Investigation of the Relationship Between Genome Wide Association Studies-derived Polymorphisms and Differentiated Thyroid Cancer Risk in a Turkish Population

Background: Thyroid cancer is the most common malignancy of endocrine system. Genome Wide Association Studies (GWAS) revealed a number of common variants associated with thyroid cancer risk. In this study, we aimed to investigate the association of these known variants with thyroid cancer risk in a Turkish population living in Trakya region. Methods: The study included 97 cases of differentiated thyroid cancer and 379 healthy controls. Real-Time Polymerase Chain Reaction (RT-PCR) method was used for the genotyping of rs965513, rs944289, rs966423 rs2439302 polymorphisms. Results: There was no statistically significant difference between patients and controls in terms of SNP genotype and allele frequencies. The distribution of cumulative genetic risk scores between patients and controls was also not significantly different. In the multiple logistic regression analysis (MLR), it was observed that the relationship of rs2439302 polymorphism GG genotype with thyroid cancer risk has a trend to be significant ((p = 0.067, 95%CI: 2.947 (0.928- 9.357 )). Conclusion: We suggest that the confirmation of the association of common variants with thyroid cancer in different populations will contribute to make a consensus on global risk alleles. The marginal significance of the association of rs2439302 with thyroid carcinoma risk shown in our study supports the need for functional studies on the role of this polymorphism in thyroid carcinoma. The role of genetic risk factors in the pathogenesis of thyroid carcinoma is widely accepted and (18; 19). The pieces of evidence of those studies are needed to be confirmed in discrete populations. Although some studies are focusing on the genetic risk factors for thyroid carcinoma in the Turkish population (20,21,22), no studies are investigating the association of previously defined GWAS-derived TC risk SNPs. In this study, we showed that the risk allele frequencies of rs96513, rs944289, and rs2439302 were higher between the thyroid cancer patients compared to healthy controls but the differences were not statistically significant for rs965513 and rs944289. However, we found a trend to be significant for the rs2439302 ( p = 0.067 ) between the DTC patients and control subjects. The risk allele frequency of rs966243, on the other hand, was higher between the controls compared to DTC patients in our study, suggesting that this polymorphism is unlikely a risk factor for thyroid carcinoma risk in the Turkish population. The first GWAS study for thyroid carcinoma, performed in a European population including 192 thyroid cancer patients and 37.196 controls revealed that rs965513 on 9q22.33 and rs944289 on 14q13.3 were related to both papillary and follicular thyroid cancer risk (8). FOXE1 gene, the nearest gene to the 9q22.33 locus was further associated with thyroid cancer in a study performed in Spanish and Italian populations. But rs1867277, rather than rs965513 have been reported to be associated with the TC risk in this study (23) In a subsequent study, rs965513 on 9q22.33 and rs944289 on 14q13.3 alleles were reported to be associated with low concentrations of thyroid-stimulating hormone (TSH), and the 9q22.33 allele is associated with a low concentration of thyroxin (T(4)) and a high concentration of triiodothyronine. This was a preliminary study suggesting an association between rs2439302


INTRODUCTION
Thyroid cancer (TC) constitutes approximately 90% of all malignancies of the endocrine system as the most common endocrine malignancy (1). Over 500000 new thyroid cancer patients have been reported In the 2018 report of the World Health Organization (2). The incidence of thyroid cancer increased both in males and females in Turkey, according to the statistics of the Turkish Cancer Research Association in 2008 (3). About 90% of all thyroid cancers are differentiated thyroid cancers (DTC) arising from follicular cells (FNMTC) (4). Papillary thyroid carcinoma (PTC) is the most common subtype of differentiated thyroid carcinoma. Several different factors including radiation exposure, smoking, iodine deficiency, hormonal factors, and genetic factors have been shown to contribute to the development of DTC (5).
Results of the GWAS studies can differ between populations due to the divergence of frequencies of previously unpublished SNPs and their associations with diseases may not be the same with the reference population of other studies. So, confirmation of the associated variants across multi-ethnic, founder, admixed, and highly consanguineous populations is important (6). Here, we aimed to evaluate the role of GWAS-derived TC variants in a population living in the Trakya region of Turkey.

Study design and samples
This was a case-control study. We selected the thyroid cancer-related polymorphisms from the previous study performed by Wang et al (12) and a priori power analysis was done by using the G-power 3.1.2 version (16) . The sample size was calculated as 412 samples based on an effect size of 0.153 at the rs944289 genetic variant (12), with an alpha level of 5%, and a power of 80%. However, we included 478 samples considering possible missing data (16). After providing written informed consent, 97 nonmedullary TC cases (13 males and 84 females) and 381 cancer-free control individuals (282 females, 99 males) were included in the study. All patients have histologically confirmed thyroid cancer patients. For controls, we prospectively scanned apparently healthy volunteers with thyroid ultrasonography, and individuals who do not have a thyroid nodule were recruited for the genetic analysis. They were cancer-free nonconsanguineous individuals who did not have any thyroid-related disorder and who did not have a first-degree relative with thyroid cancer. Ninety-four out of 97 thyroid cancer patients had papillary thyroid carcinoma whereas only 3 had follicular thyroid carcinoma. fT3, fT4, and TSH levels of patients and controls were biochemically analyzed.
The study was approved by the Trakya University Faculty of Medicine Scientific Research Ethics Committee (TÜTF-BAEK, 2015/174) and financially supported by Trakya University Scientific Research Projects Unit (TÜBAP 2016/132).

SNPs genotyping
We isolated genomic DNA from peripheral blood samples using the EasyOne Automated DNA Isolation System (Qiagen, Hilden, Germany). Quality control and purity of the isolated genomic DNA samples was determined using a NanoDrop Spectrophotometer (NanoDrop 2000C; ThermoFisherScientificInc.,Wilmington, MA, USA). Samples with a A260/280 values of between1.8-2.0 were included in while low-quality samples were re-isolated from stored blood samples. We genotyped rs2439302, rs944289, rs965513, and rs966423 SNPs using Applied Biosystems Step One Plus Real-Time PCR (Thermo Fisher Scientific, USA) with TaqMan® SNP Genotyping Assay kit (ThermoFisherScientificInc., Wilmington, MA, USA) according to the manufacturer's instructions.

Statistical analysis
All the statistical analyses were performed using SPSS 20.0 (Statistical Package for the Social Sciences 20.0; IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.). Hardy-Weinberg equilibrium was tested by chi-square tests for the four SNPs. We compared the patients and controls in terms of sex (male, female), and family history of cancer (present, absent), metastasis (absent, present) using the Chi-square test. The Student t-test was used in the comparison of age between patients and controls. Unweighted Cumulative genetic risk scores (UCGRS) were calculated as the total count of disease alleles from four SNPs studied (possible score range of 0-8) (17), samples with at least one or more missing genotypes were excluded in the genetic risk score calculations to avoid misinterpretation. The effect of SNPs on DTC was tested by using multiple logistic regression analysis (MLR).

RESULTS
Age, sex, clinical and biochemical parameters in DTC patients and controls are shown in Table 2. We studied 97 DTC patients (mean age: 50.2 ± 12.2 years). and 381 healthy control subjects (mean age: 37.1 ± 8.7 years). Of the 97 patients, 94 were diagnosed with papillary thyroid carcinoma and 3 with follicular thyroid carcinoma. The mean age was higher between the DTC patients compared to the control groups (p<0.001). the number of female patients was higher in the DTC group (p < 0.01) (Table2). Even the risk allele frequencies were higher between the patients compared to controls for rs965513, rs944289, and rs2439302 polymorphisms, the differences were not statistically significant (Table 4). It was interesting that the frequency of the risk allele of rs966423 polymorphism (T) was higher among DTC patients than the controls. Genotype distribution was not significantly different among patients in terms of lymph node metastasis or family history (Table 4). Mean Unweighted cumulative genetic risk was not statistically significant between DTC patients and controls ( Table 5). Analysis of demographic, biochemical, and genetic factors for thyroid carcinoma risk with logistic regression analysis revealed that the GG genotype of rs2439302 polymorphism has a trend to be significant (P = 0.067) between the patients with DTC and control subjects (OR=2.947 95% CI: 0.928-9.357)( Table 6).

DISCUSSION
The role of genetic risk factors in the pathogenesis of thyroid carcinoma is widely accepted and (18; 19). The pieces of evidence of those studies are needed to be confirmed in discrete populations. Although some studies are focusing on the genetic risk factors for thyroid carcinoma in the Turkish population (20,21,22), no studies are investigating the association of previously defined GWAS-derived TC risk SNPs. In this study, we showed that the risk allele frequencies of rs96513, rs944289, and rs2439302 were higher between the thyroid cancer patients compared to healthy controls but the differences were not statistically significant for rs965513 and rs944289. However, we found a trend to be significant for the rs2439302 (p = 0.067) between the DTC patients and control subjects. The risk allele frequency of rs966243, on the other hand, was higher between the controls compared to DTC patients in our study, suggesting that this polymorphism is unlikely a risk factor for thyroid carcinoma risk in the Turkish population.
The first GWAS study for thyroid carcinoma, performed in a European population including 192 thyroid cancer patients and 37.196 controls revealed that rs965513 on 9q22.33 and rs944289 on 14q13.3 were related to both papillary and follicular thyroid cancer risk (8). FOXE1 gene, the nearest gene to the 9q22.33 locus was further associated with thyroid cancer in a study performed in Spanish and Italian populations. But rs1867277, rather than rs965513 have been reported to be associated with the TC risk in this study (23) In a subsequent study, rs965513 on 9q22.33 and rs944289 on 14q13.3 alleles were reported to be associated with low concentrations of thyroid-stimulating hormone (TSH), and the 9q22.33 allele is associated with a low concentration of thyroxin (T(4)) and a high concentration of triiodothyronine. This was a preliminary study suggesting an association between rs2439302 on 8p12 and the expression of NRG1 gene, encoding for a signaling protein in the blood (11) and confirmed by a functional study (24).
FOXE1 locus was also associated with Papillary thyroid cancer (PTC) in a study focusing on patients exposed to radioactive iodine in their childhood or adolescence but the rs944289 at reported not to associated with radiationrelated PTC in the same study (9). rs944289 has been shown to predisposes to PTC through a long intergenic noncoding RNA gene (PTCSC3) (25). Association of rs944289, rs965513, rs966423 rs2439302, and PTC risk have been confirmed in a Chinese population, too (12). However, the association of rs944289 with thyroid carcinoma could not have been shown in a study performed in a population of Italian familial nonmedullary thyroid cancer patients and controls (18). rs944289 was not associated with DTC in a study performed in a German population (26) as in our study.
Cumulative effects of the risk alleles of common variants have been shown to correlate with thyroid carcinoma in some studies (17,27). In our study, cumulative genetic risk scores did not show a significant association with DTC risk. But it was an interesting finding that the number of the patients with ≥7 risk allele count were higher compared to controls, however, the difference was not statistically significant. This finding supports the need for a larger sample size to be studied for defining the possible effects of CGRS on thyroid carcinoma risk.
The strongest candidate for thyroid carcinoma risk was rs2439302 polymorphism in our study with a trend to be associated (p=0.067) with increased risk of DTC (OR 2.947 95% CI: 0.928-9.357). The risk allele (G) of rs2439302 was previously shown to be correlated with the expression of multiple NRG1 isoforms in unaffected thyroid tissue in a functional study (24). We suggest that the possible association of this polymorphism with DTC can be confirmed with a study including larger samples of DTC patients and controls.
There are some limitations of our study. First of all, benign thyroid nodules are very common in the normal population and this is restricting the inclusion of larger control groups. Second, as a disease with a higher frequency with advanced age, our control samples were younger than the patient group which may reflect the results. Finally, dietary and environmental differences which may affect thyroid cancer were not considered inpatient and control groups.
In conclusion, we suggest that the confirmation of the association of common variants with thyroid cancer performed in larger sample groups will contribute to making a consensus on global risk alleles. Especially, the marginal significance of the association of rs2439302 with thyroid carcinoma risk shown in our study urges the need for functional studies on the role of this polymorphism in thyroid carcinoma. However, our results also support that caution must be taken when extrapolating GWAS results from one population to directly predict disease risks in other populations (28,29).