We evaluated the associations of tag SNPs in candidate genes from several interrelated pathways involved in maintenance of genomic integrity, including DNA repair, epigenetic mechanisms, telomere function, apoptosis, cell cycle control, tumor suppression and the MAPK, PI3K/AKT and Wnt/beta-catenin cell-signaling pathways with risk of PTC. The group of genes as a whole was suggestively associated with risk of PTC (Poverall = 0.074/Poverall = 0.061). We found nine tag SNPs in seven gene regions that were associated with PTC at Ptrend <0.0005. The strongest associations were seen for SNPs in the histone deacetylase 4 gene HDAC4, in the DNA repair checkpoint gene HUS1 and in the apoptosis gene BAK1 (Ptrend < 0.0001). Gene region-based analyses showed that all three of these gene regions (HDAC4, HUS1 and BAK1) were significant at Pregion <0.005. While after formal correction for multiple comparisons neither individual SNP- nor gene region-based results remained statistically significant, several of our findings with suggestive P values are of potential biologic or clinical interest.
Specifically, the observed thyroid cancer associations with multiple polymorphisms in HDAC4
were intriguing. HDAC4
is a histone deacetylation gene that has the capacity to alter chromosome structure and silence gene transcription by limiting access of transcription factors to DNA, particularly tumor suppressor genes, thereby deactivating tumor suppression activity (31
). Histone deacetylase inhibitors are demonstrated anticancer agents whose main mechanism of action is the transcriptional reactivation of tumor suppressor genes that have been turned off through histone deacetylation (33
). Several studies of thyroid tumor cells have demonstrated the ability of HDAC
inhibitors to facilitate radioactive iodine uptake (34
) and suppress growth and proliferation (37
). However, no studies have been published on associations between HDAC
polymorphisms and thyroid cancer risk.
has been linked with other cancers, namely breast and ovarian (41
), and is part of the Rad9-Rad1-Hus1 (911) cell cycle checkpoint complex that plays a key role in all checkpoint responses to DNA damage (43
). In vitro
studies have shown that human cells exposed to ionizing and ultraviolet radiation have higher levels of the 911 protein complex compared with unexposed cells (44
) and that this relationship is dose dependent. No studies previously have reported a relationship between HUS1
polymorphisms and thyroid cancer.
We also found variants in two genes postulated to play a role in apoptosis, BAK1
, had suggestive P
values for an association with an increased risk of PTC. Previous research suggests that these genes may play a role in carcinogenesis of certain cancers, including testicular cancer and chronic lymphocytic leukemia (46
), myeloma (48
) and mantle cell lymphoma (49
). Moreover, BAK1
expression appears to be upregulated in thyroid tumor cells (50
Our strongest pathway-based findings for the DNA repair direct reversal of damage pathway (53
) was driven by two of our top SNP-based findings—tag SNPs rs10838192 in ALKBH3
and rs4751109 in MGMT
. While little is known about ALKBH3
in relation to cancer, several candidate gene studies have linked MGMT
polymorphisms to risk of head and neck cancer (54
), glioma (56
) and esophageal cancer (59
). Moreover, two studies have found an association between MGMT
hypermethylation and PTC (61
In interpreting the results of our study, several strengths and limitations need to be considered. Our study had high participation rates minimizing potential for selection bias. Because survival rates for PTC are exceptionally high, survival bias is unlikely. To minimize concerns about population stratification, all analyses were limited to individuals of European ancestry. Moreover, cases from the two studies were similar with regard to age at diagnosis, smoking status, tumor size and allele frequencies. Although radiation exposure is an established risk factor for thyroid cancer, our cases are unlikely to be radiation related because doses are low, most of the exposure occurred in adulthood, and no dose–response has been observed in the USRT population (63
), and only five of the UTMDACC cases were exposed to radiation from self-report of radiotherapy (12
). Thus, the results of our study should be internally valid. However, extrapolation of our findings to the general population requires caution. Thyroid cancer incidence in the USRT cohort was higher than in the general population, with a standardized incidence rate ratio of 1.7. This difference is probably due to increased screening in the USRT cohort as the proportion of small thyroid tumors was higher among the cohort (30%) compared with the general population (15%) based on the SEER registries database (23
) but comparable with that in the UTMDACC cases.
Given that PTC cases were less likely to be cigarette smokers or alcohol drinkers and had higher average body mass index that controls, we explored whether these factors may have influenced the associations of interest. When added to the models, these variables did not meaningfully change the risk estimates, and therefore, are unlikely to confound our main findings. Other strengths of our study include thorough selection of genes related to a variety of genomic integrity pathways (53
) and nearly complete representation of DNA repair genes. Relative to genome-wide association studies, the coverage of selected gene regions was higher, although we could have missed important associations with SNPs not included within the genotyping platform. While among the larger studies with respect to the number of thyroid cancer cases and controls, our study had limited power to detect weak associations, especially for less common genetic variants. Another limitation is the use of tag SNPs that are themselves unlikely to be the disease-related SNPs but are assumed to be in LD with the causal variant. To address these limitations, we excluded SNPs with minor allele frequency <10% and relied on robust gene/pathway adaptive rank-truncated product methods combining SNP-specific P
values of trend to confirm associations with risk of PTC.
In summary, our results suggest that genetic alterations in the pathway involved in maintenance of genomic integrity may contribute to thyroid cancer susceptibility.