The applicability of prognostic models under rituximab-dose-adjusted EPOCH (R-DA-EPOCH) in primary mediastinal large b-cell lymphoma (PMLBCL) : a comparison with R-CHOP

Abstract

Abstract Topic: 19. Aggressive Non-Hodgkin lymphoma - Clinical

Background: The rarity of PMLBCL has prevented the development of dedicated, powerful and validated prognostic systems. The International Prognostic Index (IPI) and its age-adjusted variant are of questionable value under R-CHOP±radiotherapy (RT). Recently, we published two potential prognostic systems for PMLBCL: The first combined the presence of extranodal disease (stage IV or simply extranodal extension; E/IV) with elevated LDH exceeding twice the upper normal limit (LDH>2x) (E/IV-LDH2x model). The second combined E/IV with bulky disease ≥10 cm (E/IV-bulk). The potentially improved outcomes with R-da-EPOCH render the identification of powerful prognostic factors even harder and the applicability of prognostic models questionable.

Aims: To assess the applicability of the E/IV-LDH2x and E/IV-bulk prognostic models in the setting of PMLBCL treated with R-da-EPOCH and minimal use of RT in a multinational real-life setting.

Methods: In Hellenic, Israeli, Turkish, Saudi, Maltese and Cypriot Centers, 258 patients (<65 years) with PMLBCL were treated under the R-da-EPOCH program. Consolidative RT was given at the treating physician’s discretion in a minority of patients (~15%) usually based on PET/CT results. Full data regarding the E/IV-LDH2x and E/IV-bulk models were available in 253 and 244 patients respectively. Survival data were compared numerically with a previously published Hellenic-Cypriot series of 332 PMLBCL patients (≤65 years old) treated with R-CHOP (Vassilakopoulos et al. Oncologist; 2021;26:597-609).

Results: The 5-year Freedom From Progression (FFP) after R-da-EPOCH was 84.7%. However, 5 patients developed therapy-related (t-)AML at 10.5-24 months from treatment initiation, while in 1st remission. The 5-year Lymphoma-Specific Survival (LSS) was 91.7%. Protocol violations were common (53%), mainly consisting of insufficient dose escalation despite the absence of prohibitive toxicity. As shown in the table, the distribution of patients with 0,1 or 2 adverse factors in the E/IV-LDH2x and E/IV-bulk were almost identical in the R-CHOP- and the R-da-EPOCH-treated populations despite the diverse ethnic origin. With less marked differences between the prognostic subgroups and less events, the prognostic models produced statistically significant results of lower magnitude after R-da-EPOCH compared to R-CHOP. Interestingly, patients with no adverse factors by either model had virtually identical 5-year FFP (88-90%) and LSS (96% for the E/IV-LDH2x and 98-99% for the E/IV-bulk model) irrespectively of treatment with R-CHOP or R-da-EPOCH. The use of RT was much more frequent with R-CHOP (data to be precisely presented in the Meeting). Patients with 1 risk factor had 7-12% better FFP in absolute terms and 4-8% better LSS with R-da-EPOCH compared to R-CHOP. Patients with 2 risk factors had 11-15% better FFP but virtually no difference in LSS when treated with R-da-EPOCH.

Summary/Conclusion: Prognostic models derived from R-CHOP-treated populations are also applicable under R-da-EPOCH, although less discriminative. This comparative study suggests that R-CHOP may be sufficient in patients with no adverse factors but the need of RT remains an open question. Patients with one or multiple adverse factors appear to do better than with R-CHOP. A final analysis of ~290 R-da-EPOCH patients will be presented. The application of the above prognostic models might facilitate the selection of optimal R-da-EPOCH candidates with the hope to avoid severe toxicity, including t-AML.