Background. Individual variations in gastric cancer risk have been associated in the last decade with specific variant alleles of different genes that are present in a significant proportion of the population. Polymorphisms may modify the effects of environmental exposures, and these gene-environment interactions could partly explain the high variation of gastric cancer incidence around the world. The aim of this report is to carry out a systematic review of the published metaanalyses of studies investigating the association between gene polymorphisms and gastric cancer risk, and describe their impact at population level. Priorities on the design of further primary studies are then provided. Methods. A structured bibliographic search on Medline and EMBASE databases has been performed to identify metaanalyses on genetic susceptibility to gastric cancer, without restriction criteria. We report the main results of the metaanalyses and we describe the subgroup analyses performed, focusing on the detection of statistical heterogeneity. We investigated publication bias by pooling the primary studies included in the meta-analyses, and we computed the population attributable risk (PAR) for each polymorphism. Results. Twelve meta-analyses and one pooled-analysis of community based genetic association studies were included, focusing on nine genes involved in inflammation (IL-1β, IL-1RN, IL-8), detoxification of carcinogens (GSTs, CYP2E1), folate metabolism (MTHFR), intercellular adhesion (E-cadherin) and cell cycle regulation (p53). According to their random- Odds Ratios, individuals carrying one of the IL-1RN *2, IL-1β -511T variant alleles or homozygotes for MTHFR 677T are significantly at higher risk of gastric cancer than those with the wild type homozygote genotypes, showing high PARs. The main sources of heterogeneity in the meta-analyses were ethnicity, quality of the primary study, and selected environmental co-exposures. Effect modification by Helicobacter pylori infection for subjects carrying the unfavourable variant of IL-1 polymorphisms and by low folate intake for individuals homozygotes for MTHFR 677T allele has been reported, while genes involved in the detoxification of carcinogens show synergistic interactions. Publication bias was observed (Egger test, p = 0.03). Discussion. The published meta-analyses included in our systematic review focused on polymorphisms having a small effect in increasing gastric cancer risk per se. Nevertheless, the risk increase by interacting with environmental exposures and in combination with additional unfavourable polymorphisms. Unfortunately meta-analyses are underpowered for many subgroup analyses, so additional primary studies performed on larger population and collecting data on environmental and genetic co-exposures are demanded.
Keywords: Gastric cancer, meta-analysis, heterogeneity, polymorphism, population attributable risk