The Interleukin-10 Promoter Polymorphism rs1800872 (-592C>A), Contributes to Cancer Susceptibility: Meta-Analysis of 16 785 Cases and 19 713 Controls

Interleukin-10 (IL-10) is a multifunctional cytokine which participates in the development and progression of various malignant tumors. To date, a number of case–control studies were conducted to detect the association between IL-10-592C>A polymorphism and cancer risk in humans. However, the results of these studies on the association remain conflicting. In an effort to solve this controversy, we performed a meta-analysis based on 70 case–control studies from 65 articles, including 16 785 cancer cases and 19 713 controls. We used odds ratios (ORs) with 95% confidence intervals (CIs) to assess the strength of the association. The overall results suggested that the variant homozygote genotype AA of the IL-10-592C>A polymorphism was associated with a moderately decreased risk of all cancer types (OR = 0.90, 95% CI = 0.83–0.98 for homozygote comparison, OR = 0.92, 95% CI = 0.86–0.98 for recessive model). In the stratified analyses, the risk remained for studies of smoking-related cancer, Asian populations and hospital-based studies. These results suggested that the IL-10-592C>A polymorphism might contribute to the cancer susceptibility, especially in smoking-related cancer, Asians and hospital-based studies. Further studies are needed to confirm the relationship.


Introduction
Cancer is a major public health problem in the world. Evidences support an important role for genetics in determining risk for cancer. Association studies are appropriate for searching susceptibility genes involved in cancer [1].
Interleukin-10 (IL-10) is a multifunctional cytokine which participate in the development and progression of various malignant tumors [2]. It has anti-inflammatory and immunosuppressive activities including the ability to downregulate the expression of macrophage costimulatory molecule. The impact of IL-10 on macrophage function appears to play a role in the growth of blood vessel, as studies showed that IL-10 might contribute to the regulation of angiogenesis in many kinds of tumors [3,4]. As an immunosuppressive molecule allowing tumor to escape from immune surveillance, IL-10 might act as a potential tumor promoter which results in a more aggressive behavior of malignant cells. Conversely, due to its immune-stimulating and anti-angiogenic properties, IL-10 is supposed to prevent or reduce the growth and distant spread of tumor [5][6][7].It has been indicated that IL-10 overexpression as well as deficiency was found under different pathophysiological conditions depending on the cancers analyzed [8].
IL-10 is encoded by a gene located on chromosome 1 (1q31-1q32) [9]. The IL-10 promoter is highly polymorphic and three important single nucleotide polymorphisms (SNPs) in the promot-er region which influence the transcription of IL-10 messenger RNA and the expression of IL-10 in vitro are rs1800896 (-1082A.G), rs1800871 (-819C.T), and rs1800872(-592C.A) [10,11]. It has been reported that the -1082A.G and haplotype (-1082_-819_-592) were associated with differential production of the protein in stimulated cells, with the ATA haplotype leading to decreased IL-10 expression and the GCC haplotype increased IL-10 expression [12].
To date, a number of case-control studies were conducted to investigate the association between IL-10 -592 C.A and cancer risk in humans . However, the results of these studies remain conflicting rather than conclusive. So, we performed the present meta-analysis to evaluate the association between IL-10 -592 C.A and cancer risk.

Identification and Eligibility of Relevant Studies
We searched the electronic literature from Pubmed for all relevant reports (the last search update was Oct 16, 2012), using the key words: (''interleukin-10'' or ''IL-10'' or ''IL10'') and (''variant'' or ''variation'' or ''polymorphism'') and (''cancer'' or ''tumor'' or ''carcinoma'' or ''malignancy''). The search was limited to English language papers. In addition, studies were identified by a manual search of the reference lists of reviews and retrieved studies. Studies were selected if there was available data for the IL-10 -592C.A polymorphism with cancer risk in a casecontrol design including retrospective or prospective and nested case-control studies. As studies with the same population by different investigators or overlapping data by the same authors were found, the most recent or complete articles with the largest numbers of subjects were included. Studies included in our metaanalysis have to meet the following criteria: (i) evaluation of the IL-10 -592C.A polymorphism and cancer risk, (ii ) use a case-control design (retrospective or prospective and nested case-control) and (iii ) contain available genotype frequency. Major reasons for exclusion of studies were (i ) only case population and (ii ) duplicate of previous publication.

Data Extraction
Two of the authors extracted all data independently complying with the selection criteria and reached a consensus on all items. In the present study, the following characteristics were collected: the first author's last name, year of publication, country of origin, ethnicity, frequencies of genotyped in cases and controls, source of control groups (population-or hospital-based controls) and cancer type. For studies including subjects of different ethnic groups, data were extracted separately for each ethnic group whenever possible [43]. Different ethnic descents were categorized as European, Asian, or African or mixed (composed of an admixture of different ethnic groups). Meanwhile, studies investigating more than one kind of cancer were counted as individual data sets only in subgroup analyses by cancer type [15,36,39,59]. One study which was duplicate of previous publication were excluded from the analysis [78].

Statistical Analysis
The strength of the association between the IL-10 -592C.A polymorphism and cancer risk was measured by odds ratios (ORs) with 95% confidence intervals (CIs). The statistical significance of the pooled OR was determined using the Ztest. Pooled estimates of the OR were obtained by calculating a weighted average of OR from each study [79]. First, we estimated cancer risks with the CA and AA genotypes, compared with the wild-type CC homozygote, and then evaluated the risks associated with CA/AA versus CC and AA versus CC/CA, assuming the dominant and recessive effects of the variant A allele, respectively. Stratified analyses were also performed by cancer types (if one cancer type contained less than three individual studies, it was combined into 'other cancers' group), ethnicity and source of controls. In consideration of the possibility of heterogeneity across the studies, I 2 was applied to assess heterogeneity between studies [80]. Values from single study were combined using models of both fixed effects and random effects [81]. We used a fixed effects model when I 2 was equal or less than 50%, and a random effects model when I 2 was greater than 50%. In the absence of heterogeneity, the two methods provide identical results, because the fixed effects model, using the Mantel-Haenszel's method, assumes that studies are sampled from populations with the same effect size, making an adjustment to the study weights according to the in-study variance; whereas the random-effects model using the DerSimonian and Laird's method assumes that studies are taken from populations with varying effect sizes, calculating the study weights both from in-study and betweenstudy variances, considering the extent of variation or heterogeneity. Sensitivity analyses were performed to assess the stability of the results, namely, a single study in the metaanalysis was deleted each time to reflect the influence of the individual data set to the pooled OR. Funnel plots and Egger's linear regression test were used to provide diagnosis of the potential publication bias [82]. All analyses were conducted using Stata software (version8.0; StataCorp LP, College Station, TX), and all tests were two sided.

Characteristics of Studies
Finally, a total of 70 studies from 65 articles that included a total of 16 785 cancer cases and 19 713 controls met the inclusion criteria ( Fig. 1). Study characteristics are summarized in Table 1. For the IL-10-592C.A polymorphism, there were 2 studies of African descendents, 20 studies of Asian descendents and 37 studies of European descendents. In our meta-analysis, most of the cancer types were gastric cancer. Cancers were confirmed histologically or pathologically in most studies. The distribution of genotype in the controls of the studies was in agreement with Hardy-Weinberg equilibrium for all except five studies [14,20,21,27,72], which were further tested in the sensitivity analyses.

Test for Heterogeneity
In the sub-analysis of gastric cancer, there was significant heterogeneity for recessive model comparison (AA versus CC/CA: I 2 = 56.60% for heterogeneity). Then, we assessed the source of heterogeneity for recessive model comparison (AA versus CC/CA) by ethnicity and source of controls. As a result, neither ethnicity (x 2 = 5.06, df = 2, P = 0.08) nor source of controls (x 2 = 0.01, df = 1, P = 0.91) was found to contribute to substantial heterogeneity.

Sensitivity Analyses
Sensitivity analyses indicated that two independent studies by Wu et al. [50] was the main origin of heterogeneity in the subanalysis of gastric cancer. The heterogeneity was effectively decreased or removed by exclusion of the study (AA versus CC/ CA: I 2 = 46.90%). Although the genotype distribution in five studies did not follow Hardy-Weinberg equilibrium, the corresponding pooled ORs were not materially altered by including the studies. In addition, no other single study influenced the pooled OR qualitatively, as indicated by sensitivity analyses, suggesting that the results of this meta-analysis are stable.

Publication Bias
Begg's funnel plot and Egger's test were performed to assess the publication bias of literatures. As shown in Fig. 3, the shapes of the funnel plots did not reveal any evidence of obvious asymmetry in all comparison models. Thus, Egger's test was used to provide statistical evidence of funnel plot symmetry. The results still did not show any evidence of publication bias (t = 21.38, P = 0.172 for CA versus CC; t = 20.86, P = 0.390 for AA versus CC; t = 21.99, P = 0.051 for AA/CA versus CC; t = 20.12, P = 0.903 for AA versus CC/CA).

Discussion
The present meta-analysis investigated the association between the IL-10-592C.A polymorphisms and cancer risk, based on 70 published case-control studies from 65 articles. The results provided evidence that the IL-10-592C.A polymorphism was associated with a significant decrease in overall cancer risk. The variant homozygote genotype AA of the IL-10-592C.A polymorphism, was associated with a modest but significantly decreased risk in homozygote comparison and recessive model. In the stratified analysis, the risk remained for studies of smokingrelated cancer, Asian populations and hospital-based studies. Given the important roles of IL-10 in carcinogenesis, it is biological plausible that IL-10 polymorphism might modulate the risk of cancer.
A common [ATA] haplotype is formed by polymorphisms at positions -1082,-819 and -592 in the promoter of IL-10 gene [83,84]. It has been reported that the -592 A variant, the -1082 A variant as well as the [ATA] haplotype was associated with lower IL-10 expression [51,85,86]. Thus, the -592 A variant can be regarded as a low-producer allele of the IL-10 gene. Research has showed that increased serum IL-10 levels could facilitate development of tumors by suppressing the expression of MHC class I and II antigens [87] and preventing tumor antigen presentation to CD8-cytotic T lymphocytes. It has been revealed that the homozygous IL-10-592AA genotype, indicating homozygosity for the [ATA] haplotype, was protective against   [51]. Similarly, elevated serum levels of IL-10 were found in non-small cell lung cancer patients; moreover, IL-10 serum levels were shown to be higher in patients with metastatic disease when compared to those with undisseminated cancer [88]. Consistently, we also found that individuals with the AA genotype which exhibits low production of IL-10 were associated with a lower cancer risk than participants with the CC genotype in our meta-analysis. Tobacco smoking is a well-established risk factor for cancers of many organs, including lung, esophagus, oral cavity, pharyngeal and kidney [89][90][91][92]. Tobacco use has been proven to affect the immune system and influence the production of IL-10 [93]. Also, studies showed that smokers have impaired T lymphocyte suppressor cell function and decreased natural killer cell activity compared with non-smokers. Furthermore, IL10 might protect tumors by inhibiting cytotoxic T lymphocyte (CTL)-mediated tumor-specific cell lysis [87,94]. Further studies are needed to investigate the relationship.
Our results indicated that the AA variant genotype was associated with decreased risk in smoking-related cancer but not for breast cancer, cervical cancer, colorectal cancer, esophageal cancer, gastric cancer, melanoma, lung cancer, hepatocellular carcinoma, Non-Hodgkin lymphoma or prostate cancer. One possible explanation for the discrepancy is that carcinogenic mechanism underlying the etiology may differ by different tumor sites and that the IL-10 genetic variants may play a different role in different cancers. Even at the same tumor site, considering the relatively small sample size in some studies and the possible small effect size of this genetic polymorphism to cancer, the discrepancy will become apparent since some of these studies may have insufficient statistical power to detect a small but real association. For example, there were only three studies included in the analysis with limited sample size for esophageal cancer, with 456 cases and 628 controls, so the results may be capricious and should be interpreted with caution.
In the subgroup analysis by ethnicity, we found an evidence for the association between the IL-10-592C.A polymorphism and cancer risk among Asians but not among Europeans or Africans, suggesting a possible role of ethnic differences in genetic backgrounds and the environment they lived in (Table 2). Several concerns may account for it. First, nine of the 20 Asian studies investigated gastric cancer (weighted 36.68% and 42.24% in  When stratifying the source of controls, a moderate strength was observed in hospital-based controls, but not the population-based controls. The discrepancy may result from a differential effect of selection criteria in different cancers, as well as the weight of each study, which was dictated by sample size in the meta-analysis. Another reason may be that the hospital-based studies have some inherent selection biases as such controls may just represent a sample of ill-defined reference population and may not be very representative of the study population or the general population, particularly when the genotypes under investigation were associated with the disease conditions that the hospital-based controls may have.
One of the most important goals of the meta-analysis is to identify the source of heterogeneity. In the sub-analysis of gastric cancer, there was significant heterogeneity for recessive model comparison. Thus, we stratified the studies on gastric cancer according to ethnicity and source of controls. Through analysis, it was found that neither ethnicity nor source of controls contribute to substantial heterogeneity. It is possible that other unmeasured characteristics in different study populations and/or inherited limitations of the recruited studies may partially contribute to the observed heterogeneity.
The meta-analysis has some strengths and limitations. This article is potentially limited in several ways. First, our result was based on unadjusted estimates, while a more precise analysis should be conducted adjusted by other factors such as age, sex, smoking and drinking status. Lack of information for data analysis might cause serious confounding bias. In addition, lacking the original data of the included studies limited our further evaluation of the potential interactions because gene-gene, gene-environment interactions and even different polymorphic loci of the same gene might modulate cancer risk. Second, some of the studies had a very small sample size and did not have adequate power to detect the possible risk for IL-10-592C.A polymorphism and the observed significant ORs in some studies of small sample size may be false association. Third, misclassifications on genotypes and disease status may influence the results because the quality control of genotyping was not well documented in some studies and cases in several studies were also not confirmed by pathology or other gold standard methods. Nonetheless, advantages in our metaanalysis should also be acknowledged. First, the statistical power of the analysis was greatly increased, because a substantial number of cases and controls were pooled from different studies. Second, the quality of case-control studies included in this meta-analysis was satisfactory according to our selection criteria. Third, we did not detect any publication bias indicating that the whole pooled result should be unbiased.
In conclusion, our meta-analysis suggests that the IL-10-592C.A polymorphism was associated with a significant decrease in overall cancer risk, especially in smoking-related cancer, Asians and hospital-based studies. However, large studies using standardized unbiased genotyping methods, enrolling precisely defined cancer patients and well-matched controls, especially in African populations, with more detailed individual and environmental data are warranted to validate the results of our meta-analysis.

Author Contributions
Conceived and designed the experiments: QD JW. Performed the experiments: QD YS. Analyzed the data: QD BF ZF. Contributed reagents/materials/analysis tools: QD LD WT FL. Wrote the paper: QD XJ.