Epstein-Barr virus positive diffuse large B-cell lymphoma predict poor outcome, regardless of the age

Epstein-Barr virus (EBV) positive diffuse large B-cell lymphoma (DLBCL) of the elderly is defined as patients older than 50 years alone. However, recent studies showed young patients with sound immune status could also be affected. In this study, we investigated the clinical features and outcomes of patients with EBV positive DLBCL in the different age groups using different EBER cut-off values. The prevalence of EBV positive DLBCL was 14.0% (35/250) and 10.4% (26/250) for EBER cut-off of 20% and 50%, respectively. With both EBER cut-off values, patients with EBV DLBCL shared many unfavorable prognostic characteristics, regardless of age. EBV positive patients, both in the elderly and young groups, showed significantly worse overall survival and progression-free survival than negative cases. Moreover, no significant differences of outcomes were identified between different age groups with EBV positive DLBCL. In conclusion, EBV positive DLBCL patients, regardless of age, shared similar poor prognostic features and showed worse outcome than negative cases. We suggest that the age criterion of EBV positive DLBCL of the elderly, and possibly the name itself, be modified in future.

Statistical analyses. Statistical analyses were performed using SPSS software, version 20.0. Chi-square and Fisher exact tests were used to compare categorical variables . OS and PFS were defined according to Cheson 2014 25 . Survival curves were plotted by using Kaplan-Meier method and were compared by using log-rank test. For all the tests, a probability value of less than 0.05 (2-sided) was considered statistically significant.

Results
Prevalence of EBV positive DLBCL in the cohort. A total of 250 cases with DLBCL were included in the analysis as the whole cohort. Using 20% as cut-off, 14.0% (35/250) cases showed EBER positivity. The prevalence of EBER positivity were 15.1% (25/166) and 11.9% (10/84) in the elderly and young group, respectively. No significant difference of incidence was observed between the two groups (P = 0.497). When a cut-off of 50% was used for EBER positivity, the incidence of EBER positive cases was 10.4% (26/250). Accordingly, 11.4% (19/166) and 8.3% (7/84) were positive for EBER in the elderly and young groups, respectively. No significant difference of prevalence was observed between the two groups (P = 0.466).

EBV positivity and clinical features.
In the whole cohort, compared with EBER negative cases, EBER positive (for both cut-off values) patients showed male predominance, advanced clinical stages (stage III/IV), poor performance status (ECOG PS status 2-4) and lower response to first-line treatment. (Table 1,2). Using 20% as cut-off, EBV positivity was significantly associated with male sex (76.0% vs. 53.9%), poor PS status (40% vs. 12.5%) and lower response to first-line treatment (64.0% vs. 84.4%), compared with negative cases. In the young group, EBER positivity was strongly associated with poor PS status (30.0% vs. 5.8%) and lower response to first-line treatment (70.0% vs. 94.6%), With regard to other clinical features, although not statistically significant, EBV positive patients more frequently showed unfavorable characteristics compared with negative cases in both elderly and young groups (Table 1). However, no significant differences of these clinical characteristics were observed between different age groups. When we analyzed above parameters with a cut-off value of 50% for EBER, although fewer Scientific RepoRts | 5:12168 | DOi: 10.1038/srep12168 factors with significant differences were observed, EBV positive cases still showed majority of unfavaorable clinical features compared with negative ones, regardless of ages (Table 2).

Number.
(%) features between different age groups. When we analyzed those parameters with the cut-off value of 50%, similar results were observed in both age groups (   (Fig. 4b,d). In addition, we also analyzed the survival differences with other age cut-offs (40 or 60 years old), however, no difference of OS or PFS was recognized between different age groups with EBV positive DLBCL (data not shown).

Discussion
While using 20% and 50% as cut-off vaues, the incidences of EBV positive DLBCL were 14.0% (35/250) and 10.4% (26/250), respectively. Our result was similar to that of a previous study in Peru (14.9% for 20% cut-off and 9.0% for 50% cut-off) 26 . However, taking different cut-off vaues used in previous studies into consideration, the incidence of EBV positive lymphoma appeared to be higher than those reported in previous studies in Asian countries 4,20 .  In current study, EBV positivity, defined by either EBER cut-off (20% or 50%), had a close association with male sex, advanced clinical stage, poor PS status and lower response to first-line treatment, which was observed in both young and elderly patients. These results indicated that EBV positive DLBCL were clinically aggressive, irrespective of age. Based on these, we then analyzed the association between EBV positivity and the clinical prognostic indicators reported previously [27][28][29][30][31][32] . The results showed that EBV positivity had a strong relationship with elevated serum CRP level, β 2M level, CA125 level, ESR level and ferritin level, regardless of age and EBER cut-off value.

Continued
We subsequently included CD10, Bcl6, MUM1, FOXP1, GCET1 and LMO2 in our study as these markers were used to establish a diagnosis and further classify DLCBL into GCB or non-GCB subtype [21][22][23][24] . Research showed the majority of EBV positive DLBCL of the elderly had a non-GCB predominance, which is a subtype with poor prognosis. In our study, we accessed the COO with four conventional IHC algorithms. Although EBV positive DLBCL of the elderly group showed less frequent FOXP1 expression and the young group demonstrated less common of LMO2 expression, the EBV positive DLBCL (both age groups), although not statistically significant, demonstrated the non-GCB subtype preference than negative ones. However, the COO difference was observed in Choi and Visco-Young algorithms alone. The poor concordance among IHC algorithms may be one of reasons to explain our Variable Number.
(%)   results, since 20-30% cases among the IHC algorithms are discrepant 22,33 . We also analyzed CD5, Myc, Bcl2, Ki-67 expression and DPE, MYC and BCL2 gene rearrangement and DHL because of their efficacy in prognostication [34][35][36] . We also chose CD30 due its significant association with EBV positivity in DLBCL that was observed in several previous studies 7, 15,37 . Moreover, EBER + /CD30 + DLBCL had significantly poorer outcomes compared with EBER − /CD30 + cases 38 . Using the cut-off values of these proteins applied in above studies 7,34-37 , we accessed the incidences of these pathological factors in DLBCL according to EBV status and age. Interestingly, EBER positivity was associated with less frequent Bcl6 expression, which was also confirmed by previous studies 7 . Moreover, Constitutive expression of EBV-derived miR-NAs, including BART3, BART7, BART9 and BART17-5p, was found to be capable of repressing Bcl6 expression, partly accounting for reduced Bcl6 expression in EBV positive lymphoma 39 . The proportions of Myc expression, DPE, MYC rearrangement and DHL were much higher in EBV positive DLBCL, which were consistent with the aggressive biological behavior of EBV positive DLBCL 1,11,13 . DPE and DHL were more frequent in EBV positive cases than EBV negative ones, which was inconsistent with the study by Ok et al. 7 , probably due to different ethnic background or geographic variation of EBV strains. Bcl2 expression was less frequent in EBV positive DLBCL. However, this was not observed in a previous study 7 . Of note, in that study, cut-off for Bcl2 was 70%, rather than 50%. Higher Ki-67 expression was observed in EBV positive patients, suggesting that EBV infection contributes the aggressiveness of EBV positive DLBCL 40 . Similar to previous studies 40 , the significant association of CD30 with EBV positivity in DLBCL was also confirmed in our study. However, the mechanism underlying this phenomenon remains to be determined. It is possible that CD30 functions synergistically with EBV to transform B lymphocytes 41 .
In present study, although the treatment for EBV positive DLBCL was more intensive than negative ones (much higher percentage of immunochemotherapy in EBV positive DLBCL), consistent with most studies 4,12,16,19 , both age groups with EBV positive DLBCL showed significantly worse OS and PFS than negative cases. Further comparison showed that EBV positive DLBCL of the elderly group had similar OS and PFS to the young counterparts. It is worth mentioning that the EBV positive DLBCL patients (both elderly and young) were treated with similar regimens. Actually, EBV positive DLBCL of the elderly was included as a provisional entity in the 2008 WHO classification of tumors of hematopoietic and lymphoid tissues, in which the WHO Working Group considered that there was insufficient evidence to recognize this entity as a distinct disease at that time 3,42 . EBV positive DLBCL of the young group that had no evidence of underlying immunosuppression had been described in previous reports 4, 16 . The identification of these young cases arised an interrogation that if EBV positive DLBCL might be an entity that was not restricted to patients who were older than 50 years old alone. However, one recent study did not recognized unfavorable baseline characteristics in young patients with EBV positive DLBCL 14 . Therefore, they concluded that EBV positive DLBCL of young group should be considered to be a distinct clinical entity different from EBV positive DLBCL of the elderly 14 . In present study, to say the least, no distinct difference was recognized in the clinical and pathological features between EBV positive DLBCL in the elderly and the young group, which was in accordance with most studies 4,12 . These unfavorable factors were shared by both groups, which indicated that EBV positive DLBCL belongs to a unique entity with aggressive clinical features 7,11,[13][14][15]20,43 , regardless of age. Since there is no uniform cut-off for EBV positivity, some reports attributed the inconsistent results to the different cut-off values used in different studies 7,11,13,14,[44][45][46] . However, by appling two most frequently-used cut-off values, our study indicated that EBER positive patients harbored more unfavorable clinical and pathological features than EBER negative ones, regardless of the cut-off values of EBV.
Several studies had reported that EBV positive patients showed an inferior prognosis compared with EBV negative cases, especially the elderly group 7,11,13,15,18,47,48 . The clinical course is often aggressive with a median survival of 2 years and an overall 5-year survival rate of approximately 25% 4,5,8 . In our study, accordantly, EBV positive DLBCL of the elderly showed worse survival than negative counterparts. Many studies also showed the young patients of EBV positivity demonstrated poor outcome with traditional immunochemotherapy 4,12,16,19 , which was confirmed in current study. However, the study by Hong et al. revealed young EBV positive DLBCL patients (6.7%, 13/195) had an outcome similar to EBV negative ones 14 .
In fact, the biological mechanism underlying similar clinical and pathological features and outcome between these age groups remains to be investigated. To elucidate molecular mechanisms involved EBV positive DLBCL, Ok et al. 7 evaluated GEP profiles signatures of DLBCL with different EBV status. It is worth noting that in the 24 EBV positive patients included in that study, 7 patients were younger than 50 years old. They revealed the NF-κ B activation and JAK/STAT activation were enhanced in EBV positive DLBCL compared with negative counterparts by gene set enrichment analysis. Thus, EBV positive DLBCL might share similar GEP signatures and common pathogenic pathways, irrespective of age. It also has been reported that ABC-like DLBCL more frequently exhibited JAK-STAT and NF-κ B pathways activation compared with GCB-like DLBCL 49  EBV negative DLBCL 47 . Using miRNA array platforms, Andrade et al. 43 showed that miR-146b and miR-222 were highly specific for EBV positive DLBCL. The targets of hsa-miR-146b and its viral counterpart are INTS6 and IPO, both of which are tumor suppressors 51 and mediators of inflammation 26 Similarly, hsa-miR-222 interferes with important proteins related to oncogenesis, cell cycle regulation, cell transcription, cell adhesion, oxidative stress, and apoptosis inhibition 52 . Besides, NF-κ B pathway can be activated indirectly by hsa-miR-222 43 . All of the above findings suggested that EBV positive DLBCL shared a common molecular basis, irrespective of age.
Additionally, using other age cut-off values, we did not recognized any significant differences of survival between elderly and the young patients of EBV positivity. This also indicated that the young group patients should not be excluded from the whole cohort of EBV positive patients and the 50 years old cut-off might not be applicable in the R-CHOP era.
In summary, EBV positive DLBCL patients shared poor prognostic features, regardless of elderly or young group. The survival analysis also showed that EBV DLBCL of the elderly showed a similar outcome to the young ones. Based on these results, we suggest that the age criterion, and possibly the name-EBV positive DLBCL of elderly itself, be modified in future.