Objective: This study aimed to evaluate the added value of volumetric body composition metrics to traditional scoring systems, Clinical Risk Score (CRS) and Tumor Burden Score (TBS), for improving risk stratification and prognosis prediction in patients with colorectal liver metastases following hepatectomy.
Methods: The information of 186 patients with colorectal liver metastases was retrospectively collected. AI-powered automated measurement of volumetric body composition was conducted on preoperative CT images (L1-L4), including skeletal muscle (SKM), subcutaneous adipose tissue (SAT), visceral adipose tissue, and intramuscular adipose tissue. A rule-based body composition signature was developed. Kaplan-Meier plots were adopted to depict overall survival (OS), disease-free survival (DFS) and hepatic-specific DFS of different risk groups.
Results: SKM and SAT were identified as significant indicators for OS with HR of 0.63 (95 % CI: 0.42-0.94) and 0.59 (95 % CI: 0.37-0.92), respectively. They were selected for developing body composition signature (BCS) (low SKM and low SAT: high risk; high SKM and high SAT: low risk; the others: medium risk). There was a significant difference among the three subgroups in OS, DFS, and hepatic DFS (all log-rank p < 0.05). When incorporated into CRS and TBS, the composite systems (CRS-BCS and TBS-BCS) improved patient risk stratification for the three survival outcomes (all log-rank p < 0.05). The composite systems had a higher time-dependent AUC than their standalone counterparts for OS, DFS and hepatic DFS.
Conclusion: Incorporating preoperative volumetric body composition signature into CRS and TBS can enhance risk stratification and prognosis prediction in CRLM patients undergoing hepatectomy.
Keywords: Artificial intelligence; Body composition; Colorectal liver metastasis; Liver resection; Prognosis; Sarcopenia.
Copyright © 2026. Published by Elsevier Ltd.