To determine the prevalence of Epstein-Barr virus (EBV)-associated gastric carcinomas in the North Region of Portugal and to study its clinicopathological characteristics.

We have performed a retrospective study including a total of 179 consecutive patients with gastric cancer (GC) submitted to gastrectomy during 2011 at the Portuguese Oncology Institute of Porto. Clinical and pathological data was collected from individual clinical records and inserted on a database with unique codification. Tumour tissues were collected from the institutional tumour bank. EBV was detected by in situ hybridization for the detection of EBV-encoded small RNAs (EBERs) and EBV latent proteins (LMP1 and LMP2A) were detected by immunohistochemistry.

The analysis showed that EBV-associated gastric carcinomas (EBVaGC) represents 8.4% (15/179) of all GC cases, with a significant differential distribution among histological types (P < 0.001): 100% (3/3) of medullary carcinomas, 100% (1/1) of adenosquamous carcinoma, 8.7% (8/92) of tubular adenocarcinomas, 8.0% (2/25) of mixed carcinomas and 2% (1/51) in poorly cohesive carcinomas. The analysis revealed a higher predominance of EBVaGC in the upper third and middle (cardia, fundus and body) of the stomach (P = 0.041), a significant lower number of regional lymph nodes invasion (P = 0.025) and a tendency for better prognosis (P = 0.222). EBV latent protein expression revealed that all EBVaGC cases were LMP1-negative, nevertheless 6 cases (40%) expressed LPM2A, which reveals that these cases show a distinct EBV-Latency profile (latency II-like).

EBVaGC represents 8.4% of all GC in the North Region of Portugal. The EBV-infected patients have specific clinic-pathological features that should be further explored to develop new strategies of management and treatment.

Gastric cancer (GC) is the fifth most common malignancy worldwide with nearly one million new cases estimated in 2012 (952000 cases) and despite the significant reduction of its incidence, GC still remains the third leading cause of death by cancer[1,2]. Portugal is one of the European countries with the higher GC incidence and mortality rates, where GC represents the fifth most common cancer with about 3000 new cases per year and with 1387 deaths in men and 898 in women[1,2].

Epstein-Barr virus (EBV) is one of the most common viruses and has been associated with several malignancies including Burkitt’s Lymphoma, Hodgkin Lymphoma, post-transplant lymphoproliferative disease, a subset of T/NK cell lymphomas and nasopharyngeal carcinoma[3]. Recently, EBV has been associated with the development of GC, with several reports pointing that E BV-associated GC (EBVaGC) might account for nearly 10% of all GC cases[4,5]. The mechanisms of EBVaGC carcinogenesis are not well understood although some authors have suggested that EBV infection can represent a late event after Helicobacter pylori (H. pylori) infection[6,7].

The presence of EBV in patients with gastric carcinoma was first reported in gastric lymphoepithelioma-like carcinomas (also classified as medullary carcinomas, gastric carcinomas with lymphoid stroma or uncommon variant), nevertheless it has been detected in different histopathological subtypes of gastric carcinoma[8,9]. The majority of GC are adenocarcinomas and are classified according to the Lauren or World Health Organization (WHO) classification system[10]. These classification systems have a reduced clinical utility and recent studies suggested a new classification based on molecular features of tumours, where EBVaGC arises as a new subtype of gastric cancer[11,12].

Despite the high incidence of both GC and EBV infection prevalence in the Portuguese population, there are no studies reporting the prevalence of EBVaGC in Portugal[2,13]. We aimed to determine the prevalence and characteristics of EBVaGC and analyse the profile of EBV latent proteins expressed in gastric tumours from the North Region of Portugal.

Gastric carcinoma is a serious public health problem worldwide with high rates of mortality and Portugal has the highest gastric cancer mortality rates in Western European countries[2,15]. The impact of classification systems in GC clinical decisions has been reduced and more recently two studies suggested a new classification based on molecular features of gastric tumours[11,12]. In this new classification arises four new subtypes of gastric cancer: Epstein-Barr Virus positive tumours (EBVaGC); microsatellite unstable tumours; genomically stable tumours; and tumours with chromosomal instability (CIN) [11,12].

Over the past 30 years, EBV infection has been reported in gastric cancer tumours[16]. Currently, it is accepted that about 10% of all GC represents a specific subset of gastric carcinomas that are associated with EBV carcinogenesis[17]. The gold-standard method for the detection of EBV in tissues is EBER-ISH and the decision to accept one case as EBV-associated should be taken considering the presence of EBERs expression in tumour cells and its absence in the normal surrounding tissue. In our study we found that 8.4% of GC in our population are EBV-associated, based on the EBERs expression in > 90 of tumour cells and its absence in non-malignant or stromal cells. To the best of our knowledge, this is the first study reporting the prevalence of EBVaGC in the North Region of Portugal. The prevalence found in our population is in line with those reported in other studies, which have reported a prevalence ranging 2%-20%[5,8,18]. Similar data was found in other countries from Europe, including Netherlands (7.8%)[18], and Denmark (7.6%)[19] and also from Asiatic countries, such as South Korea (7.8%)[20] and Japan (8%)[21]. Geographic differences have been discussed as a possible reason for the variation of EBVaGC prevalence[8,22] and some studies suggest that EBVaGC prevalence might be inversely correlated with the background incidence of GC[4]. The fact that Portugal has a higher incidence of GC when compared to other European countries seems to indicate that other factors apart from geography may be influencing the distribution of EBVaGC. Indeed, data from different meta-analyses have failed to show this association[8,22].

We found that EBVaGC was more frequent in males than in females, which is in line with several studies and suggests an association with others factors, such as life style or hormonal risks[23-25]. In contrast with other reports, we have found no association between EBV status and patients’ age[8,26]. Regarding the tumour location, our study revealed a higher predominance of EBVaGC in upper third and middle of stomach and considering that H. pylori preferentially colonises the antral region, these results suggest a possible antagonism of EBV and H. pylori in gastric mucosa[27-29]. In fact, Minoura-Etoh et al[30] described in an in vitro study that reactive products from H. pylori seem to induce EBV reactivation from latently in infected gastric epithelial cells, which would avoid EBV transformation of gastric cells in the same areas of H. pylori colonization. Despite this association remains unclear, two recent studies have suggested that H. pylori may contribute for EBV-associated gastric carcinogenesis by recruiting to gastric tissue the inflammatory cells that are already infected by EBV in the majority of adults[7,31]. This mechanism is not well understood and further studies should be made to establish if the recruitment of EBV-infected lymphoid cells might be the explanation for the infection and subsequent transformation of gastric epithelium. In our series, all EBVaGC cases had EBV only in tumour cells and not in normal tissues, nevertheless, we have found one case with EBV only in the lymphocytic infiltration, which we considered negative for EBV, and therefore more studies are required to understand if in such cases tumour cells will be later infected by EBV.

The association between EBV and gastric histological types remains controversial, indeed while one meta-analysis regarding EBV in gastric carcinoma has shown that EBVaGC was associated with diffuse histological type according to Lauren classification[5,8], two other meta-analyses did not find any association between EBV and histology subtypes[22,32]. Several studies demonstrated a strong EBV association with diffuse type while others have reported a similar prevalence between intestinal and diffuse types (classified in our study as poorly cohesive according to the WHO)[9,18,33,34]. Our study revealed that EBVaGC are more frequently associated with specific histologic subtypes. EBVaGC represent about 8% of tubular (intestinal type in the Lauren classification system) and mixed types while it was only found in 2% (n = 1/51) of poorly cohesive carcinomas (diffuse type in the Lauren classification system). These distinct features show that the differences in the GC histological distribution among different populations may impact on the prevalence of EBVaGC subtype, which should gather importance on the detection of EBV in GC. As expected, all medullary carcinomas, also known as gastric carcinoma with lymphoid stroma or lymphoepithelioma-like gastric carcinoma, were EBV positive[35]. In fact, EBV association with gastric cancer was first described in lymphoepithelioma-like gastric subtype and the literature shows that more than 80% of these cases are EBV-positive[5,22,36].

Some studies have revealed that EBVaGC is associated with poor or moderate differentiation of gastric cancer tissues[9,24]. In accordance with these reports, our study showed that 100% of cases were poorly or moderately differentiated and that these differentiation patterns seem to be associated with the histological types. This evidence may be a significant mark of EBVaGC tumours and may provide important information regarding the clinical approach to these tumours.

All these data support the evidence that recent studies have shown by considering EBVaGC as a distinct clinicopathological entity due to its individual features[11,12,18,37]. Interestingly, we found that EBVaGC had a significantly lower number of regional lymph nodes with metastasis and that none have a distant metastasis. Additionally, when analysing the overall survival of EBVaGC compared to all other cases it was observed that EBVaGC seems to have an improved survival, that despite not statistically significant may be explained by the low number of cases in our population. Several studies have reported a better overall survival of EBVaGC, especially in Asiatic populations suggesting that the explanation for this better overall survival could be due to the low frequency of lymph node involvement and distant metastasis[18,23,24,38]. In Portugal, GC remains at high mortality rates due to the late diagnosis of this disease and these patients are frequently only treated with chemo/radiotherapy. In our study we have included only patients with available tissue sample from surgical procedures and therefore we had not included patients that had received only chemo/radiotherapy, therefore our findings should be applied in patients with recommendation for total or partial gastrectomy as according to the ESMO Clinical Practice Guidelines for Diagnosis, Follow-up[39]. Moreover, others factors may be influencing patients survival and this question should be explored with further large-scale studies.

To elucidate the role of EBV in the pathogenesis of gastric carcinoma several studies have focused on the study of EBV latent proteins expression in gastric tumour cells[40-42]. As expect, in our study LMP1 was not found in any of EBVaGC cases and, in fact, literature also shows that LMP1 is generally absent in EBVaGC except for the data reported in a few studies, which have detected low albeit detectable levels of LMP1 mRNA[41-43]. The absence of LMP1 expression in EBV positive cases suggests that LMP1 may not be needed for gastric carcinogenesis or at least not for sustaining the already established malignant tumours. Regarding LMP2A, our results demonstrated that EBV LMP2A is expressed in 40% of EBVaGC cases that is in line with previous studies[41,44]. Recent studies have pointed that LMP2A seems to have an important role contributing to malignant state in epithelial cells by inducing the genome hypermethylation through up-regulation of DNMT1 and phosphorylation of STAT3 and thus contributing for the carcinogenic potential of the cells[45]. A systematic review performed by our group has shown that EBVaGC are usually associated with a distinct latency pattern, characterized by the expression of EBERs, EBNA1, LMP2A and EBV microRNAs. Furthermore, literature also shows that EBV latency I pattern, which is characterized by the expression of EBERs, EBNA 1 and EBV microRNAs[46], has been also frequently found among EBVaGC tumours[47,48]. In line with this, our study demonstrates that the majority of our cases express EBERs with absence of LMP1 and LMP2A, however, 40% of our cases also expressed LMP2A, which does not fit into the standard EBV latency patterns. In fact, literature has described this distinct pattern as latency II-like and has also reported that lytic transcripts such as BARF0 and BARF1 can also be found within EBV-associated gastric neoplasias[44,49]. Our data reinforces that the clarification of which EBV transcripts (including lytic transcripts and EBV microRNAs) are expressed in gastric tumours assumes a great importance to understand the viral carcinogenesis in gastric cancer and may contribute for the identification of new molecular targets for treatment.

In our study, EBVaGC represents 8.4% of our population with gastric cancer and it is more frequent among tubular adenocarcinomas and medullary carcinomas rather than poorly cohesive carcinomas. EBVaGCs were more common in upper and middle regions of the stomach; were correlated with a lower number of regional lymph nodes with metastasis and with no distant metastasis. The results also revealed that EBVaGC patients seem to have an improved survival and better prognosis. These features suggest that EBVaGC is a distinct subtype of gastric cancer and that the mechanisms of carcinogenesis should be further investigated for particular therapeutic targets