Telomere Length-Associated Genetic Variants and the Risk of Thyroid Cancer in Survivors of Childhood Cancer: A Report from the Childhood Cancer Survivor Study (CCSS)

Subsequent malignant neoplasm of the thyroid (thyroid SMN) is one of the most common SMNs among survivors of childhood cancer. Established risk factors for thyroid SMN include female sex, younger age at primary cancer diagnosis, and exposure to radiation (1); however, these factors do not fully explain the observed variability in thyroid SMN risk for survivors. We previously reported less telomere content detected by quantitative PCR among childhood cancer survivors who developed thyroid SMN, compared with matched survivor controls without SMN (2). Given this finding, and emerging evidence for associations between genetically predicted leukocyte telomere length (LTL) and increased risk for specific cancers (3), we extended this work to investigate the relationship between SNPs associated with LTL, as previously determined by genome-wide meta-analyses, and thyroid SMN in a cohort of childhood cancer survivors.

This study leveraged existing genotype data from participants enrolled in the Childhood Cancer Survivor Study (CCSS), a multicenter, retrospective cohort of 5-year survivors of childhood cancer diagnosed between 1970 and 1986 (4). Survivors are prospectively followed, and incidence of thyroid SMN ascertained through self-report questionnaires. SMN cases were validated by pathology reports and verified by a CCSS pathologist (M.A. Arnold). A comparison group of survivors without thyroid SMN were censored by date of other SMN, death, or date of last follow-up. Inclusion in this analysis was restricted to CCSS participants with genome-wide SNP data. Genotype data were generated by the Cancer Genomics Research Laboratory of the NCI using the Illumina HumanOmni5Exome array. Imputation was based on the 1000 Genomes Project reference haplotypes, as described previously (5). Analyses were restricted to those of European ancestry to limit bias due to population stratification. Genetically estimated LTL was determined using methods described by Machiela and colleagues (3). In brief, participant genotypes were extracted from nine common SNPs previously identified as associated with LTL by genome-wide meta-analyses (6–8). Assuming an additive genetic model, we used a multivariable Cox proportional hazards model to estimate the HR, 95% confidence interval (CI), and P value for the association between each individual SNP and thyroid SMN risk. We then estimated the HR for thyroid SMN risk using the weighted sum of nine SNPs. An unweighted genetic risk score (GRS) was determined for each participant from the number of risk alleles present (0, 1, or 2), and a weighted GRS was generated using the published beta estimate for each effect allele. On the basis of the thyroid SMN risk prediction model described by Kovalchik and colleagues (9), covariates included sex, primary cancer diagnosis, neck radiation exposure (yes/no), and alkylating agent exposure (yes/no). Results were stratified on radiation exposure and primary diagnosis, and the FDR was used to account for multiple comparisons. Power was calculated using Quanto Version 1.2.4. On the basis of a sample size of 5,324 genotyped survivors of European descent (118 with thyroid SMN and 5,206 without thyroid SMN), we were powered to detect a HR of 1.7 for thyroid SMN, using the lowest tertile of genetically estimated LTL. As this is in keeping with previous assessments evaluating LTL GRSs and subsequent cancer (3, 6–8), we were adequately powered to detect the proposed associations.

The majority of the 118 survivors with a verified diagnosis of thyroid SMN had papillary carcinoma [papillary, n = 106 (89.8%); follicular, n = 11 (9.3%); and carcinoma not otherwise specified, n = 1 (0.9%)]. Participants were predominantly survivors of Hodgkin lymphoma, leukemia, and central nervous system tumors and diagnosed before the age of 15 years (Table 1). Thyroid SMN cases were more likely to have been born before 1970 (P = 0.002), exposed to an alkylating agent (P = 0.016) or radiation (P < 0.001), and have a history of thyroid nodules (P < 0.001). Of the nine SNPs evaluated, four were directly genotyped and five were imputed, all with imputations scores that were greater than 0.89. No association was observed between the individual LTL-related SNPs and risk for thyroid SMN. When considering the total burden of the effect alleles for these SNPs on thyroid SMN risk, there was no significant relationship observed for either the unweighted or weighted GRS (Table 2). Given the strong dose-dependent relationship between radiation and risk for thyroid SMN, we then restricted the analysis to radiation to the neck-exposed individuals (yes/no, considered yes if >10% of the region was directly within the radiation field) and, similarly, no association was observed (Supplementary Table S1).

In the largest available cohort of childhood cancer survivors with comprehensive genotype data, our results suggest that variation in common SNPs influencing LTL is not strongly associated with risk for thyroid SMN in survivors of childhood cancer. No prior studies have investigated common variation in SNPs influencing LTL and risk for thyroid SMN. Of note, our observations are in contrast with evidence for genetically increased LTL in adults with primary thyroid cancer, though that study was underpowered to detect associations within this cancer subtype (10). The previously observed inverse relationship between LTL and thyroid SMN risk in survivors may therefore be due to alternative molecular mechanisms, such as rare variants interacting with therapy-related exposures, defects in specific genes related to telomere maintenance, or environmental factors (e.g. differential effects of chemotherapy and/or radiation). Investigations that expand this assessment and comprehensively evaluate the role of telomere maintenance genes on risk for thyroid SMN are ongoing.

No potential conflicts of interest were disclosed.

Conception and design: M.M. Gramatges, L.M. Morton, J.P. Neglia, G.T. Armstrong, S. Bhatia, P.J. Lupo

Development of methodology: M.M. Gramatges, W.M. Leisenring, P.J. Lupo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.M. Morton, M.A. Arnold, J.P. Neglia, C.L. Dagnall, S.J. Chanock, G.T. Armstrong, L.L. Robison, P.J. Lupo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.M. Gramatges, L.M. Morton, Y. Yasui, M.A. Arnold, M.J. Machiela, S.J. Chanock, P.J. Lupo

Writing, review, and/or revision of the manuscript: M.M. Gramatges, L.M. Morton, Y. Yasui, J.P. Neglia, W.M. Leisenring, M.J. Machiela, C.L. Dagnall, G.T. Armstrong, L.L. Robison, S. Bhatia, P.J. Lupo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.M. Morton, W.M. Leisenring, C.L. Dagnall, G.T. Armstrong, P.J. Lupo

Study supervision: M.M. Gramatges, J.P. Neglia, G.T. Armstrong, P.J. Lupo

This work was supported by the NCI (CA194473 to M.M. Gramatges, principal investigator) and a CCSS Career Development Award to P.J. Lupo. CCSS is supported by the NCI (CA55727 to G.T. Armstrong, principal investigator), the St. Jude Children's Research Hospital Cancer Center Support (CORE) grant (CA21765 to C. Roberts, principal investigator), and the American Lebanese-Syrian Associated Charities (ALSAC). Genotyping for the study also was supported by the Intramural Research Program of the NCI, NIH.