Worldwide, the relation between polymorphisms in genes associated with inflammatory factors and the risk of gastric cancer has been investigated extensively since the landmark paper by El-Omar et al. (58
) was published in 2000. Polymorphisms in several other genes related to inflammation have since been investigated in different populations in studies of variable quality. We conducted a series of meta-analyses, using a predefined protocol, in order to study the most frequently investigated genes, including those coding for interleukins (IL1B
, and IL10
) and TNFA
. We included information about IL1B
-819T, and TNFA
-238A carriers, but among these persons, we did not observe any statistically significant effects. The analyses presented revealed risk differences following stratification by histologic type, anatomic site, geographic location, and H. pylori
infection status. We believe that the simultaneous consideration of meta-analyses for multiple polymorphisms within genes with related functions results in a broader overview and allows for more detailed comparisons of the evidence base. The assessment of all studies according to predefined quality criteria was another important strength of our approach.
In this review, the clearest results were found for the association with IL1RN
2, where an increased risk of gastric cancer was observed. This appeared to be confined to non-Asian populations but was observed for both intestinal and diffuse cancers, as well as distal and, to a lesser extent, cardia cancers. Analyses restricted to high-quality studies or confined to H. pylori
-positive cases and controls were supportive of the observed effect that was found in association with both carrier and homozygote status. The IL1RN
22 genotype has been reported to cause high circulating IL-1ra and IL-1β levels, possibly resulting in a severe and prolonged inflammatory response. We believe this is mirrored in our results and indicates the importance of IL1RN
in the association with gastric cancer. Together with IL1B
-31 and IL1B
was the first polymorphism reported to be related to gastric cancer (58
-511 and IL1B
-31 were reported to be in linkage disequilibrium early, a consequence being that subsequent studies investigated only 1 of these genes in order to save money and material. Linkage disequilibrium was observed for studies in which both SNPs were analyzed; however, because of the limitation of studies not reporting on both of the SNPs for all anatomic sites or all histologic types, the corresponding odds ratios do not always indicate absolute linkage disequilibrium.
In our meta-analysis, the clearest results relating to Asian populations were observed for IL1B
-31, where C carrier status was associated with a reduced overall risk. This effect was also observed in analyses restricted to high-quality studies, and there is some evidence that the association is confined to cancer at distal subsites and of intestinal histology. The reduced risks may be explained by the high incidence of H. pylori
infection in the Asian populations. The polymorphisms in genes related to IL-1β production might encourage a stronger reaction against the infection that further reduces the damage to the gastric cells and increases apoptosis of the epithelial cells, preventing the development of gastric cancer (72
By contrast, in an association apparently confined to non-Asians, IL1B
-511T carriers were found to have an increased risk of cardia gastric cancer specifically. This association was also observed in analyses restricted to high-quality studies but was less evident in association with distal cancers or in relation to any specific histologic subtype. Here, a stronger inflammatory reaction may instead increase the risk of cancer through damage to gastric cells and bacterial overgrowth and accumulation of toxic byproducts (72
In a recently published meta-analysis by Kamangar et al. (6
), no association was found between the IL1B
-31, and IL1RN
polymorphisms and risk of developing gastric cancer independent of histologic type. These conclusions differ from our results, possibly because of fewer studies in the analysis and the lack of further stratification by study quality. On the other hand, Camargo et al. (5
) and Wang et al. (7
) suggested that IL1B
-511T and IL1RN
2 carriers had an increased risk of developing intestinal gastric cancer. Those studies (5
), which also stratified the meta-analysis into Caucasian and Asian populations and included similar numbers of studies, showed results similar to ours.
Interestingly, the results for IL10
-1082G showed entirely contrasting effects in Asians (increased risk) and non-Asians (decreased risk), although this contrast was not specifically confirmed within high-quality studies, possibly because of the low numbers of studies included. Recently, a meta-analysis confirmed the increase in Asian populations, but a limited number of Caucasian studies showed nonsignificant results (9
). Finally, in carriers of TNFA
-308A, there was good evidence for increased risk specific to diffuse cancers in non-Asians (supported by high-quality studies, but with no effect of H. pylori
In recently published meta-analyses (4
), investigators reported an increased risk for gastric cancer in carriers of TNFA
-308A among Caucasians, while increased risks for distal and intestinal cancers were also found in another meta-analysis (8
), though with no further stratification by population.
There are some difficulties in the interpretation of our results. This systematic review and meta-analysis was built upon data collected from the current literature and through personal contact with authors when relevant results were not available from the publication. Not all of the authors contacted had access to data or could provide information for all polymorphisms, anatomic sites, or histologic types. As a result, some studies only contributed information to the overall gastric cancer analysis, whereas others contributed additionally to subanalyses (e.g., anatomic site-specific results). Therefore, the overall results cannot always be expected to reflect a weighted average result for the component anatomic locations. Few high-quality studies were conducted within Asian populations, and when we stratified by H. pylori
infection status, some studies had to be excluded because of the lack of relevant information among controls. Differential misclassification of H. pylori
infection could have occurred if, for example, the presence of gastric cancer results in atrophy, which can make detection less likely (73
). Differences in our results as compared with earlier published meta-analyses may be due to the data collected from the included papers. Here we collected raw data based on information presented in the published articles or obtained from the contacted authors, and we recalculated the odds ratios and confidence intervals. Another problem resides in the fact that definitions (mainly concerning histologic type and site localization) may vary in different parts of the world (e.g., in Japan vs. the West), which can lead to inconsistency when study results are pooled together.
In conclusion, the importance of stratifying by gastric cancer type, site, H. pylori infection, geographic location, and study quality needs to be considered in future studies, together with consideration of associations with multiple genes with similar functions. The underlying genetic background of the population specifically must also be understood. Study quality considerations, both laboratory and epidemiologic, can also affect results and may help explain the variability in findings that have been published to date.