Gastric Cancer: Epidemiologic Aspects

Although the rates of gastric cancer (GC) have been declining over the past 50 years in most Western countries, GC is still the fourth most common malignancy and the second leading cause of death due to cancer worldwide. In 2008, more than 990,000 incident cases were recorded (7.8% of new cancer cases) with 738,000 deaths. There were approximately 870,000 noncardia GC cases and 74.7% of them have been attributed to Helicobacter pylori infection 1. More than two-thirds of GC occur in developing countries. The highest incidence rates are observed in Far East Asia, Andean regions of South America, and Eastern Europe, and the lowest in North and East Africa, Northern Europe and North America 1.

Gastric cancer has two main anatomical localizations: tumors arising in the cardia and those in the distal stomach (noncardia). While the incidence of noncardia GC has declined in most countries, the rates of cardia GC have remained stable or risen in several European countries, Japan, and North America. However, recent data from the Netherlands 2 found that the incidence of cardia GC between 1989 and 2008 remained stable in females and slightly decreased in males. Age-standardized incidence rates are about twice as high in men as in women 1. This gender difference is consistent across populations with different prevalences of environmental risk factors. A multifactorial and multistep model of human gastric carcinogenesis is currently accepted, according to which different environmental and genetic factors are involved at different stages in the cancer process. The aim of this article is to review the most relevant information published from April 2012 to May 2013 on the relative contribution of genetic factors and environmental factors in humans.

Genetic Factors

Individual variations in cancer risk have been consistently associated with specific variant alleles on different genes (polymorphisms) that are present in a significant proportion of the normal population. Single nucleotide polymorphisms (SNPs) in a wide variety of genes may modify the effect of environmental exposure, and these gene-environmental interactions could explain the high variation in the GC incidence observed around the world. Individual genetic susceptibility may be critical in a variety of processes relevant to gastric carcinogenesis, including mucosal protection against H. pylori, an inflammatory response to the infection, carcinogen detoxification and antioxidant protection, DNA repair processes, and cell proliferation ability.

Regarding genes involved in the inflammatory response, a meta-analysis based on 18 studies 3 found an association between IL8 promoter−251 AA genotype and GC risk, mainly in Asian populations and for intestinal-type GC, but only using a codominant model. Another meta-analysis 4 on IL10-1082 promoter polymorphism and GC risk, based on 22 studies, including 4289 GC cases, found a significant negative association (overall OR = 0.049, p < .001). The association was observed in Asians, Caucasians but not in Latin-American populations. A third meta-analysis 5 on LAT (TNF-β) rs909253 GA genotype and GC risk, based on 12 studies including 2074 GC cases, found a positive and significant association in Asian populations.

Regarding polymorphisms in DNA repair genes, in a Chinese case–control study 6 including 1125 cases and 1196 controls, the effect of three functional SNPs of XPG (xeroderma pigmentosum group) belonging to the group of nucleotide excision repair genes was investigated. The rs873601A variant was significantly and positively associated with an increase in GC. These results were confirmed in subsequent mRNA expression analyses as well as on subjects from different ethnicities. Another case–control study in China 7 investigated the association of two functional polymorphisms in the promoter region of the excision repair cross-complementing group 5 gene (ERCC5) on 400 GC cases and 400 healthy controls. A significant and positive association was found for both, rs751402 (AA vs GG) genotype (OR 1.99; p = .008) and rs2296147 (CC vs TT) genotype (OR 2.17; p = .039). The effect was stronger for the diffuse subtype of GC.

In relation to tumor-suppressor genes, a Chinese case–control study based on 311 cases and 425 controls 8 found a significant inverse association between the human DOC2/DAB2 interactive protein (HDAB2IP) gene and GC risk, for the minor allele C, confirmed by haplotype analysis. An updated meta-analysis including 28 case–control studies confirm a modest decreased risk of GC among Asians for TP53 codon 72 Arg/Arg genotype 9.

Regarding metabolic genes, in a two-phase (discovery and replication) genetic association study, the role of genes involved in the steroid hormone biosynthesis pathway and progesterone receptor (PGR) was investigated in a Korean cohort 10. Several SNPs in the CYP19A1, which encodes aromatase, were associated with an increased GC risk. In a nested case–control study in the EPIC cohort, based on 365 incident GC cases and 1284 controls 11, the H63D variant of the hemochromatosis gene (HFE) appears to be associated with noncardia intestinal GC risk, possibly due to its association with iron overload. This is consistent with the association with red meat intake. In the same EPIC cohort 12 on GC, results showed that genetic variants of alcohol dehydrogenase (ADH1A) and aldehydrogenase (ALDH2) may influence GC risk, and alcohol intake may further modify the effect of ADH1A rs1230025.

Regarding genes involved in cell biology or in the regulation of gene expression, genetic polymorphisms of E-cadherin gene (CDH1) may affect GC risk by altering gene transcriptional activity of epithelial cells, but so far results of CDH1 variants in sporadic GC are inconsistent. A case–control study in China 13 based on 387 incident cases found a positive association with two SNPs for diffuse GC. Another study 14, based on 311 cases and 425 controls, found that EZH2 (enhancer of zeste) gene variants were associated with GC risk. EZH2 encodes a histone methyltransferase that may produce epigenetic silencing of genes. A meta-analysis on a common functional polymorphism of the survivin gene 15 involved in the regulation of survivin expression found a positive association with GC risk. A few years ago, prostate stem cell antigen (PSCA) polymorphism was associated with GC in a genome-wide association study (GWAS) analysis of an Asian population. In a meta-analysis including nine case–control studies 16, it was found that rs2294008 and rs2976392 were associated with GC risk in both Asian and European populations, the risk being higher for noncardia diffuse type GC.

MicroRNAs (miRNAs) are stable small RNAs that have an important role in regulating gene expression and their analysis in tumor samples and biologic fluids is of growing interest as useful biomarkers for early stage diagnosis and prognosis of cancer. Some studies have been published on GC in the last years, although more comprehensive studies are required. In a large cohort in China 17, three miRNAs (miR-221; miR-744, and miR-376c) were identified as being capable of distinguishing GC cases from controls with 82.4% sensitivity and 58.8% specificity. Another study 18 showed that the has-miR-335 had the potential to recognize the recurrence risk and could be related to the prognosis of GC patients. Genetic polymorphisms in several microRNA genes, such as miR-27a, miR-181a and miR-196a2, have also been found associated with GC and its prognosis 19-21 during the last year. Furthermore, polymorphisms in the miRNA-binding site of specific target genes have also been found associated with GC 22, 23.

Other genetic variants that have been associated with noncardia GC through GWAS and further replication analyses are rs2494938 at 6p21 and rs2285947 at 7p15.3, which also have a role in the susceptibility to other cancers 24. Similarly, potentially functional variants at PLCE1 have been confirmed to be associated with cardia GC 25.

Environmental Risk Factors

It is well known that dietary factors play a role in gastric carcinogenesis. High consumption of fruit and vegetables has been associated with a reduction in GC risk, but mainly from case–control studies, while the effect from cohort studies seems to be weaker. In a reanalysis in the EPIC cohort 26, based on 683 gastric adenocarcinomas, an inverse and significant association between the total vegetable and fruit intake and the GC risk was observed, between fresh fruit intake and the risk of diffuse type GC, and between citrus fruit intake and the risk of cardia GC. In the same study, a negative association was also found with dietary total antioxidant capacity 27 for both cardia and noncardia GC. In another study on the same EPIC cohort, a significant inverse association between total flavonoid intake and GC risk was found in women but not in men 28. In a systematic review of cohort and case–control studies among the Japanese population 29, a decrease in GC was associated with the consumption of green tea in women but not in men. Green tea is one of the sources of flavonoid intake.

It is believed that salt and salt-rich foods probably increase the risk of GC. A meta-analysis of prospective studies 30 found a positive and significant association between the amount of habitual salt intake and GC risk, with a progressively increasing risk across consumption levels. The effect was stronger in Japanese studies. There is some evidence that high intake of pickled foods in Far East Asia increases the risk of GC. A systematic review and meta-analysis 31 confirmed this association, suggesting a potential 50% higher risk of GC associated with intake of pickled vegetables/foods and perhaps stronger associations in Korea and China.

In the last years, several studies have shown a positive association between GC (mainly noncardia) and red meat intake, although the mechanism is not yet clear. Red meat intake is the most important source of endogenous formation of nitrosamines, probably due to the heme-iron content. In a population-based case–control study in Nebraska, USA 32, including 154 GC cases and 449 controls, a significant positive association between a high rate of GC and a high intake of heme iron and total iron from meat was observed. In a prospective study in Finland 33 in which prediagnostic serum iron, ferritin, and unsaturated iron-binding capacity were measured, a “u”-shaped relationship with total iron-binding capacity and an inverse association between serum ferritin and serum iron was observed in patients with GC. In the Netherlands cohort 34 including 497 noncardia GC, 166 cardia GC and 110 esophageal squamous cell carcinoma (ESCC), a positive association between N-nitrosodimethylamine intake (the most important nitrosamine, considered as a probable carcinogen for humans) and noncardia GC and ESCC in men was observed. Heme-iron intake was associated with ESCC but not with noncardia GC. On the other hand in a prospective study (EPIC Spain), a positive association between aromatic DNA adducts from leukocytes and GC risk was observed 35. Aromatic compounds are formed during cooking of meat but also occur in tobacco smoking.

There is important evidence showing that regular aspirin use may reduce the progression of preneoplastic lesions and reduce the incidence of GC and other gastrointestinal cancers. A wide systematic review comparing results from observational and randomized trials 36 confirms this evidence. Regular use of aspirin reduces the long-term risk of GC and also the risk of distant metastasis. Results were consistent among both types of studies.

There is strong evidence showing the positive association between esophageal adenocarcinoma and general and abdominal obesity, but it remains unclear whether there is an association with GC. In a large prospective study in the USA 37 including 191 cardia and 125 noncardia GC, a positive association between cardia GC and BMI (HR highest vs referent 3.67, 95% CI 2.0–6.7) and waist circumference HR 2.22, 95% CI 1.4–3.5) was observed. However, as expected, obesity was not associated with noncardia GC.

It is well known that people infected with human immunodeficiency virus have an increased risk of some cancers, but little is known about the effect on GC. In a large study in the USA 38, the risk of GC in patients with AIDS and those from the general population was compared. There was a positive association for both cardia and noncardia GC.

In a meta-analysis of 29 case–control studies in Latin America (so far no cohort studies have been published) from countries with high GC incidence, the role of different GC risk factors was investigated 39. Consistent with multifactorial pathogenesis and with results from other populations, the GC risk factors were the following: smoking, alcohol drinking, consumption of red and processed meat, salt, and carriage of ILRN*2. Conversely, a higher level of education, fruit, and vegetable intake were associated with a decreased risk. The only region-specific factor was the positive association with chili pepper reported in at least five studies.