Epstein–Barr virus (EBV) infection is common worldwide. Although more than 90% of Japanese are infected with EBV early in life (Mizuno, 1990), most remain asymptomatic throughout their lifetimes (Rickinson and Kieff, 1996), which seems to belie the established role of EBV in the aetiology of Burkitt's lymphoma, nasopharyngeal carcinoma and some forms of Hodgkin's disease (Anon., 1997).
Recently, EBV involvement has been noted in a portion of gastric adenocarcinoma cases. Using in situ hybridization assay, the uniform presence of EBV-encoded small RNA (EBER) has been demonstrated in the carcinoma cells of gastric tumours but not in surrounding normal cells (Shibata and Weiss, 1992). Using Southern blot analysis, the episomal monoclonality of EBV has been demonstrated for the fused termini of EBV DNA (Imai et al., 1994). These findings suggest that EBV is involved in the development of gastric carcinomas having EBER in the carcinoma cells.
Tokunaga et al. (1993) investigated 1795 gastric carcinoma cases and found that among Japanese patients approximately 7% of gastric carcinoma cases are EBER-positive. Elsewhere in the world, the prevalence of EBV-positive cases is 16% in North America (Shibata and Weiss, 1992), 8.7% in Russia (Galetsky et al., 1997), and 8.2% in Mexico (Herrera-Goepfert et al., 1999). It remains unknown why only a portion of gastric carcinoma cases is EBV-positive and why prevalence of EBV-associated gastric carcinoma differs from country to country.
A possible explanation is specific immune susceptibility in certain individuals. Levine et al. (1995) reported that specific altered antibody reactions to EBV occur a few years before the development of EBV-associated gastric carcinoma. At least 3 years before cancer diagnosis, Levine et al. observed significantly elevated levels of antibodies against viral-capsid antigen in the serum of patients diagnosed with EBV-positive gastric carcinoma. Our previous study of a Japanese population also suggests in vivo EBV reactivation (Shinkura et al., 2000). It is quite likely that cellular immune reactions to EBV are restricted by the major histocompatibility complex, as has been suggested in other EBV-associated malignancies (Chan et al., 1983;Masucci et al., 1987;Lu et al., 1990;Andersson et al., 1991).
To assess whether the major histocompatibility complex contributes to the development of EBV-associated gastric carcinoma, we examined the frequencies of human leukocyte antigen (HLA) class I (A and B loci) and class II (DR and DQ loci) among Japanese patients diagnosed with gastric carcinoma. We then compared HLA frequencies in EBV-positive and EBV-negative gastric carcinoma cases.
Materials and methods
The gastric carcinoma cases were ascertained at three major hospitals and one clinical laboratory in Kagoshima Prefecture, Japan, from June 1993 to March 1997. Tissue samples from 2966 gastric carcinoma cases were sent to Kagoshima City Hospital for laboratory assays. After reconfirming pathological diagnosis on the basis of the guidelines of the Japanese Research Society for Gastric Cancer, we examined, using in situ hybridization assay, all gastric carcinoma cases for the presence of EBER in the carcinoma cells. A case was considered EBV-positive on the basis of a positive signal in the carcinoma cells under microscopy and EBV-negative on the basis of a negative signal as described previously (Tokunaga et al., 1993). One hundred sixty cases (5.4%) were EBV-positive. Of these, we did not do HLA typing on 50 cases (31%) because the patients had not donated blood before being discharged from the hospital. One hundred fifty-five EBV-negative cases were randomly selected for HLA analysis from a list of EBV-negative gastric carcinoma cases diagnosed at the same three hospitals during the same study period. Geographic distributions of EBV-positive and EBV-negative carcinoma cases differed only slightly; the proportion of EBV-positive patients living outside Kagoshima City was 51% and 44% for EBV-negative cases. Demographic information for all subjects was obtained from pathological examination reports or medical records.
Laboratory procedures for in situ hybridization and HLA typing
Paraffin-blocked sections from biopsy samples were examined by in situ hybridization assay using a digoxigenin-labelled EBER-1 oligonucleotide probe, as described previously (Chang et al., 1992). Each paraffin section was deparaffinized, rehydrated, predigested with pronase, and hybridized with a concentration of 25 ng digoxigenin-labelled probe at 37°C overnight. After washing with 0.5× standard sodium citrate (3 mol/l NaCl, 0.3 mol/l sodium citrate), hybridization was detected using an anti-digoxigenin antibody–alkaline phosphatase conjugate (Boehringer Mannheim, Germany). Lymph node tissue from a patient with infectious mononucleosis and a sense probe for EBER-1 were used for positive and negative controls, respectively.
From each gastric carcinoma patient, a 10-ml heparinized peripheral blood sample was collected during hospitalization. Within 24 hours after drawing blood, serological HLA typing was performed using peripheral lymphocytes from both EBV-positive and EBV-negative patients. The peripheral lymphocytes were separated by Ficoll-Hypaque density gradient centrifugation. We selected T and B cells by positive selection using Dynabeads (Dynal, Norway). The fresh T and B cells were examined at the HLA class I (HLA-A and B) and class II (HLA-DR and DQ) loci, respectively. We applied the standard US National Institutes of Health microlymphocytotoxicity test (Terasaki et al., 1978) with commercially available HLA trays (Veritus, USA).
To assess the association of a specific HLA type and EBV-positive gastric carcinoma, the frequency of each antigen was compared using the chi-squared test (likelihood ratio test) and Fisher's exact test (Colton, 1974). Maximum likelihood estimates of odds ratios (OR) and corresponding 95% confidence intervals (CI) were calculated. All P -values presented were two-tailed, and statistic significance was defined as P < 0.05. Furthermore, we calculated adjusted P -values using Hommel's procedure to reduce the possibility that significant associations were due to chance caused by multiple comparisons (Wright, 1992).
Characteristics of the subjects
shows sex and age distributions of the EBV-positive and EBV-negative carcinoma cases. The odds of EBV positivity in males was more than twice as high as in females (OR 2.4; 95% CI 1.35–4.27). Among EBV-positive and EBV-negative males, the mean age did not differ significantly (P = 0.43). Among females, however, the EBV-positive cases were nearly 5 years older than the EBV-negative patients (P = 0.10).
HLA frequencies in the EBV-positive and EBV-negative cases
We were able to analyse at least one allele of the HLA-A, HLA-B, HLA-DR and HLA-DQ loci in 264, 263, 255 and 257 subjects, respectively. The failure of typing class II alleles was mainly due to an insufficient quantity of B cells to recognize any specific reaction with class II antisera.
The antigen frequencies of HLA-A, HLA-B, HLA-DR and HLA-DQ among the gastric carcinoma cases are shown in Table 2
, along with their univariate ORs and 95% CIs. HLA-A frequencies in EBV-positive and EBV-negative carcinoma patients did not differ significantly (Table 2). As shown in Table 3, B59 frequency in EBV-positive carcinoma cases was higher than in EBV-negative carcinoma cases (OR 3.06; 95% CI 1.02–9.23), whereas no significant difference was observed for other B antigens. However, after adjusting the P -value for multiple comparisons, B59 frequencies in EBV-positive and EBV-negative carcinoma cases did not differ significantly (P = 0.88). For analysis at the HLA-DR locus, the frequency of DR9 in the EBV-positive carcinoma cases was higher than in EBV-negative carcinoma cases (OR 1.93; 95% CI 1.11–3.37), whereas DR11 frequency in the EBV-positive carcinoma cases was lower (OR 0.10; 95% CI 0.01–0.79) (Table 4). DR11 antigen was observed in 13 EBV-negative carcinoma cases but in only one EBV-positive carcinoma case. Among the five DQ antigens, DQ3 frequency in the EBV-positive carcinoma cases was significantly higher than in EBV-negative carcinoma cases (OR 1.94; 95% CI 1.16–3.24) (Table 5). Even after adjustment for multiple comparisons, the DR11 frequency in the EBV-positive carcinoma cases was still significantly lower (P = 0.04), and an elevated DQ3 frequency in the EBV-positive carcinoma cases was also marginally significant (P = 0.05).
We investigated the association between specific HLA types and the risk of EBV-positive gastric carcinoma in Japan using a case–series design (Begg and Zhang, 1994). A deficiency of DR11 in the EBV-positive carcinoma cases was statistically significant (P = 0.04) after adjusting the P -value for multiple comparisons, and elevation of DQ3 frequency was marginally significant (P = 0.05). Among EBV-negative carcinoma cases reported in a large-scale, multicentre study conducted in Japan (Imanishi et al., 1992), antigen frequencies did not differ significantly from the antigen frequencies noted in our EBV-negative gastric carcinoma cases except for B54, DR1, DR13, DR14 and DQ4 (Table 6
). However, after adjusting for multiple comparisons, only B54 and DR14 frequencies in the EBV-negative carcinoma cases were higher than those reported in the multicentre study. Lacking any published report of higher B54 and DR14 frequencies among gastric carcinoma cases, we speculate that such high frequencies of B54 and DR14 could be genetic markers of a source population.
DR9 is known to be associated with one of the DQ3 alleles at the DNA level, and both antigens showed positive association with EBV-positive carcinoma cases in our study. On the other hand, DR11 is also known to be strongly associated with DQ7, and the two antigens showed a negative association with EBV-positive carcinoma cases; an association with DQ7 was not statistically significant. Thus, these results were consistent with the disequilibrium between DR and DQ antigens.
DQ3 is known to have two alleles at the DNA level, DQB1*0302 and DQB1*0303, and each allele is mainly associated with DR4 and DR9, respectively. Since both DR4 and DR9 showed positive associations with EBV-positive carcinoma cases, we could not deny the possibility that an increased DQ3 frequency was affected by the high frequencies of both alleles.
This is the first study reporting significant differences between specific HLA frequencies in EBV-positive and EBV-negative gastric carcinoma cases. Qiu et al. (1997) reported that the frequency of HLA-O2-8 in 16 EBV-associated gastric carcinoma cases (60%) was higher than the published data for local populations in Shanghai and Japan (30.3% and 23.7%, respectively), but the sample size was small. They observed no correlation between HLA-O2-8 and the severity of lymphoid cell infiltration and concluded that HLA-O2-8 was not associated with gastric carcinoma carrying the EBV genome. In our study, the HLA-O2-8 frequencies among the EBV-positive and EBV-negative carcinoma cases were similar, which is consistent with the findings of Qiu and co-workers.
Khanna et al. (1995) succeeded in isolating class II (DR1)-restricted CTLs against EBNO1-8. They also proved that EBNO1-8 with the Gly-Ala repeated region was recognized by DR1-restricted CTLs through an HLA-DM-independent pathway. However, we found that the DR1 frequency in EBV-positive carcinoma cases was elevated, though the association was not significant.
The aetiological relationship between EBV and gastric carcinoma remains unclear. Although EBV infection occurs worldwide, only a portion of gastric carcinoma cases is EBV-positive, and the prevalence of EBV-positive gastric carcinoma differs geographically. If EBV plays a role in the development of EBV-positive gastric carcinoma, variable individual susceptibility to EBV reactivation or to the development of EBV-positive gastric carcinoma may explain the prevalence discrepancy. Our study suggests an association between EBV-positive gastric carcinoma and DR11, and possibly with DQ3. These results raise the following possibilities: DR11-restricted CTLs or genetic markers indicated by DR11, and possibly DQ3, are related to EBV-associated gastric carcinoma. However, we need to confirm the association of DR11 and DQ3 with EBV-positive gastric carcinoma at the DNA level since both antigens have more than two alleles. In addition, further studies in other large populations are needed because our findings in this study are exploratory and the number of the subjects was relatively low.
This work was performed at Department of Public Health, Faculty of Medicine, Kagoshima University and Department of Pathology, Kagoshima City Hospital. We thank Professors Suminori Akiba and Nancy Mueller for their scientific advice on this manuscript, Dr James Warram for helpful advice on the statistical analysis, Drs Shinji Yashiki and Fumio Ohtani for advice on the HLA typing technique, Drs Kuniaki Aridome, Tetsuhiko Itoh and Junkou Arikawa for their support. We are particularly grateful to Ms Katsue Yamashita, Ms Hiroko Nagasato, Ms Midori Tsuruta, Ms Satoko Hori, Ms Michiko Hata, Ms Mihoko Teshima, Ms Kiyoko Chochi, Mr Seihou Oyamada, Mr Hitomi Fujisaki and Mr Kiyohiro Kubo for their technical support. This work was supported by Grants-in-Aid for Scientific Research of the Ministry of Education, Science, Sports and Culture of Japan (09253103), and the Ministry of Health and Welfare of Japan (H9-gann-014), and the Kodama Foundation for Research of Medical Science.
1. Andersson M , Stam NJ , Klein G , Ploegh HL , Masucci MG ( 1991 ). Aberrant expression of HLA
class-I antigens in Burkitt lymphoma cells . Int J Cancer 47 : 544 – 550 .
2. Anon. ( 1997 ). Epstein-Barr virus: summary of data reported and evaluation . IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 70 IARC Press , Lyon , pp. 255 – 262 .
3. Begg CB , Zhang Z ( 1994 ). Statistical analysis of molecular epidemiology studies employing case-series . Cancer Epidemiol Biomarkers Prev 3 : 173 – 175 .
4. Chan SH , Day NE , Kunaratnam N , Chia KB , Simons MJ ( 1983 ). HLA
and nasopharyngeal carcinoma in Chinese. A further study . Int J Cancer 32 : 171 – 176 .
5. Chang KL , Chen YY , Shibata D , Weiss LM ( 1992 ). Description of and in situ
hybridization methodology for detection of Epstein-Barr virus RNA in paraffin-embedded tissues, with a survey of normal and neoplastic tissues . Diagn Mol Pathol 1 : 246 – 255 .
6. Colton T ( 1974 ). Statistics in Medicine Little, Brown and Company , Boston .
7. Galetsky SA , Tsvetnov VV , Land CE et al. ( 1997 ). Epstein-Barr-virus-associated gastric cancer in Russia . Int J Cancer 73 : 786 – 789 .
8. Herrera-Goepfert R , Reyes E , Hernandez-Avila M et al. ( 1999 ). Epstein-Barr virus-associated gastric carcinoma in Mexico: analysis of 135 consecutive gastrectomies in two hospitals . Mod Pathol 12 : 873 – 878 .
9. Imai S , Koizumi S , Sugiura M et al. ( 1994 ). Gastric carcinoma: Monoclonal epithelial malignant cells expressing Epstein-Barr virus latent infection protein . Proc Natl Acad Sci USA 91 : 9131 – 9135 .
10. Imanishi T, Akaza T, Kimura A, Tokunaga K, Gojobori T (1992). Allele and haplotype frequencies for HLA
and complement loci in various ethnic groups. In HLA 1991, Proceedings of the Eleventh International Histocompatibility Workshop and Conference, Vol 1
(eds Tsuji K, Aizawa M, Sasazuki T); Oxford University Press, Oxford, pp. 1065–74.
11. Khanna R , Burrows SR , Stegerwald-Mullen PM et al. ( 1995 ). Isolation of cytotoxic lymphocytes from healthy seropositive individuals specific for peptide epitopes from Epstein-Barr virus nuclear antigen 1: implications for viral persistence and tumor surveillance . Virology 214 : 633 – 637 .
12. Levine PH , Stemmermann G , Lennettem ET et al. ( 1995 ). Elevated antibody titers to Epstein-Barr virus prior to the diagnosis of Epstein-Barr-virus-associated gastric adenocarcinoma . Int J Cancer 60 : 642 – 644 .
13. Lu S , Day NE , Degos L et al. ( 1990 ). Linkage of a nasopharyngeal carcinoma susceptibility locus to the HLA
region . Nature 346 : 470 – 471 .
14. Masucci MG , Torsteinsdottir S , Colombani J et al. ( 1987 ). Down-regulation of class I HLA
antigens and of the Epstein-Barr virus-encoded latent membrane protein in Burkitt lymphoma lines . Proc Natl Acad Sci USA 84 : 4567 – 4571 .
15. Mizuno F ( 1990 ). EB virus . Nippon Rinsho 48 ( Suppl ): 272 – 276 .
16. Qiu K , Tomita Y , Hashimoto M et al. ( 1997 ). Epstein-Barr virus in gastric carcinoma in Suzhou, China and Osaka, Japan: association with clinico-pathologic factors and HLA
-subtype . Int J Cancer 71 : 155 – 158 .
17. Rickinson AB , Kieff E ( 1996 ). Epstein-Barr Virus . In Fields Virology, Vol 2, 3rd edn (eds Fields BN , Knipe DM , Howley PM ); et al. ( 1996 ). Lippincott-Raven Publishers , Philadelphia , pp. 2397 – 2446 .
18. Shibata D , Weiss LM ( 1992 ). Epstein–Barr virus-associated gastric adenocarcinoma . Am J Pathol 140 : 769 – 774 .
19. Shinkura R , Yamamoto N , Koriyama C et al. ( 2000 ). Epstein-Barr virus-specific antibodies in Epstein-Barr virus-positive and -negative gastric carcinoma cases in Japan . J Med Virol 60 : 411 – 416 .
20. Terasaki PI , Bernoca D , Park MS , Otzurk G , Iwaki Y ( 1978 ). Microdroplet testing for HLA
-A, -B, -C, and -D antigens . Am J Clin Pathol 69 : 103 – 120 .
21. Tokunaga M , Uemura Y , Tokudome T et al. ( 1993 ). Epstein-Barr virus related gastric cancer in Japan: a molecular patho-epidemiological study . Acta Pathol Jpn 43 : 574 – 581 .
22. Wright SP ( 1992 ). Adjusted P-values for simultaneous inference . Biometrics 48 : 1005 – 1013 .
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
EBV; HLA; stomach neoplasms.