Construction and evaluation of clinical prediction model for immunotherapy-related adverse events and clinical benefit in cancer patients receiving immune checkpoint inhibitors based on serum cytokine levels

Abstract

Abstract Background: Immune checkpoint inhibitors (ICIs) have revolutionized the therapeutic landscape of cancer. The aim of this study was to develop novel risk classifiers to predict the risk of irAEs and probability of clinical benefits of these individuals. Methods: The cancer patients received ICIs from the First Affiliated Hospital of Xi 'an Jiaotong University from November 2020 to October 2022 were collected and followed up. The logistic regression analyses were adopted to identify independent predictive factors of irAEs and clinical response. Two nomograms were developed to predict the irAEs and clinical response of these individuals, with receiver operating characteristic curve (ROC) and calibration curve being generated to assess their predictive ability. Besides, decision curve analysis (DCA) was performed to estimate the clinical utility of the nomograms. Results: This study included 583 cancer patients from 2434 cancer patients. Among them, 111 patients (19.0%) developed irAEs. The multivariate analysis indicated that duration of treatment (DOT)＞3 cycles, Hepatic-metastases, IL2＞2.225pg/ml, and IL8＞7.39pg/ml were correlated with higher irAEs risk. Overall, 347 patients were included in the final efficacy analysis, with an overall clinical benefit rate of 39.7% being observed. The multivariate analysis indicated that DOT＞3cycles, non-hepatic-metastases, irAEs and IL8＞7.39pg/ml were independent predictive factors of clinical benefit. Ultimately, two nomograms were successfully established to predict the probability of irAEs and clinical benefits. ROC curves yield acceptable performance of nomograms. Calibration curves showed satisfying consistencies between actual and predicted probability. DCA supported that the nomograms could provide more significant net clinical benefits to these patients. Conclusion: Specific baseline serum cytokines are closely correlated to irAEs and clinical response in these individuals. We established two nomograms that could effectively predict the risk of irAEs and probability of clinical response by integration of common clinicopathological parameters and serumcytokines.