Molecular Epidemiology of Papillary Thyroid Cancer: In Search of Common Genetic Associations

T hyroid cancer is the most common endocrine malignancy and will account for approximately 37,000 new cancers in 2009 (1). There has been a steady increase in the incidence thyroid cancer over the past 15–20 years, particularly for papillary thyroid carcinoma (PTC) in developed countries (2,3). While some have suggested that these trends are principally attributable to a screening bias, others have argued that environmental exposures must be considered (2,3). While ionizing radiation exposure (particularly in childhood or adolescence) is the well-established risk factor for PTC, epidemiologic studies have also attempted to identify associations between thyroid cancer incidence and other potential risk factors, including diet, smoking and alcohol use, and hormonal factors, but no consistent findings have emerged (reviewed by Grubbs et al. [4]). Certainly increases in environmental ionizing radiation exposures could account for the observed increases in thyroid cancer incidence in the U.S. population, but there is not currently data to support this concept. Regardless, identifying markers of susceptibility to thyroid carcinogenesis or risk of thyroid cancer could help modify both prevention and screening.

While families with a strong phenotype of PTC exist, these strict Mendelian defined families are relatively rare and several distinct loci have been identified (5,6). Similarly, inherited syndromes associated with nonmedullary thyroid cancers, such as Werner's syndrome, familial adenomatous polyposis, and Cowden disease, are well-characterized with known germline mutations of specific genes (WRN, APC, and PTEN); however, these syndromes are extremely rare, making their utility for broader screening programs or to understanding population level risks limited (reviewed by Grubbs et al. [4] and Harb and Sturgis [7]). At the same time, it is well understood that PTCs are frequently (up to 87%) multifocal and bilateral, and these cancers often appear to arise as independent clonal events suggesting that an underlying predisposition or genetic association exists (8,9). Furthermore, studies using large population databases in the United States and Europe, found a higher familial or genetic component to the risk for thyroid cancer than for any other major cancer (10,11). These clinical, histopathologic, molecular, and epidemiologic data suggest that a genetic susceptibility to PTC is likely. Unfortunately, while it is well understood that the MAP kinase pathway is central to PTC tumorigenesis, germline mutations in this pathway have yet to be defined as com-mon susceptibility loci for this cancer (12,13). Unlike major susceptibility loci with high penetrance as found in cancer susceptibility syndromes (such as germline RET mutations in multiple endocrine neoplasia syndromes or germline APC mutations in familial adenomatous polyposis syndrome), the genetic variants hypothesized to influence PTC risk are very likely common in the population yet of low penetrance. Furthermore, it may be that multiple common variants found together create a risk profile of PTC susceptibility, but identifying such profiles will be even more complicated.

The high prevalence of these variants in the general population, the likelihood that these variants interact to create genetic profiles of risk, and the potential of major confounding and bias, have created a field of study that demands large sample sizes. To identify these genetic risks factors for what has traditionally been viewed as a sporadic cancer, epidemiologic researchers have taken three general approaches. A candidate gene approach had this ability in clearly genetic syndromes such as multiple endocrine neoplasia or familial adenomatous polyposis or in families with a strong cancer phenotype. However, for a sporadic disease such as the overwhelming majority of PTCs, researchers have also used a pathway-directed approach or a broad-based genome screen approach to compare the frequencies of genetic variants between cancer patients and controls. In this issue of Thyroid, Bastos et al. (14) report a pathway-based case–control study design. The authors report the frequency of common single nucleotide polymorphisms in the homologous recombination DNA repair pathway in 109 nonmedullary thyroid carcinoma patients (78 PTC patients) and 217 cancer-free hospital-based controls. The authors have assembled a homogeneous case population from the central referral hospital for thyroid cancer in the Lisbon region. Furthermore, the cases and controls were closely matched in age and sex, and all subjects lacked a family history of thyroid cancer or known radiation exposures. The authors chose the homologous recombination DNA repair pathway for study, because it is the central mechanism by which cells repair the principal genetic damage induced by ionizing radiation (i.e., double DNA strand breaks). The premise of this work is that while none of the participants had known radiation exposures, all humans experience ionizing radiation exposure by virtue of living on earth, and it is variations in the ability to repair the genetic damage of ionizing radiation that are most likely to be markers for PTC risk. While several genes make up this pathway, the current analysis is of three polymorphisms affecting amino acid changes (XRCC2 R188H, XRCC3 T241M, and NBS1 E185Q) and a 5′ untranslated region polymorphism of RAD51, the gene that encodes the central protein of the homologous repair pathway. Within the control population, all polymorphisms were distributed within expectations of the Hardy–Weinberg equilibrium, supporting that the selection of the control population had not introduced an unusual bias in the distribution of these variant genotypes. Very consistent with an earlier report, Bastos et al. found that the homozygous polymorphic XRCC3 genotype was associated with a twofold increased risk for nonmedullary thyroid cancer (and PTC specifically) as compared to those with the wild-type allele (14,15). Furthermore, the authors found a twofold risk of nonmedullary thyroid cancer associated with the homozygous polymorphic RAD51 genotype, and when the XRCC3 and RAD51 genotypes were combined into a risk profile there was a significant trend of increasing risk for nonmedullary thyroid cancer (and PTC specifically) with an increasing number of polymorphic alleles. While the study is relatively small and the risk estimates broad, the findings are both confirmatory to previous work and also extend the field into the exploration of genetic profiles of thyroid cancer risk. Furthermore, the authors should be commended for assembling a uniform case population, and it is hoped that their future efforts will expand their database to allow even more sophisticated analyses and to explore other potential genetic risks for PTC.

Several lines of evidence support the authors' choice of homologous recombination DNA repair pathway to explore for polymorphisms associated with PTC. Firstly and most simplistically, ionizing radiation is the only clearly documented environmental risk factor for PTC, and DNA damage (chiefly, double strand breaks) induced by ionizing radiation is repaired via this pathway. It would seem logical that individuals having variations in this pathway could have differential risk for PTC. Further supporting this concept is that the common molecular alteration in PTC known to be associated with ionizing radiation exposure is a somatic RET/PTC rearrangement, an event that requires strand breaks to occur (16 –18). Other evidence has shown that the RET proto-oncogene has greater sensitivity to fragmentation after in vitro exposure to ionizing radiation, and this sensitivity was greater when thyroid-derived cells were used in such assays as opposed to lymphoblasts (19). Additionally, younger age of onset, often considered a marker of cancer susceptibility, appears associated with the RET/PTC rearrangement (16,18,20). In vitro assays utilizing peripheral blood lymphocytes have shown that both endogenous cytogenetic damage as well as such damage induced by gamma irradiation is significantly higher for nonmedullary thyroid cancer patients than for cancer-free controls, and suggests that pathways responsible for repair or removal of such damage are less efficient in thyroid cancer patients (21 –23). Finally, several authors have suggested that germline variants of DNA repair genes contribute to the risk of nonmedullary thyroid cancer (reviewed by Adjadj et al. [24]) (15,25 –31). While some authors have confirmed the findings here with regard to the risk associated with the XRCC3 T241M polymorphism, others have not; data on the other three polymorphisms reported here do not exist (15,25,31).

Certainly, other pathways of interest to the pathogenesis of PTC exist and deserve study whether associated with the DNA repair process or not: DNA damage recognition, base excision repair, nucleotide excision repair, cell cycle control, apoptosis, hormonal metabolism and signaling, kinase-dependent signaling, and endogenous and exogenous carcinogen metabolism (reviewed by Adjadj et al. [24]) (15,25 –38). Several of these have received early attention and appear to be associated with PTC risk based on some confirmatory results in more than one study (in particular, p53 codon 72Pro, XRCC1 Codon 194Trp and 399Arg, and GSTM1+T1null) (reviewed by Adjadj et al. [24]) (25,29,33,35,37,38). Genome-wide association studies have been conducted and are a powerful tool to identify novel loci that may affect risk of various cancers. In a recent high-impact publication, Gudmundsson et al. (39) used a genome-wide screening approach too and identified common variants on 9q22.33 and 14q13.3, which together were associated with an almost sixfold risk of nonmedullary thyroid cancer. The authors demonstrated that the 9q22.33 variant was associated with alterations in thyroid-related hormone levels and suggest that these variations may reflect altered functions of the nearest gene (FOXE1 [TTF2]). Future efforts will need not only confirmation of current pathway-based and genome-wide–based results but also greater exploration of haplotypes, gene–gene, and gene–environment interactions to better model PTC risk. Such efforts are daunting as they require much larger sample sizes and more sophisticated analyses. Additionally, PTC is clearly a mixture of diseases demographically (young vs. old), environmentally (radiation exposure vs. no known exposure), and molecularly (BRAF mutations vs. RET/PTC rearrangements), and we encourage future reports of PTC genetic susceptibility to stratify cases by age, exposure, and mutation profile. Nonetheless, such stratifications and/or genome-wide association studies are extremely demanding in terms of sample size and validation, and these limitations emphasize the need for international consortia to study thyroid cancer susceptibility.