A Comparison of 11 Clinical Risk Scores for Prediction of Survival After Curative-Intent Resection of Colorectal Liver Metastases.
[BACKGROUND] Colorectal liver metastases (CRLM) are the main determinant of survival in colorectal cancer, and radical resection offers the best oncological outcomes.
- p-value p = 0.001
- p-value p < 0.001
APA
Schmidt F, Meister FA, et al. (2026). A Comparison of 11 Clinical Risk Scores for Prediction of Survival After Curative-Intent Resection of Colorectal Liver Metastases.. Annals of surgical oncology. https://doi.org/10.1245/s10434-026-19627-9
MLA
Schmidt F, et al.. "A Comparison of 11 Clinical Risk Scores for Prediction of Survival After Curative-Intent Resection of Colorectal Liver Metastases.." Annals of surgical oncology, 2026.
PMID
42045727
Abstract
[BACKGROUND] Colorectal liver metastases (CRLM) are the main determinant of survival in colorectal cancer, and radical resection offers the best oncological outcomes. However, heterogeneous clinicopathological features make appropriate patient selection essential. Numerous clinical risk scores (CRS) have been proposed to predict outcomes after liver resection for CRLM. This study evaluated the prognostic performance of 11 established CRS regarding survival after curative-intent resection of CRLM.
[METHODS] This retrospective study included patients who underwent curative-intent liver resection for CRLM at University Hospital RWTH Aachen, Germany, between 2010 and 2021. The following CRS were analyzed: Fong, Nordlinger, Nagashima, Konopke, Basingstoke Predictive Index, Tumor Burden Score, Resection Severity Index, Kulik, RAS-mutation CRS, Comprehensive Evaluation of Relapse Risk (CERR) score, and the Genetic and Morphological Evaluation score. Overall survival (OS) was compared using Kaplan-Meier analysis and log-rank testing. Predictive accuracy was assessed using the Akaike information criterion, Harrell's C-index for OS, and area under the curve (AUC) analyses for 1- and 5-year survival.
[RESULTS] A total of 528 patients were included, with a median OS of 26 months (95% confidence interval [CI] 23-28). All CRS except the Resection Severity Index significantly stratified patients according to OS. CERR consistently ranked among the top three scores for both the Akaike information criterion (1725) and the C-index (0.61) and had the highest accuracy for predicting 1-year survival (AUC 0.654, p = 0.001) and 5-year survival (AUC 0.62, p < 0.001).
[CONCLUSIONS] Although the CERR demonstrated the most consistent predictive performance, 10 of 11 evaluated CRS effectively stratified patients according to long-term survival after CRLM resection.
[METHODS] This retrospective study included patients who underwent curative-intent liver resection for CRLM at University Hospital RWTH Aachen, Germany, between 2010 and 2021. The following CRS were analyzed: Fong, Nordlinger, Nagashima, Konopke, Basingstoke Predictive Index, Tumor Burden Score, Resection Severity Index, Kulik, RAS-mutation CRS, Comprehensive Evaluation of Relapse Risk (CERR) score, and the Genetic and Morphological Evaluation score. Overall survival (OS) was compared using Kaplan-Meier analysis and log-rank testing. Predictive accuracy was assessed using the Akaike information criterion, Harrell's C-index for OS, and area under the curve (AUC) analyses for 1- and 5-year survival.
[RESULTS] A total of 528 patients were included, with a median OS of 26 months (95% confidence interval [CI] 23-28). All CRS except the Resection Severity Index significantly stratified patients according to OS. CERR consistently ranked among the top three scores for both the Akaike information criterion (1725) and the C-index (0.61) and had the highest accuracy for predicting 1-year survival (AUC 0.654, p = 0.001) and 5-year survival (AUC 0.62, p < 0.001).
[CONCLUSIONS] Although the CERR demonstrated the most consistent predictive performance, 10 of 11 evaluated CRS effectively stratified patients according to long-term survival after CRLM resection.