Comparative Effectiveness of Bendamustine Plus Rituximab (BR) and Rituximab Plus Gemcitabine and Oxaliplatin (R-GemOx) in Relapsed/Refractory Diffuse Large B-Cell Lymphoma

Introduction:

The treatment regimens bendamustine and rituximab (BR), and rituximab, gemcitabine, and oxaliplatin (R-GemOx) have proven to be efficacious and have manageable safety profiles for transplant-ineligible patients (pts) with relapsed/refractory diffuse large B-cell lymphoma (R/R DLBCL; NCCN guidelines 2020). However, there is a paucity of published data from randomized clinical trials (RCTs) to determine the comparative efficacy of R/R DLBCL treatments (Vander Velde et al. Hematol Oncol 2019) and limited evidence from real world studies (Ionescu-Ittu et al. J Comp Eff Res 2019). One of the most recent RCTs to evaluate BR as a treatment backbone in this setting is the Phase Ib/II GO29365 study (NCT02257567). In this study, polatuzumab vedotin combined with BR (Pola+BR) significantly improved progression-free survival (PFS) and overall survival (OS), compared with BR alone in pts with R/R DLBCL (Sehn et al. J Clin Oncol 2020). In order to compare survival of pts treated with BR or R-GemOx in the second line (2L) setting, we performed a retrospective analysis using real-world data (RWD) from the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) cancer registry linked to Medicare enrollment data and insurance claims.

Methods:

Pts with cancer diagnoses from 2004 to 2016 were identified using the SEER Medicare database. Pts diagnosed with DLBCL not otherwise specified (ICD-O-3 histology code: 9680) who received 2L BR or R-GemOx alone, were included. Additional inclusion criteria were: pts aged >65 years at diagnosis, with DLBCL as their first or only primary cancer diagnosis, and other standard SEER Medicare analysis criteria such as minimum Medicare enrollment of 12 months, to ensure sufficient data for analysis. Pts with evidence of prior hospice care were excluded. Accurate identification of transplant-ineligible pts is a limitation of using the database; transplant ineligibility was assumed for this population based on their age and associated risk factors. Survival was assessed by Kaplan–Meier and Cox regression analysis. The inverse probability-of-treatment weighting (IPTW) method was applied to balance baseline characteristics such as age at the start of 2L treatment, gender, stage of disease, race, Charlson Comorbidity Index (CCI), relapsed or refractory status, time from initial treatment to 2L treatment initiation, calendar year of 2L start, and health maintenance organization. Analyses of pts treated in the third line (3L) setting were also performed.

Results:

Of the 3,606 pts with DLBCL within the database, 439 pts (BR: n=308; R-GemOx: n=131) met all study criteria and were included in the analysis. Median follow-up was 12.42 months (interquartile range [IQR]: 4.78–28.41) for pts treated with BR, and 7.72 months (IQR: 3.02–21.14) for pts treated with R-GemOx. Pts treated with R-GemOx were more likely than pts treated with BR to be younger or male (Table). There was a similar proportion of pts who were primary refractory in each treatment group (BR: 20.8%; R-GemOx: 21.4%; p=0.99). The unadjusted median duration of 2L treatment was 11.4 weeks (IQR: 4.29–20.04) with BR and 8.14 weeks (IQR: 4.07–14.71) with R-GemOx. Median OS was 16.39 months (95% confidence interval [CI]: 13.01–19.48) with BR and 8.74 months (95% CI: 7.00–12.98) with R-GemOx (Figure A). After adjustment for covariates, median OS was 16.39 months (95% CI: 12.88–19.48) with BR and 9.26 months (95% CI: 7.10–14.36) with R-GemOx (Figure B); the hazard ratio (HR) for adjusted OS in 2L was 1.24 (95% CI: 0.97–1.58) for R-GemOx compared with BR. After propensity score adjustment, the HR for OS in 3L was 0.996 (95% CI: 0.70–1.42) for R-GemOx compared with BR.

Conclusions:

OS was not significantly different between pts with R/R DLBCL treated with BR or R-GemOx in this RWD analysis from the SEER Medicare database. This type of analysis is limited by the assumption that all important variables have been accounted for in the propensity scoring, and by the inclusion of only pts who were recorded in the database, which may impact how these results can be generalized. Nonetheless, in the absence of comparative data from RCTs, this RWD analysis demonstrates similar real-world effectiveness of the two regimens in R/R DLBCL.