CAR-T 细胞治疗的真实世界皮肤不良事件：FDA 不良事件报告系统十年期不成比例分析

Background: Chimeric antigen receptor (CAR) T-cell therapy has transformed haematological-malignancy care, but its dermatologic safety profile remains incompletely characterised across products and indications. Methods: We analysed 8,431,841 deduplicated FDA Adverse Event Reporting System reports from 2016 Q3 to 2026 Q1. Seven approved CAR T-cell products were identified. The primary outcome was any dermatologic adverse event, defined using the MedDRA Skin and subcutaneous tissue disorders system organ class. Secondary outcomes included broad severe cutaneous adverse reactions, narrow Stevens-Johnson syndrome/toxic epidermal necrolysis, and 14 phenotype-specific categories. Multivariable models adjusted for demographics, polypharmacy, cancer, immune checkpoint inhibitor exposure, lymphodepleting chemotherapy and cytokine release syndrome. Additional sensitivity analyses evaluated HSCT/GVHD co-reporting proxies, infection and cytopenia/bleeding proxies, severe-event clinical characteristics, comparator robustness and multiplicity correction. Results: Dermatologic adverse events were identified in 996,654 reports, including 425 CAR-T-associated cases.

CAR T-cell exposure was associated with reduced adjusted reporting odds for the primary outcome (adjusted odds ratio 0. 13, 95% confidence interval 0. 09-0. 20) and broad severe cutaneous adverse reactions (0. 35, 0. 23-0. 52). The primary SKIN_ANY reduced-reporting pattern was consistent across all-FAERS, haematological-malignancy and active haematology-oncology comparators. HSCT/GVHD proxy co-reporting was uncommon and did not materially alter estimates. Severe dermatologic reports frequently co-mentioned CRS and serious outcomes. The tisagenlecleucel vascular cutaneous signal was nominally significant but attenuated after excluding infection-attributable and cytopenia/bleeding-proxy reports.

Conclusions: Within spontaneous reporting systems, CAR T-cell therapy showed reduced relative reporting of dermatologic adverse events across broad, phenotype-specific and product-level analyses. These results should be interpreted as differences in reporting behaviour, not as evidence of reduced true clinical incidence or lower patient-level risk.

Early severe cutaneous reports frequently overlapped with cytokine release syndrome, while infection, cytopenia/bleeding proxies and supportive-care drugs were important alternative explanations for selected cutaneous signals.