Teclistamab and glofitamab carry higher non-relapse mortality signals than mosunetuzumab in BsAb meta-analysisNew data reveals safety risks for certain lymphoma and myeloma drugs

Bispecific antibodies (BsAbs) redirect T-cells to tumor antigens in an MHC-independent manner. These universally available, off-the-shelf products have demonstrated unprecedented efficacy in heavily pretreated patients with relapsed/refractory (R/R) hematological malignancies (HMs). The indefinite treatment duration associated with most BsAb products, coupled with neutropenia, B-cell aplasia, and hypogammaglobulinemia result in significant immune-related toxicities and infections. Hence, non-relapse mortality (NRM) which represents the cumulative burden of treatment-related toxicities, defined as death from any cause other than disease relapse or progression, is a patient-centric endpoint across cellular therapy platforms. While recent NRM estimates with CAR T-cell therapy have identified infections as major drivers of NRM, real-world NRM estimates remain limited with BsAbs. To mitigate the limitations associated with each of the methods, we conducted a systematic review and meta-analysis of the existing literature, supplemented concurrently by a pharmacovigilance analysis using the FAERS database, to estimate NRM with BsAbs and compare real-world findings between commercially available BsAb products approved for relapsed/refractory multiple myeloma (RRMM) and NHL. The primary endpoint was NRM; secondary endpoints included CRS, ICANS, infections, and death. Teclistamab demonstrated a high toxicity burden across both analytic methods, with the highest ICANS incidences, significantly greater infection rates (particularly grade > 3), and a non-negligible NRM estimate in the SRMA as well in FAERS. Glofitamab was associated with severe toxicities, including increased reporting odds for NRM and overall mortality. Mosunetuzumab showed lower pooled estimates and lower reporting odds, as seen in both analytic approaches. Low pooled incidences of infections, and ICANS were calculated in the SRMA. FAERS analyses also suggested reduced reporting odds for NRM and neurotoxicity. While trial-level data and real-world reporting largely converged for severe toxicities, divergences in other agents underscore the importance of continuous post-marketing surveillance. Given the risks associated with prolonged treatment, future protocols must prioritize infection management and adopt NRM as a standardized endpoint to guide appropriate therapeutic selection and sequencing.