Contrast-Enhanced CT Imaging Features Combined with Blood Inflammatory Factors to Predict the Efficacy and Prognosis for Immunotherapy of NSCLC

Abstract Background Immunotherapy has shown significant clinical efficacy in patients with non-small-cell lung cancer (NSCLC) and has been utilized in various therapeutic regimens. However, it is still difficult to identify patients who might benefit from the treatment, which would limit the use of immunotherapy. Therefore, it is extremely important to find predictors for immunotherapy. Purpose This study aimed to develop clinical benefit prediction models for NSCLC patients undergoing immunotherapy via the most routine and affordable medical tests and diagnostic procedures. Methods A total of 112 patients with advanced NSCLC receiving immunotherapy were retrospectively included. Prior to immunotherapy, the lesion’s CT radiomics features were extracted. A radiomics signature for short-term effectiveness or overall survival (OS) prediction was developed using the least absolute shrinkage and selection operator (LASSO) method. Additionally, the prognostic performance of the clinical factors was evaluated. Finally, a nomogram for short-term effectiveness prediction was established using multivariate logistic regression, and a nomogram for OS prediction was built using multivariate Cox regression. Results For short-term effectiveness prediction, the area under the receiver operating characteristic curve values (AUCs) for the radiomics signature in the training and validation cohorts were 0.740 and 0.708, respectively. While that for the nomogram model were 0.793 and 0.740, respectively. As for OS prediction, the Harrell’s concordance indexes (C-index) of the radiomics signature in the training and validation cohorts were 0.750 and 0.831, respectively. While that of the nomogram model were 0.789 and 0.862, respectively. Conclusions CT radiomics signature integrated with clinical factors, could be used in the identification of patients likely to benefit from immunotherapy. Models developed in this study might be promising in facilitating more precise medicine and tailored treatment for advanced NSCLC patients.