Sulfotransferase SULT1A1 Arg213His Polymorphism with Cancer Risk: A Meta-Analysis of 53 Case-Control Studies

Background The SULT1A1 Arg213His (rs9282861) polymorphism is reported to be associated with many kinds of cancer risk. However, the findings are conflicting. For better understanding this SNP site and cancer risk, we summarized available data and performed this meta-analysis. Methods Data were collected from the following electronic databases: PubMed, Web of Knowledge and CNKI. The association was assessed by odd ratio (OR) and the corresponding 95% confidence interval (95% CI). Results A total of 53 studies including 16733 cancer patients and 23334 controls based on the search criteria were analyzed. Overall, we found SULT1A1 Arg213His polymorphism can increase cancer risk under heterozygous (OR = 1.09, 95% CI = 1.01–1.18, P = 0.040), dominant (OR = 1.10, 95% CI = 1.01–1.19, P = 0.021) and allelic (OR = 1.08, 95% CI = 1.02–1.16, P = 0.015) models. In subgroup analyses, significant associations were observed in upper aero digestive tract (UADT) cancer (heterozygous model: OR = 1.62, 95% CI = 1.11–2.35, P = 0.012; dominant model: OR = 1.63, 95% CI = 1.13–2.35, P = 0.009; allelic model: OR = 1.52, 95% CI = 1.10–2.11, P = 0.012) and Indians (recessive model: OR = 1.93, 95% CI = 1.22–3.07, P = 0.005) subgroups. Hospital based study also showed marginally significant association. In the breast cancer subgroup, ethnicity and publication year revealed by meta-regression analysis and one study found by sensitivity analysis were the main sources of heterogeneity. The association between SULT1A1 Arg213His and breast cancer risk was not significant. No publication bias was detected. Conclusions The present meta-analysis suggests that SULT1A1 Arg213His polymorphism plays an important role in carcinogenesis, which may be a genetic factor affecting individual susceptibility to UADT cancer. SULT1A1 Arg213His didn't show any association with breast cancer, but the possible risk in Asian population needs further investigation.


Introduction
Sulfotransferase (SULT) enzymes catalyze the sulfate conjugation of a broad range of substrates and play an important role in metabolism of endogenous and exogenous compounds including thyroid and steroid hormones, neurotransmitters, drugs and procarcinogens [1,2]. There are many isoforms of the SULTs supergene family, each with different amino acid sequence identity and substrate specificity [3]. SULT1A1 is an important member of the sulfotransferase family involving in the pathogenic process of various cancers [3][4][5].
The SULT1A1 gene is located on chromosome 16p12.1-p11.2 [6]. Previous study indicated that exon 7 of the SULT1A1 gene contained a G to A transition at codon 213 (rs9282861) that causes an Arg to His amino acid substitution [4]. Some studies have shown that this genetic polymorphism leads to a decrease in enzymatic activity of SULT1A1 and the sulfonation efficiency thus associating with susceptibility to several cancers [7,8]. Although the specific role of SULT1A1 Arg213His polymorphism in carcinogenesis has been investigated in numerous case-control studies, the results have been inconclusive, even conflictive. In order to give a comprehensive and precise result, we performed this meta-analysis study to analyze the association between this polymorphism and cancer risk.

Identification of eligible studies
The meta-analysis was conducted following the criteria of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Checklist S1). In this study, we did an exhaustive literature search on studies that examined the association of the SULT1A1 gene polymorphisms with cancer risks. All eligible studies were identified by searching the following databases: PubMed, Web of Knowledge and China National Knowledge Infrastructure (CNKI, http://www.cnki.net/). The following terms were utilized: ''sulfotransferase, SULT or SULT1A1'', ''polymorphism, variation, variant or mutation'' and ''cancer or carcinoma''. In the CNKI database, we searched with these corresponding key words in Chinese characters. Included studies should meet the following criteria: (1) evaluating the association between SULT1A1 Arg213His polymorphism and cancer risk; (2) study designed as case-control; (3) sufficient data available to estimate an odd ratio (OR) with its 95% confidence interval (95% CI).

Data extraction
Two investigators extracted data independently and reached consensus on the following characteristics of the selected studies: first author's name, the year of publication, ethnicity of the study population, matching criteria, number of participants, genotype distribution and control source.

Statistical analysis
Hardy-Weinberg equilibrium was assessed by Chi-square test. Crude odd ratio (OR) and 95% confidence interval (CI) were used to estimate the association between SULT1A1 polymorphism and cancer susceptibility under the dominant model (Arg/His+His/ His vs. Arg/Arg), recessive model (His/His vs. Arg/Arg + Arg/His), homozygous model (His/His vs. Arg/Arg), heterozygous model (His/Arg vs. Arg/Arg) and allelic model (His vs. Arg). The heterogeneity among the studies was evaluated by Q-test and I 2 value ranging from 0% to 100% to describe the percentage of between-study variation caused by heterogeneity. P value for the Q-test less than 0.10 indicates existing heterogeneity among studies. And then the pooled OR was measured by a random effect model (the DerSimonian-Laird method). Otherwise, a fixed effect model (the Mantel-Haenszel method) was chosen.
Subgroup analyses were performed according to cancer type (breast cancer, colorectal cancer, urothelial cancer, prostate cancer, lung cancer, upper aero digestive tract (UADT) cancer, ovarian cancer and gastric cancer), ethnicity (Caucasian, East Asian, Indian and African) and source of controls (hospital based and population based). When heterogeneity was detected, a multivariable meta-regression analysis including cancer type, ethnicity, control source and year of publication to explore potential source of heterogeneity and sensitivity analysis were performed.
The potential publication bias was estimated using Egger's linear regression test by visual inspection of the funnel plot. P , 0.05 was considered statistically significant, and all P values were twosided. Analyses were performed using the software Review

Characteristics of eligible studies
The flow diagram of literature search was given in Figure 1. A total of 91 studies focusing the association between the SULT1A1 Arg213His polymorphism and cancer risks were identified. 25 of them were ruled out because of unavailable data or repeated data. Thus, the allele and genotype frequencies of the SULT1A1 Arg213His polymorphism were extracted from 66 articles. However, 18 articles didn't meet with Hardy-Weinberg equilibrium and were abandoned (Excluded list S1). As a result, 53 studies of 48 articles, involving 16733 cases and 23334 controls were included in the pooled analyses .

Subgroup Analyses
We analyzed the association in cancer type subgroup. By control source, significant association was observed in hospital based study, but not the population based study.

Meta-regression analysis
To find potential source of heterogeneity, multivariable metaregression analyses were conducted in total group and subgroups including cancer type, ethnicity, control source and publication year. In the breast cancer subgroup, ethnicity (heterozygous model, P = 0.027; recessive model, P = 0.020) and publication year (heterozygous model, P = 0.019; recessive model, P = 0.012) are significant sources of heterogeneity (Table S1). Other variables don't affect heterogeneity.

Sensitivity analysis
The sensitivity analysis was constructed by repeating the metaanalysis sequentially removing each study. In the recessive model, two studies [26,57] were found to affect the pooled OR and the heterogeneity when removed. The study conducted by Khvostova was focused on breast cancer and Sun's study was focused on colorectal cancer among Caucasians, so further sensitivity analyses were conducted in total data set and breast cancer, colorectal cancer and Caucasian subgroups after removing the two studies (Table 4 and Table S2). In total group, the heterogeneity was significantly decreased (I 2 = 58.2, 42.2, 63.5, 33.1 and 66.4, respectively). In the subgroup sensitivity analyses, removing the two studies can significantly decrease the heterogeneity among studies, most I 2 values less than 50%. And this polymorphism didn't show any obvious correlation with breast cancer risk (Figure 4). At last, we conducted the sensitivity analyses on the remaining studies and the result was stable.

Publication bias
Funnel plots and Egger's test were carried out to assess publication bias. The shapes of funnel plots indicated no obvious asymmetry ( Figure 5). Egger's test found no publication bias in the

Discussion
SULT1A1 enzyme encoded by SULT1A1 gene plays an important role in xenobiotic metabolism. The Arg213His polymorphism, the most widely studied polymorphism within SULT1A1 gene, can reduce enzyme activity and thermostability, and consequently results in an individual's susceptibility to cancer [7,8].
There have been a few meta-analyses focusing on this mutation and cancer risk [58][59][60]. However, most of these analyses were conducted before the year 2012 and a new meta-analysis is needed to give a comprehensive conclusion due to the increasing data of case-control studies.
This present meta-analysis, including 16733 cases and 23334 controls from 53 case-control studies, explored the association between the SULT1A1 Arg213His polymorphism and cancer risk. This is the largest scale meta-analysis so far. Our results suggested that the SULT1A1 Arg213His was associated with UADT cancer risk. As the upper aero digestive tract is exposed to numerous potential carcinogens such as phenolic xenobiotics, polycyclic aromatic hydrocarbons and heterocyclic aromatic amines contained in cigarette smoking, environmental pollutants and some food, this result manifests that the mutation within SULT1A1 causes the low SULT1A1 activity and is associated with high susceptibility to cancers related with environment.
In the sensitivity analyses, the study conducted by Khvostova influences the pooled estimates and the heterogeneity most in breast cancer subgroup. And after removing this study, the significant association between SULT1A1 Arg213His and breast cancer risk became null (Figure 2 and Figure 4). We further checked data from Khvostova and observed the percentage of wild homozygous genotype in Khvostova's study was obviously lower than that in other studies thus causing great heterogeneity. At last a robust result was achieved and failed to reveal significant association in breast cancer subgroup. This result is similar to Wang, Lee and Jiang [61][62][63], but they found a positive association of this polymorphism with breast cancer susceptibility among Asians. While in our meta-analysis, we only recruited one paper focused on breast cancer among Asians because other papers on Asians deviate from HWE and were excluded. This is a limitation of this meta-analysis and more independent case-control studies conducted on Asians are needed to conclude a more comprehensive result.
In the ethnic subgroup analysis, we found that the genotype distributions of the SNP site are different in ethnic groups. When calculating the percentage of alleles in every ethnic, we found that His allele in Asians (9.58%) is significantly less than in Caucasians (35.2%). Different ethnicities may have different genetic backgrounds, thus causing different genotype frequencies in Asian and other ethnic groups which may influence cancer susceptibility.
Li and Kotnis have conducted meta-analyses focused on environment-related cancers, such as tobacco-related cancers and found cancer risk could be modulated by interaction between genetic variants and environmental factors [58,59]. As exposed environmental factors are different according to cancer types, for example smoking leads to lung cancer, while the intake of meat influences breast cancer and colorectal cancer [64,65] and our analysis took many kinds of cancer into account, we decided not to include environmental factors. Moreover, the definitions of exposed environmental factors were not consistent in the studies, which could cause great heterogeneity. Our estimates were based Table 4. Meta-analysis in breast, colorectal and Caucasian subgroups after omitting studies of Khvostova  There were several sources bringing in heterogeneity, such as study design, age and sex distribution, and ethnicity. Metaregression analysis was conducted to find source of heterogeneity. In the breast cancer subgroup, publication year could cause great heterogeneity and further attention was paid to years. We found all the recruited studies were carried out before 2005 or after 2010, and there were no studies between 2006 and 2009. The His allele was 29.6% in the studies before 2005 and 33.0% after 2010, which was significantly different (P = 0.02). This may be caused by the different study population, and needs more case-control studies to illustrate.
In conclusion, our meta-analysis suggests that the SULT1A1 Arg213His polymorphism may contribute UADT cancer risk. As the result was calculated through sampling statics and statistical difference is not the same as clinical difference, the result can be used for clinical reference, not for clinical diagnosis of cancer. Further detailed investigation with larger number of worldwide participants is needed to clarify the role of this polymorphism in cancer risk. Figure S1 Forest plot on the association between SULT1A1 Arg213His polymorphism and overall cancer risk in dominant model.

(TIF)
Table S1 The P-value of meta-regression in overall and breast cancer groups. (DOCX) Excluded list S1 Excluded studies list with reasons. (XLS)