Development and Validation of a Radiomics-Based Model to Predict Overall Survival after Definitive Chemoradiotherapy in Patients with Esophageal Cancer.

Abstract

This study aimed to evaluate the predictive potential of contrast-enhanced computed tomography (CT)-based radiomics for the treatment outcomes of patients with esophageal squamous cell carcinoma (ESCC) after definitive chemoradiotherapy (dCRT). This retrospective study included 122 locally advanced ESCC patients who received dCRT. Eligible patients were randomly divided into training cohort (n = 85) and validation cohort (n = 37). Least absolute shrinkage and selection operator (LASSO) regression was performed to select optimal radiomic features to calculate Rad-score for predicting overall survival (OS) in the training cohort. Univariate and multivariate analyses were performed to identify the predictive clinical factors and hematologic parameters for developing a nomogram model. The C-index was used to assess the performance of the predictive model and calibration curve was used to evaluate the accuracy. Tenradiomic features were selected by LASSO regression analysis to calculate Rad-score for predicting OS. The patients with Rad-score>0.47 had high risk of death, and those with a Rad-score≤0.47 had low risk of death. Tumor location and neutrophil-to-monocyte ratio (NMR) were significantly associated with OS in univariate analysis. Multivariate analysis showed that NMR and Rad-score were independent predictive factors for OS. A nomogram model was built based on the result of multivariate analysis. The C-index of the nomogram was 0.619 (95% CI 0.518-0.720) in training cohort and 0.573 (95% CI 0.385-0.760) in validation cohort. The 2-year OS rate predicted by the nomogram model was highly consistent with the actual 2-year OS rate both in training and validation cohorts. We developed and validated a prediction model based on radiomic features and hematologic parameters, which could be used to predict OS of ESCC patients after dCRT. This model is conducive to identifying the patients with ESCC benefited more from dCRT.