CT-Based Radiomics Combined with Genomics may Predict Survival in Esophageal Cancer Patients Receiving Definitive Chemoradiotherapy.

Abstract

Definitive chemoradiotherapy (dCRT) is a standard treatment option for locally advanced stage inoperable esophageal squamous cell carcinoma (ESCC). Evaluating clinical outcome prior to dCRT remains challenging. The purpose of this study was to evaluate the predictive power of contrast-enhanced computed tomography (CT)-based radiomics in combination with genomics for the treatment efficacy of ESCC patients after dCRT. A total of 118 ESCC patients who received dCRT were enrolled in this retrospective study. These patients were randomly divided into the training group (N = 82) or the validation group (N = 36). Radiomic features were derived from the region of primary tumor on pretreatment CT images, also known as the region of interest (ROI), and clinical features were generated from medical records. The least absolute shrinkage and selection operator (LASSO) regression was conducted to select optimal radiomic features, and Rad-score was calculated to predict progression-free survival (PFS) in the training group. Genomic DNA was extracted from formalin-fixed and paraffin-embedded pre-treatment biopsy tissue. The univariate and multivariate COX analyses were undertaken to determine the predictors for developing models. The area under receiver operating characteristic curve (AUC) and the C-index were used to evaluate the predicting performance and the discriminating ability of prediction models, respectively. A total of 851 radiomic features were extracted from each CT image. The final Rad-score were constructed from 6 radiomic features to predict PFS. Multivariate analysis demonstrated that Rad-score and HRR pathway alterations were independent prognostic factors correlated with PFS. The C-index of integration model in combination with radiomics and genomics was better than that of radiomics or genomics model in the training group (0.616 vs 0.587 or 0.557) and the validation group (0.649 vs 0.625 or 0.586). Kaplan-Meier survival analysis also showed significant differences between different risk subgroups in the training and validation groups. The Rad-score based on pre-treatment CT and HRR pathway alterations could predict PFS for patients with ESCC after dCRT, with better predictive efficacy of combined radiomics and genomics models.