Abstract
BACKGROUND: The classical prognostic indicators of lung cancer are no longer sufficient for prognostic stratification and individualized treatment of highly heterogeneous non-small cell lung cancer (NSCLC). This study aimed to establish a radiomics model to predict CD74 expression level in NSCLC patients and to explore its role in the tumor immune response and its prognostic value. METHODS: The prediction model was developed based on 122 NSCLC transcriptome samples, including 68 paired enhanced CT and transcriptome samples. Survival analysis, gene set variation analysis, and immune cell infiltration analysis were used to investigate the relationship between CD74 expression and tumor immune response. Logistic regression (LG) and support vector machine (SVM) analysis were used to construct the prediction model. The performance of the model was assessed with respect to its calibration, discrimination, and clinical usefulness. RESULTS: High CD74 expression is an independent prognostic factor for NSCLC and is positively correlated with antigen presentation and processing gene expression and antitumor immune cell infiltration. The radiomics prediction models for CD74 expression demonstrated good predictive performance. The areas under the receiver operating characteristic curves for the LG and SVM radiomics models were 0.778 and 0.729, respectively, in the training set and 0.772 and 0.701, respectively, in the validation set. The calibration and decision curve analysis curves demonstrated good fit and clinical benefit. CONCLUSION: CD74 expression significantly impacts the prognosis of NSCLC patients. The radiomics model based on contrast-enhanced CT exhibits good performance and clinical practicability in predicting CD74 expression.