Immune-effector and infection-associated reporting patterns of CD20×CD3 bispecific antibodies in B-cell lymphoma: a label-anchored FAERS analysis with descriptive JADER contextualization.

Background: CD20×CD3 bispecific antibodies are associated with acute immune-effector toxicities and infections, but drug-specific postmarketing co-reporting and report-level onset patterns remain incompletely characterized in B-cell lymphoma. Methods: We performed a label-anchored FAERS pharmacovigilance analysis with descriptive JADER contextualization. Deduplicated FAERS reports from 2022 Q4 through 2026 Q1 were restricted to primary-suspect epcoritamab, glofitamab, or mosunetuzumab reports in B-cell lymphoma, yielding 2,879 exposed reports. Prespecified MedDRA endpoints were evaluated against a disease-restricted comparator. Primary adjusted reporting odds ratios (aRORs) accounted for demographics, disease subgroup, reporting region, and calendar year. We mapped report-level phenotype intersections, summarized time to onset (TTO), and performed model-specification, restriction, comparator-exclusion, and TTO-completeness analyses. Results: The cohort included 1,463 epcoritamab, 944 glofitamab, and 472 mosunetuzumab reports. CRS had higher primary adjusted reporting odds for epcoritamab (aROR 4.70, 95% CI 4.15 to 5.32), glofitamab (3.43, 2.94 to 4.00), and mosunetuzumab (6.23, 4.97 to 7.81). ICANS had higher reporting odds for epcoritamab (2.28, 1.87 to 2.80); the glofitamab estimate crossed 1 in the primary model (1.26, 0.93 to 1.70) and varied with covariate specification, while mosunetuzumab ICANS was sparse. Epcoritamab had higher reporting odds for CMV (4.54, 3.66 to 5.62) and narrow opportunistic infection (2.20, 1.84 to 2.63). Mosunetuzumab had higher reporting odds for pneumonia/lower respiratory infection (1.74, 1.32 to 2.31) and COVID-19 (1.66, 1.25 to 2.21), although these infection-specific patterns attenuated in some geographic and calendar restrictions. Immune-effector and infectious phenotypes co-occurred in reports containing serious outcomes. TTO was calculable for 33.7%-61.2% of reports and suggested earlier CRS than infection-associated reporting, but report characteristics differed by TTO completeness. Conclusions: CRS was the clearest shared reporting pattern across the three CD20×CD3 antibodies. Neurotoxicity and infection-associated findings were more drug- and analysis-dependent. Report-level co-reporting supports concurrent consideration of inflammatory toxicity and infection when presentations overlap, while later infection-associated TTO distributions support continued awareness beyond step-up dosing. Spontaneous-reporting data cannot estimate incidence, causality, or comparative treatment risk.