A CT-based Radiomics Model to Predict Survival-graded Fibrosis in PDAC

This retrospective, bicentric study included 295 pretreated patients with PDAC. Tumor fibrosis was assessed using the collagen fraction (CF). Clinical-pathological variables were gathered, and radiological features were evaluated by three radiologists in consensus. The patients were followed up at 1, 3, 6, and 9 months postoperatively and every 3-6 months thereafter. All follow-up examinations included carbohydrate antigen 19-9 (CA 19-9) measurements and imaging (contrast-enhanced CT, contrast-enhanced MRI, ultrasonography, or positron emission tomography). The overall survival (OS) and disease-free survival (DFS) were also recorded. Cox regression analysis was used to evaluate the associations of CF with OS and DFS. Receiver operating characteristic (ROC) analyses were used to determine the rounded threshold of CF. An integrated model (IM) was developed by incorporating selected radiomic features and clinical-radiological characteristics. The predictive performance was validated in the test cohort (Center 2). It was hypothesized that tumor fibrosis could be classified into two survival-graded groups and that the incorporation of radiomics features and clinical-radiological features would help predict the status of fibrosis. Thus, the association between fibrosis and overall survival (OS)/disease-free survival (DFS) was determined, and the patients with PDAC were divided into high- and low-CF groups in this study. In addition, a CT-based radiomics model was developed and validated to non-invasively predict fibrosis before treatment in patients with PDAC from two centers. It was hypothesized that the performance of the integrated model would be superior to that of the clinical-radiological model.