Diabetes therapy burden as proxy of impairment of immune checkpoint inhibitors efficacy

Background: Chronic hyperglycemia is known to induce immune dysfunctions and multiple studies has identified resistance/adherence to insulin therapy as barriers to achieving an optimal glycaemic control in patients with diabetes mellitus (DM) after front line metformin failure. How DM affects the efficacy of immune checkpoint inhibitors (ICI) is yet to be defined. Methods: We evaluated the impact of DM in a retrospective cohort of patients with advanced cancer treated with ICI at 22 Institutions. The diabetes medication burden (DMB) at ICI initiation was used as a proxy of long-term/poorly controlled DM. Patients with a high DMB were on high dose metformin (> 1000 mg) either alone or in combination with insulin therapy and/or other oral antidiabetics. Patients with low DMB were on oral antidiabetics/insulin with or without low dose metformin. In a subgroup of 133 patients the inter-relationships between the median baseline glycemia (MBG) (computed upon up to 3 fasting blood glucose tests within 3 months of ICI initiation) and the neutrophil-to-lymphocyte ratio (NLR) were assessed. Results: From June 2014 to November 2020, 1395 patients treated with CTLA-4 (2.5%) and PD-1/PD-L1 (97.5%) inhibitors were included. Median age was 68 years; male/female ratio 888/507. Primary tumours were: NSCLC (54.7%), melanoma (24.7%), renal cell carcinoma (15.0%) and others (5.6%). 148 (10.6%) and 78 (5.6%) were on low and high DMB, respectively. The DMB was proportionally associated to both an increasing MBG (5.6, 7.5 and 9.5 mmol/L) and median NLR (3.8, 4.1 and 5.6). After adjusting for gender, age, BMI, primary tumour, treatment line, burden of disease, performance status and corticosteroids, patients on high DMB were confirmed to have shorter progression free survival (PFS) (HR 1.39[95%CI: 1.09-1.78] p¼0.0075) and overall survival (OS) (HR 1.44 [95%CI:1.09-1.90] p ¼ 0.0087) as compared to non- DM. MBG significantly predicted for the NLR [F(1,131) ¼ 4.09, p ¼ 0.04), R2 of .030], and was associated to a high NLR (  4) even after adjusting for corticosteroids (OR ¼ 1.68[95%CI: 1.23 e 2.29], p ¼ 0.0011). Conclusions: Long-term/poorly controlled diabetes may impair ICI efficacy. Strategies to improve glycaemic control should be pursued in patients about to start an immunotherapy.