Thyroid cancer (TC) is a rare neoplasia, but is the most frequent endocrine malignancy (1). In general, it originates from thyroid follicular cells and its most common histological types are papillary carcinoma (≈70–80%) and follicular carcinoma (≈10–20%) (2). Papillary and follicular TC are often categorized together as non-medullary well-differentiated thyroid cancer (DTC), and, in contrast to undifferentiated (anaplastic) TC, have indolent behaviour and can be treated with high survival rates, particularly if they are localized and small-sized (1). TC can occur in any age group but its incidence increases with age (1). This type of cancer (particularly DTC) is 3 times more likely to occur in women than in men and, in the past 2 decades, its incidence has increased (1). The best-established cause of thyroid carcinogenesis is exposure to ionizing radiation, although other candidate risk factors such as dietary iodine deficiency, hormonal factors, benign thyroid conditions and familial history have also been noted (2).

DTC frequency is significantly higher in relatives (particularly first-degree) of DTC patients compared to the general population (2,3). However, familial DTC accounts for only a minor percentage of cases (2), suggesting that other genetic risk factors could be involved. Identifying such individual genetic differences so that these may be used as genetic susceptibility markers for DTC in the remaining sporadic cases is therefore an important challenge. In line with this, several DNA polymorphisms in genes involved in endobiotic or xenobiotic metabolism (including GST and CYP superfamilies), in hormonal and iodine metabolism (such as TG), in cell-cycle control and regulation of apoptosis (such as TP53), in kinase-dependent signaling pathways (such as RET) and in DNA repair (among others) have been associated with differential susceptibility to DTC [reviewed in reference (3)]. The first genome-wide association study (GWAS) performed on TC identified 2 other polymorphisms located near the FOXE1 (TTF-2) and NKX2-1 (TTF-1) genes (which encode for thyroid-specific transcription factors) as strong genetic risk markers of sporadic DTC in European populations (4).

Since patients with papillary TC present a significant increase in DNA damage (5) and DNA repair mechanisms are important in correcting such damage, it is reasonable to assume that defective DNA repair capacity may contribute to DTC risk. Variants in DNA repair genes may affect the DNA repair capacity and, in fact, several single nucleotide polymorphisms (SNPs) in almost all DNA repair pathways have been shown to incrementally contribute to cancer risk (6). Regarding TC, polymorphisms in DNA repair genes such as XRCC1(7–9) and possibly MUTYH(10) (BER pathway), Ku80(11) (NHEJ pathway), BRCA1(12), XRCC3 and possibly RAD51(13) (HR pathway) have been associated with either TC or, more specifically, DTC risk [reviewed in references (3,14,15)].

Nucleotide excision repair (NER) is a versatile DNA repair mechanism capable of repairing UV light-induced lesions, bulky DNA adducts, distorting interstrand crosslinks and even certain oxidative lesions (16). A significant association between an ERCC2 haplotype (rs13181/rs1799793) and TC risk (mainly papillary) was previously reported by our team (17), suggesting that NER polymorphisms may also be relevant for thyroid carcinogenesis. However, to our knowledge, no other published study has thus far focused on a possible role of NER pathway SNPs in DTC susceptibility.

Therefore, we carried out an exploratory hospital-based case-control study in a Caucasian Portuguese population to evaluate the potential modifying role of a panel of 11 NER pathway SNPs (CCNH rs2230641, CDK7 rs2972388, RAD23B rs1805329, ERCC1 rs3212986, ERCC4 rs1800067, ERCC5 rs17655 and rs2227869, ERCC6 rs4253211 and rs2228529 and XPC rs2228000 and rs2228001) on the individual susceptibility to non-familial DTC.

This study comprised 106 DTC patients and 212 age- and gender-matched controls. The histological classification of DTC cases was 73.6% papillary tumours (78 patients) and 26.4% follicular tumours (28 patients). Table II lists the general characteristics of both case and control populations. In the case group, the frequency of females (90 patients) was significantly higher than the frequency of males (16 patients), in accordance with the worldwide estimation for gender distribution in DTC (1). Case and control populations are not statistically different in respect to age distribution, gender and smoking habits.

This report includes a set of 11 SNPs in 8 NER pathway genes (Table I). The MAF and genotypic frequencies of these SNPs, in both DTC cases and controls, are depicted in Table III. All SNPs considered are in Hardy-Weinberg equilibrium (P≥0.05), both in case and control populations, except for rs2972388 (CDK7) and rs2228000 (XPC) that show a significant deviation in the control (rs2972388) or case population (rs2228000).

Genotypic distributions were compared among cases and controls (Table III). A significant difference was observed only for rs2230641 (CCNH) (P=0.04). When considering a dominant model of inheritance, the difference in rs2230641 (CCNH) genotypic frequency between cases and controls was even more significant (P=0.02).

Through logistic regression analysis (Table III), we observed that the rs2230641 (CCNH) heterozygous genotype was significantly associated with increased DTC risk (adjusted OR=1.89, 95% CI=1.14–3.14, P=0.01). Similar results were verified when considering at least one variant allele (adjusted OR=1.79, 95% CI=1.09–2.93, P=0.02), further supporting an association between rs2230641 (CCNH) and DTC risk. Also, an almost significant association between the rs2227869 (ERCC5) heterozygous genotype and reduced DTC risk was observed (adjusted OR=0.39, 95% CI=0.16–1.00, P=0.05).

Following stratification according to histological criteria (Table IV), the DTC risk increase observed for rs2230641 (CCNH) remained in the follicular subset (adjusted OR=2.72, 95% CI=1.19–6.22, P=0.02, for heterozygous; adjusted OR=2.44, 95% CI=1.07–5.55, P=0.03, for variant allele carriers) and almost reached significance in the papillary subset (adjusted OR=1.74, 95% CI=0.99–3.07, P=0.05, for heterozygous; adjusted OR=1.69, 95% CI=0.98–2.92, P=0.06, for variant allele carriers), supporting the idea that this polymorphism may influence DTC susceptibility, irrespective of tumour type. The risk of papillary TC was significantly increased in rs2228001 (XPC) homozygous variant individuals (adjusted OR=2.33, 95% CI=1.05–5.16, P=0.04) and significantly reduced in rs2227869 (ERCC5) variant allele carriers (adjusted OR=0.25, 95% CI=0.07–0.86, P=0.03) and heterozygous individuals (adjusted OR=0.26, 95% CI=0.08–0.90, P=0.03). No significant differences in genotypic frequencies or adjusted ORs were observed for the remaining SNPs, either when considering overall DTC cases or its histological subsets, suggesting that these SNPs alone do not contribute to individual susceptibility to DTC.

To assess the effect of combined genotypes, further statistical analysis was applied to those SNPs that are located in the same gene (rs17655 and rs2227869 on ERCC5; rs4253211 and rs2228529 on ERCC6; rs2228000 and rs2228001 on XPC). Also, since SNPs in different NER genes may influence the way their expression products interact (hence, their repair activity), we also analyzed SNP-SNP interactions between different genes, as long as these interactions were biologically plausible. The genotype distribution of the rs17655/rs2227869 (ERCC5) combination was significantly different in cases and controls (P=0.01); however, no specific ERCC5 genotype combination was associated with altered DTC risk (data not shown), probably due to the low number of patients included in each genetic subgroup. None of the remaining genotype combinations showed association with disease (data not shown).

We conducted a hospital-based case-control study in a Caucasian Portuguese population to evaluate the potential modifying role of a comprehensive selection of 11 SNPs in 8 NER pathway genes in individual susceptibility to non-familial DTC. Overall, we observed that NER polymorphisms such as rs2230641 (CCNH) and, possibly, rs2227869 (ERCC5) and rs2228001 (XPC) were associated with DTC susceptibility. A consistent risk increase was observed for rs2230641 (CCNH) heterozygous and variant allele carriers, compared to wild-type individuals, both when considering all DTC cases and only the follicular subset. When considering only the papillary subset, the risk association almost reached significance for rs2230641 (CCNH) but was significant for the rs2228001 (XPC) variant genotype (increased risk). Also, a protective effect was observed for rs2227869 (ERCC5) in heterozygous and variant allele carriers, in the papillary subset. The association between the rs2227869 (ERCC5) heterozygous genotype and reduced DTC risk almost reached significance when all DTC cases were considered. No other significant correlation was observed for the remaining SNPs.

To our knowledge, this is the first study to assess the effect of these SNPs on DTC susceptibility. Our previous study reporting an association between an ERCC2 haplotype (rs13181/rs1799793) and increased TC (mainly papillary) risk (17) is the only other association study that we are aware of, focusing on the role of NER pathway SNPs in TC susceptibility. However, studies correlating the polymorphisms considered in this study with cancer risk have been published for other types of cancer.

Concerning rs2230641 (CCNH), studies on oesophageal (22) bladder (23) and renal cell carcinoma (24) have yielded mostly negative results. Two significant associations have been reported, with opposite findings; according to Enjuanes et al(25), the minor allele is associated with decreased risk for chronic lymphocytic leukaemia; on the contrary, in line with our results, Chen et al(26), observed, in ever smokers, a significant association for the rs2230641 variant allele with bladder cancer risk and an almost 30-fold increased risk in carriers of the rs2230641 (CCNH), rs2228526 (ERCC6) and rs1805329 (RAD23B) variant alleles.

Previous evidence on the role of XPC polymorphisms on cancer risk is also conflicting; both rs2228001 and rs2228000 have been extensively investigated in case-control cancer association studies but, when considered separately, results are mostly negative, possibly due to insufficient sample size. Increasing the power through performance of meta-analysis has revealed small but significant increases in overall (27–29), bladder (27–31) and breast (29) cancer risk for rs2228000 and in overall (28,29), lung (27–30), bladder (29) and colorectal (29) cancer risk for rs2228001. Meta-analyses yielding negative results have also been published (32,33). The effect of XPC polymorphisms may be best represented by its haplotype since significantly increased cancer risk has been observed more frequently when considering both polymorphisms together, as a haplotype block (22,34), or in combined analysis with other NER variants (23,35,36).

With respect to rs2227869 (ERCC5), notably, results similar to our own have been reported by Hussain et al(37) in the only cancer association study we retrieved from our literature review; decreased stomach cancer risk was observed in heterozygous individuals, according to this study.

As for the remaining SNPs, rs3212986 (ERCC1), rs1800067 (ERCC4), rs17655 (ERCC5) and rs1805329 (RAD23B) have been extensively analyzed in prior cancer association studies, mostly with negative results. For rs3212986 (ERCC1), rs1800067 (ERCC4) and rs17655 (ERCC5) these findings are further corroborated by several meta-analyses that, in agreement with our report, demonstrated no clear association with overall (38–41), lung (42) or breast (43) cancer risk. rs2972388 (CDK7), rs4253211 and rs2228529 (ERCC6) have received little attention in the context of cancer susceptibility but the unique reports we found for each of these SNPs are substantially different from ours, as rs2972388 (CDK7) and rs2228529 (ERCC6) were associated, respectively, with increased breast cancer risk in a Korean population (44) and increased non-melanoma skin cancer risk in an American population (45). For rs4253211 (ERCC6), the variant allele seems to confer a protective effect towards laryngeal (36) and oesophageal (22) cancer, but no association with bladder cancer was observed (23).

CCNH codes for Cyclin H, a protein that, together with CDK7 and MAT1, forms the cyclin-activated kinase (CAK) complex. CAK integrates TFIIH, a larger complex implicated in DNA denaturation prior to damage excision. CAK can also phosphorylate nuclear receptors such as the retinoic acid or the estrogen receptors and a very different range of substrates (16). It is also involved in cell cycle regulation (46). Although data on the functional consequences of rs2230641 is lacking, the pleiotropic effects of CCNH on NER, cell cycle regulation and oestrogen receptor phosphorylation, among others, confer biological plausibility to our hypothesis that CCNH variants (namely, rs2230641) may be involved in cancer susceptibility. Its role in oestrogen receptor phosphorylation could be of particular significance for DTC, which, as previously noted, is an endocrine tumour 3 times more prevalent in women than in men. Discrepancies with prior studies could therefore be explained on the basis of the specific hormonal involvement on DTC, insignificant for the types of cancer previously evaluated.

XPC codes for a DNA binding protein that, together with RAD23B and centrin 2, forms the distortion-sensing component of NER, thus playing a central role in the process of early damage recognition (16,47). XPC is also involved in DNA damage-induced cell cycle checkpoint regulation and apoptosis, removal of oxidative DNA damage and redox homeostasis (47,48). XPC deficiency has been correlated with decreased DNA repair capacity (48), and hypersensitivity to DNA-oxidizing agents such as X-rays (47), a well-known DTC risk factor, providing the rational basis for a putative involvement of XPC polymorphisms in DTC susceptibility. Notably, multinodular TC was recently reported as the most frequently observed internal tumour in Xeroderma pigmentosum type C patients (49), further substantiating the potential role of XPC in thyroid carcinogenesis. rs2228001 originates an aminoacid substitution in the interaction domain with TFIIH. In silico analysis indicates that rs2228001 may possibly be damaging and in vitro evidence demonstrates differential repair capacity [reviewed in reference (27)]. rs2228000 is located in the interaction domain of XPC with RAD23B and, although its functional significance remains unclear, it is predicted to be benign through in silico analysis (27). It is possible that both these variants, particularly their haplotypes, may alter NER capacity, thereby modulating cancer susceptibility. Despite the numerous case-control studies and meta-analyses that exist, clinical evidence is conflicting and, thus, further studies are warranted.

ERCC5 codes for an endonuclease that exerts its activity at the 3′ side of the damaged strand (16). ERCC5 also plays a structural role, stabilizing the TFIIH complex; in its absence, the CAK complex and the ERCC2 subunit dissociate from the TFIIH core (50). Point mutations in ERCC5 may give rise to Xeroderma pigmentosum and Cockayne syndrome, highlighting its importance for effective DNA repair.

It is possible that NER polymorphisms, through impairing oxidative DNA damage repair, may contribute to DTC development. In addition, it is possible that the pleiotropic actions of some NER proteins (such as CCNH or XPC, demonstrated to be involved in cell cycle regulation, apoptosis or hormone signalling), may convey these specific proteins a relevant role in carcinogenesis, particularly DTC.

Discrepancies from prior studies may have originated from the inherent characteristics of each cancer and respective organ. Divergent genetic background and environmental exposure of study populations may have also contributed. The low statistical power inherent to small sample use may explain some of our negative results. The histology-dependent differences observed for some SNPs could derive from different carcinogenesis pathways (hence, different genetic risk factors) among DTC histological types, as occurs between well-differentiated and anaplastic TC, or, more likely, from small sample size on stratified analysis.

Indeed, the main limitation of our study was sample size; small samples may be underpowered to detect modest effects of low penetrance genes and, on the other hand, may increase the probability that findings are attributable to chance, particularly after stratification (small numbers in the subgroups). The success of more sophisticated statistical analysis, such as haplotype analysis and evaluation of gene-gene and gene-environment interactions, is also limited. Moreover, we cannot exclude the possibility that other variants in linkage disequilibrium with the ones considered here could be responsible for the observed associations.

In conclusion, our study provides, for the first time, insight into the potential role of CCNH, ERCC5, XPC and other NER polymorphisms in DTC susceptibility. Additional studies with larger sample sizes are necessary to validate our findings and to provide conclusive evidence for associations between these and other NER variants and DTC risk. Such studies should be powered to allow for haplotype analysis and evaluation of gene-gene and gene-environment interactions. Functional studies are also warranted, as well as a broader analysis of the involvement of NER variants in DTC progression and therapy response.