Endocrine adverse events following CAR-T cell therapies: A pharmacovigilance analysis in the real-world setting utilizing the faers database

Abstract Background: CAR-T cell therapies have transformed treatment for hematologic malignancies. While hematologic and cytokine-related toxicities are well described, real-world data on endocrine adverse events (AEs) remain sparse. Given expanding clinical use, better characterization of immune-mediated endocrine toxicities is needed. Methods: We conducted a retrospective pharmacovigilance study using deduplicated FAERS data from 2017 to 2025. Reports were limited to primary suspect (PS) cases. Endocrine AEs were defined using exact MedDRA preferred terms across Pituitary, Thyroid, Adrenal, Diabetes, and calcium-parathyroid axes. CAR-T products analyzed included axicabtagene ciloleucel, tisagenlecleucel, brexucabtagene autoleucel, ciltacabtagene autoleucel, idecabtagene vicleucel, and lisocabtagene maraleucel. Signal detection employed both Bayesian disproportionality analysis using the Information Component (IC) with 95% credibility intervals and Hierarchical Bayesian Models (HBMs) to account for low event counts and data sparsity. A signal was considered significant when IC025 > 0. Traditional measures like Reporting Odds Ratios (RORs) were not used due to their limited utility in rare-event settings. All modeling was performed in Python using PyMC and the R software, an open-source probabilistic programming framework. Results: Among 55,720 Adverse Events for CAR-T therapies, 56 (0.1%) endocrine AEs were reported with CAR-T therapies. Axicabtagene ciloleucel contributed the most endocrine events (n=30) and showed the strongest signal for endocrine toxicities (IC=7.95 [IC025: 2.31, IC975: 13.6]), followed by tisagenlecleucel (n=12 IC=7.04 [0.53–13.55])). Both agents showed elevated IC estimates, though credibility intervals were wide (Axicabtagene: IC=8.14, IC025=–2.86; Tisagenlecleucel: IC=6.96, IC025=–5.73). HBMs supported these as the strongest signal clusters, particularly for pituitary and thyroid-related events. The most frequent AEs included hypophysitis or hypopituitarism (n=15), hypothyroidism (n=12), adrenal insufficiency (n=8), and diabetes-related events including diabetic ketoacidosis (n=5). Axicabtagene ciloleucel was most associated with hypopituitarism (n=9), while DKA was reported with tisagenlecleucel and brexucabtagene autoleucel. Thyroiditis and hypothyroidism occurred across multiple agents, suggesting a delayed autoimmune thyroid pattern. Rare events such as hypocalcemia and Cushing syndrome appeared as isolated reports. While many IC025 values did not meet the significance threshold due to low power, HBMs revealed signal enrichment even in sparse subgroups. Conclusions: This FAERS-based real-world analysis identifies relevant endocrine toxicities linked to CAR-T therapies, with strongest associations observed for axicabtagene ciloleucel and tisagenlecleucel. Immune-mediated dysfunction in the pituitary and thyroid axes appears most prominent. These findings support the need for increased endocrine monitoring post-CAR-T and suggest a class-wide effect with further prospective study.