One of the main risk factors contributing to gastric cancer is infection with H. pylori, which causes a chronic active gastritis [4, 21]. In South Africa, gastric cancer is infrequent, while the prevalence of H. pylori infection is very high. Although differences in genotypes of H. pylori exist in different geographic areas, this African enigma can not only be explained by differences in virulent strains of H. pylori [22–24]. High prevalence of vacA s1b strain is observed in South Africa as well as in Brazil and Portugal, countries with high incidences of gastric cancer,[25–27] and frequencies of CagA antibodies were similar between patients with gastric neoplasia compared to healthy controls . The prevalence of the different virulent strains in the present study is unknown. Since H. pylori is thought to play a major role in the initiation phase of gastric cancer development and most often already disappeared at time of gastric cancer diagnosis, it is impossible to accurately retrieve this information.
Besides the virulence of the infecting H. pylori strain, other factors influence gastric cancer risk, including environmental factors such as diet and socioeconomic status, and host factors, such as polymorphisms, which are involved in the inflammatory response to the infection [29, 30]. Knowing that the prevalence of H. pylori infection and incidences of gastric cancer are different in South Africa and Western Europe, we aimed to study if this would reflect in different patterns of gastric carcinogenesis.
The concept of the African enigma has been challenged since it has been suggested that the enigma could be explained due to lack of infrastructure and access to hospitals and care in African countries resulting in incomplete reporting of gastric cancer. However the incidence of gastric cancer would have been so dramatically underestimated that it has been stated that under-reporting by itself could not explain the lower frequencies of gastric cancer in African countries . Also, when using the proportional frequency of gastric cancer compared to other cancer types in Africa, gastric cancer incidence remains very low . Another criticism on the African enigma has been the high prevalence of HIV infection. A relatively large part of the African population would die of HIV before the age in which gastric cancer becomes more frequent. However, the low gastric cancer incidence in Africa was described before the HIV epidemic.
South African patients showed significantly more microsatellite instable gastric cancers compared to Western European patients. Also at the level of chromosomal instability clear differences were found, reflected by a significant correlation between cluster membership and geographical tumor origin, i.e. UK, native SA and Caucasian SA. Microsatellite instable gastric cancers are described to have fewer chromosomal aberrations compared to microsatellite stable gastric cancers [32, 33]. To rule out that tumors from South African patients cluster together by hierarchical cluster analysis due to the fact that these tumors show higher frequencies of microsatellite instability, only microsatellite stable gastric cancers were included in the hierarchical cluster analysis.
Not much has been reported about microsatellite status in gastric cancers from African patients. One study reported infrequent microsatellite instable gastric cancers in South African patients which is in contrast with our findings which show a higher frequency of microsatellite instability in gastric cancers from SA patients compared to UK patients. Based on the present data, MSI does play an important role in gastric carcinogenesis in South Africa.
Several chromosomal aberrations are common in the three different tumor groups analyzed, including gains of chromosomes 7, 8q, 9q, 17q, 19 and 20 and losses of 3p and 4q, while other chromosomal changes are specific for each tumor group. In addition, gastric cancers from UK patients showed a significantly higher number of clones showing a loss compared to gastric cancers form South African patients. These results indicate different patterns of chromosomal instabilities in gastric cancers correlating to geographical origin of the patient.
The chromosomal aberrations of the UK tumors are comparable to other array CGH studies analyzing Western European tumors [12, 35–37]. To the best of our knowledge, this is the first array CGH study on gastric cancers from South African patients. Since several chromosomal regions are significantly different between gastric cancers from different geographical origin, and each region comprises multiple genes, further studies are needed to pinpoint candidate genes contributing to the differences in genomic profiles.
The higher frequency of diffuse gastric cancers from UK patients compared to the SA patients in the present study could be considered as a confounding factor. Contradicting results have been published either describing different or similar patterns of DNA copy number aberrations between intestinal and diffuse type gastric cancers [7, 33, 37–39]. In the context of the present study we believe that the differences in DNA copy number aberrations between UK and SA gastric cancers are independent of the histological tumor type. When repeating the cluster analysis with intestinal type carcinomas only, cluster membership again was significantly correlated with geographical origin of the tumors. Also supervised data analysis, i.e. testing copy number status of all genomic loci, did not reveal any significant differences in DNA copy number changes between intestinal and diffuse gastric cancers from UK patients. Furthermore, hierarchical cluster analysis including UK gastric cancers only did not separate intestinal and diffuse type gastric cancers. We therefore do not believe our findings to be influenced by distribution of histological types in this series. Question remains why diffuse type gastric cancers were more frequently observed in gastric cancers from UK patients compared to SA patients. Besides being a confounding factor, we can hypothesize that mutation of E-cadherin (CDH1), or other mechanisms disrupting the CDH1 gene function such as epigenetic mechanisms or miRNAs, playing an important role in diffuse type gastric cancer, might play a minor role in SA gastric cancer patients due to different pathways of carcinogenesis, as shown in the present study by differences in patterns of DNA copy number aberrations. Also, the prevalence of H. pylori infection is very high in South Africa, and H. pylori infection mainly plays a role in intestinal type gastric cancers. This could also explain the higher number of intestinal type gastric cancers in SA patients.
Further, with respect to copy number changes in relation to histological types, chromosomal gains of 8q and 17q and losses of 3p have been described to be associated with intestinal type gastric cancers . On the other hand, gains of 8q and 17q have been reported to be altered predominantly in diffuse type gastric cancers . In the present study gains on chromosomes 8q and 17q and losses on 3p were common to both intestinal and diffuse type gastric cancers. In addition, these aberrations also were common in tumors from both UK and SA patients. Gains on chromosomes 13q and 19q have been found more frequently in diffuse type gastric cancers [33, 36, 38]. Again, in the present study, gains of these chromosomes were observed equally in intestinal and diffuse type gastric cancers. Gain of 19q was frequently observed in tumors from both geographical origins. Although gain of 13q was observed less frequently in tumors from SA patients compared to tumors from UK patients, still around 20-25% of the tumors of native SA patients show a gain of chromosome 13q, making it unlikely that tumor type has influenced cluster membership.
A limitation of the present study is the fact that native SA gastric cancer patients were significantly younger compared to Caucasian SA and UK gastric cancer patients. We previously showed that gastric cancers of young and elderly patients have different patterns of chromosomal aberrations . We cannot rule out that also in these series, age might contribute to differences in DNA copy number profiles, however cluster analysis showed that gastric cancers from native SA patients were more similar to cancers from Caucasian SA patients, who have similar age as UK patients, indicating that cluster membership is independent of age in this respect. Overall, most differences were observed between UK and native SA tumors.
We realize that the present study is based on a heterogeneous group of gastric cancer patients, with different genetic background and different environmental factors, including H. pylori, diet and socioeconomic status, influencing gastric cancer risk. Statements on genotype influencing gastric cancer are very difficult to make since the degree of heterogeneity within each different patient group, i.e. UK, native SA and Caucasian SA, is unknown.
The patterns of genomic alterations in gastric cancers from UK and SA patients could gain clinical relevance in the future. In addition to surgery, gastric cancer treatment increasingly includes (neo)adjuvant chemotherapy and/or radiotherapy, however still without patients being stratified based on biological characteristics of their tumors. Clinical trials are underway in which also the value of genetic markers for predicting response to therapy are studied. In the end, stratification for therapy may include genomic alterations observed in tumors of patients from different geographical origin.