Evaluating the safety of Janus kinase inhibitors in the management of severe immune-related adverse events

Immune checkpoint inhibitors (ICIs) have revolutionized the treatment of cancer. However, ICIs are associated with potentially life-threatening immune-related adverse events (irAEs). Janus kinase inhibitors (JAKi) are a promising steroid-sparing agent for treating corticosteroid-refractory severe irAEs. However, little is known about their safety in the ICI population, limiting clinical adoption due to concerns of reduced anticancer immunity and ICI response. In this retrospective multi-institutional cohort study, we aim to evaluate the safety of JAKi in the management of refractory irAEs requiring escalation beyond the standard of care with systemic corticosteroid therapy. Patients who received both ICI and JAKi at Mass General Brigham Healthcare System and the Dana-Farber Cancer Institute between 2018 and 2025 were manually reviewed for irAE presence, severity, treatment response, and survival (n=33). A comparator cohort of patients with irAEs treated with systemic corticosteroids (n=99). JAKi-treated patients were matched 1:4 to controls using nearest-neighbor matching with a 0.4 caliper based on age, cancer type, number of organs affected by ICI toxicity, and ICI regimen, yielding a matched cohort of 27 cases and 83 controls. Associations between JAKi use, progression-free survival, and overall survival were evaluated with multivariable Cox regression with landmark analyses to account for immortal time bias. To address limited power for survival analyses, we performed an external validation using the TriNetX research network, generating a 1:1 matched multi-institutional cohort of 297 JAKi-treated and 297 non-JAKi-treated irAE patients. Kaplan-Meier analysis showed no statistically significant difference in overall survival (median 34 vs 45 months) or progression-free survival (median 5.6 vs 9.6 months) in JAKi-treated versus control groups. These findings remained consistent with multivariable Cox proportional hazard models. Across all landmark times, the association between JAKi use and survival outcomes remained similar between the two treatment groups. Similarly, in the larger TriNetX cohort, JAKi therapy was not associated with inferior overall survival compared with matched non-JAKi-treated patients. In these complementary multi-institutional cohorts, no large adverse signal was observed with JAKi versus standard-of-care systemic corticosteroid therapy for severe irAEs in this real-world cohort. These results merit further investigation in prospective studies.