Immune checkpoint inhibitor–induced diabetes mellitus across NCI trials.

2668 Background: Immune checkpoint inhibitors (CPI) are known to rarely cause new onset diabetes (CPI DM). While rare, this adverse event is quite challenging for patients and clinicians to manage. Therefore, it is important to identify risk factors and clinical characteristics. This is the first comprehensive, multi-institutional study of CPI-DM across multiple agents and cancer types. Methods: The NCI Cancer Therapy Program (CTEP) database of adverse events (AEs) was queried for AEs related to diabetes (Grade 3 or 4 hyperglycemia, Acidosis including Diabetic Ketoacidosis (DKA), glucose intolerance and diabetes mellitus) among 6,925 patients who had been treated with a PD-(L)1 inhibitor alone or in combination from 6/2015 to 12/2019. Each AE report was reviewed and classified as due to CPI DM, new onset type 2 diabetes mellitus (T2DM), T2DM exacerbation without medication non-compliance, existing DM with medication non-compliance, or association with steroids (SDM). CPI DM was diagnosed based on: evidence of insulin deficiency either through presentation in DKA or low c-peptide with need for long term basal bolus insulin to maintain euglycemia and/or positive islet autoantibodies. Results: In total, there were 82 cases with at least one of these AEs; 41 had CPI-DM, 22 had SDM, 1 had new T2DM, 4 had T2DM exacerbation, 3 had medication noncompliance and 11 had acidosis not attributable to diabetes or had insufficient data. After excluding non-hyperglycemic acidosis, 57.8% had CPI-DM. Furthermore, if not on steroids and in good compliance with diabetes medications, 89.1% had CPI-DM. The incidence of CPI-DM was 0.59%; it was most common on combination PD-1/CTLA-4 inhibitor therapy (0.85%, 15/1767), followed by PD-(L)1 inhibitor monotherapy (0.54%, 18/3354), followed by CPIs combined with additional agents including chemotherapy and targeted agents (0.44%, 8/1804)(p = 0.25). Hospitalization was required for 87.5% of CPI-DM cases with 74.3% of those requiring an inpatient endocrine consult. All but one CPI-DM case had an endocrine consult at as either an inpatient or outpatient. Conclusions: While rare, this cohort shows the large health care burden of CPI-DM and that any hyperglycemia, and especially marked hyperglycemia, should be treated as CPI-DM until proven otherwise. [Table: see text]