The clinical spectrum and causal relationship assessment of checkpoint inhibitor‐associated autoimmune diabetes mellitus (CIADM): A retrospective observational study

AIMS: This study aims to characterize the clinical spectrum, elucidate the temporal relationship and assess the causality of checkpoint inhibitor-associated autoimmune diabetes mellitus (CIADM) in a cohort of patients receiving anti-programmed cell death-1 (PD-1)/anti-programmed death ligand-1 (PD-L1) inhibitor therapy. METHODS: Individuals who developed CIADM were identified among 4382 hospitalized patients receiving PD-1/PD-L1 inhibitor therapy between 2020 and 2024. Demographic characteristics, oncological history, immune checkpoint inhibitor regimens, laboratory parameters (including glycemic control, C-peptide and islet autoantibodies) and clinical outcomes were systematically collected. Causality was evaluated using the Naranjo algorithm. RESULTS: Among 4382 patients who received PD-1/PD-L1 inhibitors, seven (0.16%) were diagnosed with CIADM during the study period. Seven patients were predominantly male, with a median age of 52 years (range: 44-62), and hepatocellular carcinoma was the most common primary malignancy (3/7). All patients developed CIADM after a median of 8 cycles (range: 5-23) of PD-1/PD-L1 inhibitors (sintilimab, camrelizumab, envafolimab or atezolizumab). Notably, 86% (6/7) of patients presented with diabetic ketoacidosis (DKA) at onset. Severe pancreatic β-cell dysfunction was universal, with undetectable fasting and postprandial C-peptide levels. Islet autoantibodies were positive in only one patient. The Naranjo score indicated a 'probable' to 'highly probable' causal relationship in all cases. All patients required lifelong insulin therapy. Following glycemic stabilization, immunotherapy was successfully reinitiated in six patients without exacerbating hyperglycemia. CONCLUSION: CIADM is a severe immune-related adverse event (irAE) characterized by abrupt onset, a high rate of DKA, and profound insulin deficiency, often in the absence of conventional islet autoantibodies. A definitive causal link exists between PD-1/PD-L1 inhibitors and CIADM. Our findings underscore the necessity for proactive glycemic monitoring in patients undergoing ICI therapy, especially in populations such as middle-aged males with cancers known to have specific demographic profiles (e.g., hepatocellular carcinoma). Importantly, the resumption of immunotherapy is feasible with careful management, highlighting the critical role of a multidisciplinary approach to optimize oncological and metabolic outcomes.