Bispecific antibody vs non-bispecific antibody systemic therapy as immediate next line treatment after chimeric antigen receptor T-cell (CART) failure in large B-cell lymphoma

Abstract Background For patients with large B-cell lymphoma (LBCL) who relapse after CART, optimal next line therapy is unknown. Bispecific antibodies (BsAbs) redirect T cells to target malignant B cells and are commonly used post-CART. Real-world patient selection and comparative efficacy for BsAbs vs other regimens after CART failure is limited. We evaluated outcomes of BsAb- vs non-BsAb-based salvage therapies following CART failure. Methods Among 935 relapsed/refractory (R/R) LBCL patients (pts) treated with CART between 2015-2024 across 15 academic institutions, pts with CART failure were grouped according to BsAb-based vs non-BsAb salvage therapy. Differences between variables were evaluated with Wilcoxon rank-sum test and pooled t-test (two-sided P <.05). Time-to-event curves were established by Kaplan-Meier method. Cox multivariable regression analysis (MVA) evaluated clinical impact. Results In 490 pts with CART failure, 311 received systemic salvage therapy. Median age was 59 years (range 19–85), 62% were male, 78% were Caucasian, and histology included de novo DLBCL (72%), transformed follicular lymphoma (13%) or other LBCLs (15%). Most common CART product was axi-cel (56%), followed by tisa-cel (21%), liso-cel (22%), and others (1%). 73 pts received BsAbs ,with BsAb monotherapy (B-M) in 60 pts and BsAb + targeted therapy (B+T) in 13 pts. Non-BsAb regimens (n=238) included chemoimmunotherapy (CIT, n=45), lenalidomide +/- tafasitamab (len +/-taf, n=75), polatuzumab-bendamustine-rituximab (pola-BR; n=48), checkpoint inhibitors (CPI, n=25), loncastuximab (lonca, n=7) or other targeted therapies (TT, n=38). Baseline characteristics were balanced between the B-M and non-BsAb groups, including age, sex, cell of origin, double-hit lymphoma (DHL), primary refractory disease to frontline CIT (PRD), prior autologous transplant, prior bendamustine, and time to salvage <90 days post–CAR-T. However, B-M pts had a higher rate of EN disease as compared to non-BsAb pts (86% vs 70%, p=0.01). Pts who received CIT were typically younger (p<0.05) with de novoDLBCL (p=0.03) compared to all other treatment groups with the exception of pts receiving CPI. For B-M vs non-BsAb, median duration of follow-up was 12.5 vs 40.8 mo. Median PFS was 4.7 vs 2.4 mo (p=0.057); mOS was similar, 8.9 vs 9.9 mo (p=0.7). Comparing across all treatment regimens: mPFS (in mo) was 4.7 (B-M), 7.4 (B+T), 1.1 (CIT), 2.7 (len +/- taf), 4.6 (pola-BR), 1.8 (CPI), 1.4 (lonca), and 1.9 mo (TT; p=0.003). Median OS (in mo) was 8.9 (B-M), 10 (B+T), 5.4 (CIT), 12.5 (len +/- taf), 10.7 (pola-BR), 9.9 (CPI), 18.7 (lonca), and 8.5 mo (TT; p=0.8). In PRD, longest mPFS was associated with B+T (7.4 mo), pola-BR (3.6 mo), and B-M (3.2 mo) as compared to len +/- taf, lonca and CPI (mPFS range 0.5-2.7 mo, p<0.001) with no difference in mOS (p=0.6). In DHL, OS and PFS did not differ by regimen (mPFS range 0.9-13.6 mo, p=0.3; mOS range 0.9-4.8 mo, p=0.14). With relapse >180 days post-CART (n=91): B-M vs B+T vs len +/- taf performed best for mPFS (NR vs 16.3 vs 22.4 mo respectively, p=0.005). With relapse <180 days post-CART (n=211): mPFS (in mo) from highest to lowest was 7.4 (B+T), 3.8 (pola-BR), 2.6 (B-M), 1.9 (len +/- taf), 1.8 (CPI), 1.4 (TT) and 1 mo (either CIT or lonca, p=0.03). With relapse >180 days post-CART (n=91), B-M vs B+T vs len +/- taf performed best (mPFS NR vs 16.3 vs 22.4 mo respectively, p=0.005). For mOS for this subset was not different by regimen (range 4.5-9.4 mo, p=0.8). In MVA, no regimen offered PFS or OS benefit post-CART failure over B-M. However, pts who received CIT [HR=2.0 (95% CI 1.1-3.7)], CPI [HR=2.0 (95% CI 1.2-3.4)] or lonca [HR=4.4 (95% CI 1.7-11.6)] had worse PFS (p=0.003). PRD and relapse <180 days post-CART were independently associated with decreased mPFS (p<0.007) and mOS (p<0.009). Conclusion In pts with R/R LBCL post-CART failure, BsAb-based salvage demonstrated superior mPFS without significant OS benefit compared to most non-BsAb therapies, particularly in late relapse. Survival with BsAb and non-BsAb treatments within 6 months of CART were diminished, suggesting that timing of relapse influences effectiveness of salvage regimens. Treatment-era bias may affect results, as many non-BsAb pts were potentially treated before BsAb FDA approval, but our results support integration of BsAbs in the post-CART setting. Prospective validation is warranted to define optimal sequencing of BsAbs and other treatment options in post-CART treatment paradigms.