Gastric cancer, the second leading cause of death from cancer throughout the world, is an important health problem. A 2005 analysis of the worldwide incidence of and mortality from cancer showed that 934,000 cases of gastric cancer occurred in 2002 and that 700,000 patients die annually of this disease.1 Despite the overall decline in gastric cancer rates in most of the Western World, gastric cancer remains a serious fatal disease throughout much of the rest of the world. Thirty-eight percent of worldwide cases occur in China, where it remains the most common cancer in both sexes as it is elsewhere in Eastern Asia.2 Conversely, the incidence rates of adenocarcinomas of the proximal stomach and distal esophagus have been increasing, particularly in the Western World.3 Gastric carcinogenesis is a complex, multistep and multifactorial process, in which many factors are implicated. The majority of gastric cancers are thought to be caused by environmental factors that result in damage to the mucosa and that inhibit its ability to repair itself. This response is regulated, in part, by inhibitory and stimulatory factors that are products of proto-oncogenes and tumor suppressor genes.4
The p53 tumor suppressor gene, located on chromosome 17p13, is one of the most commonly mutated genes in all types of human cancer.5 It contains 11 exons, and encodes a 53 kDa phosphoprotein that is a transcription factor for genes that induce cell cycle arrest or apoptosis. The p53 acts as a tumor suppressor gene, negatively regulates the cell cycle and requires loss of function mutations for tumor formation.6 Although p53 contains several polymorphic sites, only those in exon 4 have been examined in gastric cancer. Exon 4 contains 2 polymorphic sites, 1 at codon 36 and another at codon 72. Of these, the codon 72 polymorphism is by far more common. The polymorphism consists of a single base pair change of either arginine or proline which creates 3 distinct genotypes: homozygous for arginine (Arg/Arg), homozygous for proline (Pro/Pro) and a heterozygote (Pro/Arg).7 P53 codon 72 polymorphisms have been reported to be associated with cancers of the lung,8 esophagus,9 colorectum,10 breast,11 bladder12 and cervix.13
Over the last two decades, a number of case–control studies were conducted to investigate the association between p53 codon 72 polymorphism and gastric cancer risk in humans. But these studies reported conflicting results. No quantitative summary of the evidence has ever been performed. The purpose of this meta-analysis was to quantitatively summarize the evidence for such a relationship.
Material and methods
Literature search strategy
Search was applied to the following electronic databases: the Cochrane Library (third quarter, 2006), MEDLINE (1966 to January 2007), EMBASE (1980 to January 2007) and Chinese Bio-medicine Database (1979 to January 2007). The following key words were used: ‘p53’ or ‘codon 72’, ‘gastric’ or ‘stomach,’ ‘carcinoma’ or ‘cancer’ or ‘tumor.’ The search was without restriction on language, conducted on human subject. The reference lists of reviews and retrieved articles were hand searched at the same time. We did not consider abstracts or unpublished reports. If more than 1 article was published by the same author using the same case series, we selected the study where the most individuals were investigated.
Inclusion and exclusion criteria
We reviewed abstracts of all citations and retrieved studies. For inclusion in the meta-analysis, the identified articles have to provide information on: (i) the number of gastric cancer cases and controls studied; (ii) the number of individuals homozygous for arginine (Arg/Arg), proline (Pro/Pro) and heterozygote (Pro/Arg) in cases and controls. Major reasons for exclusion of studies were (i) no control; (ii) duplicate; (iii) no usable data reported.
All data were extracted independently by 2 reviewers (Zhou Y and Li N) according to the prespecified selection criteria. Disagreement was resolved by discussion. The following data were extracted: study design and period, statistical methods, population, number of gastric cancer cases and controls studied and results of studies.
The statistical analysis was conducted using STATA 8.2 (StataCorp, College Station, Tex), p < 0.05 was considered statistically significant. Dichotomous data were presented as odds ratio (OR) with 95% confidence interval (CI). Statistical heterogeneity was measured using the Q statistic (p < 0.10 was considered representative of significant statistical heterogeneity).14 Heterogeneity was also assessed through visual examination of L'Abbe plots. Fixed effects model was used when there was no heterogeneity of the results of the trials. Otherwise, the random effects model was used. For dichotomous outcomes, patients with incomplete or missing data were included in sensitivity analyses by counting them as treatment failures. To establish the effect of clinical heterogeneity between studies on meta-analysis' conclusions, subgroup analysis was conducted on the basis of race and the location, stage, Lauren's classification, histological differentiation of gastric cancer.
Several methods were used to assess the potential for publication bias. Visual inspection of asymmetry in funnel plots was conducted. The Begg rank correlation method and the Egger weighted regression method were also used to statistically assess publication bias (p < 0.05 was considered representative of statistically significant publication bias).
There were 1,254 papers relevant to the searching words (Fig. 1). Through the step of screening the title, 1,079 of these articles were excluded (246 were not English or Chinese Languages, 98 were not case–control studies, 735 were not conducted in humans). Abstracts from 175 articles were reviewed and an additional 158 trials were excluded (55 were not case-control studies, 103 were not conducted in humans), leaving 17 studies for full publication review. Of these, 5 were excluded [2 were not case–control studies,7, 15 2 did not report usable data,16, 17 1 was duplicate18]; thus, 12 papers,19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 which included 1,665 gastric cancer cases and 2,358 controls, were found to conform to our inclusion criteria. Twelve studies, including 4 population-based case-control studies and 8 hospital-based case–control studies were included in this meta-analysis. Studies were carried out in Japan, UK, China, Mexico, USA, Korea and Russia. Characteristics of studies included in this meta-analysis are presented in Table I.
Table I. Characteristics of Studies Included in the Meta-Analysis
No. of cases
No. of controls
Arg/Arg of cases
Pro/Arg of cases
Pro/Pro of cases
Arg/Arg of controls
Pro/Arg of controls
Pro/Pro of controls
HCC, hospital-based case–control; PCC, population-based case–control; DNR, data not reported.
The combined results based on all studies showed that there was no significant difference in genotype distribution [Arg/Arg (OR = 0.96, 95% CI = 0.79, 1.16); Pro/Pro (OR = 1.21, 95% CI = 0.92, 1.58); Pro/Arg (OR = 0.95, 95% CI = 0.79, 1.14)] between gastric cancer and noncancer patients. When stratifying for race, results were similar except that patients with gastric cancer had a significantly lower frequency of Arg/Arg (OR = 0.84, 95% CI = 0.72, 0.99) than noncancer patients among Asians (Figs. 2–4).
When we stratified the various studies by the location of gastric cancer, no statistically significant results was observed for all analysis except that patients with cardia gastric cancer had a significantly higher frequency of Pro/Pro (OR = 3.20, 95% CI = 1.46, 7.01) than those with noncardia gastric cancer among Asians. When we stratified the various studies by the stage of gastric cancer, no statistically significant results was observed for all analysis except that patients with advanced (stage III/IV) gastric cancer had a significantly higher frequency of Arg/Arg (OR = 1.48, 95% CI = 1.01, 2.16) than those with early (stage I/II) gastric cancer among Asians. When we stratified the various studies by the histological differentiation of gastric cancer, no statistically significant results was observed for all analysis except that patients with poor differentiation had a significantly lower frequency of Pro/Pro (OR = 0.13, 95% CI = 0.03, 0.64) than those with well differentiation among Caucasians. When we stratified the various studies by the Lauren's classification of gastric cancer, no statistically significant results was observed for all analysis (Table II).
Table II. Meta-Analysis of P53 Codon 72 Polymorphism and Gastric Cancer
Statistically significant heterogeneity was observed between trials for all analysis with the Q statistic (Arg/Arg p = 0.04; Pro/Pro p = 0.01; Pro/Arg p = 0.06). In addition, L'Abbe plots did show evidence of heterogeneity (Fig. 5). Review of funnel plots could not rule out the potential for publication bias for all analysis. Publication bias was not evident when the Begg rank correlation method (Arg/Arg p = 0.73; Pro/Pro p = 0.94; Pro/Arg p = 0.30) and the Egger weighted regression method (Arg/Arg p = 0.70; Pro/Pro p = 0.94; Pro/Arg p = 0.07) were used (Figs. 6 and 7).
A genetic predisposition to gastric cancer has been suggested by both epidemiological studies and case reports of gastric cancer families.31 Recent studies suggest that single nucleotide polymorphisms may be related to the tumorigenesis of gastric cancer.32 Individual genetic susceptibility may be critical in a variety of processes relevant to gastric cancer tumorigenesis, such as (i) mucosal protection in the face of Helicobacter pylori infection and other carcinogens, (ii) the inflammatory response, which conditions the maintenance, severity and outcome of the Helicobacter pylori infection, (iii) the functioning of carcinogen detoxification and antioxidant protection, (iv) cell proliferation ability, (v) the intrinsic variability of DNA repair processes; and (vi) the cell apoptotic pathway.26, 33 The mechanism of human gastric tumorigenesis is still relatively unknown, and single nucleotide polymorphisms can be used as a tool in searching for genetic variations of the disease gene and susceptibility, and to increase understanding of the disease mechanism.34
A number of studies have reported the role of p53 codon 72 polymorphisms in gastric cancer. The genotype frequency in the initial study was as follows: Arg/Arg (54%); Pro/Arg (33%); and Pro/Pro (14%). The genotype differed significantly with race (p = 0.0001): 64% of whites had the Arg/Arg genotype compared with 24% of African Americans. There was no statistical significance for tumor location or histological tumor type.7 Another study indicated that codon 72 Arg p53 might be associated with a prolonged survival for patients who had had gastric adenocarcinoma, especially non-cardia adenocarcinoma.21 The summary OR from our meta-analyses revealed that patients with gastric cancer had a significantly lower frequency of Arg/Arg (OR = 0.84, 95% CI = 0.72, 0.99) than noncancer patients among Asians. Stratified the various studies by the location, stage, Lauren's classification, and histological differentiation of gastric cancer, we found that (i) patients with cardia gastric cancer had a significantly higher frequency of Pro/Pro (OR = 3.20, 95% CI = 1.46, 7.01) than those with noncardia gastric cancer among Asians; (ii) patients with advanced (stage III/IV) gastric cancer had a significantly higher frequency of Arg/Arg (OR = 1.48, 95% CI = 1.01, 2.16) than those with early (stage I/II) gastric cancer among Asians; (iii)patients with poor differentiation had a significantly lower frequency of Pro/Pro (OR = 0.13, 95% CI = 0.03, 0.64) than those with well differentiation among Caucasians.
A number of studies have shown significant differences in the biochemical properties of the p53 protein, depending on the particular polymorphic form. There is little information, however, on their respective biologic activities. In Pim's study, they have used an inducible switch system for expressing both polymorphic forms of p53 within Saos-2 cells. Cell cycle analysis postinduction of p53 function reveals striking differences in how the 2 forms of p53 bring about a cessation of cell growth. Thus, the Arg72 form of p53 is significantly more efficient than the Pro72 form at inducing apoptosis. In contrast, the Pro72 form appears to induce a higher level of G1 arrest than the Arg72 form. These results demonstrate significant differences in how the codon 72 polymorphism affects the biological activity of p53.35 Dumont et al. also found that in cell lines containing inducible versions of alleles encoding the Pro72 and Arg72 variants, and in cells with endogenous p53, the Arg72 variant induces apoptosis markedly better than does the Pro72 variant. Their data indicate that at least one source of this enhanced apoptotic potential is the greater ability of the Arg72 variant to localize to the mitochondria; this localization is accompanied by release of cytochrome c into the cytosol. These data indicate that the two polymorphic variants of p53 are functionally distinct, and these differences may influence cancer risk or treatment.36 From our meta-analyses, we found that patients with gastric cancer had a significantly lower frequency of Arg/Arg than noncancer patients among Asians (p = 0.04). But patients with advanced (stage III/IV) gastric cancer had a significantly higher frequency of Arg/Arg than those with early (stage I/II) gastric cancer among Asians (p = 0.04). We also found that the study by Yi and Lee30 may be the reason why although those with Arg/Arg had the low risk of gastric cancer, they were susceptible to advanced gastric cancer, because only this study had the statistical significance (OR = 1.95, 95% CI = 1.15, 3.32) and weighted 45.8% in these 4 included studies.
P53 codon 72 polymorphisms also have been extensively studied for many other cancers. Individuals with the p53 Pro/Pro genotype have been shown to be more likely to develop lung cancer (especially in smokers), and to have slightly worse outcomes.37, 38, 39, 40 The Pro allele has also been found in increased frequency in breast cancer patients.41, 42, 43 Conversely, recent studies with HPV-induced cervical cancer show that more cases have the homozygous Arg allele.44, 45 These studies suggest that p53 codon 72 polymorphisms may serve as risk factors for many different types of cancers, and may play a role in the modulations of certain environmental risk factors.
Statistically significant heterogeneity was observed between trials for all analysis with the Q statistic (Arg/Arg p = 0.04; Pro/Pro p = 0.01; Pro/Arg p = 0.06). The most important factor that contributed to the heterogeneity was whether or not the genotype frequencies were in Hardy-Weinberg equilibrium, because the equilibrium may not hold among a case group if the genotype is truly associated with disease. Observed departures from equilibrium therefore suggest possible issues with the control group, or the study population in general, that might have generated less than ideal circumstances for the investigation of the p53 polymorphism and gastric cancer.46 A departure from Hardy-Weinberg equilibrium can also imply possible ethnic admixture in the population, if the polymorphic site varies in genotype by race.47, 48 In fact, race-specific variation in the distribution of genotypes in the p53 codon 72 polymorphism has been demonstrated.7, 49 Because race may be related to disease, either through common risk factors or other genes in linkage disequilibrium with p53, confounding by race, or population stratification, may have biased results in studies conducted on ethnically diverse populations that did not account for possible confounding.50 In this meta-analysis, subgroup analysis was conducted on the basis of race. In fact, 8 studies were conducted among Asians, and 4 studies were conducted among Caucasians. When stratifying for race, patients with gastric cancer had a significantly lower frequency of Arg/Arg (OR = 0.84, 95% CI = 0.72, 0.99) than noncancer patients among Asians.
There are some limitations to this meta-analysis. First, only published studies were included in the meta-analysis; therefore, publication bias may have occurred, even though the use of a statistical test did not show it. Second, we could not obtain information from most studies on the presence or absence of a history of infection with Helicobacter pylori, a strong risk factor for gastric cancer. Third, as in most meta-analyses, these results should be interpreted with caution because the population from 7 countries and controls were not uniform. Fourth, our meta-analysis is based on unadjusted estimates, while a more precise analysis could be performed if individual data were available, which would allow for an adjustment estimate (by age and sex). To be made, however, this approach requires the authors of all of the published studies to share their data. Finally, meta-analysis remains retrospective research that is subject to the methodological deficiencies of the included studies. We minimized the likelihood of bias by developing a detailed protocol before initiating the study, by performing a meticulous search for published studies, and by using explicit methods for study selection, data extraction, and data analysis.
In conclusion, this meta-analysis suggests that the p53 codon 72 polymorphism may be associated with gastric cancer among Asians, and that difference in genotype distribution may be associated with the location, stage, and histological differentiation of gastric cancer. Since more than half of the included studies were based on a limited number of cases (<150), it is critical that larger and well-designed multicentric studies based on the same ethnic group confirm our results.