The present study detected EBV-GCs in 3% of gastric carcinomas and suggested that the majority of gastric carcinomas in Iran are EBV-negative. There were no LELCs, which are well known to be strongly associated with EBV . Lace pattern, which is a unique morphology in EBV-positive early gastric carcinoma , was not seen in our study. This frequency is among the lowest frequencies in the world. As mentioned above, according to WHO classification, the north of Iran is considered as one the regions with high gastric carcinoma frequency in the world . The sparse studies in Iran also show this fact except that they show more or less homogeneity throughout the country [48, 62, 63]. Thus, the low EBV-GC frequency in Iran supports the previous hypothesis that high risk countries for GC have low rates of EBV-GC . This also confirms the belief that it is low in Asia as the prevalence of 3% in Korea , 6% in Japan , 6% in china , 5% in India , 2% in Pakistan  and maximal 10% in Malaysia , Taiwan  and Kazakhstan . However, this is much lower than Kazakhstan frequency, (our nearby country) with many similarities in the custom and socioeconomic conditions.
The prevalence of EBV-GC in diffuse type gastric carcinoma was more in our study, a fact which was seen in most of studies of Latin America [16, 32] and some countries in Asia such as Korea , China  and India . However, there are many contradictory studies which did not show any relationship in this concept [15, 17, 19, 23, 33]. Koriyama et al  in their classic study proposed that there are many other factors which have effect on the incidence of diffuse type gastric cancer such as age, gender and location and the conflicting results in different studies could be explained by the difference in distribution of these factors.
Although the ratio of male/female was approximately 4 in our study, it was not significant in statistical analysis. The absence of gender preference was seen in other studies in Mexico [32, 37] Chile  and one area in China . However, we should consider that the low number of positive cases in our study might be the reason of insignificant result in the analysis. Meanwhile, Koriyama et al  show that the gender preference for EBV is restricted to diffuse type GC and is not seen in intestinal type. Nevertheless, regarding the low number of our cases, we could not perform the analysis separately for each type.
Increase in cancer of upper portion of the stomach despite decrease in total incidence of gastric cancers is seen in many studies around the world [65, 66]. There are also many evidences that this type of gastric cancer has different risk factors [67–69]. EBV is one of them [15–31]. Abdirad et al  also show that there is such increase in cancer of upper third of stomach in Iran. Yet, despite our expectation, this study did not show any preference of EBV for upper or middle third of stomach. It is important to notice that in spite of many confirmatory studies in this context, there is also no such relation in our nearby countries, i.e. Kazakhstan  and India .
EBV in our study did not show any relation with age and depth of invasion. Some studies indicate that there is tendency of EBV-GC to occur in lower age [19, 26, 33, 38]. However, Koriyama et al  revealed that only intestinal type of EBV-GC shows this age preference, and proposed that difference in age of exposure to specific cofactors for each of these types is the probable reason.
One of the interesting findings of the present study is the fact that 6 out of our 9 EBV-GC cases which were selected randomly, were born during the period between 1928 and 1930. Although we could not perform any analysis in regard of the low number of positive cases, this finding seems important since an epidemiological study suggested that EBV infection at early childhood may be related to the development of EBV-GC in adulthood . Although we have no evidence, some hypothesis could be considered. For example, specific conditions or events, leading to EBV infection in early childhood or resulting to exposure to some specific co-risk factors which might have existed around the late 1920s. Therefore, we propose a case control study focusing on gastric carcinoma in patients born around the late 1920s to compare the prevalence of EBV in these patients with the others.
Genotyping of the virus revealed that type 1 is the exclusive type in all of our EBV-GC. The predominance of type 1 is in agreement with other studies [18, 45, 58, 40]. Type 2 is mostly seen in Africa  and there is a belief that it is weaker than type 2 and is mostly seen in immunodeficient patients [40, 72]. The prototype F at BamHI-F was the most common type in this study. This type has a worldwide distribution except in southern China which "f" variant is prevalent and causes nasopharyngeal carcinoma [43, 45]. Our finding is the same as similar studies in gastric cancers in India , Japan , and Chile . BamHI-W1/I1 region polymorphism analysis shows that 90% of our cases were type "i". Previous studies revealed that type I is the most prevalent in Asia [45, 54] and type "i" is the most in western countries . This pattern of distribution had been also seen in similar studies on EBV-GC [18, 45, 73]. However, our finding seems to be similar to western countries like Corvalan findings in Colombia and Chile . XhoI+ was the most prevalent type at XhoI restriction site in exon one of LMP1 gene in the present study. This finding is also similar to Corvalan finding in EBV-GC in Colombia and Chil  and in contrast to findings in eastern Asia .
The most important finding in this study was the combination of "i"/XhoI+ in 6 of our cases. Corvalan et al had shown that this combination is significantly more in EBV-GC than healthy people . They propose that the existence of certain gene with transforming capacity in the vicinity of these sites is the probable explanation of more tumorigenecity of this variant. This finding is against the belief that EBV strains are geographically distributed but not disease restricted . However, we need first to evaluate the genotype of EBV in healthy people in Iran before any conclusion. To our knowledge, the genotype distribution among healthy population in Iran is not reported in the literature.