Background
Gastric remnant carcinoma (GRC) is defined as a carcinoma occurring in the gastric stump 5 or more years after distal gastric resection for benign disease such as gastric ulcer and duodenal ulcer or gastric malignancy [
1,
2]. The etiology and pathogenesis associated with malignant transformation in the gastric stump are poorly understood.
An oncogenic association between gastric carcinoma (GC) and Epstein–Barr virus (EBV) is well known [
3]. EBV-associated-gastric carcinoma (EBVaGC) prevalence varied from 2 to 18% in different countries [
4]. Lee et al. have reported a meta-analysis indicating that the clinicopathological and molecular characteristics of EBVaGC were quite different from those of conventional gastric adenocarcinoma [
5]. As early as 1994, Yamamoto et al. have published the relationship between GRC and EBV, which revealed that the EBV may play an important role in the carcinogenesis of GRC [
6].
EBV is a gamma herpesvirus, which is a ubiquitous cause of infection in humans worldwide. Evidence of prior infection is present in adults throughout the world, with over 90% showing a serologic response [
7]. In addition to GC, EBV is also closely related to Burkitt’s lymphoma (BL), Hodgkin lymphoma (HD) and nasopharyngeal carcinoma (NPC) [
8‐
10]. The exact role of EBV in the pathogenesis of EBV-associated-gastric remnant carcinoma (EBVaGRC) remains to be determined. Previous study has reported that different genotype EBV expressed in GRC, which may play a key role in the pathogenesis of EBVaGRC [
11]. However, there were still some controversies.
In this study, we aimed to explore whether there was a higher EBV infection rate in GRC compared to conventional GC (CGC). The exploration of related clinicopathologic factors such as Lauren classification between EBVaGRC and EBV-negative-gastric remnant carcinoma (EBVnGRC) were also one of the purposes in the study. In addition, we also analyzed EBV genome polymorphisms in EBVaGRC in order to reveal some relevance.
Materials and methods
Data selection and data abstraction
We searched articles published until July 15, 2020, in the PubMed, MEDLINE and Embase databases, and also searched Chinese database named CNKI. In order to cover as many articles as possible, following terms were used to search: ((residual gastric cancer) OR (residual gastric carcinoma) OR (gastric remnant carcinoma) OR (gastric remnant cancer) OR (gastric stump cancer) OR (gastric stump carcinoma)) AND ((Epstein–Barr virus) OR EBV OR (EB virus) OR EBNA1 OR EBNA2 OR EBNA3C OR (30 bp deletion in LMP1) OR BamHI-F OR BamHI-W1/I1). EBNA1, EBNA2, EBNA3C, 30 bp deletion in LMP1, BamHI-F and BamHI-W1/I1 are all EBV genotypes. After the primary election, the inclusion and exclusion criteria were used for the next screening. Articles would be included if they met all the following criteria: (1) Published as full articles; (2) Related data of GRC and EBV can be calculated; (3) The initial malignant tumor was cured before GRC; (4) EBV positivity was identified by in situ hybridization; (5) No other coexisting malignant tumors. Articles which did not meet the above criteria or duplicate publication were excluded.
Main study characteristics for analysis such as EBVaGRC data, EBV genotypes data were extracted to Microsoft Excel (2019 edition; Microsoft, Redmond, WA, USA) for effective organization. In addition, clinical features such as reconstruction style, initial diagnosis, location of GRC, Lauren classification, sex of patients, lymphocytic infiltration between EBVaGRC and EBVnGRC were also abstracted to assess risk factors.
Studies were independently screened by title and abstract by CL and XY. Both authors subsequently performed full articles. If there was a disagreement, XYW would evaluate the disagreement again and form a final result after the trade. Risk of bias and Quality assessment were assessed using elements of the STROBE checklist for studies included [
12]. Once an overall gene effect was confirmed, the genetic model-free approach was used to estimate the genetic effects and mode of inheritance [
13]. In addition, the work was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [
14].
Statistical analysis
The odds ratio (OR) combined 95% confidence intervals (CIs) was used to describe dichotomous variable, which contained EBVaGRC data, EBVnGRC data, clinical features and genome polymorphisms data. For individual studies of genome polymorphisms with no events in one or both groups, a continuity correction of 0.5 would be added to each cell for each effect measure, as implemented in Review Manager 4.2 [
15], and OR was used to describe it. Statistical heterogeneity was assessed by the I
2 statistic and Cochran’s Q-test. Then pooled estimates were obtained using the fixed-model (Mantel and Haenszel) method (if I
2 ≤ 50%, P > 0.1) or random-model (M-H heterology) method (if I
2 > 50%, P ≤ 0.1) [
16]. What we need to explain here was that random-model method was more widely used. Regardless of heterogeneity of pooled estimates, random-model method can be selected. When heterogeneity was high, random-model method was the only choice. While heterogeneity was low, both random-model and fixed-model model can be selected. In this study, fixed-model model was selected [
16]. Furthermore, we used sensitivity analysis, which was conducted by metainf application, to evaluate whether the meta-analysis results were stable and reliable. Metainf investigated the influence of a single study on the overall meta-analysis estimate. And this command showed graphically the results of an influence analysis. Publication bias was estimated by Begg’s test (P > 0.05 suggests no publication bias). All analyses were carried out through the application of STATA 15 (StataCorp., College Station, TX, USA).
Discussion
In this study, we firstly reported that EBV infection rate in the GRC was higher than CGC. EBVaGRC had higher rate of Billroth-II, carcinoma in anastomotic site and diffuse type (Lauren classification) compared to EBVnGRC, but it is not associated with sex, initial diagnosis and lymphocytic infiltration. In addition, EBV genome polymorphisms analysis showed that expression of “V-val” subtype of EBNA1 and “C” subtype of BamHI-W1/I1 were higher in EBVaGRC than control. We comprehensively analyzed EBV infection and genome polymorphisms on GRC through systematic review and meta-analysis for the first time.
Up until now, the role of EBV in carcinogenesis of the gastric remnant or stomach is not completely understood. It has been suggested that EBVaGC have distinct molecular characteristics, as well as clinicopathological characteristics [
5]. The distinguish in EBVaGRC was also found in this study. The loss of
p16 protein [
28], or over-expression of bcl-2 [
29] was detected in EBV-infected tumors or EBVaGC, which may be related to the pathogenesis of GRC. In addition, Imai et al. revealed that EBV is uniformly present in all carcinoma cells of GRC, but nearly negative in normal epithelium or stroma [
3]. The findings indicated that EBV infection may influence carcinogenesis, perhaps triggering a clonal expansion of infected cells [
3].
In the study, three clinicopathological factors were observed related to EBVaGRC. EBVaGRC was more likely to occur in anastomotic sites and after Billroth-II reconstruction. The changes of anatomical circumstances may be the main factor. The reflux of bile and pancreatic juice is more common in Billroth-II reconstruction, which can directly change the physiological environment and pH-value of the gastric remnant. These could act as a cofactor mediating EBV infection of the epithelial cells or facilitate EBV entering the mucosa epithelia, for instance by inducing fusion of EBV carrying B cells and epithelial cells. In addition, more diffuse type of GRC can be found with EBV infection. This type of cancer usually presents shorter duration and worse prognosis compared with the intestinal type [
30]. Therefore, EBV infection may aggravate the severity of the cancer. Although some articles have found a correlation between sex, initial diagnosis, lymphocytic infiltration and EBVaGRC, the results of this study revealed no correlation by data Integration.
In addition, we paid attention to survival rate of GRC, but no articles reported it. Liu et al. reported that EBV had a favorable impact on GC patient’s survival, especially in an Asian population [
31]. In GRC, whether EBV can influence the survival rate, we need more studies to support evidence.
We also found “V-val” subtype of EBNA1 and “C” subtype of BamHI-W1/I1 were higher expressed in EBVaGRC. These two subtypes may be closely related to oncogenesis and be as potential disease predictor. It is known that EBNA1 can induce genomic instability and reactive oxygen species (ROS)-mediated DNA damage response [
32], which may act as a potential oncogene [
33]. EBNA1 contains two prototypes (P-ala and P-thr) and three variants (V-pro, V-leu and V-val) [
34]. It was reported that in EBV-associated neoplasms, “V-val” was predominant in individuals from Asian regions [
35], while NPC, EBVaGC and nasal NK/T cell lymphoma are more prevalent in this area [
36]. This phenomenon may suggest that “V-val” subtype might be more aggressive than other subtypes. The possible reason was that most of amino acid alterations of “V-val” subtype (codons 439, 487, 499, 502, 524, 528 and 533) were located in the function domains of EBNA1, which contained DNA binding domain (aa 459,487), dimerization domain (aa 501,532) and transactivation domain (aa 450,641) [
26]. Thus “V-val” subtype was more likely to cause changes of EBNA1’s function. In addition, previous studies revealed that “C” subtype prevails in NPC patients from Southern China where incidence of NPC is high. It was suggested that the presence of “C” subtype is also associated with oncogenesis. Therefore, “V-val” subtype and “C” subtype can be tested to predict high risk of EBVaGRC.
Although we have achieved some encouraging results, this article still had some defects. First, included articles had limitations due to article source and type. Articles were mainly from the Asian region and had no global representation. In addition, we can only clarify the relationship between EBV and GRC, but there was not enough evidence of causality. Enough randomized controlled trials (RCTs) were needed to clarify the causality. Second, the credibility of results was weakened because of insufficient data, especially analysis of EBV genome polymorphisms. Third, since the ORs of the included studies were crude ORs which were calculated by authors, there was of some heterogeneity. This part of heterogeneity cannot be ruled out, because we cannot obtain data directly from authors. And next, in this study, it was not certain that GRC originated from a recurrence of the previous GC or a “de novo” EBV-related GC. Homology of cancer is also crucial for patient treatment and follow-up. Finally, we didn’t know whether EBV eradication can change the process of GRC. If EBV eradication weakened oncogenesis of gastric remnant, it will provide strong evidence that EBV is an important causative agent to GRC. Such articles were really needed.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.