Impact of prior B-cell-directed immunotherapy on the outcome of CD19 CAR T-cell therapy in aggressive B-cell lymphoma - an analysis of the EBMT and the GoCART coalition

Abstract Introduction: Treatment outcomes of diffuse large B-cell lymphoma (DLBCL) and related large B-cell lymphoma (LBCL) have been revolutionized by novel B-cell-directed immunotherapies (BCDI), such as CD19-targeting chimeric antigen receptor T-cell therapy (CAR-T), polatuzumab vedotin (pola), and bispecific antibodies (BSA). However, it remains unclear how BCDI-exposure prior to CAR-T therapy affects outcomes, and prolonged B-cell depletion may contribute to CAR-T-associated non-relapse mortality (NRM). Here, we compare outcomes of patients (pts) receiving BCDI (excluding rituximab) versus conventional treatments prior to CD19 CAR-T in a large European multicenter cohort. Methods: We collected retrospective data on the use and type of BCDI, as well as outcomes from 1154 adult pts across 38 European centers who received CD19 CAR-T between 2018 and 2023 for LBCL, identified in the European Group for Blood and Marrow Transplantation (EBMT) registry. We categorized BCDI as BSA, pola, and other, applied prior to T-cell apheresis or as bridging therapy. Results: Of 1154 CAR-T pts, 380 (32.9%) received BCDI other than rituximab (BCDI group), while 768 (66.6%) pts were not pretreated with BCDI (non-BCDI group). Twenty-four of 380 pts (6.3%) received BSA, 305 (80.3%) pola-based regimens, and 51 (13.4%) other BCDI (including obinutuzumab, 26 pts; brentuximab, 16 pts). The BCDI and non-BCDI groups were balanced for lymphoma diagnosis and IPI variables at diagnosis and lymphodepletion. In the BCDI group, more pts had received 3+ prior treatment lines than in the non-BCDI group (54.5% vs. 42.3%, p<0.001), and more pts had complete or partial remission at lymphodepletion (40.6% vs. 28.9%, p<0.001). At a median follow-up of 36 months, no significant differences were observed in 3y-OS (BCDI vs. non BCDI: 41.9% [36.6-48] vs. 50.1% [46.5-54.1], p=0.08), progression-free survival (35.7% [30.7-41.5] vs. 37.8% [34.2-41.7], p=0.27), relapse incidence (51.4% [45.9-56.6] vs. 51% [47.1-54.7], p=0.62), and NRM (12.9% [9.3-17.1] vs. 11.2% [8.9-13.8], p=0.24). There were no differences in the incidence and severity of cytokine-release syndrome (83.2% vs. 84.4%, p=0.64) and immune effector cell-associated neurotoxicity syndrome at 15 days after infusion (30.9% vs. 32.8%, p=0.52). Furthermore, there was no difference in the 2y-incidence of grade III/IV infections (22.6% vs. 21.7%, p=0.92). Multivariable analysis was performed to adjust for remission, LDH and ECOG at CAR-T, age, sex, transformed DLBCL, prior treatment lines, and CAR-T product type. Interestingly, this analysis showed a trend towards inferior OS for the use of BCDI prior to CAR-T treatment (HR 1.21, 95% CI [1-1.47], p=0.06), while no differences in other outcomes were observed. Concerning type of BCDI, 3y-OS was 48% [31.3-73.7%], 40.6% [34.7-47.4%], and 59.9% [47.3-75.9%] in pts receiving BSA, pola-based regimens, or other BCDI, respectively (p=0.03). 22 BSA pts could be matched with 40 pola pts on number of treatment lines, sex, age, ECOG, LDH, and remission status prior to CAR-T as propensity-score matching covariates. The difference in 2y-OS was 52.5% vs. 36.7%, showing relevant effect size, but no significance due to the low number of patients in the BSA group (HR 1.68, p=0.12). Higher OS for pts receiving other BCDI may be explained by higher proportion of PMBCL (33.3% vs. 3% in the pola group). In 39 other BCDI pts matched to 74 pola pts on transformed LBCL, LDH at CAR-T, age, CAR-T product and year of infusion, higher 2y-OS showed a relevant effect size but was not significant (2y-OS 64.3% vs. 56.7%, HR 0.61, p=0.11). We finally analyzed timing of BCDI application relative to apheresis. Of 359 pts with available data, 121 (33.7%) received BCDI prior to apheresis, 172 (47.9%) as bridging therapy, and 66 (18.4%) both. 3y-OS compared to no BCDI (50.3%) was 47.1% for BCDI as bridging, 41.2% for BCDI only prior to apheresis, and 22.6% for pts receiving both (global logrank-test, p=0.007). Conclusions: Our data suggest that BCDI treatment prior to CAR-T therapy negatively impacts OS but not NRM. While pts receiving BCDI as bridging show OS indistinguishable from the non-BCDI group, application of BCDI prior to apheresis was associated with lower OS in univariate analysis. In matched subgroups, there was a signal towards inferior OS in pts receiving pola-based regimens; however, the effect of combination partners, retrospective selection bias and small size of subgroups needs to be considered.