Outcomes of CAR T-Cells Therapy in High-Grade B-Cell Lymphomas Compared to Diffuse Large B-Cell Lymphomas: A Weighted Comparison Analysis from CAR-T SIE Study

Background: High-grade B-cell lymphomas (HGBL) account for 10% of all aggressive lymphomas and include double-hit/triple-hit (DH/TH), characterized by myc/bcl2 or myc/bcl2/bcl6 rearrangements and HGBL not otherwise specified (NOS). This category previously included also diffuse large B-cell lymphomas (DLBCL) with myc/bcl6 rearrangements. Due to biological and clinical characteristics, the risk of relapse remains high following first-line intensive treatments. Approval of anti-CD19 Chimeric Antigen Receptor (CAR) T-cells for third-line aggressive lymphomas has resulted in long-term remission in up to 40% of pts. Recent studies have demonstrated similar efficacy of CAR T-cells therapy in the HGBL subgroup versus DLBCL. However, these studies included few HGBL pts with limited long-term follow-up results. The present study compares the outcomes of patients affected by HGBL and DLBCL treated with CAR T-cells using a weighted comparison analysis and evaluates the response to salvage therapy following treatment failure. Methods: Pts diagnosed with relapsed/refractory HGBL and DLBCL who received commercial CAR T-cells in the third line or beyond betweenFebruary 2020 and January 2024 were included. All data were collected from the CAR-T SIE prospective observational study. When available, histological reports and FISH analyses were centrally collected. Propensity score inverse probability weighting (IPW) was used to create a balanced covariate distribution between pts with HGBL and those with DLBCL. Covariates used for propensity score construction included age, sex, C-reactive protein (CRP), ECOG, refractory vs relapsed, International Prognostic Index, number of prior lines of treatment, previous ASCT, response to bridge, bulky disease, CAR T-cells product, and vein-to-vein time. Results: We analyzed a population of 420 pts: 66 HGBL [n=41 HGBL DH/TH, n=20 HGBL NOS, n=5 DLBCL myc/bcl6] and 354 DLBCL [n=278 de-novo DLBCL (79%), n=76 transformed DLBCL (21%)]. All pts were treated with axicabtagene ciloleucel (axicel) (n=206, 49%) or tisagenlecleucel (tisacel) (n=214, 51%). The median follow-up was 23.5 months for HGBL and 18.1 months for DLBCL. Before weighting, DLBCL pts exhibited higher CRP at infusion, number of previous lines >2, and higher use of tisacel than HGBL pts. Overall, 62 of 66 (94%) and 279 of 354 (79%) received bridging therapy with no statistically significant difference in overall response rate (31% in HGBL vs 38% in DLBCL, p=0.250). In univariable analysis, the OS was significantly lower in HGBL than in DLBCL, (24-month OS 39% vs. 49%, log-rank p=0.025) but not PFS (24-months PFS 35% vs. 33%, log-rank p=0.529). Interestingly, we observed a better outcome for transformed DLBCL than de-novo DLBCL (24-month OS: 60% versus 45%, p=0.008). Notably, the 24-month non-relapse mortality was superimposable in HGBL and DLBCL pts (11% vs 11%, p=0.830). In the univariable logistic model, any grade CRS and ICANS (but not severe grade) were significantly higher in HGBL compared to DLBCL [CRS 92% vs 82%, odds ratio (OR) 0.37, (p=0.022); ICANS 35% vs 19%, OR 0.43, (p=0.005)]. We conducted a weighted comparison analysis on 132 pts (n=66 HGBL, n=66 DLBCL) to assess the impact of biological subtype on outcomes. The 24-month weighted PFS and OS were not significantly different between HGBL and DLBCL [PFS: 35% vs 31%, weighted log-rank p=0.937; OS: 39% vs 44%, weighted log-rank p=0.147]. Similarly, no differences were observed in weighted logistic regression for CRS and ICANS of any grade in the two groups (weighted p-values 0.112 and 0.075, respectively). A total of 234 pts (56%) experienced PD following CAR T-cells: 176 (75%) were treated with salvage therapy, including bispecific antibodies in 57 (32%). Following CAR T-cells failure, DLBCL pts had a better outcome than those diagnosed with HGBL (1-year post-relapse OS: 33% vs 16%, p=0.028). In univariable Cox models for OS following CAR T-cells therapy failure, pts treated with salvage and those affected by DLBCL had a better prognosis (HR=0.30, p<0.001; HR=0.64, p=0.038). Conclusions: Our weighted comparison study showed that CAR T-cells therapy has similar efficacy in HGBL and DLBCL with an acceptable toxicity profile. Post-relapse OS was significantly lower in pts affected by HGBL compared to those with DLBCL due to inferior efficacy of salvage therapies. Higher PFS and OS may be expected using CAR T-cells therapy in the second line.