In this large-scale case-control study, we investigated the associations between five tagging SNPs of the DNA repair gene XPC and risk of lung cancer in a Chinese population. Our results showed that the rs3731055 AG+AA genotype was associated with a decreased overall risk of lung cancer, especially among young subjects (age ≤ 60 years old), non-smokers, and patients with lung AC, but an increased risk of SCLC. When we evaluated the haplotypes derived from all 5 tagging SNPs, we also found that the haplotype ACCCA containing the rs3731055 A allele was significantly associated with a decreased risk of lung AC but an increased risk of SCLC. Considering both potential biological functions and use of the tagging SNPs representative of other untyped SNPs, our results may be due to the rs3731055 SNP that is in LD with A499V and K939Q (r2 = 0.17 and 0.21, respectively, in this study population, stronger than 0.025 and 0.040, respectively, obtained from the mixed populations in the NIEHS SNP database), or it is likely that the rs3731055 SNP may be in LD with other untyped disease-causing SNPs. In addition, studies showed that the XPC promoter region contains some binding sites of transcription factors, such as p53 , AP1, and EGR1 ; thus, rs3731055 G > A change might alter the effect on these protein-DNA interactions. However, the functional relevance of rs3731055 SNP needs further investigations.
XPC is an important damage-recognition protein that recognizes a variety of bulky DNA damage, including UV-induced photolesions and chemical carcinogen-induced DNA adducts, that are repaired by both transcription-coupled and global genome repair processes [29, 30]. XPC can also interact with many other important proteins, such as the transcription factor IIH (i.e., TFIIH) [31, 32] and the centrisome protein Centrin 2 (CEN2) . In addition to its role in DNA repair, XPC also play an important role in cell-cycle arrest and activation of the p53 pathway . Furthermore, reduced XPC mRNA and protein levels were more frequently observed in both XP heterozygotes  and lung cancer patients , suggesting that the amount of XPC may modulate susceptibility to cancer.
Although the XPC protein is known to play an important role in the NER pathway, the results of published association studies on XPC SNPs and risk of lung cancer remain inconsistent. There are only a few published studies that investigated the role of XPC SNPs in the etiology of lung cancer, mostly in Asian populations. For example, Hu et al. reported that compared with the 499 CC (i.e., rs2228000) and 939 AA (i.e., rs2228001) wild-type homozygotes, subjects carrying 499 CT + TT and 939 AC + CC respectively had a 1.57-fold and 1.21-fold increased the risk of lung cancer in a Chinese population , but this association was not observed in another Chinese study . More recently, Lee et al. found that rs3731055 AA genotype was associated with a 2.1-fold increased risk for lung SCC compared to the rs3731055 GG genotype in a Korean population of 432 lung cancer patients and 432 healthy controls. These differences in risk associations may be due to different etiology and mechanisms of lung cancer in the study populations with different ethnic background. In a Spanish population of 359 lung cancer patients and 375 healthy controls, Marin et al. found that the frequency of XPC PAT+ allele was 45.0% in cases and 39.5% in controls, the difference being statistically significant (P = 0.032) . Similarly, Vogel et al.  also reported that XPC Lys939Gln, which is linked with XPC PAT, may be risk factor for lung cancer in another Europe cohort study. In order to verify the association, we also conducted this large-scale study and did not find any significant association on Lys939Gln (37.5% vs. 35.8%). This difference may be due to the different ethnic background or small sample size with limited statistical power.
Some recent studies had shown that mutations in the epidermal growth factor receptor gene, which often took place among the patients with lung AC, were more frequent in never smokers and women in eastern populations, whereas such mutations were more frequent in smokers and men in western populations [40, 41]. These observations suggested that the arising incidence of lung AC may be associated with not only environmental risk factors, such as N-nitrasomines or other carcinogens in the air pollutions[42, 43], but also genetic susceptibility factors in different ethnic groups and possible different smoking behaviors.
A recent animal study had shown that 100% of XPC -deficient mice develop spontaneous lung tumors, the majority of which were adenomas; furthermore, when the mice had XPC and Gadd45a deleted at the same time, their lung adenomas were progressing to non-small cell lung adenocarcinomas . These results suggested that genetic alterations in XPC, in interaction with environmental factors, could result in altered susceptibility to different histological types of lung cancer, particularly in the presence of other genetic susceptibility factors. Indeed, the finding that rs3731055 AG+AA genotype or haplotype ACCCA were associated with an increased risk of SCLC in the present study suggests that different histopathological types may have different etiologies. Recently, Hollander et al. reported that some allelic loss of XPC in the lung of mice, coupled with carcinogens such as polycyclic aromatic hydrocarbons, resulted in highly frequent small cell lung cancer and some non-small cell lung cancer . However, the result on SCLC may be due to chance because of the relatively small number of observations in the subgroup of patients with SCLC.
In the present study, we found that the protective effect of rs3731055 AG+AA genotype was more pronounced among young people (≤ 60 years old), suggesting that such a protective effect may have been diminished because of prolong exposure, as age increased, to N-nitrosomines or other carcinogens When the subjects were divided into three subgroups according to cumulative cigarette consumption (i.e., 0 pack-years, < 30 pack-years, and ≥ 30 pack-years of smoking), we observed that this protective effect was more evident in the never smokers. This result further suggests that cigarette smoking may not be the major pathogenic agent involved in the initiation of lung AC but that some as-yet-unidentified carcinogens may have played a major role in the development of lung AC in this study population. This is consistent with a previous study in which lung AC were more frequent in never smokers than in ever smokers in eastern Asians. However, it is also possi ble that these findings may be due to chance because of the small sample size in the subgroup.
Although the present study was considerably larger than previous studies, it was a hospital-based study that has several limitations. First of all, the participation rate was still relatively low for both cases (77.8%) and controls (81.3%), and about seven percent of DNA samples failed in the genotyping for each locus, which may have increased the probability of selection bias. However, the general demographics and tobacco-exposure information of subjects included in the final analysis were similar with those of people who were excluded, and all lung cancer patients and controls were matched on age, sex, and residential area, which may have minimized the selection bias and confounding factors. Second, because some DNA samples failed in the genotyping, we used the Bayesian statistical method to infer the most probable haplotypes, which may have potential errors. However, the difference in haplotype frequencies between the Stochastic-EM algorithm and the Bayesian method were not significantly different in either cases or controls, increasing the reliability of haplotype estimation. Finally, although we consider both the relevance of biological functions and the representativeness of other untyped SNPs in selecting tagging SNPs of the XPC gene, this study may be limited because of excluding some non-synonymous SNPs with low frequencies, which may be more important in the etiology of lung cancer.