report from the Childhood Cancer Survivor Study

Infertility concerns have a significant impact on quality of life in survivors of childhood cancer (1). Emerging technologies and greater awareness are expanding the pool of childhood cancer patients who may be able to preserve fertility despite intensive therapy, though predicting risk of infertility remains challenging (1). Previous studies have found clear associations between gonadotoxic therapeutic exposures and subsequent risk of infertility (2). In addition, methods such as the cyclophosphamide equivalent dose (CED) for quantifying exposure to alkylating agents allow comparison of infertility risk across different treatment regimens, independent of primary cancer type (3). In the general population, genome-wide association studies (GWAS) have explored the individual risk of infertility (4). However, genome-wide association analysis to delineate the likelihood of future pregnancy in childhood cancer survivors has not been previously attempted.

We performed a GWAS utilizing the Childhood Cancer Survivor Study (CCSS), to examine the association between SNPs and pregnancy or siring a pregnancy. The CCSS is a multi-center, retrospective cohort of 25,665 5-year survivors of childhood cancer diagnosed between 1970 and 1999, who are followed prospectively for the development of late-effects (2). We attempted replication for SNPs with a P < 10^–7, utilizing the St. Jude Lifetime Cohort (SJLIFE) that includes individuals treated for childhood cancer at St. Jude Children's Research Hospital (Memphis, TN) who had survived ≥5 years after diagnosis between 1962 and 2012 (5). Participants were excluded if they received radiation to the pelvis, testes, or brain, had a sterilizing surgical procedure prior to 5-year survival, were missing a CED, or were survivors of non-European genetic ancestry (Supplementary Fig. S1; refs. 2, 3). If patients had a sterilizing procedure after 5-year survival, analysis was censored at the time of the sterilizing procedure.

The CCSS conducted genotyping and imputation on 4.1 million loci (Illumina; Infinium Human Omni5 Exome-4 v1.0 array with imputation using the 1000 Genomes Phase 3 data as reference; ref. 6). SNPs passing quality control steps were included for analysis; SNPs were excluded if the minor allele frequency (MAF) was less than 0.05, were missing from more than 20% of subjects, or showed extreme deviation from Hardy–Weinberg equilibrium (P < 10^–6). For imputed SNPs the INFO score needed to be ≥0.5 and the certainty core ≥0.95. For the SJLIFE cohort, whole-genome sequencing and quality control was performed as previously described (7).

We created sex-specific multivariable Cox proportional hazard models to relate SNPs with pregnancy using attained age as the time axis, starting the at-risk time from 5 years postdiagnosis or at age 15 years, whichever was later, and ending at the earliest of pregnancy, death last survey, or age 44. The models included CED (0; >0), birth year (<1965; ≥1965), and the first three principle coordinates of genotypes. Due to the association of marital status on pregnancy we restricted the analysis to married subjects. Power analyses were conducted using the R package (survSNP) for the CCSS cohort (8).

The data analyzed in this study were obtained from the CCSS (https://ccss.stjude.org/develop-a-study/gwa-data-resource.html). Clinical Data is maintained in a database managed by the CCSS and GWAS data was downloaded from dbGaP (https://dbgap.ncbi.nlm.nih.gov/aa/wga.cgi?page=login). CCSS dbGaP Study Accession: phs001327.v2.p1 (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs001327.v2.p1). SJLIFE data including genotype data are available through: https://www.stjude.cloud/research-domains/cancer-survivorship.

The demographic and clinical characteristics of the two cohorts were generally similar with a few notable exceptions. CCSS participants were more likely to be born prior to 1965, less likely to have a sterilizing surgical procedure subsequent to 5-year survival, and had a higher rate of pregnancy/siring a pregnancy (Table 1). For male CCSS participants, our study had a power of 81% to detect a HR = 1.61 for MAF = 0.2 at a genome wide significance level of 5 × 10^–8, assuming an additive model; the power was >90% for MAF > 0.3. For female CCSS participants, our study had 81% power to detect a HR = 1.52 for MAF = 0.2.

In the CCSS cohort, using our GWAS model, one SNP had a P < 5.0 × 10^–8 and eight SNPs were in the range of P < 10^–7 with HRs ranging from 1.35 to 1.84 (Table 2). None of the nine SNPs was associated with a statistically significant impact on pregnancy/siring a pregnancy in the replication cohort.

We performed a GWAS to determine the association of SNPs with the likelihood of pregnancy or siring a pregnancy in a large cohort of childhood cancer survivors. Although several SNPs were potentially associated with pregnancy in the discovery cohort, none was replicated in an independent cohort. We were unable to examine the risk of clinical infertility or biochemical markers of subfertility. Infertility is multifactorial, and our study may not have been able to incorporate all pertinent factors. Indeed the ability to conceive (if desired by the patient) is the outcome with the greatest clinical relevance; however, data on patient desires or clinical infertility were not available for analysis. Likewise, another limitation of the study was that the questionnaire asked participants if they had ever been married, but did not specify whether the marriage was heterosexual. Although models in both cohorts made adjustments for birth cohort and CED, the two cohorts differed in regards to birth cohort, sterilizing procedures and overall rate of pregnancy. We did not include factors such as “desire to become pregnant” or use of assisted reproductive technology; these could have differed by birth cohorts. Overall, the SJLIFE cohort was born more recently and may have had greater access to assisted reproductive techniques compared with the CCSS cohort. Together these differences might explain why the findings in the CCSS cohort did not replicate. These limitations notwithstanding, our study demonstrates a need to continue to explore the role of genetic susceptibility in determining the risk of infertility among childhood cancer survivors.

S.J. Rotz reports grants from National Center for Advancing Translational Sciences during the conduct of the study; personal fees from Resource for Clinical Investigation in Blood and Marrow Transplantation (RCI BMT) outside the submitted work. M.M. Hudson reports grants from NCI U01 CA195547 during the conduct of the study. K.C. Oeffinger reports other support from GRAIL, LLC outside the submitted work. L.L. Robison reports grants from NCI during the conduct of the study. D. Sahoo reports other support from Shanvi outside the submitted work. G.T. Armstrong reports grants from St. Jude Children's Research Hospital during the conduct of the study. No disclosures were reported by the other authors.

S.J. Rotz: Conceptualization, resources, data curation, formal analysis, supervision, investigation, writing–original draft, project administration, writing–review and editing. S. Worley: Data curation, software, formal analysis, investigation, writing–review and editing. B. Hu: Software, formal analysis, investigation, visualization, methodology, writing–review and editing. P. Bazeley: Data curation, software, formal analysis, methodology, writing–review and editing. J.L. Baedke: Data curation, software, formal analysis, investigation, methodology, writing–review and editing. M.M. Hudson: Conceptualization, resources, supervision, funding acquisition, validation, writing–review and editing. D.J. Kuo: Conceptualization, validation, investigation, methodology, writing–review and editing. K.C. Oeffinger: Conceptualization, resources, methodology, writing–review and editing. L.L. Robison: Conceptualization, resources, funding acquisition, methodology, writing–review and editing. D. Sahoo: Conceptualization, data curation, software, methodology, writing–review and editing. F. Wang: Data curation, software, formal analysis, investigation, writing–review and editing. Y. Yasui: Conceptualization, resources, data curation, formal analysis, supervision, validation, methodology, writing–review and editing. G.T. Armstrong: Conceptualization, resources, data curation, formal analysis, supervision, investigation, methodology, project administration, writing–review and editing. S. Bhatia: Conceptualization, resources, validation, investigation, methodology, project administration, writing–review and editing.

This work was supported by the NCI (grant no. CA55727 to G.T. Armstrong, principal investigator). Support to St. Jude Children's Research Hospital was also provided by the NCI (grant no. NCI U01 CA195547 to M.M. Hudson, principal investigator), Cancer Center Support (CORE) grant (grant no. CA21765 to C. Roberts, principal investigator) and the American Lebanese-Syrian Associated Charities (ALSAC). Support to Seth Rotz was provided from the National Center for Advancing Translational Sciences (2KL2TR002547, Raed Dweik, principal investigator).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.