1 School of Health Sciences, University of Wolverhampton, Wolverhampton; 2 Cancer Research UK Institute for Cancer Studies, University of Birmingham, Birmingham; 3 Department of Primary Care and Population Sciences, University College London, Royal Free Campus, London; 4 Department of Pathology, Division of Cancer Studies, University of Birmingham, Birmingham, UK
Received 28 February 2002; revised 4 September 2002; accepted 17 September 2002
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Conflicting data on the effect of the EpsteinBarr virus (EBV) on outcome in Hodgkins disease (HD) might be due to the heterogeneous nature of this disease. In this study we have investigated whether the effect of EBV status on outcome is different between aetiologically defined age groups (1534, 3544, 45+ years) and also between males and females.
Patients and methods:
Paraffin-embedded sections from 273 patients with advanced HD from two related clinical trials were analysed for the presence of EBV using in situ hybridisation.
Results:
EBV was detected in 78 (29%) of cases. For all patients, after a median follow-up of 5 years, there were no significant differences in survival by EBV status although there was a trend towards longer failure-free survival times for EBV-positive patients. Multivariate analyses suggested that EBV and sex, when in combination, were prognostic factors for failure-free survival (P = 0.06 for both). For subgroups, the effect of EBV on failure-free survival was significant for males and 1534 years age group (P = 0.05 and P = 0.03, respectively).
Conclusion:
This study suggests that with a median follow-up of 5 years, EBV status does not affect survival but being EBV-positive may be beneficial in terms of failure-free survival, particularly for males and younger adults.
Key words: EpsteinBarr virus, Hodgkins disease, subgroup analysis, survival
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Initial investigations of the association of EBV with survival in HD in the early 1990s relied on polymerase chain reaction-based detection of EBV [8] and would, therefore, be expected to detect EBV in non-malignant cells as well as in the H-RS cells. However, data from studies that have employed EBER in situ hybridisation or immunohistochemistry for detection of LMP1, both of which localise EBV detection to the malignant cells, are conflicting. Some studies of predominantly adult patients have shown significantly improved failure-free intervals or survival for patients with EBV-positive disease [911] and this was also seen in a small study of childhood cases [12]. However, Vestlev et al. [13] found that patients with EBV-positive HD had a slightly increased risk of progression with a relative risk of 1.15, although there was no significant difference in 5-year progression-free survival. A detrimental effect, albeit non-significant, of EBV positivity on overall survival and disease-free rate was also observed in a population-based study [14]. Recently, what appears to be the first finding of a significantly adverse effect of EBV positivity on overall survival was found in a population-based series of women [15]. Furthermore, a study of comparable size to those that have shown improved survival for EBV-positive cases of HD has indicated similar survival patterns for EBV-positive and EBV-negative HD patients [16].
Of those studies that have investigated survival differences by age, significant age-dependent differences have been observed by some [15, 17] but not others [14]. For women, no survival differences between EBV-positive and EBV-negative disease were observed in those aged 1944 years, but survival was significantly poorer for those with EBV-positive HD in the age group 4579 years [15]. Another study of adult HD revealed significantly longer disease-free survival for EBV-positive patients aged 30 years and below [17], although when all patients were included no differences in overall or disease-free survival based on EBV status were observed.
Conflicting data on the effect of EBV on outcome in HD might be due to differences in the age or sex distribution of the patients included in each study. Originally MacMahon [18] suggested that HD comprised three aetiologically different diseases based on the age groups 014 years, 1534 years and 50+ years. This has been modified recently by Armstrong et al. [19] who described differences in the EBV association and distribution of histological subtypes between age groups. Their three disease model proposes that HD of childhood (014 years) and older adults (45+ years) is commonly EBV-associated and of mixed cellularity (MC) subtype, whereas disease in young adults (1534 years) is generally not EBV-associated, and is usually of nodular sclerosis (NS) subtype. The intermediate age group (3544 years) is considered to be a mixture of the young adult and older adult disease entities.
Further investigation of the association of EBV with survival in each of these different age groups is therefore required. In this study, we have explored potential age- and sex-related differences in the effect of EBV on survival and failure-free survival by studying a large cohort of adult patients with advanced HD who were recruited to one of two related clinical trials.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Patient data (date of birth, date of entry to trial, sex, residential postcode at entry to trial, histological subtype, disease stage, presence or absence of B symptoms, history of previous treatment, treatment during trial, response to treatment, additional treatment, date of death, date of relapse, date of last follow-up) were extracted from trial records. Response to treatment was recorded as complete remission if all disease disappeared for a minimum of 1 month, or partial remission if at least 50% of known disease disappeared [20, 21].
For quantifying material deprivation, the Townsend score [22], a 1991 census-based deprivation measure (standardised for England and Wales), was employed. This score is derived from the enumeration district (the smallest unit of population for which data were available) in which the patient resided at entry to the trial, as determined from the patients postcode. The Townsend score is a widely accepted indicator of material deprivation [23].
Histological review
Pathology was reviewed at entry to each trial [20, 21] and the Rye classification system [24] was used for histological subtyping. The subtypes were NS, MC, lymphocyte depletion (LD) and lymphocyte predominance (LP). Tissue blocks were available for EBV analysis in 273 histologically confirmed HD cases from the 959 patients (280 in HO2001, 679 in HO3001) randomised in these two trials.
Detection of latent EBV infection
Paraffin-sections (4 µm) were prepared, deparaffinised and washed in phosphate-buffered saline pH 7.6. In situ hybridisation for the detection of EBERs was performed according to modified standard methodology [25] and was used to detect the presence of latent EBV infection in all HD samples. Positive controls for EBER in situ hybridisation were paraffin wax sections of pelleted B95.8 cells and a known EBER-positive HD case. U6 and sense control probes were included in all runs and their use has been previously described elsewhere [25]. Negative controls consisted of consecutive test sections, processed using hybridisation buffer without the addition of the probe. Tumour specimens were recorded as either EBV-positive (EBERs detected within H-RS cells) or EBV-negative (EBERs not detected in H-RS cells).
Survival and failure-free survival time
Survival was calculated as the time from entry to the trial to the date of death from any cause or date of the last known follow-up. For those patients who responded to treatment (i.e. achieved a complete or partial remission), failure-free survival was calculated as the time from entry to the trial to either the date of relapse, or in those patients with no previous documented relapse, either the date of death from any cause or date of the last known follow-up. For those patients who did not respond to treatment (i.e. did not achieve a complete or partial remission at any time), failure-free survival was recorded as zero.
Statistical methods
Differences in baseline patient and disease characteristics between EBV-positive and EBV-negative HD cases were compared using either chi-square tests with Yates continuity correction (for 2 x 2 tables) or Fishers exact test (where numbers were small). The MannWhitney U test was used to detect differences between EBV-positive and EBV-negative groups for the Townsend score. Fishers exact test was also used to compare treatment and response to treatment based on EBV status and the MannWhitney U test was used to compare the number of courses of ChlVPP and PABlOE received by each group.
Differences in survival and failure-free survival between EBV-positive and EBV-negative patients were investigated with the use of KaplanMeier curves and log-rank tests. Two-year and 5-year survival rates were compared for the two groups using a test of proportions based on KaplanMeier estimates of survival rates. Univariate analyses were repeated for subgroups of patients defined by age (1534, 3544, 45+ years) and sex. In addition, hazard ratio (HR) plots were used to compare EBV-positive against EBV-negative patients both overall and within subgroups in terms of survival and failure-free survival. The HR is a measure of the relative survival or failure-free survival experience of the EBV-positive group compared with the EBV-negative group over the whole follow-up time. HRs were pooled across subgroups within age and sex to give overall HRs stratified by that factor. HRs less than 1.0 indicate that survival or failure-free survival was superior for EBV-positive patients compared with EBV-negative. Chi-square tests for heterogeneity (or interaction) were used to assess the variability of the HRs across subgroups within age and sex. HRs, HR plots and tests of heterogeneity were produced using SAS software version 8 (SAS Institute Inc., Cary, NC, USA).
Cox proportional hazards regression analysis was used to assess the prognostic significance of a number of covariates in a multivariate setting. The covariates considered for inclusion in the Cox regression model were age (1534, 3544, 45+ years), sex, EBV, histological subtype (NS, MC, LD, LP), clinical stage (I/II, III/IV), presence of B symptoms and Townsend score (as quartiles). A forward selection procedure was used to find the combination of covariates that were or tended towards significance for survival and failure-free survival (P <0.10). Terms representing the interaction of EBV with age and sex were included to assess whether the effects of EBV differed for different age and sex groups. Cox regression analysis was carried out for all patients and also for each subgroup defined by age and sex. For subgroup analysis by age, LD was not included as a covariate due to very small numbers of cases. Survival and failure-free survival analyses were stratified by trial. Except where otherwise stated, analyses used SPSS Version 10 (SPSS Inc., Chicago, IL, USA). Differences were deemed statistically significant if two-sided P 0.05.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
EBV status and treatment
EBV-positive and EBV-negative groups were comparable in terms of the trials they came from and the treatment they received before and during the trial (Table 1). The EBV-positive group was associated with a slightly higher frequency of complete remission in response to chemotherapy, albeit non-significantly (P = 0.16, Table 1).
Univariate analyses of survival and failure-free survival for all patients
The median length of follow-up for the 210 patients still alive at last follow-up was 60 months (interquartile range 4096 months). Of these cases, four had follow-up for less than 12 months, three of these were EBV-positive. The median length of follow-up for the 60 EBV-positive patients and the 150 EBV-negative patients still alive at last follow-up was the same.
On analysis of all patients, no difference in overall survival by EBV status was identified (Table 2, diamonds in Figure 1). This was also the case when 2- and 5-year survival rates were compared. When failure-free survival was investigated, there was a trend towards longer failure-free survival in the EBV-positive group (Table 2, diamonds in Figure 2), but this was not statistically significant (P = 0.18, Table 2). Two-year failure-free survival rates were 86% for EBV-positive patients and 76% for EBV-negative patients (P = 0.07, Table 2), reflecting the trend for improved failure-free survival in EBV-positive patients. The same pattern was observed for 5-year failure-free survival rates (77% for EBV-positive patients compared with 63% for EBV-negative patients, P = 0.08).
|
|
|
|
|
Multivariate analyses of survival and failure-free survival
Cox regression analysis of all patients showed age group to be significantly prognostic for survival (1534 years, P <0.001; 3544 years, P = 0.06, Table 5), with younger age groups having a better prognosis than the 45+ year age group. LD subtype was of borderline significance in this analysis (P = 0.07). EBV was not found to be a significant factor for survival (P = 0.96). Similarly, age group and LD were significant prognostic factors for failure-free survival (1534 years, P <0.001; 3544 years, P = 0.02; LD, P = 0.03, Table 5) with sex and EBV status being of borderline significance (male, P = 0.06; EBV-positive, P = 0.06, Table 5). Sex and EBV status only tended towards significance when they were added to the model in combination rather than as single factors. For failure-free survival, the HRs show that being in the younger age groups or being EBV-positive reduces the hazard of failure, with the hazard for the younger age groups being approximately half that for the 45+ year age group, and for the EBV-positive group being a little over half that for the EBV-negative group. Also, having LD subtype or being male increases the hazard of failure with the hazard for LD being almost three times that for not having LD, and for males it is more than one and half times that for females. The interaction between EBV and age did not significantly add to the base model for failure-free survival shown in Table 5 (1534 years, P = 0.29; 3544 years, P = 0.75), neither did the interaction between EBV and sex (P = 0.54). There is, therefore, no evidence to suggest that the relative effect of EBV on failure-free survival differed for different age or sex groups.
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The research presented here is the largest study of male and female HD patients to investigate the relationship between EBV status and survival or failure-free survival. This is a more in-depth analysis of our previous work [10] on a subset of these patients, which became possible through augmentation of the case series by 100 additional patients. These patients had advanced HD and were from two related clinical trials. Although not a population-based study, EBV-positive and EBV-negative patients had received similar treatments and thus any observed differences between the groups in terms of outcome were not likely to be due to treatment.
We have shown that survival does not vary significantly with EBV status in this group of patients or in any subgroups defined by age and sex. This is consistent with some studies [16], but not with others [9, 11], none of which analysed subgroups in terms of EBV status. Age-dependent survival differences based on EBV status have been observed [14, 15]. Enblad et al. [14] found a non-significant trend for poorer survival in EBV-positive patients that was only observed in patients >60 years of age. Clarke et al. [15] showed significantly lower overall survival for women with EBV-positive HD aged 4579 years, but no such effect in those aged <45 years; we did not have sufficient female patients to investigate these findings. Multivariate analysis of all patients revealed age and possibly LD, but not EBV status, to be prognostic factors for survival. Increasing age and LD are recognised adverse factors for survival [26].
Although our data do not give any evidence of an effect of EBV on survival after a median follow-up of 5 years for patients with advanced HD, we have observed a trend towards improved failure-free survival for EBV-positive patients when compared with their EBV-negative counterparts. This trend was not statistically significant for the univariate analysis of all patients although differences in 2- and 5-year rates were of borderline significance, as was the inclusion of EBV in the multivariate analysis. Survival of HD patients has been found to decrease with increasing deprivation [27] and we have recently shown that EBV-positive HD patients were from more materially deprived areas than EBV-negative patients [28]. Therefore, improved survival or failure-free survival for patients with EBV-positive HD is the reverse of what might be expected. Some studies have demonstrated significantly improved failure-free intervals for EBV-positive patients [9], whereas others suggest adverse effects of EBV positivity on failure-free survival [13, 14].
On multivariate analysis of all patients, the effect of EBV on failure-free survival was still present even after adjusting for the effects of LD, age and sex. The inclusion of LD, age and sex in the model was not unexpected as they are well-recognised prognostic factors for relapse in HD [26]. Interestingly, EBV only approached significance when in conjunction with sex, implying that it is the combination of EBV status and sex that is an important prognostic factor, with being EBV-positive and female having the best prognosis and being EBV-negative and male the worst. The effects of sex and EBV status have approximately equal but opposite signs in the model suggesting that being EBV-positive and male is equivalent to being EBV-negative and female. This observation does not give evidence of a subgroup effect; it is the inclusion of interaction terms in the model that would provide this. In fact, the effects observed in separate analyses of each subgroup were not confirmed by the inclusion of interaction terms in the overall Cox model. This may be because there are no real subgroup effects or because the numbers of patients were too small to detect them. Univariate and multivariate analyses of males and females separately revealed that the trend for improved failure-free survival in EBV-positive patients was only significant in males, and it may be that the trend is clearest in this subgroup because it was the largest. Analysis of each age group showed that the effect of EBV was greatest in the 1534 year age group, with the HR from the HR plots showing the greatest difference from 1.0 and EBV being significant only in this age group in multivariate analysis. The significant effects seen in males and the 1534 year age group may be because these were the largest subgroups and hence interpretation requires caution.
In summary, our study suggests that, with a median follow-up of 5 years, being EBV-positive does not affect survival but may be beneficial in terms of failure-free survival. This effect was more apparent for males and younger adults. Additional research is recommended to investigate further the effect of EBV status on outcome and to confirm our findings.
![]() |
Acknowledgements |
---|
![]() |
Footnotes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Weiss LM, Movahed LA, Warnke RA, Sklar J. Detection of EpsteinBarr viral genomes in ReedSternberg cells of Hodgkins disease. N Engl J Med 1989; 320: 502506.[Abstract]
3. Wu TC, Mann RB, Charache P et al. Detection of EBV gene expression in ReedSternberg cells of Hodgkins disease. Int J Cancer 1990; 46: 801804.[ISI][Medline]
4. Niedobitek G, Kremmer E, Herbst H et al. Immunohistochemical detection of the EpsteinBarr virus-encoded latent membrane protein 2A in Hodgkins disease and infectious mononucleosis. Blood 1997; 90: 16641672.
5. Deacon EM, Pallesen G, Niedobitek G et al. EpsteinBarr virus and Hodgkins disease: transcriptional analysis of virus latency in the malignant cells. J Exp Med 1993; 177: 339349.[Abstract]
6. Murray PG, Young LS, Rowe M, Crocker J. Immunohistochemical demonstration of the EpsteinBarr virus-encoded latent membrane protein in paraffin sections of Hodgkins disease. J Pathol 1992; 166: 15.[ISI][Medline]
7. Pallesen G, Hamilton-Dutoit SJ, Rowe M, Young LS. Expression of EpsteinBarr virus latent gene products in tumour cells of Hodgkins disease. Lancet 1991; 337: 320322.[ISI][Medline]
8. Fellbaum C, Hansmann M-L, Niedermeyer H et al. Influence of EpsteinBarr virus genomes on patient survival in Hodgkins disease. Am J Clin Pathol 1992; 98: 319323.[ISI][Medline]
9. Naresh KN, Johnson J, Srinivas V et al. EpsteinBarr virus association in classical Hodgkins disease provides survival advantage to patients and correlates with higher expression of proliferation markers in ReedSternberg cells. Ann Oncol 2000; 11: 9196.[Abstract]
10. Murray PG, Billingham LJ, Hassan HT et al. Effect of EpsteinBarr virus infection on response to chemotherapy and survival in Hodgkins disease. Blood 1999; 94: 442447.
11. Morente MM, Piris MA, Abraira V et al. Adverse clinical outcome in Hodgkins disease is associated with loss of retinoblastoma protein expression, high Ki67 proliferation index, and absence of EpsteinBarr virus-latent membrane protein 1 expression. Blood 1997; 90: 24292436.
12. Engel M, Essop MF, Close P et al. Improved prognosis of EpsteinBarr virus associated childhood Hodgkins lymphoma: study of 47 South African cases. J Clin Pathol 2000; 53: 182186.
13. Vestlev PM, Pallesen G, Sandvej K et al. Prognosis of Hodgkins disease is not influenced by EpsteinBarr virus latent membrane protein. Int J Cancer 1992; 50: 670671.[ISI][Medline]
14. Enblad G, Sandvej K, Sundström C et al. EpsteinBarr virus distribution in Hodgkins disease in an unselected Swedish population. Acta Oncol 1999; 38: 425429.[CrossRef][ISI][Medline]
15. Clarke CA, Glaser SL, Dorfman RF et al. EpsteinBarr virus and survival after Hodgkin disease in a population-based series of women. Cancer 2001; 91: 15791587.[CrossRef][ISI][Medline]
16. Axdorph U, Porwit-MacDonald A, Sjöberg J et al. EpsteinBarr virus expression in Hodgkins disease in relation to patient characteristics, serum factors and blood lymphocyte function. Br J Cancer 1999; 81: 11821187.[CrossRef][ISI][Medline]
17. Glavina-Durdov M, Jakic-Razumovic J, Capkun V, Murray P. Assessment of the prognostic impact of the EpsteinBarr virus-encoded latent membrane protein-1 expression in Hodgkins disease. Br J Cancer 2001; 84: 12271234.[CrossRef][ISI][Medline]
18. MacMahon B. Epidemiology of Hodgkins Disease. Cancer Res 1966; 26: 11891200.[ISI][Medline]
19. Armstrong AA, Alexander FE, Cartwright R et al. EpsteinBarr virus and Hodgkins disease: further evidence for the three disease hypothesis. Leukemia 1998; 12: 12721276.[CrossRef][ISI][Medline]
20. Hancock BW, Gregory WM, Cullen MH et al. ChlVPP alternating with PABlOE is superior to PABlOE alone in the initial treatment of advanced Hodgkins disease: results of a British National Lymphoma Investigation/Central Lymphoma Group randomized controlled trial. Br J Cancer 2001; 84: 12931300.[CrossRef][ISI][Medline]
21. Cullen MH, Stuart NSA, Woodroffe C et al. ChlVPP/PABlOE and radiotherapy in advanced Hodgkins disease. J Clin Oncol 1994; 12: 779787.[Abstract]
22. Townsend P, Phillimore P, Beattie A. Health and Deprivation, Inequality and the North. London, UK: Croom Helm 1988.
23. Morris R, Carstairs V. Which deprivation? A comparison of selected deprivation indexes. J Public Health Med 1991; 13: 318326.[Abstract]
24. Lukes RJ, Carver LF, Hall TC et al. Report of the nomenclature committee in symposium: obstacles to the control of Hodgkins disease. Cancer Res 1966; 26: 1311.[ISI]
25. Barletta JM, Kingma DW, Ling Y et al. Rapid in situ hybridization for the diagnosis of latent EpsteinBarr virus infection. Mol Cell Probes 1993; 7: 105109.[CrossRef][ISI][Medline]
26. Löffler M, Mauch P, MacLennan K et al. The second symposium on Hodgkins disease: review on prognostic factors. Ann Oncol 1992; 3: S63S66.[Medline]
27. Coleman MP, Babb P, Damiecki P et al. Cancer Survival Trends in England and Wales, 19711995: Deprivation and NHS Region. London, UK: The Stationery Office 1999.
28. Flavell K, Constandinou C, Lowe D et al. Effect of material deprivation on EpsteinBarr virus infection in Hodgkins disease in the West Midlands. Br J Cancer 1999; 80: 604608.[CrossRef][ISI][Medline]