A Composite Score to Predict Survival in Patients Undergoing Resection because of Colorectal Liver Metastases (CRLM)

Abstract

Introduction: Several scoring systems exist to predict survival in patients with CRLM. Many of these, however, are cumbersome, and do not take into account the role of inflammation and recent treatment strategies. Thus there is a need for an up-to-date robust prognostic model to predict overall survival in these patients. Method: Data on patients that underwent liver resection for the first time because of CRLM between 2005 and 2015 at two hepatobiliary centers were included. Univariable and multivariable analyses were performed, and a Cox regression model was developed. Results: In total 849 patients were included. Predictive factors included age > 70 years, per-operative ablation, extended and two-stage resections, as well as a high Glasgow Prognostic Score (GPS), and a Tumour Burden Score (TBS). No correlation between GPS and TBS was observed (Spearman, r= 0.168). Based on the predictive factors, GPS and TBS a Composite Score was developed, which classified patients into separate cohorts. Survival differences between the cohorts were significant, and outperformed both GPS and TBS. Median overall survival according to the Composite Score was: 1 = 53 months, 2 = 31 months and 3 = 20 months. Conclusion: The Composite Score offers a good prognostic tool in the assessment of overall survival in patients undergoing resection after CRLM. Notably, the model identifies a group (Composite score 3), which may not benefit from surgery since it performs worse than the median estimated survival in palliative patients with CRLM. The model is currently undergoing external validation.