Association of potentially functional variants in the XPG gene with neuroblastoma risk in a Chinese population

Abstract XPG gene plays a critical role in the nucleotide excision repair pathway. However, the association between XPG gene polymorphisms and neuroblastoma risk has not been investigated. In this study with 256 neuroblastoma cases and 531 cancer‐free controls, we investigated the effects of five potentially functional polymorphisms (rs2094258 C>T, rs751402 C>T, rs2296147 T>C, rs1047768 T>C and rs873601G>A) on neuroblastoma risk. We calculated odds ratio (OR) and 95% confidence interval (CI) to evaluate the association between the five selected polymorphisms and neuroblastoma risk. False‐positive report probability (FPRP) was utilized to determine whether significant findings were noteworthy or because of a chance. We also performed genotype–phenotype association analysis to explore the biological plausibility of our findings. We found that the rs2094258 T allele was significantly associated with decreased neuroblastoma risk (CT versus CC: adjusted OR = 0.65, 95% CI = 0.47–0.90, P = 0.010; and CT/TT versus CC: adjusted OR = 0.71, 95% CI = 0.53–0.97, P = 0.030) after adjusting for age and gender. The association was more prominent for subjects with retroperitoneal tumour or early‐stage tumour. We also found that carriers of the 2–3 risk genotypes had a significantly increased neuroblastoma risk when compared to carriers of the 0–1 risk genotypes. The association with risk genotypes was more predominant in older children, females and subjects with retroperitoneal tumour or early stage. Our results were further supported by FPRP analysis and genotype–phenotype association analysis. In conclusion, our study verified that the XPG gene rs2094258 C>T polymorphism may contribute to neuroblastoma susceptibility. Our findings require further validation by studies with larger sample size and concerning different ethnicities.


Introduction
Neuroblastoma has been recognized as the most common childhood extracranial solid tumour, and the third leading cause of tumourrelated death in children [1]. The incidence peaks in infancy with a median age at diagnosis of approximately 17 months [2]. Despite the utilization of multiple modality treatment involving intensive chemotherapy, radiotherapy and autologous bone marrow transplantation, cure rates for high-risk patients remain 40% or less. Survivors are frequently subject to serious lifelong coexisting conditions and have poor outcomes [3]. With the incidence rate of about 7.7 per million, neuroblastoma is the fourth most frequently diagnosed solid tumour in Chinese children after CNS tumours, lymphomas and germ cell tumours (incidence rate: 23.8, 11.0 and 7.8 per million, respectively) [4]. Neuroblastoma is a worldwide public health problem. To date, common environmental exposures that can increase neuroblastoma risk have not been well documented [5,6]. It was suggested that fathers who exposed to some factors (e.g., wood dust, solders, radiation sources and hydrocarbons) were more likely to have children with neuroblastoma [5,6]. However, even if fathers were exposed to the same risk environment, only a small proportion of their offspring finally developed neuroblastoma, indicating that genetic factors (e.g., polymorphism) may also contribute to neuroblastoma [6].
Genome-wide association studies (GWASs) have provided evidence that common genetic variants are associated with neuroblastoma susceptibility, with several neuroblastoma susceptibility loci identified [7][8][9][10][11]. In the first GWAS for neuroblastoma, a total of 1032 neuroblastoma patients and 2043 controls of European descent were genotyped by using Illumina HumanHap550 BeadChip covering approximately 500,000 single nucleotide polymorphisms (SNPs). Sequentially, an addition of 720 patients and 2128 controls were tested to further validate significant SNPs identified using first-stage data. The two-stage GWAS discovered that three SNPs (rs6939340 A>G, rs4712653 T>C and rs9295536 C>A) within the CASC15 gene at 6p22 were associated with neuroblastoma susceptibility [7]. The association between these three SNPs and neuroblastoma susceptibility has been verified by studies with different ethnicities, such as African-Americans [12], Italians [13], Northern Chinese population [14] and Southern Chinese children [15]. We previously confirmed the association of the three SNPs with decreased neuroblastoma risk in Southern Chinese children with a total of 201 neuroblastoma cases and 531 cancer-free controls [15]. GWASs serve as a powerful and important tool to discover inherited common genetic variations for human diseases including cancer [16,17]. It can identify genes previously not implicated in cancer susceptibility by utilizing an agnostic approach. Thus, because of the adoption of restricted P-value (1 9 10 À5 ), most of the GWAS-identified SNPs only have modest risk effects, usually with odds ratio (OR) ranging from 1.1 to 1.5 [17]. Moreover, the majority of them were located in the intron regions without function. Polymorphisms in the candidate genes, especially functional SNPs, represent a more meaningful way to investigate the role of genes in cancer risk. Previous studies have identified several neuroblastoma risk loci in Chinese populations, such as functional polymorphisms in the FAS/FASL system genes [18] and TGFBR3L gene [19].
DNA repair genes play critical roles in maintaining the integrity and stability of genomic DNA, and about 130 genes have been reported to be involved in the five major DNA repair pathways, including nucleotide excision repair (NER) pathway [20]. The NER pathway are mainly responsible for the removing of DNA adducts and lesions [21]. In humans, xeroderma pigmentosum group G/excision repair cross-complementation group 5 (XPG/ERCC5, MIM:133530) is a member of the xeroderma pigmentosum complementation groups (XPA, XPB, XPC, XPD, XPE, XPF and XPG) [22]. It functions as an endonuclease and cuts the damaged DNA at the 3 0 ends of lesions during the process of DNA repair [23]. Thus far, investigations for the association between XPG gene polymorphisms and cancer susceptibility have mainly focused on rs17655G>C polymorphism (Asp1104His) and conclusions were controversial [24]. The associations of potentially functional polymorphisms in the XPG gene with neuroblastoma susceptibility have not been investigated yet. With these in mind, we performed this study including 256 neuroblastoma cases and 531 cancer-free controls from Southern Chinese children to investigate the association between five potentially functional polymorphisms (rs2094258 C>T, rs751402 C>T, rs2296147 T>C, rs1047768 T>C and rs873601G>A) and neuroblastoma susceptibility.

Study subjects
The current retrospective hospital-based case-control study consisted of 256 patients with newly diagnosed and histopathologically confirmed neuroblastoma. Patients were enrolled from the Department of Pediatric Surgery, Guangzhou Women and Children's Medical Center, mainly between February 2010 and November 2015. We also recruited 531 age-, gender-, and ethnicity-matched cancer-free controls from the same geographical regions (Table S1) as described previously [15,25]. Written informed consent was obtained from each subject or his/her guardian. The response rate was about 90% for neuroblastoma subjects and 95% for cancer-free controls. The study protocol was approved by the Institutional Review Board of Guangzhou Women and Children's Medical Center.

SNP selection and genotyping
As shown in Table S2, we chose five potentially functional XPG SNPs (located in the 5 0 -flanking region, 5 0 UTR, exon and 3 0 UTR) in this study. Of them, three (rs2094258 C>T, rs2296147 T>C and rs873601G>A) were reported as potentially functional SNPs in a previous investigation [26]. The rs751402 C>T polymorphism located in the 5 0 UTR was also reported in another study [27]. Besides, we chose rs1047768 T>C since it can lead to splicing alteration as predicted by online website SNPinfo (http://snpinfo.niehs.nih.gov/). The widely studied rs17655G>C polymorphism was not included in this study because of strong evidence of linkage disequilibrium between this SNP and rs873601G>A polymorphism (R 2 = 0.91). All of these chosen polymorphisms had a minor allele frequency >5% in Chinese Han subjects. Single nucleotide polymorphism genotyping was performed in 384-well plates using Taqman methodology as described elsewhere [15,26].

Genotype-phenotype correlation analysis
We performed genotype-phenotype correlation analysis in order to provide biological evidence for our findings. We explored the effects of XPG SNPs on gene expression by evaluating the correlation between genotype data of 270 individuals from HapMap phase II release 23 data set containing genotypes of 3.96 million SNPs from four populations (http://hapmap.ncbi.nlm.nih.gov/), and corresponding XPG gene transcripts from EBV-transformed B lymphoblastoid cell lines of the same 270 subjects from SNPexp (http://app3.titan.uio.no/biotools/help.php? app5snpexp). The four populations were as follows: CEU: 90 Utah residents with ancestry from northern and western Europe; CHB: 45 unrelated Han Chinese in Beijing; JPT: 45 unrelated Japanese in Tokyo; YRI: 90 Yoruba in Ibadan, Nigeria. For more details on the study subjects and methodology, refer to previous publications [26,28,29].

Statistical analysis
Distributions of demographic variables and genotype frequencies in neuroblastoma cases and controls were evaluated using chi-squared test. Hardy-Weinberg equilibrium (HWE) was calculated for controls using the goodness-of-chi-squared test. Odds ratios and corresponding 95% confidence intervals (CIs) were used to estimate the association between selected polymorphisms and neuroblastoma risk. Adjusted ORs were calculated by multivariate analysis with unconditional logistic regression, with adjustment for age and gender. We also computed the false-positive report probability (FPRP) values for the significant findings. We set 0.2 as FPRP threshold and adopt a prior probability of 0.1 to detect OR of 1.50/0.67 (risk/protective effects) as described previously [30,31]. The association that reached the FPRP threshold of <0.2 was considered noteworthy. All statistical tests were two-sided, with a significance level of P < 0.05. All statistical analyses were performed using SAS software (Version 9.1; SAS Institute, Cary, NC, USA).

Associations between XPG gene polymorphisms and neuroblastoma risk
Of the enrolled subjects, 248 neuroblastoma cases and 531 cancerfree controls were successfully genotyped. The allele and genotype frequencies of the selected SNPs and their associations with neuroblastoma risk are summarized in Table 1. All of the genotype distributions for the five selected polymorphisms were consistent with the HWE (P = 0.701 for rs2094258 C>T, P = 0.380 for rs751402 C>T, P = 0.583 for rs2296147 T>C, P = 0.409 for rs1047768 T>C and P = 0.686 for rs873601G>A) in control subjects. When compared with the rs2094258 CC genotype, the variant CT and CT/TT genotypes were significantly associated with decreased neuroblastoma risk (CT versus CC: adjusted OR = 0.65, 95% CI = 0.47-0.90, P = 0.010; and CT/TT versus CC: adjusted OR = 0.71, 95% CI = 0.53-0.97, P = 0.030) after adjusting for age and gender. We also observed a borderline statistically significantly increased neuroblastoma risk in the carriers of the rs1047768 C allele (CC versus TT: adjusted OR = 1.73, 95% CI = 0.93-3.21, P = 0.083; and CC versus TT/CT: adjusted OR = 1.68, 95% CI = 0.92-3.08, P = 0.092). However, there was no significant association or trend observed between the rest of the three polymorphisms and neuroblastoma risk (Table 1). Compared to the subjects without risk genotype, we found a signifi-cant trend toward increased risk for subjects carrying 0-3 risk genotypes (adjusted OR = 1.29, 95% CI = 1.06-1.56, P = 0.011). We also observed carriers of 2-3 risk genotypes were significantly more predisposed to developing neuroblastoma when compared to carriers of 0-1 risk genotypes (adjusted OR = 1.47, 95% CI = 1.08-1.99, P = 0.013).
As shown in Table 3, at the prior probability level of 0.1 and FPRP threshold of 0.2, the association between XPG rs2094258 CT genotype and neuroblastoma risk remained noteworthy (FPRP = 0.136). Regarding stratified analysis, the significant increase in neuroblastoma risk for carrier of CT/TT genotypes was noteworthy in the subgroup with early-stage tumour (FPRP = 0.186). Moreover, in the combined analysis, the associations between 2 and 3 risk genotypes and neuroblastoma risk also reached the FPRP threshold of <0.2, and were considered deserving of attention: the association in the whole study population (FPRP = 0.154), the association in the subgroup with retroperitoneal tumour (FPRP = 0.127) and subgroup with early-stage tumour (FPRP = 0.043).

Genotype-based mRNA expression analysis
Results of the genotype-based XPG mRNA expression analysis for rs2094258 C>T and rs1047768 T>C polymorphisms were shown in Table 4. We found the XPG mRNA expression levels in rs2094258 CT genotype carriers were significantly elevated when compared to the CC genotype carriers in Africans (P = 0.004) and Chinese subjects (P = 0.044). We also found a higher XPG mRNA expression for rs2094258 CT/TT genotype carriers for Africans (P = 0.005) and Chinese subjects (P = 0.027). For all subjects, we found a trend toward increased XPG mRNA expression (P = 0.074). These results were consistent with our findings from the association study. The decreased neuroblastoma susceptibility might be partially attributed to the decrease in XPG mRNA expression levels. As to the rs1047768 T>C polymorphism, we only found significantly higher expression in  the CC and CT/CC genotype carriers than in the TT genotype carriers for all the mixed subjects (P = 0.029 and P = 0.040, respectively).

Discussion
In the present hospital-based case-control study, we explored the associations between five potentially functional polymorphisms in the XPG gene and the risk of neuroblastoma. We observed significant association between rs2094258 C>T polymorphism and neuroblastoma susceptibility among Southern Chinese children. We found that individuals with rs2094258 CT and CT/TT genotypes were at significantly decreased neuroblastoma risk when compared with those with the CC genotype. We also found that the 2-3 risk genotype carriers had a significantly higher risk of neuroblastoma than the 0-1 risk genotype carriers. The FPRP analysis strengthened the significant associations, while genotype-based mRNA expression analysis further provided biological evidence for our findings.
The human XPG gene is located on chromosome 13q33 and comprises a 30-kb coding region with 15 exons and 14 introns. This gene encodes a 1186 amino acid structure-specific endonuclease [33]. XPG protein takes part in the transcription-coupled repair [34], as well as the global genomic NER [35]. It also plays an important role in RNA transcription through the interaction with other transcription activator complexes. Improper repair of DNA damage may lead to mutagenesis and cell death [36,37]. The XPG protein can cleave damaged oligonucleotide at the 3 0 end. It also cooperates with the XPF/ERCC1 complex that excises damaged oligonucleotide at the 5 0 end during NER process. Finally, this protein can stabilize the binding of DNA repair complex to damaged DNA [38][39][40]. The sequence variations in the DNA repair genes may alter DNA repair capacity, consequentially causing inter-individual differences in the predisposition to cancer [41]. The XPG gene is highly polymorphic. According to the dbSNP database (http://www.ncbi.nlm.nih.gov/SNP), there are at least 2432 identified SNPs in the XPG gene region, and 654 of them are coding region SNPs. SNPs may lead to the encoded amino acids changes (non-synonymous), may be silent (synonymous) or may occur in non-coding regions (transcription factor binding sites or miRNA binding sites). The non-synonymous SNPs may affect gene function and phenotype and predispose to diseases [42]. The most widely investigated SNP in the XPG gene is rs17655G>C, which leads aspartate to histidine alteration at codon 1104 [24]. The association with Asp1104His polymorphism has been widely investigated in different types of cancers, including breast cancer, skin cancer, lung cancer, bladder cancer, head and neck cancer, colorectal cancer and non-Hodgkin lymphoma [24].
Moreover, three potentially functional XPG polymorphisms (rs2094258 C>T, rs2296147 T>C and rs873601G>A) were studied for the association with gastric cancer in a case-control study with 1125 cases and 1196 controls by He et al. [26]. They found that the rs873601G>A polymorphism (located in the 3 0 UTR) was significantly associated with increased gastric cancer risk. They also found that the rs873601 A allele were associated with decreased mRNA expression. Zhu et al. [43] also explored the association between these three polymorphisms and oesophageal squamous cell carcinoma (ESCC) susceptibility. They reported a significant association between the rs2296147 C allele and decreased ESCC susceptibility. Duan et al. [27] explored the association of rs751402 C>T and rs2296147 T>C polymorphisms with gastric cancer in 403 cases and 403 controls. They found that both of two polymorphisms were associated with an increased gastric cancer risk. Yang et al. [44] also genotyped these three polymorphisms in 337 stomach cancer cases and 347 controls. Data showed that the rs2296147 T>C polymorphism was associated with decreased gastric cancer risk, while the rs2094258 C>T  polymorphism was associated with increased gastric cancer risk. In a study with 241 prostate cancer cases and 264 controls [45], the rs2296147 T>C polymorphism was shown to associate with increased prostate cancer risk. However, the association with the rs2094258 C>T polymorphism was not replicated in the same study. In the study with 325 breast cancer cases and 325 controls, Na et al. [46] tested all of the five polymorphisms included in our study and found significant association between the rs2094258 TT genotype and increased breast cancer risk. Sun et al. [47] failed to find any significant association between rs2094258 C>T polymorphism and laryngeal cancer susceptibility with a total of 271 cases and 271 controls.
To the best of our knowledge, this is the first investigation on the association between XPG gene polymorphisms and neuroblastoma susceptibility. In this study, we found the rs2094258 C>T polymorphism was associated with decreased neuroblastoma risk. This protective effect of the SNP may be ascribed to the resultant upregulation of the XPG gene expression. As predicted by SNPinfo (Table S2), the rs2094258 C>T polymorphism is located in the 5 0 near region of XPG gene and is within a transcription factor binding site. The C>T alteration may lead to an increase in the XPG mRNA expression (Table 4) in Chinese subjects as well as in overall individuals. The rs2094258 C>T polymorphism-induced increase in XPG gene expression may enhance an individual's DNA repair capacity, which support the protective association between the rs2094258 T allele and decreased neuroblastoma. Previous studies reported that this polymorphism could increase or decrease cancer risk. The different results may be ascribed to the fact that different cancers may have different environment exposures as well as the difference in sampling in each investigation. Besides, gene-environment interaction may also play important roles in the tumourigenesis [48]. It's important to perform FPRP analysis to verify if the significant findings were chance findings or really noteworthy. Moreover, further functional studies are needed to explore the specific mechanisms by which this polymorphism modifies cancer susceptibility.
Though this is the first and largest study to investigate the association of DNA repair gene polymorphisms with neuroblastoma in Chinese children, several limitations in this study should be addressed. First, only 256 cases and 531 controls were included in this study. The relatively small sample size, especially for the cases, may result in limited statistical power. Second, selection bias may exist; since all the subjects were enrolled only from our hospital and restricted to a Chinese Han population, we might miss a larger number of neuroblastoma cases who did not visit our hospital for treatment during the same period. Third, we only included five potentially functional polymorphisms in this study. In the future, all functional SNPs in XPG gene should be investigated in different ethnicities, which will yield a meaningful conclusion. Finally, some important information was not available in our study, such as the paternal exposures, living environment and dietary intake for the included children, which limited our ability to perform gene-environmental interactions analysis in neuroblastoma susceptibility.
In conclusion, this study provides evidence that potentially functional polymorphisms in the XPG gene, especially the  The significant results were in bold, if the 95% CI excluded 1 or P <0.05. *Genotyping data and mRNA expression levels for XPG by genotypes were obtained from the HapMap phase II release 23 data from EBV-transformed lymphoblastoid cell lines from 270 individuals. † Two-side Student's t-test within the stratum. ‡ P-values for the trend test of XPG mRNA expression among three genotypes for each SNP from a general linear model.

Conflicts of interest
The authors declare no competing financial interests.

Supporting information
Additional Supporting Information may be found in the online version of this article: Table S1 Frequency distribution of selected characteristics in neuroblastoma patients and controls.