Prognostic value of immuno-nutritional status in patients with cancer treated by immune checkpoint inhibitors.

226 Background: The treatment efficacy of immune checkpoint inhibitors (ICIs) may relate to tumor factors or host factors, such as systemic inflammation status or nutritional status. In this study, we investigated the prognostic value of immuno-nutritional status in cancer patients treated by ICIs, and develop nomogram models for predicting overall survival (OS). Methods: We used Chang Gung Research Database to retrospectively evaluate cancer patients who received ICIs during January 2015 to December 2021. Patients had double cancer or received only one cycle of ICI were excluded. All patients were divided into a training cohort and a validation cohort. The training cohort was used to analyze the risk factors and develop nomogram models. The models were validated by the validation cohort. The univariate analysis, stepwise multivariate analysis, receiver operator characteristics (ROC) analysis were used to find out the independent risk factors. The calibration plot were used to evaluate the reliability of the models. Results: All patients (3219 cases) could be categorized into lung cancer (22.1%), genitourinary cancer (14.9%), upper gastrointestinal tract cancer (8.1%), colorectal cancer (3.1%), hepatocellular carcinoma (24.6%), pancreatobiliary cancer (3.8%), head and neck cancer (12.9%), melanoma (4.0%), breast cancer (1.9%) and lymphoma (0.5%). 80.2% of patients received anti-PD1, 19.8% of patients received anti-PD-L1. In training cohort, the cut-off value was set for neutrophil-to-lymphocyte ratio (NLR), total lymphocyte count (TLC), lymphocyte-to-monocyte ratio (LMR), platelet-to-lymphocyte ratio (PLR), systemic inflammation index (SII), prognostic nutrition index (PNI) and hemoglobin-albumin-lymphocyte-platelet (HALP) score. The above risk factors showed significant difference in median OS in validation cohort respectively (Table). The multivariate analysis demonstrated that NLR, PNI and HALP score were independent risk factors in predicting OS. The hazard ratio were 1,439 (95% CI 1.149-1.801), 0.689 (95% CI 0.549-0.865) and 0.784 (95% CI 0.626-0.982) respectively. Conclusions: We developed nomogram models to investigate the immuno-nutritional status and predict the OS of cancer patients who received ICIs therapy. High NLR, low PNI and low HALP score were associated with worse OS. [Table: see text]