Real-world multicenter outcomes of CD3xCD20 bispecific antibody therapies in patients with central nervous system (CNS) lymphoma

Abstract Introduction: CNS involvement in B-cell non-Hodgkin lymphoma (B-NHL) results in poor outcomes and remains an unmet need. The CD3xCD20 bispecific antibodies (BsAb) are effective in systemic B-NHL, but data on the CNS activity of BsAbs are limited. Herein, we report real-world outcomes of CD3xCD20 BsAb in patients (pts) with CNS involvement by B-NHL. Methods: This is a multicenter retrospective study of adult pts with B-NHL with active or history of CNS involvement treated with commercial CD3xCD20 BsAb at 5 US medical centers as part of the Collaborative US Bispecific Consortium (CUBIC). Responses were determined using PET/CT for systemic involvement and brain MRI and/or CSF analysis for CNS involvement. Progression-free survival (PFS) was calculated from BsAb initiation to disease progression (PD) or death, and overall survival (OS) from BsAb initiation to death. Results: We included 20 pts who received BsAb (glofitamab n=15, epcoritamab n=3, mosunetuzumab n=2) from 9/2023 to 11/2024. Twelve pts received BsAb alone and 8 combined with systemic therapy (acalabrutinib/zanubrutinib n=4, polatuzumab n=3) and/or radiation (RT, n=4). Median age was 68 years, 55% were male, and 45% had ECOG ≥2. Twelve pts had diffuse large B cell lymphoma (DLBCL), 3 primary CNS lymphoma (PCNSL), 2 high-grade B cell lymphoma (HGBL), 2 mantle cell lymphoma (MCL), and 1 Burkitt (BL). At BsAb initiation, 60% (n=12) of pts had CNS disease: 92% (n=11) with parenchymal disease and 33% (n=4) with leptomeningeal disease (3 had both). Three pts (25%) had isolated CNS disease and 75% (n=9) had concurrent systemic disease. Eight pts (40%) had a history of CNS disease without active involvement at time of BsAb initiation. Pts received a median of 2 (range, 2-8) prior lines of therapy, including chimeric antigen receptor (CAR) T cell therapy in 65% (n=13; median 7 months from CAR T to BsAb), CNS RT in 60% (n=12; median 4 months from RT to BsAb), and intrathecal chemotherapy in 70% (n=14). The best overall response rate (ORR) (CNS + systemic, 1 unevaluable) was 58% (11/19) including complete response (CR) in 37% (7/19) and partial response (PR) in 21% (4/19). The ORR and CR rate were 58%/33% in DLBCL, 67%/67% PCNSL, 50%/50% HGBL, 100%/0% MCL (1 pt evaluable), and 0% BL. ORR and CR rates were similar irrespective of prior CAR T (p=.31, p=.68), choice of BsAb (p=.92, p=.50), or receipt of BsAb alone vs in combination (p=.09, p=.36). In pts with CNS disease at BsAb initiation (n=12), the best ORR (systemic + CNS) was 50% (6/12) including CR in 25% (3/12) and PR in 25% (3/12). The ORR and CR rate were 50% (3/6) and 33% (2/6) with BsAb monotherapy, and 50% (3/6) and 17% (1/6) with BsAb combination. Evaluating CNS-specific responses (n=11, 1 unevaluable), the best CNS ORR was 55% (6/11) including CR in 27% (3/11) and PR in 27% (3/11). Median time to first CNS response was 53 days (range, 19-113), with 83% of responding pts achieving response within 60 days. In pts without active CNS disease, (n=7; 1 unevaluable), the best ORR was 71% (5/7) including CR in 57% (4/7) and PR in 14% (1/7). The ORR and CR rate were 80% (4/5) and 60% (3/5) with BsAb monotherapy, and 50% (1/2) and 50% (1/2) with BsAb combination. With median follow-up of 7 months (95% confidence interval (CI), 5-not reached), among the whole cohort, the 6-month PFS and OS were 39% (95% CI, 21-71) and 75% (95%CI, 58-97), respectively. The 6-month PFS and OS were similar with BsAb monotherapy (38% [95% CI, 18-81] and 75% [95% CI, 54-100]) compared with BsAb combination (42% [16-100%] [p=.99] and 75% [50-100%] [p=.91]), respectively. PFS and OS were similar irrespective of lymphoma subtype (PFS, p=.61; OS, p=.78), prior CAR T (PFS, p=.34; OS, p=.72), or BsAb agent (PFS, p=.89; OS, p=.62). Among pts with CNS disease at BsAb initiation, the 6-month CNS PFS was 65% (95% CI, 38-100%). No patients achieving CNS CR had CNS relapse at last follow-up. Among pts without active CNS disease at BsAb initiation, only 1 pt had CNS relapse. Any grade (G) CRS occurred in 35% (n=7, all G1-2). Any G ICANS occurred in 15% (n=3; 2 G1, and 1 suspected G4). Conclusion: Our study shows significant clinical activity for the CD3xCD20 BsAb, alone and in combination, in B-NHL with CNS involvement (CNS-specific ORR and CR rate of 55% and 27%, respectively) without new safety concerns. Larger studies, ideally prospective, are needed to confirm the efficacy and safety of CD3xCD20 BsAb in pts with CNS involvement by B-NHL.