Safety evaluation of CD3×CD20 bispecific antibodies: a pharmacovigilance study using FDA adverse event reporting...

To characterise post-marketing adverse event (AE) reporting patterns associated with CD3×CD20 bispecific antibodies (BsAbs) using multi-method disproportionality analysis and to identify potential safety signals warranting further investigation.

Retrospective pharmacovigilance study using disproportionality analysis with multiple complementary methods.

Food and Drug Administration Adverse Event Reporting System database.

A total of 2237 AE reports associated with CD3×CD20 BsAbs (1135 for epcoritamab, 673 for glofitamab and 429 for mosunetuzumab) were submitted from the fourth quarter of 2022 to the first quarter of 2025.

Primary outcomes included identification of statistically significant AE signals using reporting odds ratio (ROR) and Bayesian confidence propagation neural network with information component as the principal signal detection methods, while proportional reporting ratio was only used as a supplementary measure to enhance the robustness of signal detection. Secondary outcomes included subgroup analyses by sex and age, analysis of fatal outcomes and time-to-onset patterns using Kaplan-Meier analysis and Weibull distribution.

A total of 93 AE signals were identified for epcoritamab, 65 for glofitamab and 41 for mosunetuzumab. The three disproportionality methods showed near-complete agreement (>98% concordance) in signal detection. Signals were detected for several AEs not currently listed in product labelling. For epcoritamab, unlabelled signals included progressive multifocal leukoencephalopathy (ROR (95% CI) 13.28 (5.51 to 31.97)), haemophagocytic lymphohistiocytosis (ROR (95% CI) 18.19 (10.75 to 30.79)) and disseminated intravascular coagulation (ROR (95% CI) 8.76 (3.28 to 23.39)). For glofitamab, unlabelled signals included disseminated tuberculosis (ROR (95% CI) 53.30 (17.10 to 166.13)) and cerebral haemorrhage (ROR (95% CI) 6.08 (2.28 to 16.22)). For mosunetuzumab, tumour lysis syndrome (ROR (95% CI) 29.91 (11.19 to 79.97)) and uveitis (ROR (95% CI) 10.75 (3.46 to 33.41)) were detected. With respect to patient characteristics, cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome were more frequently reported in younger patients treated with epcoritamab, while pyrexia was more commonly reported among elderly patients receiving glofitamab or mosunetuzumab. Among reports with available dates, Kaplan-Meier analysis showed that most reported onset times fell within the first month following treatment initiation. In the indication-restricted sensitivity analysis, most signals remained detectable, whereas certain signals, including haemophagocytic lymphohistiocytosis with epcoritamab, were sensitive to comparator selection.

This study provides a systematic characterisation of post-marketing AE reporting patterns for CD3×CD20 BsAbs, identifying several signals for AEs not currently listed in product labelling. However, disproportionality signals do not establish causality, and clinical correlation is essential. These findings may help inform post-marketing pharmacovigilance and prioritise signals for further clinical and epidemiological validation.