Introduction

Cytokines have been investigated for decades and many important cytokines are involved in human diseases, such as interleukin-1 and osteoarthritis1,2. In 1989, Mosmann and colleagues3 first reported a cytokine named “cytokine synthesis inhibiting factor (CISF)”, which was secreted by T helper 2 (Th2) clones and inhibited synthesis of interferon-γ (IFN-γ) in Th1 clones. The CISF is now known as interleukin-10 (IL-10).

IL-10 is a cytokine with potent anti-inflammatory activity4, produced by macropahges, T helper 2 cells and B lymphocytes5,6. IL-10 is a multifunctional cytokine involved in both innate and adaptive immune response4,6. As an inflammatory cytokine, IL-10 participates in the development of various diseases, such as kidney disease7, heart failure8, chronic infection9,10 and cancer10. Although IL-10 has been extensively studied, the exact role of IL-10 in cancer is still elusive, since evidence suggested that IL-10 could mediate both anticancer immune response and immune-mediated rejection of cancer6.

The gene encoding IL-10 is located on chromosome 1 (1q31-1q32). Three single nucleotide polymorphisms (SNPs) have been confirmed in the promoter region of IL-10: -592C>A (rs1800872), -819C>T (rs1800871) and -1082A>G (rs1800896). Previous studies have shown that the three polymorphisms could affect the expression of IL-1011,12 and alter the susceptibility to digestive cancers13,14,15. In addition to the elusive role of IL-10 in cancer development, the relationship between functional polymorphisms in IL-10 promoter region and cancer risk is also mysterious. Several meta-analyses have been performed to evaluate the association between IL-10 polymorphisms and cancer risk16,17; however, the association between -1082A>G polymorphism and digestive cancer has not been assessed. Thus, this meta-analysis was performed to investigate the association between 1082 polymorphism and digestive cancer and assess the influence of confounding factors.

Methods

Searching strategy

This meta-analysis were conducted and reported in corresponding to the PRISMA guidelines of systematic reviews and meta-analyses (see Supplementary Table S1 online)18. Online databases of PubMed, EMBASE and CNKI were searched. The following terms were used: “Interleukin-10” or “IL10”, “polymorphisms, single nucleotide” or “SNPs” and “cancer” or “ neoplasm”. Both plain text and medical subheadings of above key words were used for searching. No limitation of origin, languages, or other items was placed. To identify additional studies, references of previous meta-analyses and reviews were also manually searched.

Inclusion criteria

Records identified from databases were first screened by titles and abstracts and then full-text articles were further reviewed. Eligible studies were judged by the following criteria: (1) case-control studies; (2) investigating the association between IL-10 -1082A>G polymorphism and digestive cancer risk; (3) available genotype distribution data. According to the inclusion criteria, 2 authors (LC and TW) extracted eligible studies independently. The two authors reached consensus on each records.

Data extraction

Name of first author, year of publication, country where the study was carried out, cancer type, ethnicity, the source of control, number of cases and controls, genotype frequency in cases and controls were collected from eligible studies. Ethnicity was simply classified as Asian, Caucasian and Latino (Table 1). Included studies were defined as hospital-based (HB) and population-based (PB) according to the source of control. Sample size of eligible studies was classified as large (>500) or small (<500). All data were extracted by two authors (LC and TW) independently with a predesigned data-collection form. Two authors reached consensus on each item.

Table 1 Baseline Characteristics of Eligible Studies
Full size table

Quality assessment

“Methodological quality assessment scale” (the scale can be found as Supplementary Table S2 online), a quality scale modified form previous meta-analyses19, was used to evaluate methodological quality of eligible studies. Briefly, the following items were assessed: the representativeness of cases, source of controls, ascertainment of relevant cancer, sample size, quality control of genotyping methods and Hardy-Weinberg equilibrium (HWE). Quality scores ranged from 0 to 10 (0: the lowest; 10: the highest).

Statistical analysis

Odds ratios (ORs) with 95% confidence intervals (95% CIs) were calculated to estimate the association strength between IL-10 -1082A>G polymorphism and digestive cancer risk. Deviation from the Hardy–Weinberg equilibrium (HWE) among controls subjects was tested by a x2-test and a P<0.05 was considered as significant disequilibrium. The pooled ORs were calculated for allele comparison (G vs. A), homozygote comparison (GG vs. AA), heterozygote comparison (GA vs. AA) and dominant models (GG/GA vs. AA, considering the dominant effect of the IL-10 -1082G allele). For some studies only combined genotype (GG/G) data was reported20,21, thus, only dominant comparison models were conducted for these studies. Heterogeneity between studies were determined by chi-square based on Q test and the random-effects model was used when there was significant heterogeneity (P<0.1); otherwise, the fixed-effects model was applied22. Sub-group analyses were conducted according to cancer types, ethnicities, source of control, HWE and sample size. Sub-group analysis was not performed for those subgroups with less than 2 studies. When significant heterogeneity presented, meta-regression was performed to detect the source of heterogeneity. Egger's test and Begg's test were used to test publication bias and a p < 0.05 was significant23. Sensitivity analysis was performed to assess individual studies' effect on the pooled results. All meta-analyses were calculated by STATA (version 10.0; Stata Corp, College Station, Texas USA). And all P values are two-side.

Results

Overview of eligible studies

According to our searching strategy, 752 records were retrieved and screened. After primary screening, 38 full-text papers were retrieved for further assessment13,14,15,20,21,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56. The study reported by Zhou SZ et al was excluded for lacking of detail genotype distribution data56. In the studies reported by El-Omar EM29, Guo W32 and Savage SA14, both gastric cancer and esophageal cancer were reported and the data were presented independently and each kind of the cancer was treated as a separate study. Thus, 40 eligible studies were included in this meta-analysis13,14,15,20,21,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55. The process of study selection was shown in Figure 1.

Figure 1
figure 1

Flow Chart of Study Selection.

Both gastric cancer and esophageal cancer were reported in 3 studies and each kind of cancer was treated as a separate study. Thus, 37 studies were included in qualitative synthesis and 40 studies were included in quantitative synthesis.

Full size image

Of the 40 eligible studies, 7195 patients of digestive cancer and 11755 controls were enrolled. Baseline characteristics of those studies were shown in Table 1. Most studies were performed among Asian population (24 studies) and Caucasian population (11 studies).

Methodological quality of eligible studies was assessed by a quality scale reported by previous studies. Generally, quality of eligible studies was acceptable, with an average score of 7.3. Of 40 analyzed studies, 23 were hospital-based and 17 studies were population based. As for HWE, 24 studies were in agreement with HWE, 13 studies were in disagreement with HWE and it was unable to test in 3 studies20,21,44 due to combined data (Table 2). Since no genotyping error was reported, all studies were included in quantitative synthesis and stratified analysis was performed to assess the influence of disagreement of HWE.

Table 2 Meta-analysis Results
Full size table

Meta-analysis Results

By pooling all eligible studies, compared with the wild -1082A allele, we found the variant IL-10 -1082G allele was associated with significantly increased risk of digestive cancer in all four comparison models (G vs. A: OR = 1.181, 95% CI: 1.057–1.319; Heterogeneity, P<0.001; Figure 2 and Figure 3; Table 2).

Figure 2
figure 2

Forest plot of IL-10 -1082A>G polymorphism and risk of digestive cancer, G vs. A.

Full size image
Figure 3
figure 3

Forest plot of IL-10 -1082A>G polymorphism and risk of digestive cancer, GGGA vs. AA.

Full size image

Further sub-group analyses were conducted to assess the effects of potential confounding factors. When stratified by cancer types, we found the variant G allele only increased risk of gastric cancer (G vs. A: OR = 1.281, 95% CI: 1.102–1.488; Heterogeneity, P<0.001) but did not alter the risk of colorectal cancer (G vs. A: OR = 0.937, 95% CI: 0.805–1.090; Heterogeneity, P = 0.710), hepatocellular carcinoma (G vs. A: OR = 1.104, 95% CI: 0.797–1.530; Heterogeneity, P = 0.283) or esophageal cancer (G vs. A: OR = 0.982, 95% CI: 0.820–1.175; Heterogeneity, P = 0.591). As for ethnicities, significant association was only found among Asians (G vs. A: OR = 1.399, 95% CI: 1.188–1.646; Heterogeneity, P<0.001), while the -1082 polymorphism did not alter digestive cancer risk in Caucasians (G vs. A: OR = 1.016, 95% CI: 0.930–1.111; Heterogeneity, P = 0.796). HWE also significantly affected the pooled analysis. In the sub-groups classified according to source of control and sample size, meta-analysis results were quite consistent.

Heterogeneity and publication bias

Notably, significant heterogeneity was observed in most comparisons. Thus, meta-regression was performed to detect the source of heterogeneity. The results indicated that sample size (P<0.001), HWE (P<0.001), source of control (P<0.001) and cancer types (P = 0.03) contributed heterogeneity, while ethnicity (P = 0.207) did not. Publication bias was detected by Egger's test and Begg's test and no evidence of publication bias was found (P = 0.133 for Begg's test and P = 0.524 for Egger's test; Figure 4). As shown in Figures 2,3,4, the study reported by Alpizar-Alpizar W24 was an outlier. Thus, sensitivity analysis was performed to assess individual study's effect by omitting one study each time. Sensitivity analysis revealed that the pooled results were not affected by the study by Alpizar-Alpizar W et al or any other studies (see supplementary Figure S3 online), indicating our results were stable and reliable.

Figure 4
figure 4

Funnel plot of IL-10 -1082A>G polymorphism and risk of digestive cancer, G vs. A.

Circles represent the weight of studies.

Full size image

Discussion

In this meta-analysis, we identified 40 eligible studies, including 7195 cases and 11755 controls. By pooling all eligible studies, we found the variant IL-10 -1082G allele significantly increased the susceptibility to digestive cancer, especially to gastric cancer and among Asian population.

By pooling all eligible studies, we found the IL-10 -1082A>G polymorphism was associated with significantly increased risk of digestive cancer in all comparison models. Then stratified analysis showed that the increased risk was mostly contributed by gastric cancer, since significant association was observed only in gastric cancer and ORs in the sub-groups of gastric cancer were similar with those in overall analysis. It has been proposed that inflammation is a risk factor of tumorigenesis57. In the process of chronic gastric inflammation, different types of cytokines are secreted by activated neutrophils and mononuclear cells and altered cytokine levels have been observed36. Thus, it is biological plausible that IL-10 polymorphism increased risk of gastric cancer. Sub-group analysis revealed that the IL-10 -1082A>G polymorphism was only associated with gastric cancer and no association was found for other digestive cancers, indicating that the role of IL-10 varied among cancers.

During sub-group analysis for ethnicities, we found ethnicity significantly affect the association between IL-10 -1082A>G polymorphism and digestive cancer risk. Since the variant -1082G allele was only associated with increased risk in Asian population and no significant association was found in Caucasian population. This ethnicity difference is common for meta-analysis and may be explained by different environmental exposure, life style and genetic background. Of note, since the incidence of gastric cancer was higher in Asian population, most Asian studies were about gastric cancer (19 of 24 studies, as shown in table 1). The higher prevalence of gastric cancer in Asian population might be another explanation for the ethnicity difference.

In the process of statistical analysis, we found the study reported by Alpizar-Alpizar W24 and colleagues was an outlier. This could be explained by ethnicity difference, since the study was conducted among Latinos. Additionally, the frequency of IL-10 -1082G allele was very low in Alpizar-Alpizar's study24. Specifically, the GG and GA genotype was not detected in cases while the GG was not detected in controls and only one participants carried the heterozygote GA genotype in controls (0% in cases and 1.14% in controls), which would led to relatively wide confidence intervals as shown in Figure 2 and Figure 3. It should also be highlighted that this study was conducted in high-risk population, which might be also related with the low frequency of G allele.

In this meta-analysis, we included 40 studies with 18950 participants. The sample size was large enough to provide enough statistical power. Additionally, no publication bias was detected by Egger's test and Begg's test, suggesting our results were unbiased. On the other hand, limitation of this meta-analysis should also be also highlighted. Firstly, heterogeneity was significant in this meta-analysis. Due to the significant heterogeneity, we used random-effects model to calculate the pooled ORs, which could provide stable results. To identify the source of heterogeneity, meta-regression was conducted and revealed that sample size, HWE, source of control and cancer types were the sources. And stratified analyses were also performed to evaluate the influence of these confounding factors. Secondly, individual data were missed and we could not assess the effects of other factors, like environmental factors, life habit and family history.

In summary, in this meta-analysis of 40 studies and 18950 participants, we found the variant IL-10 -1082G allele significantly increased susceptibility to digestive cancer, especially for gastric cancer and in Asian population.