In the present study, 81 cases of GC from patients of the south region of Tunisia were investigated for the presence of EBV and the prevalence of EBVaGC was 14.8% (12 out of 81). It was well known that the prevalence of EBVaGC varies widely from one geographical region to another and the highest frequency was noted in Germany (18%) whereas the lowest one (3.9%) was detected in Peru [13–20, 31].
The EBER in situ hybridization was the most reliable method reported in the literature to detect latent EBV in GC. In fact, all the above mentioned studies were conducted following this method, including our study. The variable proportions of nuclear tumor cells that express EBER showing in our study were also deduced by Truong et al., who suggest to be related with EBV infection occurs in oncogenic process of EBVaGC . However, further investigations must be conducted to clarify this observation
In addition, we showed that the heterogeneous data of EBER expression could be correlated with the number of EBV DNA copies as estimated by Q-PCR supporting that this method is reliable as reported previously .
With regards to the clinico-pathological features of EBVaGC, strong association with age at diagnosis was observed (P = 0.009, Table 3). In fact, the EBVaGC was more frequently found in the group of patients aged between 45 and 60 years old, which is in line with previous studies [19, 34]. No other statistical association was found, except a tendency with tumor differentiation and venous invasion (P = 0.075 and P = 0.09 respectively). Previous reports have indicated that EBVaGC were frequent in male, proximal stomach and tumors of diffuse type [9, 14, 16, 35–39].Recently, in a large meta-analyse, Carmago et al., confirms these associations in addition to the age at diagnosis . In our study, we didn't find difference in the distribution of EBV in male compared to female patients, in proximal vs distal stomach and in diffuse vs intestinal histotype. These variations between data could be explained by the contribution of local risk factors in the pathogenesis of EBV and also by the size and charasteristics of the cohort.
Polymorphism analysis of the 12 EBVaGC cases show exclusively the type D, prototype F, XhoI-retention and the 30 bp del-LMP1 variant. Concerning the polymorphism of the EBNA-3C gene, we found the A type in all EBVaGC cases and a combination of types A and B was found in only one case. These findings are in agreement with recent study conducted on four EBVaGC cases from the central region of Tunisia .
The predominance of type A and prototype F was also shown in previous reports independently to the geographic origin of EBVaGC as in southern China , southern Japan , and Latin American countries . Interestingly, the variant-f was especially described in NPC associated EBV strains of southern Asia . In this area, a predominance of C type and loss-XhoI variant was observed in patients with EBVaGC or NPC [35, 40], in contrast to our present finding and previous data on Tunisian NPC patients [29, 43].
To better define the genotype of EBVaGC strains and compare them with those associated with the NPC, we carried out the partial sequencing of exon 3 of the BNLF1 gene and compared them to the prototype B95-8 and others published sequences from Asian and Tunisian NPC EBV strains [29, 43]. The Q334R, L338S and L338P were found in all our EBV isolates and were already reported in EBV strains associated to NPC originating from China or Tunisia [29, 44]. However, the S366A is specific to EBV isolates associated with Tunisian NPC and GC.. Indeed, and based on the previous report of Edwards et al., describing seven phylogenetically distinct strains of LMP1, we can propose that Tunisian isolates constitute a supplementary group with a specific signature (T366A) but this hypothesis needs confirmation by sequencing of the entire BNLF1 gene in these isolates. According to this date, we suggest that polymorphism in the BNLF1 gene constitute an additional argument in line with the others EBV polymorphisms (D type, prototype F and XhoI+) supporting the dissimilarity of EBV strains between the two geographic regions.
On the other hand, we have described previously others del-LMP1 variants (69 bp and 81 bp deletion spanning codons 334-353 and 345-371, respectively) in Tunisian NPC patients [29, 43]. These variants were not found and only the 30 bp del-LMP1 variant was identified in all EBVaGC cases as already reported by Chen et al., in a large study conducted on Chinese GC patients .
Our polymorphism analysis of EBV isolates in healthy carriers revealed an identical genotype to those of EBVaGC and NPC suggesting that common strains are geographically distributed but not associated with a specific malignancy. In fact, the development of EBV associated malignancies could be correlated to variation in potential immune recognition in distinct populations and individuals, as proposed by Edwards et al, .