Biology-Driven Scoring Is Redefining Risk Stratification in Colorectal Liver Metastases: ESMO Open

Introduction

Colorectal liver metastases (CRLM) remain a major determinant of mortality in colorectal cancer, with hepatic resection offering the best chance of long-term survival in selected patients. However, outcomes after surgery remain highly variable, and recurrence is common, highlighting the need for more precise tools to guide treatment selection and timing beyond conventional clinicopathologic assessment.

Problem Statement

Traditional clinical risk scores for CRLM, although widely used in surgical decision making, are limited by their reliance on static anatomical and morphological variables. These models inadequately capture tumor biology, treatment sensitivity and real-time disease evolution, restricting their ability to predict recurrence, guide systemic therapy or support precision treatment strategies in modern multidisciplinary care.

Summary

This review outlines the transition from conventional anatomy-based risk scoring toward a more dynamic and biologically informed framework for CRLM management. Classical clinical risk scores remain useful for baseline prognostication, but their predictive value is increasingly constrained in the era of molecular oncology. Emerging biomarkers—including tumor genomics, histological growth patterns and circulating tumor DNA (ctDNA)—provide clinically relevant insight into tumor behavior, metastatic biology and residual disease. Molecular alterations such as RAS, BRAF and mismatch repair status refine prognosis and increasingly influence treatment strategy, particularly with the expansion of targeted therapies and immunotherapy. Histological growth patterns offer additional prognostic value by distinguishing biologically favorable from aggressive liver metastatic phenotypes, while ctDNA has emerged as one of the most promising dynamic biomarkers for real-time monitoring of recurrence risk, treatment response and minimal residual disease after resection. The authors propose that integrating static clinical variables with dynamic molecular and liquid biopsy data can enable adaptive risk stratification across the treatment continuum. This biology-driven approach represents a major step toward precision oncology in CRLM, with the potential to improve patient selection, optimize perioperative therapy and better align surgery with tumor biology.