Objectives: While tumor necrosis factor inhibitors (TNFis) are widely used in pregnancy, assumption of a uniform class-wide safety profile warrants scrutiny. This pharmacovigilance study aimed to explore potential heterogeneity in safety signals among five TNFis using real-world data, generating hypotheses for future validation.
Methods: We analyzed 7022 pregnancy exposure reports from the FDA Adverse Event Reporting System (FAERS) database (January 1, 2004 - June 30, 2025). Employing a multi-phase signal detection framework, we utilized four disproportionality algorithms and indication-stratified modeling to mitigate confounding. Novelty assessment and Weibull modeling were applied to characterize signal strength and temporal risk patterns, acknowledging the inherent limitations of spontaneous reporting.
Results: Analysis revealed distinct reporting patterns across agents. A notable signal for lactation impairment was observed for four TNFis (reporting odds ratios [RORs] 21.3-130.3). Temporal modeling suggested structure-dependent trends: agents with high placental transfer (adalimumab, etanercept) showed increasing reporting rates over time (Weibull shape parameter β > 1), whereas certolizumab pegol (minimal transfer) exhibited a constant risk pattern (β ≈ 1). However, these findings must be interpreted with caution given potential biases such as the Weber effect and lack of denominator data.
Conclusions: Our exploratory analysis suggests potential heterogeneity in the safety profiles of TNFis during pregnancy and lactation, particularly regarding lactation-related events. Rather than confirming causal differences, these data generate important hypotheses linking molecular structure to adverse event patterns. The observed signals underscore the need for prospective studies to validate these findings and support individualized risk assessment.
Keywords: FAERS; Lactation; Pharmacovigilance; Pregnancy; Signal detection; Tumor necrosis factor inhibitors.
Copyright © 2026 Elsevier Ltd. All rights reserved.