We read with interest the report by Jarrett et al1 on the association between latent Epstein-Barr virus (EBV) infection of Hodgkin-Reed-Sternberg (HRS) cells and clinical outcome of patients with classical Hodgkin lymphoma (cHL). Using Epstein-Barr virus-encoded RNA (EBER) in situ hybridization, latent infection was detected in 33% of patients and was associated with inferior overall survival (OS) and disease-specific survival (DSS). When adjusted for sex, stage, and B symptoms by multivariate analysis, EBV infection was not associated with inferior DSS in the whole population but was associated with inferior OS only in patients older than 50 years.
The prognostic association of EBV infection of HRS cells in cHL is controversial.1-3 Although the study by Jarrett et al1 provides potential explanations for the diversity of published results, some issues need to be considered. First, treatment was not described and probably was not uniform. Second, OS is not an optimal end point for prognostication in cHL because it is affected by salvage management after relapse and the frequency of unrelated or treatment-related deaths (neutropenic infection, bleomycin pulmonary toxicity, anthracycline cardiotoxicity, and secondary myelodysplasia/acute leukemia and solid tumors). Similarly, DSS is affected by treatment after relapse, and it is unclear how to score treatment-related toxic deaths, which have no obvious relation to disease biology. Amore optimal end point, related to the biologic aggressiveness of the disease, would be failure-free survival (FFS), reflecting tumor control, as has been used in the International Prognostic Score.4
Nevertheless, the suggestion that EBV infection may be associated with worse outcome in the oldest age group is very interesting.1 Therefore, we re-examined the updated database of the International Hodgkin Study Group.2 The follow-up is now longer with a median of 10.2 years versus 5.4 years2 for patients treated with Adriamycin, bleomycin, vinblastine, dacarbazine (ABVD) or equivalent regimens. The primary end point was FFS, whereas OS was analyzed as a secondary end point. The results are presented in Table 1. EBV expression, as evaluated by LMP-1 immunostaining, is equivalent to EBER in cHL, because these neoplasms have a latency type II infection. EBV infection was not associated with significantly different FFS, either for all 575 patients or for the age-defined subgroups identified in the Jarrett et al study.1 Only a small difference in 10-year OS in favor of patients negative for EBV was noted (78% ± 4% versus 81% ± 2%, P = .04). This difference disappeared after adjustment for stage and age (P = .51). Because FFS was not different in the same group, this numerically small difference could be due to variable management after relapse or to deaths from toxicity, second malignancies, or unrelated causes. The results were identical for the 303 patients treated with ABVD or equivalent regimens with or without radiotherapy, thus eliminating the confounding effect of variable therapy (Table 1). Multivariate analysis with adjustment for stage, age, and sex did not reveal any independent effect of EBV expression on outcome (not shown).
Effect of LMP-1 expression on FFS and OS of patients with cHL after stratification with respect to age
Outcome measure . | All ages . | 14-34 y . | 35-49 y . | 50 y and older . |
|---|---|---|---|---|
| All patients, no. of all patients/failures/deaths | 575/126/115 | 356/74/45 | 134/28/25 | 85/24/45 |
| 10-y FFS, LMP-1+ vs LMP-1– | 77% ± 4% vs 78% ± 2%, P = .68 | 78% ± 5% vs 79% ± 2%, P = .60 | 82% ± 7% vs 77% ± 5%, P = .62 | 70% ± 9% vs 70% ± 6%, P = .89 |
| 10-y OS, LMP-1+ vs LMP-1– | 78% ± 4% vs 81% ± 2%, P = .04 | 91% ± 4% vs 87% ± 2%, P = .69 | 83% ± 7% vs 82% ± 4%, P = .82 | 46% ± 10% vs 48% ± 7%, P = .56 |
| ABVD and equivalent with or without radiotherapy: no. of patients/failures/deaths | 303/65/57 | 194/40/28 | 66/11/7 | 43/14/22 |
| 10-y FFS, LMP-1+ vs LMP-1– | 73% ± 6% vs 80% ± 3%, P = .26 | 72% ± 9% vs 81% ± 3%, P = 0.25 | 83% ± 9% vs 82% ± 6%, P = .97 | 65% ± 12% vs 68% ± 9%, P = .95 |
| 10-y OS, LMP-1+ vs LMP-1– | 77% ± 6% vs 80% ± 3%, P = .56 | 92% ± 6% vs 82% ± 4%, P = .61 | 85% ± 10% vs 89% ± 5%, P = .95 | 50% ± 13% vs 49% ± 11%, P = .56 |
Outcome measure . | All ages . | 14-34 y . | 35-49 y . | 50 y and older . |
|---|---|---|---|---|
| All patients, no. of all patients/failures/deaths | 575/126/115 | 356/74/45 | 134/28/25 | 85/24/45 |
| 10-y FFS, LMP-1+ vs LMP-1– | 77% ± 4% vs 78% ± 2%, P = .68 | 78% ± 5% vs 79% ± 2%, P = .60 | 82% ± 7% vs 77% ± 5%, P = .62 | 70% ± 9% vs 70% ± 6%, P = .89 |
| 10-y OS, LMP-1+ vs LMP-1– | 78% ± 4% vs 81% ± 2%, P = .04 | 91% ± 4% vs 87% ± 2%, P = .69 | 83% ± 7% vs 82% ± 4%, P = .82 | 46% ± 10% vs 48% ± 7%, P = .56 |
| ABVD and equivalent with or without radiotherapy: no. of patients/failures/deaths | 303/65/57 | 194/40/28 | 66/11/7 | 43/14/22 |
| 10-y FFS, LMP-1+ vs LMP-1– | 73% ± 6% vs 80% ± 3%, P = .26 | 72% ± 9% vs 81% ± 3%, P = 0.25 | 83% ± 9% vs 82% ± 6%, P = .97 | 65% ± 12% vs 68% ± 9%, P = .95 |
| 10-y OS, LMP-1+ vs LMP-1– | 77% ± 6% vs 80% ± 3%, P = .56 | 92% ± 6% vs 82% ± 4%, P = .61 | 85% ± 10% vs 89% ± 5%, P = .95 | 50% ± 13% vs 49% ± 11%, P = .56 |
No difference was statistically significant in the multivariate analysis after adjustment for age (all ages subgroup), sex, and stage (early IA, IIA versus advanced IB, IIB, III, IV).
Given these results, it would be interesting to see whether latent EBV infection is associated with different FFS in the population and subgroups defined in the study by Jarrett et al,1 particularly under treatment with current standard therapy.
Impact of tumor EBV status on presenting features and outcome in age-defined subgroups of patients with classic Hodgkin lymphoma
We are grateful to Herling and colleagues for their comments regarding our recent publication. We agree that their data differ, in some respects, from those of our study. The literature in this field is conflicting; our discussion dealt with several issues, which may explain these differences, including the contrasting results found in population-based and non-population-based studies and geographic variation in the proportion of Epstein-Barr (EBV)-associated cases.1
It is difficult to directly compare the data of Herling et al2 with our own results because the case series are fundamentally different. Our study focused on a population-based group of patients with biopsy samples available from 461 (87%) of 531 individuals and collected over 4.5 years. Herling's group describes outcome data from 575 (34%) of 1686 patients treated at 4 academic centers over 16 years. Differences between patients studied are highlighted by the observation that 85 (14.8%) of 575 of Herling's patients were age 50 years or older, compared with 127 (27.5%) of 461 patients in our study. In addition, the 2 studies vary significantly in both the overall proportion of EBV-associated cases (P < .001) and proportion of EBV-associated mixed cellularity cases (P = .003), with lower values in the study by Herling and coworkers.
Herling et al correctly state that our study population was not treated uniformly. The principles of patient management are, however, broadly agreed upon across the Scotland and Newcastle Lymphoma Group. Younger patients (< 60 years) are typically managed using a risk-adapted strategy, which has been described previously.3 Therapy for older patients is inconsistent, as highlighted in a previous publication by our group.4 Only 6% of patients in our study received doxorubicin (Adriamycin), bleomycin, vinblastine, and dacarbazine (ABVD); small numbers preclude statistical analysis of the ABVD-treated subgroup. In addition, we are concerned that the restriction of a study to ABVD-treated patients is likely to result in the selection of older individuals destined to have a comparatively favorable outcome; fitter patients, considered able to tolerate this relatively intensive chemotherapy, will be selected.
Herling et al favor the use of failure-free survival (FFS) over disease-specific survival (DSS) and overall survival (OS). While DSS and OS are affected by salvage therapies after relapse, in our opinion both outcome measures are robust and clinically meaningful for patients and physicians. The use of OS and DSS enabled us to explore the relationship between EBV association and comorbidity, because our interest was not confined to disease aggressiveness. Our results were also explicit with respect to the causes of non-Hodgkin lymphoma-related deaths.1 While DSS and OS may be poor outcome measures in studies with relatively short followup, in a study with a median follow-up of 93 months they provide useful information. We believe the use of FFS to study outcome in our population could be misleading, because it is established that patients who relapse after treatment with radiotherapy alone have an excellent prognosis after salvage chemotherapy.
Comparisons between population-based and single or multicenter non-population-based data are difficult. The main issues discussed herein only serve to highlight the importance of performing population-based studies.
Correspondence: Department of Haematology, Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne, NE1 4LP, United Kingdom; e-mail: g.l.stark@ncl.ac.uk.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal