Immune-related adverse events, survival outcomes, and next steps in patients with stage III melanoma: Real-world experience at an NCI comprehensive cancer center

Abstract Purpose Immune checkpoint inhibitors (ICIs) improve recurrence outcomes in stage III melanoma but can cause immune-related adverse events (irAEs) of varying severity. While some studies suggest irAEs correlate with improved survival, findings remain inconsistent. Optimal management of recurrent stage III melanoma after adjuvant ICI therapy is also unclear. Here, we evaluate the impact of irAEs on survival and describe subsequent treatments in patients with recurrent stage III melanoma. Patients and Methods We identified 171 patients with stage III melanoma treated with adjuvant ICIs since 2017. Clinical endpoints included overall survival, progression-free survival, distant metastasis-free survival, melanoma-specific survival, and time-to-irAE, which were analyzed using Kaplan-Meier method, Cox model and Fine-Gray’s method. Cox model with time-varying covariate modeling addressed immortal-time bias. Subgroup analysis examined survival by irAE type, grade, and treatment resumption. Results irAEs occurred in 43.86% of patients, with dermatitis (17.64%), colitis (16.67%) and hypothyroidism (15.69%) the most common. irAE presence did not impact survival outcomes. 31 grade 3+ irAEs occurred, which were linked to increased short-term melanoma-specific mortality. Immune-related hepatitis was associated with higher mortality risk in multivariable modeling. Recurrence or progression occurred in 28.65% of patients, with 38.78% located at a distal site. Among those who started additional medical treatment, 42.86% received dual-checkpoint inhibitor therapy and 14.29% enrolled in clinical trials. Conclusion Our findings do not support an association between irAEs and survival in patients with stage III melanoma. However, specific types of events, particularly hepatitis, may increase mortality. Prospective studies are needed to clarify optimal treatment after recurrence.