Genetic variants in the nucleotide excision repair pathway genes and gastric cancer susceptibility in a southern Chinese population

Abstract

Background

Potentially functional polymorphisms can modulate protein activities and host’s DNA repair capacity, thereby influencing cancer susceptibility. The association of the polymorphisms in the nucleotide excision repair core pathway genes and gastric cancer susceptibility remains largely unknown.

Methods

Here, we systematically analyzed the associations between nine polymorphisms in four key genes (XPA, ERCC1, ERCC2, and ERCC4) in the nucleotide excision repair pathway and gastric cancer risk in a Chinese population including 1142 patients and 1173 controls. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to estimate the risk associations.

Results

We observed that ERCC1 rs2298881 CA variant genotype was associated with an increased gastric cancer risk (CA vs. CC: adjusted OR [AOR]=1.33, 95% CI=1.09–1.62; dominant model: AOR=1.32, 95% CI=1.10–1.60). However, ERCC1 rs3212986 AA variant genotype was identified as a protective factor for gastric cancer (AA vs. CC: AOR=0.73, 95% CI=0.54–0.98; recessive model: AOR=0.72, 95% CI=0.54–0.96). Genotype-based mRNA expression analysis further indicated that the rs2298881 A allele was associated with decreased ERCC1 mRNA expression.

Conclusion

In all, these results indicated that the ERCC1 polymorphisms may affect the risk of gastric cancer in the Chinese Han population.

Keywords:

• gastric cancer
• DNA repair
• NER
• polymorphism
• susceptibility

Introduction

Gastric cancer, one of the most lethal malignancies, is the fourth most common cancer and the second leading deadly cancer in the world.^1,2 According to statistics of the National Central Cancer Registry of China, gastric cancer ranks second in both incidence and mortality of cancers in China.³ Despite remarkable progress, the current treatments for gastric cancer are still not efficacious with overall 5-year survival rates <30%.⁴ One of the main reasons for such a predicament might be that most patients were diagnosed at advanced stages of the disease.⁵ Understanding the underlying mechanisms of gastric cancer initiation and progression may promote biomarker development for early detection of cancer.

Increasing evidence has proven that both environmental and genetic factors contribute to the occurrence and development of gastric cancer.⁶ Helicobacter pylori infection is a well-established risk factor for gastric cancer, affecting >60% of all gastric cancer cases.^7,8 However, not all the H. pylori-infected patients finally develop gastric cancer. Many other factors also play roles in gastric carcinogenesis, including micronutrient deficiencies, high body mass index, a high salt or a low fiber diet, over consumption of tobacco or alcohol, as well as genetic risk factors.^9–11 Increasing numbers of genetic variations have been found to influence susceptibility to gastric cancer in the previous epidemiological studies.^12,13

The integrity and stability of the genome are primarily maintained by DNA repair systems, which include base excision repair, double strand break repair, mismatch repair, and nucleotide excision repair (NER).^14,15 Among these systems, NER system plays a major role in monitoring and repairing DNA damages caused by exogenous and endogenous factors.¹⁶ Defects in the NER system might threaten the integrity of genome and thus lead to the development of disease.¹⁷ It is elucidated that reduced DNA repair capacity is most frequently associated with various human diseases including cancer.¹⁸ NER process consists of four main steps: damage recognition, damage unwinding, damage incision, and new strand ligation.¹⁹ There are at least eight key proteins (complementation groups XP-A to G and ERCC1) identified to limit the rate of NER process.²⁰ Specifically, XPA and XPC play critical roles in recognizing the DNA damage^21,22; XPD and XPB are responsible for the process of damage unwinding^23,24; ERCC1, XPF, and XPG are all essential components for the DNA damage incision.^25,26

Thus far, several studies have been reported concerning the association between the polymorphisms in the NER pathway genes and the outcomes of gastric cancer.^27,28 However, the association of these polymorphisms with gastric cancer risk was not fully elucidated. Therefore, the aim of this study was to further identify the association between these polymorphisms and gastric cancer susceptibility. In this study, we systematically analyzed the association between nine potential functional single nucleotide polymorphisms (SNPs) in the NER pathway genes (XPA, ERCC1, ERCC2, and ERCC4) and gastric cancer risk using 1142 patients and 1173 cancer-free controls in a southern Chinese population.

Materials and methods

Study population

This study was approved by the Institutional Review Board of Sun Yat-sen University Cancer Center, Guangzhou, Guangdong. The case group comprised 1142 patients with histologically confirmed gastric cancer enrolled from Sun Yat-sen University Cancer Center from February 2002 to September 2013. The control group consisted of 1173 healthy controls randomly recruited from the same region.^29,30 Enrollment was restricted to unrelated ethnic Han Chinese population from South China. Detailed information was obtained on all subjects, including demographic characteristics (e.g., age and sex), and lifestyle habits (e.g., smoking habits and alcohol drinking). The classification criteria for smoking status and drinking status were described elsewhere.³¹ Written informed consent was acquired from each participant, accompanying with a donation of 5 mL of venous blood sample.

Polymorphism selection and genotyping

The potentially functional polymorphisms of main genes in NER pathway were selected from dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP). Specifically, the following items were set as the selection criteria: 1) located at the 5′ untranslated regions (UTR), upstream promoter region, coding region, and 3′ UTR of genes; 2) the minor allele frequency was >5% in Chinese Han populations; 3) no obvious linkage between paired SNPs in linkage disequilibrium (R²<0.8). We also adopted SNPinfo (http://snpinfo.niehs.nih. gov/snpfunc.htm) to predict the potential functions of those polymorphisms; they could affect the activity of transcription factor binding sites or microRNA binding sites. As a result, the following polymorphisms were included: XPA (rs1800975 G>A, rs3176752 C>A); ERCC1 (rs2298881 C>A, rs11615 G>A, rs3212986 C>A); ERCC2 (rs3810366 C>G, rs238406 G>T, rs13181 T>G); and ERCC4 rs2276466 C>G.

DNA was extracted from the blood samples using QIAamp DNA Blood mini kit (QIAGEN Inc, Valencia, CA, USA). Genotyping were performed by the Taqman real-time PCR method on 7900 Sequence Detection System (Applied Biosystems, Foster City, CA, USA), as previously described.^31–34 For quality control purposes, four duplicate positive controls and four negative controls (without DNA) were used in each of 384-well plates. Moreover, 10% of the samples were randomly selected to re-genotype. There was 100% genotype concordance for each polymorphism among duplicates.

Statistical analysis

First, we adopted goodness-of-fit χ²-test to check whether genotype frequencies of each polymorphism in controls were in Hardy–Weinberg equilibrium (HWE). Then the clinical and demographic characteristics were compared between cases and controls, using the two-sided χ²-test. To investigate the association of the polymorphisms with gastric cancer risk, odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Multivariate analysis using unconditional logistic regression model was performed to calculate adjusted ORs (AORs), with adjustment for age, sex, pack-years, smoking and drinking status. Genotype-based mRNA expressions were also conducted as we described previously.^31,35 All statistical analyses were carried out using version 9.1 SAS software (SAS Institute, Cary, NC, USA). A two-sided P-value <0.05 was used as a criterion of significance.

Results

Population characteristics

This study consisted of 1142 cases of gastric cancer and 1173 healthy controls, whose individual characteristics are shown in Table S1. With regard to sex, there was no statistically significant difference between cases and controls (65.59% male vs. 67.26% male, P=0.393). However, significant differences were observed between cases and controls, regarding age, smoking status, drinking status, and pack-years. Thereafter, these variables were further adjusted for in the subsequent multivariate analyses. Overall, 12.26% (140), 28.81% (329), 39.93% (456), and 19.00% (217) of patients had TNM stage I, II, III, and IV tumors, according to the 7th Edition of the American Joint Committee on Cancer.³⁶

Associations between selected polymorphisms and gastric cancer risk

The raw data in this paper has been successfully uploaded and locked onto Research Data Deposit with a RDD number of RDDA2018000557. The genotype frequencies of all the selected gene polymorphisms among cases and controls are summarized in Table 1. All observed genotype frequencies among the controls were conformed to the HWE. In the single locus analysis, we observed a significantly increased gastric cancer risk associated with the ERCC1 rs2298881 A variant allele (CA vs. CC: AOR=1.33, 95% CI=1.09–1.62; dominant model: AOR=1.32, 95% CI=1.10–1.60; and additive model: AOR=1.20, 95% CI=1.04–1.38). However, ERCC1 rs3212986 A variant allele contributed to decreased gastric cancer risk (AA vs. CC: AOR=0.73, 95% CI=0.54–0.98; recessive model: AOR=0.72, 95% CI=0.54–0.96). There were no significant associations between the rest of all SNPs and gastric cancer risk.

Table 1 Associations between selected polymorphisms and gastric cancer risk

Genotypes	Cases (N=1141)	Controls (N=1173)	P-valuea	OR (95% CI)	P-value	AOR (95% CI)	P-valueb
XPA rs1800975 G>A
GG	296 (25.94)	327 (27.88)		1.00		1.00
GA	575 (50.39)	590 (50.30)		1.08 (0.89–1.31)	0.458	1.01 (0.81–1.26)	0.954
AA	270 (23.66)	256 (21.82)		1.17 (0.92–1.47)	0.197	1.05 (0.81–1.37)	0.693
Dominant	845 (74.06)	846 (72.12)	0.294	1.10 (0.92–1.33)	0.295	1.02 (0.83–1.26)	0.843
Additive model			0.435	1.08 (0.96–1.21)	0.197	1.03 (0.90–1.17)	0.702
Recessive	871 (76.34)	917 (78.18)	0.291	1.11 (0.91–1.35)	0.291	1.05 (0.84–1.31)	0.665
XPA rs3176752 C>A
CC	801 (70.20)	824 (70.25)		1.00		1.00
CA	316 (27.70)	318 (27.11)		1.02 (0.85–1.23)	0.814	1.03 (0.84–1.27)	0.760
AA	24 (2.10)	31 (2.64)		0.80 (0.46–1.37)	0.410	0.92 (0.50–1.71)	0.794
Dominant	340 (29.80)	349 (29.75)	0.981	1.00 (0.84–1.20)	0.981	1.02 (0.84–1.25)	0.821
Additive model			0.677	0.98 (0.84–1.15)	0.818	1.01 (0.84–1.21)	0.908
Recessive	1117 (97.90)	1142 (97.36)	0.394	0.79 (0.46–1.36)	0.395	0.91 (0.50–1.69)	0.771
ERCC1 rs2298881 C> A
CC	461 (40.40)	540 (46.04)		1.00		1.00
CA	548 (48.03)	500 (42.63)		1.28 (1.08–1.53)	0.005	1.33 (1.09–1.62)	0.005
AA	132 (11.57)	133 (11.34)		1.16 (0.89–1.52)	0.276	1.31 (0.96–1.78)	0.087
Dominant	680 (59.60)	633 (53.96)	0.006	1.26 (1.07–1.48)	0.006	1.32 (1.10–1.60)	0.003
Additive model			0.018	1.14 (1.01–1.29)	0.035	1.20 (1.04–1.38)	0.010
Recessive	1009 (88.43)	1940 (88.66)	0.862	1.02 (0.79–1.32)	0.862	1.13 (0.85–1.51)	0.404
ERCC1 rs11615 G>A
GG	594 (52.06)	592 (50.47)		1.00		1.00
GA	465 (40.75)	489 (41.69)		0.95 (0.80–1.12)	0.537	0.94 (0.78–1.14)	0.533
AA	82 (7.19)	92 (7.84)		0.89 (0.65–1.22)	0.467	0.86 (0.60–1.22)	0.392
Dominant	547 (47.94)	581 (49.53)	0.444	0.94 (0.80–1.11)	0.444	0.93 (0.77–1.12)	0.418
Additive model			0.691	0.95 (0.83–1.08)	0.391	0.93 (0.81–1.08)	0.344
Recessive	1059 (92.81)	1081 (92.16)	0.549	0.91 (0.67–1.24)	0.550	0.88 (0.62–1.24)	0.468
ERCC1 rs3212986 C> A
CC	477 (41.81)	478 (40.75)		1.00		1.00
CA	535 (46.89)	535 (45.61)		1.00 (0.84–1.19)	0.981	1.02 (0.83–1.24)	0.878
AA	129 (11.31)	160 (13.64)		0.81 (0.62–1.05)	0.114	0.73 (0.54–0.98)	0.037
Dominant	664 (58.19)	695 (59.25)	0.606	0.96 (0.81–1.13)	0.606	0.95 (0.78–1.14)	0.565
Additive model			0.236	0.93 (0.82–1.05)	0.227	0.90 (0.78–1.03)	0.125
Recessive	1012 (88.69)	1013 (86.36)	0.090	0.81 (0.63–1.03)	0.090	0.72 (0.54–0.96)	0.023
ERCC2 rs3810366 C>G
CC	331 (29.01)	379 (32.31)		1.00		1.00
CG	560 (49.08)	554 (47.23)		1.16 (0.96–1.40)	0.129	1.21 (0.98–1.50)	0.079
GG	250 (21.91)	240 (20.46)		1.19 (0.95–1.50)	0.134	1.20 (0.92–1.55)	0.181
Dominant	810 (70.99)	794 (67.69)	0.085	1.17 (0.98–1.39)	0.086	1.21 (0.99–1.48)	0.067
Additive model			0.219	1.10 (0.98–1.23)	0.110	1.10 (0.97–1.26)	0.138
Recessive	891 (78.09)	933 (79.54)	0.393	1.09 (0.89–1.33)	0.393	1.06 (0.85–1.33)	0.598
ERCC2 rs238406 G>T
GG	296 (25.94)	343 (29.24)		1.00		1.00
GT	556 (48.73)	564 (48.08)		1.14 (0.94–1.39)	0.181	1.20 (0.96–1.49)	0.112
TT	289 (25.33)	266 (22.68)		1.26 (1.00–1.58)	0.048	1.26 (0.97–1.63)	0.081
Dominant	845 (74.06)	830 (70.76)	0.076	1.18 (0.98–1.42)	0.077	1.22 (0.99–1.50)	0.063
Additive model			0.134	1.12 (1.00–1.26)	0.046	1.12 (0.99–1.28)	0.075
Recessive	852 (74.67)	907 (77.32)	0.135	1.16 (0.96–1.40)	0.136	1.12 (0.90–1.39)	0.295
ERCC2 rs13181 T>G
TT	971 (85.10)	982 (83.72)		1.00		1.00
TG	161 (14.11)	187 (15.94)		0.87 (0.69–1.09)	0.235	0.85 (0.66–1.10)	0.220
GG	9 (0.79)	4 (0.34)		2.28 (0.70–7.41)	0.173	1.37 (0.38–4.99)	0.636
Dominant	170 (14.90)	191 (16.28)	0.359	0.90 (0.72–1.13)	0.360	0.87 (0.67–1.12)	0.262
Additive model			0.175	0.94 (0.76–1.16)	0.557	0.89 (0.70–1.13)	0.335
Recessive	1132 (99.21)	1169 (99.66)	0.150	2.32 (0.71–7.56)	0.162	1.40 (0.38–5.11)	0.609
ERCC4 rs2276466 C>G
CC	663 (58.11)	726 (61.89)		1.00		1.00
CG	418 (36.63)	383 (32.65)		1.20 (1.004–1.42)	0.045	1.12 (0.92–1.36)	0.272
GG	60 (5.26)	64 (5.46)		1.03 (0.71–1.48)	0.889	0.96 (0.64–1.46)	0.860
Dominant	478 (41.89)	447 (38.11)	0.063	1.17 (0.99–1.38)	0.063	1.10 (0.91–1.32)	0.348
Additive model			0.130	1.11 (0.97–1.27)	0.148	1.05 (0.90–1.23)	0.530
Recessive	1081 (94.74)	1109 (94.54)	0.833	0.96 (0.67–1.38)	0.833	0.93 (0.62–1.39)	0.709

Notes:

a Chi-square test for genotype distributions between cases and controls.

b Adjusted for age, gender, smoking, and drinking status. Bold represents any values with a 95% CI excluding 1 or P<0.05.

Abbreviations: AOR, adjusted odds ratio; OR, odds ratio.

Stratification analysis

Stratified analysis was performed to further analyze the association of two independent ERCC1 rs2298881 C>A, rs3212986 C>A polymorphisms and gastric cancer risk by age, sex, smoking status, pack-years, drinking status, tumor sites, and TNM stage (Table 2). The risk association with the ERCC1 rs2298881 CA/AA genotypes remained significant in the following subgroups: males (AOR=1.37, 95% CI=1.08–1.73), never-smokers (AOR=1.40, 95% CI=1.09–1.79), 0 pack-year (AOR=1.40, 95% CI=1.09–1.79), ≤30 pack-years (AOR=1.74, 95% CI=1.19–2.54), never drinkers (AOR=1.36, 95% CI=1.09–1.69), non-cardia (AOR=1.31, 95% CI=1.08–1.60), stage I/II (AOR=1.42, 95% CI=1.11–1.82), and stage III/IV (AOR=1.28, 95% CI=1.03–1.59). Moreover, the ERCC1 rs3212986 C>A polymorphism AA variant significantly reduced gastric cancer risk in the following subgroups: age ≤58 years (AOR=0.66, 95% CI=0.47–0.93), males (AOR=0.65, 95% CI=0.46–0.92), never drinkers (AOR=0.70, 95% CI=0.50–0.98), and noncardia (AOR=0.72, 95% CI=0.53–0.97).

Table 2 Stratification analysis of ERCC1 gene variant genotypes with gastric cancer risk

Variables	rs2298881 (cases/controls)		AOR (95% CI)	P-valuea	rs3212986 (cases/controls)		AOR (95% CI)	P-valuea
	CC	CA/AA			CC/CA	AA
Median age, years
≤58	250/470	348/546	1.23 (0.99–1.52)	0.062	540/877	58/139	0.66 (0.47–0.93)	0.017
>58	211/70	332/87	1.29 (0.89–1.85)	0.176	472/136	71/21	0.97 (0.57–1.65)	0.914
Gender
Male	310/371	439/418	1.37 (1.08–1.73)	0.009	663/677	86/112	0.65 (0.46–0.92)	0.016
Female	151/169	241/215	1.26 (0.92–1.72)	0.159	349/336	43/48	0.90 (0.55–1.47)	0.679
Smoking status
Never	298/305	436/357	1.40 (1.09–1.79)	0.008	644/571	90/91	0.76 (0.53–1.10)	0.143
Ever	163/235	244/276	1.26 (0.93–1.70)	0.134	368/442	39/69	0.68 (0.42–1.08)	0.102
Pack-years
0	298/305	436/357	1.40 (1.09–1.79)	0.008	644/571	90/91	0.76 (0.53–1.10)	0.143
≤30	102/182	170/201	1.74 (1.19–2.54)	0.004	248/331	24/52	0.56 (0.31–1.01)	0.053
>30	61/53	74/75	0.71 (0.42–1.21)	0.205	120/111	15/17	0.92 (0.41–2.04)	0.833
Drinking status
Never	377/282	556/318	1.36 (1.09–1.69)	0.007	827/516	106/84	0.70 (0.50–0.98)	0.035
Ever	84/258	124/315	1.28 (0.88–1.86)	0.201	185/497	23/76	0.79 (0.45–1.41)	0.429
Tumor site
Cardia	102/540	138/633	1.36 (0.99–1.87)	0.059	212/1013	28/160	0.76 (0.47–1.22)	0.252
Non-cardia	359/540	542/633	1.31 (1.08–1.60)	0.007	800/1013	101/160	0.72 (0.53–0.97)	0.030
TNM stage
I/II	184/540	285/633	1.42 (1.11–1.82)	0.006	414/1013	55/160	0.74 (0.51–1.07)	0.112
III/IV	277/540	395/633	1.28 (1.03–1.59)	0.024	598/1013	74/160	0.72 (0.52–1.00)	0.050

Notes:

a Obtained in logistic regression models with adjustment for age, sex, pack-years, smoking, and drinking status, omitting the corresponding stratification factor. Bold represents any values with a 95% CI excluding 1 or P<0.05.

Abbreviation: AOR, adjusted odds ratio.

We also performed a stratification analysis for the ERCC2 gene rs3810366 C>G and rs238406 G>T polymorphisms (Table 3). Both the rs3810366 (AOR=1.32, 95% CI=1.04–1.68) and rs238406 (AOR=1.32, 95% CI=1.03–1.69) polymorphisms conferred gastric cancer susceptibility in never drinkers.

Table 3 Stratification analysis of ERCC2 gene variant genotypes with gastric cancer risk

Variables	rs3810366 (cases/controls)		AOR (95% CI)	P-valuea	rs238406 (cases/controls)		AOR (95% CI)	P-valuea
	CC	CG/GG			GG	GT/TT
Median age, years
≤58	173/331	425/685	1.19 (0.94–1.50)	0.145	157/298	441/718	1.17 (0.92–1.48)	0.196
>58	158/48	385/109	1.06 (0.72–1.57)	0.755	139/45	404/112	1.17 (0.79–1.75)	0.437
Gender
Males	227/266	522/523	1.20 (0.93–1.54)	0.155	201/242	548/547	1.22 (0.95–1.58)	0127
Females	104/113	288/271	1.25 (0.88–1.76)	0.215	95/101	297/283	1.24 (0.87–1.77)	0.234
Smoking status
Never	211/202	523/460	1.19 (0.91–1.55)	0.207	194/182	540/480	1.20 (0.91–1.58)	0.188
Ever	120/177	287/334	1.27 (0.92–1.75)	0.142	102/161	305/350	1.32 (0.95–1.83)	0.103
Pack-years
0	211/202	523/460	1.19 (0.91–1.55)	0.207	194/182	540/480	1.20 (0.91–1.58)	0.188
≤30	80/137	192/246	1.43 (0.97–2.12)	0.075	70/126	202/257	1.44 (0.96–2.16)	0.075
>30	40/40	95/88	1.07 (0.61–1.87)	0.824	32/35	103/93	1.19 (0.66–2.16)	0.564
Drinking status
Never	264/199	669/401	1.32 (1.04–1.68)	0.022	238/181	695/419	1.32 (1.03–1.69)	0.027
Ever	67/180	141/393	1.03 (0.69–1.53)	0.888	58/162	150/411	1.13 (0.74–1.69)	0.602
Tumor site
Cardia	75/379	165/794	1.19 (0.85–1.67)	0.318	67/343	173/830	1.20 (0.84–1.70)	0.318
Non-cardia	256/379	645/794	1.19 (0.97–1.48)	0.103	229/343	672/830	1.21 (0.97–1.50)	0.093
TNM stage
I/II	143/379	326/794	1.18 (0.91–1.53)	0.223	126/343	343/830	1.22 (0.93–1.60)	0.152
III/IV	188/379	484/794	1.22 (0.97–1.54)	0.088	170/343	502/830	1.21 (0.95–1.54)	0.116

Notes:

a Obtained in logistic regression models with adjustment for age, gender, pack-years, smoking, and drinking status, omitting the corresponding stratification factor. Bold represents any values with a 95% CI excluding 1 or P<0.05.

Abbreviation: AOR, adjusted odds ratio.

Correlation analysis for ERCC1 mRNA expression levels and genotypes

We further conducted the ERCC1 genotype expression correlation analysis (Table S2), aiming to explore underlying molecular mechanisms. The genotype data for 270 individuals were collected from HapMap. ERCC1 mRNA expression levels of lymphoblastoid cell lines from the same 270 individuals were extracted from SNPexp. We observed that genotypes of the rs2298881 C>A polymorphism were significantly correlated with decreased ERCC1 mRNA expression in Chinese subjects (P=0.003), Africans (P<0.0001), and combined subjects (P<0.0001). However, no genotype expression correlation was found for the rs3212986 C>A and rs11615 G>A polymorphisms in combined subjects.

Discussion

In the present hospital-based case-control study, we investigated the association between the polymorphisms in the NER genes and gastric cancer risk in a southern Chinese population. We observed a significantly increased gastric cancer risk associated with the ERCC1 rs2298881 A variant allele. However, we found that ERCC1 rs3212986 A variant allele was associated with decreased risk of gastric cancer. We also confirmed that the ERCC1 rs2298881 C>A polymorphism was associated with a decrease in ERCC1 mRNA expression. However, no association with gastric cancer risk was detected for the polymorphisms in the XPA, XPD, and XPF genes. To the best of our knowledge, this is by far the most comprehensive study investigating the association between the NER pathway genes and gastric cancer risk.

ERCC1 gene is located on chromosome 19q32.32, consisting of 10 exons and encoding a 297 amino acid protein. The ERCC1 protein is an indispensable component of the NER pathway.^37,38 It interacts with XPA, XPF, and/or RPA, and catalyzes the 5′ cleavage of DNA lesions.³⁹ Given the critical role of ERCC1 protein in NER, it is biologically plausible that potentially functional ERCC1 gene variants could modify gastric cancer risk. Our findings are in accordance with others. For instance, He et al reported that ERCC1 rs11615 G>A was associated with an increased risk of breast cancer.⁴⁰ Likewise, the ERCC1 rs11615 G>A polymorphism was shown to increase the risk of developing lung cancer.⁴¹ It is worth mentioning that we previously observed that ERCC1 rs11615A and rs2298881C variant alleles were associated with increased gastric cancer risk in an eastern Chinese population.⁴² Moreover, patients with 2–3 ERCC1 risk genotypes had a significantly increased risk of gastric cancer compared with those with 0–1 ERCC1 risk genotypes.⁴² However, the previous study did not detect an association between the rs3212986 polymorphism and gastric cancer risk. The discrepant results between the former study and the present study might be due to the different population selected. Our previous study population was recruited from East China, while the current study population was recruited from South China. Apart from our studies, two published studies regarding ERCC1 polymorphisms and gastric cancer risk were conducted in Italian population with relatively small sample sizes.^43,44 One study included 314 cases and 548 controls, and the other included 126 cases and 144 controls. No significant association was detected in these two studies. However, all the included polymorphisms of ERCC1 in these two studies were not under investigation in the present study.

In the stratification analysis, our data suggested that the risk effect of ERCC1 rs2298881 CA/AA genotypes remained significant in males, never-smokers, pack-year of 0, pack-years ≤30, never drinkers, non-cardia, stage I/II, and stage III/IV subgroups. The association between decreased gastric cancer risk and ERCC1 rs3212986 was more evident in subgroups of median age ≤58 years, males, never-drinkers, and non-cardia tumor. This phenomenon can be explained by the concept that susceptible individuals are likely to have a light exposure to risk factors. Young individuals, never smokers, or never drinkers are tended to be exposed to less environmental carcinogens. Thus, the role of genetic variants might not be outweighed by carcinogens in carcinogenesis in such subgroups. Considering the reduced sample sizes in the stratification analysis, some results might be just chance findings. Therefore, these results should be interpreted with caution. We further adopted the public data on ERCC1 genotypes and mRNA levels for the genotype–phenotype association analysis. A significant correlation between ERCC1 mRNA levels and rs2298881 C>A genotypes was observed, which provide further evidence that rs2298881 C>A may associate with gastric cancer by mRNA expression alteration, sequentially DNA repair capacity alteration. Therefore, additional larger case-control studies with functional analysis are warranted to explore the exact role of ERCC1 in gastric cancer risk.

We failed to detect any relationship between other polymorphisms and gastric cancer risk. Lack of an association of gastric cancer susceptibility with single NER pathway gene variants was also reported by other studies. For instance, in a case-control study including 246 cases and 1175 controls, no significant association was observed between the analyzed polymorphisms in the MSH2, MLH1, XRCC1, OGG1, and ERCC2 genes and gastric cancer risk.⁴⁵ However, some previous studies have demonstrated that some polymorphisms including rs11615 G>A were independent risk factors for gastric cancer.⁴² Such a discrepancy among studies might be partly due to the limited sample sizes; small sample studies may not have sufficient statistical power to reveal an association. Another possible explanation was that the effect of each single variant was too weak to be detected. Moreover, the potential effect of polymorphisms in gastric cancer risk may be dissimulated by other complex exposures or environmental–genetic interactions.

Although we extensively analyzed a number of polymorphisms in the NER core pathway genes, some limitations still existed in this study. First, due to the nature of a retrospective study, selection bias and recall bias could not be completely avoided. To minimize such biases, we further performed multivariate logistic regression analysis on potential confounding factors such as age, smoking, and drinking status. Second, gastric cancer is a heterogeneous disease affected by multiple factors including H. pylori infection, environmental exposures, and diet habits, yet these data were not available for further analysis. Third, the sample size in the subgroup analysis was relatively small, which might limit the statistical power in the stratification analysis. Fourth, we adopted only the public data to preliminarily investigate the correlation between ERCC1 genotype and mRNA expression. The findings should be validated in gastric tissues in the future. We failed to quantify the ERCC1 mRNA levels in the target tissue of the included subjects due to tissue access constraints. Finally, as all participants were recruited from a hospital in South China, special caution should be paid in extrapolating the results to other populations.

In conclusion, we found that the ERCC1 gene rs2298881 C>A and rs3212986 C>A polymorphisms were associated with gastric cancer susceptibility in a southern Chinese population. Well-designed studies with larger sample sizes and functional analysis are required to further verify our findings.

Acknowledgments

This study was supported by grants from National Natural Science Foundation of China (Grant No. 81502046), the Natural Science Foundation of Guangdong Province (Grant No. 2015A030310324), and the National Science Fund for Distinguished Young Scholars (Grant No. 81325018).

Supplementary materials

Table S1 Clinical and demographic characteristics of gastric cancer cases and cancer-free controls

Variables	No. of cases (%)	No. of controls (%)	P-valuea
All subjects	1142 (100.0)	1173 (100.0)
Gender
Male	749 (65.6)	789 (67.3)	0.393
Female	393 (34.4)	384 (32.7)
Age, years	15–86	16–80
Mean±SD	56.3±12.5	45.2±11.6	<0.0001
≤50	334 (29.3)	789 (67.3)
51–60	362 (31.7)	285 (24.3)
61–70	312 (27.3)	73 (6.2)
>70	134 (11.7)	26 (2.2)
Smoking status
Never	735 (64.4)	662 (56.4)	<0.0001
Ever	407 (35.6)	511 (43.6)
Drinking status
No	934 (81.8)	600 (51.2)	<0.0001
Yes	208 (18.2)	573 (48.8)
Pack-years
0	735 (64.4)	662 (56.4)	<0.0001
≤30	272 (23.8)	383 (32.7)
>30	135 (11.8)	128 (10.9)
Sites
Cardia	240 (21.0)	–
Non-cardia	902 (79.0)	–
TNM stages
I	140 (12.3)	–
II	329 (28.8)	–
III	456 (39.9)	–
IV	217 (19.0)	–

Note:

a Two-sided chi-square test for distributions between gastric cancer cases and cancer-free controls.

Table S2 ERCC1 mRNA expression by the genotypes of polymorphisms, using data from the HapMapa

Population	mRNA expression (rs2298881)				mRNA expression (rs3212986)				mRNA expression (rs11615)
	Genotypes	No.	Mean±SD	P-valueb	Genotypes	No.	Mean±SD	P-valueb	Genotypes	No.	Mean±SD	P-valueb
CHB	CC	15	6.81±0.08	0.003c	CC	20	6.74±0.13	0.442c	GG	29	6.73±0.11	0.044c
	AC	20	6.76±0.09	0.126	AC	19	6.77±0.09	0.416	AG	12	6.79±0.10	0.144
	AA	10	6.68±0.13	0.006	AA	5	6.77±0.07	0.664	AA	4	6.83±0.07	0.111
	AC/AA	30	6.73±0.11	0.026	AC/AA	24	6.77±0.08	0.377d	AG/AA	16	6.80±0.09	0.054
JPT	CC	9	6.81±0.07	0.242c	CC	31	6.75±0.09	0.442c	GG	21	6.75±0.10	0.872c
	AC	26	6.74±0.11	0.067	AC	13	6.77±0.12	0.442	AG	22	6.76±0.10	0.846
	AA	10	6.76±0.08	0.118	AA	0	–	–	AA	2	6.76±0.06	0.976
	AC/AA	36	6.74±0.10	0.060	AC/AA	13	6.77±0.12	0.442d	AG/AA	24	6.76±0.10	0.848
CEU	CC	79	6.77±0.12	0.370c	CC	52	6.77±0.13	0.725c	GG	6	6.85±0.13	0.447c
	AC	11	6.74±0.18	0.370	AC	35	6.74±0.12	0.279	AG	49	6.76±0.14	0.168
	AA	0	–	–	AA	3	6.95±0.04	0.026	AA	35	6.77±0.11	0.111
	AC/AA	11	6.74±0.18	0.370	AC/AA	38	6.76±0.13	0.620	AG/AA	84	6.76±0.13	0.129
YRI	CC	76	6.80±0.09	<0.0001c	CC	39	6.77±0.10	0.208c	GG	87	6.79±0.10	0.137c
	AC	11	6.71±0.07	0.002	AC	45	6.81±0.09	0.046	AG	3	6.71±0.05	0.137
	AA	2	6.61±0.003	0.004	AA	6	6.76±0.05	0.976	AA	0	–	–
	AC/AA	13	6.70±0.07	0.0001d	AC/AA	51	6.80±0.09	0.065	AG/AA	3	6.71±0.05	0.137
All	CC	179	6.79±0.10	<0.0001c	CC	142	6.76±0.11	0.095c	GG	143	6.78±0.10	0.599c
	AC	68	6.74±0.11	0.001	AC	112	6.78±0.11	0.243	AG	86	6.76±0.12	0.385
	AA	22	6.71±0.11	0.001	AA	14	6.80±0.09	0.162	AA	41	6.77±0.10	0.793
	AC/AA	90	6.73±0.11	<0.0001d	AC/AA	126	6.78±0.10	0.149d	AG/AA	127	6.77±0.12	0.435

Notes:

a ERCC1 genotyping data and mRNA expression levels for ERCC1 by genotypes were obtained from the HapMap Phase II release 23 data from EBV-transformed lymphoblastoid cell lines from 270 individuals, including 45 unrelated CHB.

b Two-sided Student’s t-test within the stratum.

c P-values for the trend test of ERCC1 mRNA expression among three genotypes for each polymorphism from a general linear model.

d There were missing data because genotyping data were not available. Bold represents any values P<0.05.

Abbreviations: CEU, Utah residents with ancestry from northern and western Europe; CHB, Han Chinese in Beijing, China; JPT, Japanese in Tokyo; YRI, Yoruba in Ibadan, Nigeria.

Disclosure

The authors report no conflicts of interest in this work.