Thyroid carcinogenesis is hypothesized to result from mutational events combined with growth stimulation (4
), a hypothesis consistent with the marked inverse association observed between radiation dose-response and exposure age. All the SNPs and genes were selected because they could affect the likelihood of mutations from DNA misrepair or might constitutively influence cellular proliferation. In this study we identified two polymorphisms, RET
S836S and GFRA1
-193 C>G, the variant forms of which were associated with an increased risk of both thyroid nodules and papillary thyroid cancer. Two other polymorphisms were associated with decreased risk of thyroid nodules, TGFB1
T263I and XRCC1
R194W, with TGFB1
T263I also associated with decreased risk of papillary thyroid cancer. XRCC1
R194W and BRIP1
P919S modified the radiation-related risk of thyroid nodules. The results suggest that, at least in the population studied here, several SNPs may be related to thyroid nodule and cancer risk, or that other variants in linkage disequilibrium, are risk alleles.
S836S polymorphism has been associated with both differentiated thyroid cancer and benign thyroid disease, with adjusted odds ratios of 2.5 (95% CI 1.0-6.6) and 2.2 (0.7-6.8) observed, respectively, for the minor allele (13
). However, Lesueur et al (12
) and Lönn et al (14
) did not find an association with RET
S836S and sporadic papillary thyroid cancer, but did observe an increased risk associated with RET
L769L and RET
G691S, respectively. It may be that the synonymous RET
polymorphisms exert their effects by affecting protein folding or altering RNA secondary structures resulting in reduced protein function (reviewed in (27
)). Two RET
haplotypes have been associated with risk of benign thyroid disease, but not with differentiated thyroid cancer (13
). In our RET
haplotype analysis, we found no additional information beyond that obtained from analyzing SNPs individually.
RET activation requires binding with one of the four RET ligands that interact preferentially with four high-affinity glycosyl-phosphatidylinositol co-receptors called GFRA1 through GFRA4 (reviewed in (28
)). Polymorphic variants in the genes GFRA1-GFRA4
have been analyzed in relation to sporadic medullary thyroid cancer, (29
) but medullary thyroid cancer is histologically distinct from and probably uninformative for comparisons with papillary thyroid cancer. To our knowledge, GFRA1
gene polymorphisms have not been evaluated previously in persons with benign thyroid nodules.
Several of the villages studied in Kazakhstan are in mild to moderately iodine deficient areas (17
). In geographic regions of iodine deficiency, the thyroid glands of residents are exposed to mild chronic TSH stimulation that could increase H2
and free radical production (reviewed in (33
)), causing damage to DNA. We evaluated several gene variants in DNA damage repair pathways as we hypothesized these would be important among persons with marginal iodine intake and exposure to radioactive fallout. We included the base excision repair (BER) pathway because oxidative and some radiation damage to nucleotides would be repaired by this mechanism. In radiation challenge assays, frequencies of chromatid breaks per cell after lymphoctye γ-radiation were higher among papillary thyroid cancer cases than controls and benign thyroid disease cases were intermediate between the two groups (34
), suggesting higher radiation sensitivity or compromised radiation-induced damage repair ability was a risk factor for thyroid tumors. XRCC1 is a key BER scaffolding protein and the XRCC1
R194W minor allele was associated with a statistically significantly decreased risk of thyroid nodules and the common allele with a suggestive radiation-related increased thyroid nodule risk. In other words, the more common C allele appeared to be the risk allele, associated both with thyroid nodules and nodules related to radiation exposure to the thyroid gland. BRIP1 is a BRCA1 interacting protein for which one SNP (P919S) modified the radiation-related thyroid nodule risk. However, absent a main effect, it is more difficult to ascribe importance to the possible gene-radiation interaction observed for BRIP1
P919S when a chance finding is also possible.
Transforming growth factor β is a cytokine with multiple regulatory actions and responses to cellular damage (including from radiation) and can promote or suppress carcinogenesis depending on the timing and specific function mediated by its extracellular signaling (reviewed in (35
)). We found that the TGFB1
T263I minor allele was associated with statistically significantly decreased risk of thyroid nodules; the minor allele was not represented among any thyroid cancer cases compared to 4.9% of controls. Further detailed evaluation of the polymorphic variation in the TGFβ1
gene and thyroid tumors, such as could be obtained from collaborative efforts, should be considered.
Strengths of this study were the relatively large sample size, having 84% power to detect an OR of 1.5 with alpha set at 0.002 (Bonferroni correction for 23 tests) if the allele frequency was ≥ 30%, and that the thyroid nodules were detected within a screened population under standard conditions. We were able to test for variation across two diverse populations and for two thyroid disease endpoints. While not an exhaustive set of risk factors, we were able to assess the impact of selected co-variates such as TSH levels, thyroid volume, body weight, and the matching variables. We found several SNP and thyroid nodule associations for which one might argue the papillary thyroid cancer results provided some corroboration of the findings. Confounding by population stratification is unlikely because thyroid nodule prevalence did not differ by ethnicity and no inconsistencies were found when we conducted our analyses stratified by ethnicity.
Several limitations should also be considered, including the large number of tests conducted and the relatively modest genotypic associations seen. Using a Bonferroni correction, none of the p-values for the genotype associations with thyroid nodule risk would be considered significant (p < 0.002). Power for detecting less common variants was low: 38% to detect an OR of 1.5 with alpha set at 0.002 and an allele frequency of 10%. However, the gene-radiation interaction analyses were more exploratory in nature and require replication as the primary safeguard against false positive associations. The radiation doses to the thyroid gland are uncertain and efforts have been made to re-contact village residents to collect additional key data on milk consumption besides from cows, housing construction, and patterns of animal husbandry (location of grazing areas around the villages). Analyses of uncertainties are planned after the doses are revised based on incorporating the new information.
In summary, the associations found with thyroid nodules and papillary cancer suggest polymorphic variation in RET S836S, GFRA1 -193 C>G, and TGFB1 T263I, were related to risk. The common allele of XRCC1 R194W was related to risk of thyroid nodules and suggestively modified the radiation-associated nodule risk. Other pathways, such as genes in double strand break repair, apoptosis, and additional genes related to proliferation should also be pursued.