Eosinophilia during treatment of immune checkpoint inhibitors (ICIs) to predict succeeding onset of immune-related adverse events (irAEs).

Abstract

e14110 Background: ICIs such as anti-PD-1/-L1 Inhibitor and anti-CTLA-4 Inhibitor have been used in many types of cancer, and irAEs occur in a considerable proportion of patients. Predictive biomarkers could help us to manage irAEs promptly. Methods: We retrospectively analyzed the consecutive patients treated with nivolumab, pembrolizumab, atezolizumab or ipilimumab, not used as investigational drugs, between March 2015 and December 2018 based on the electric medical records in our institution. We evaluated the relationship between clinical and laboratory factors (gender, age, body mass index, lymphocyte-to-monocyte ratio and eosinophilia) and the occurrence of irAEs using univariate and multivariate analyses. Eosinophilia was defined as absolute eosinophil count over 500/mm3 at least once between initiation of treatment and the occurrence of irAEs. We graded irAEs according to CTCAE v. 5.0. Results: Among 139 patients analyzed in this study, 136, 13 and 6 patients were treated with PD-1, PD-L1 and CTLA-4 inhibitors, respectively (cumulative total of 155 patients). This study included non-small cell lung cancer (n = 78; 56.1%), renal cell carcinoma (n = 17; 12.2%), melanoma (n = 16; 11.5%), gastric cancer (n = 15; 10.8%), Hodgkin lymphoma (n = 7; 5.0%), urothelial carcinoma (n = 3; 2.2%), head and neck carcinoma (n = 3; 2.2%) and mesothelioma (n = 1; 0.7%). Any grade and grade 3-5 irAEs occurred in 60 (38.7%) and 23 (14.8%) cumulative patients, respectively. The most common irAEs in any grade were endocrine disorders including thyroid dysfunction and adrenal insufficiency (n = 21) and skin disorders (n = 20). Eosinophilia occurred in 43 cumulative patients (27.7%). Eighteen patients out of them had eosinophilia from baseline. Among the clinical and laboratory factors, only eosinophilia was significantly associated with the occurrence of any grade of irAEs in both univariate (p = 0.0193) and multivariate (p = 0.0464) analysis. Notably, endocrine irAEs were significantly related to eosinophilia (p = 0.0287). Conclusions: Eosinophilia after the initiation of treatment with ICIs predicts succeeding onset of irAEs.