RADIOMIC FEATURES OF TUMOR AND OF LIVER-TUMOR INTERFACE IN PATIENTS WITH COLORECTAL LIVER METASTASES. IDENTIFICATION OF NEW PROGNOSTIC BIOMARKERS.

BACKGROUND: Surgery with perioperative chemotherapy offers a potentially curative treatment for colorectal liver metastases (CRLM), but a subset of resectable patients does not achieve long-term benefit. Patient selection relies on survival prediction, but available prognostic factors have limited reliability. PURPOSE: To investigate the potential of preoperative CT-based radiomics for survival prediction in CRLM patients, focusing on the impact of CT-surgery interval and comparison with clinical scores. METHODS: This single-center retrospective study included all consecutive patients undergoing resection for CRLM (2010-2020) with high-quality contrast-enhanced CT scan performed ≤60 days before surgery and at least one detectable CRLM ≥10 mm. Manual tumor segmentation (Tumor-VOI) and automatic 5-mm peritumoral expansion (Margin-VOI) were performed on portal phase images. From each VOI, 110 IBSI-compliant radiomic features were extracted using LIFEx. Three models were developed to predict overall survival (OS): Clinical, Clinical+Tumor-radiomics, Clinical+Tumor/Margin-radiomics. Feature selection was performed using Boruta algorithm, followed by Random Forest classification with 10-fold cross-validation. Model performance was evaluated in the entire cohort and in patients with CT-surgery interval ≤30 days. In patients with multiple 2 metastases, the impact of inter-tumor heterogeneity (up to five lesions per patient) was evaluated using unsupervised clustering algorithms. RESULTS: 306 patients were included (mean age 63 years; 187 men). 5-year survival was 40.9% (mean follow-up 34 months). At internal validation, the clinical model achieved Cindex=0.629. Radiomics provided modest improvement in the entire cohort, with greater impact in the 212 patients with a CT-surgery interval ≤30 days: the Clinical+Tumorradiomics model reached C-index=0.691, increasing to 0.717 with Margin-VOI features. Clinical–radiomic models outperformed established scores (Fong, GAME, m-CS; Cindices range=0.553–0.613). Inter-tumor heterogeneity did not improve OS prediction. CONCLUSIONS: Radiomic features of CRLM and peritumoral tissue from preoperative CT scans enhance survival prediction beyond clinical scores. A short CT-surgery interval impacts performance, being associated with better prediction.