Construction and validation of 'WCH‐nomogram' for predicting the prognosis after resection of colorectal liver metastases

Abstract Background The prognostic predictive tool for patients with colorectal liver metastasis (CRLM) is limited and the criteria for administering preoperative neoadjuvant chemotherapy in CRLM patients remain controversial. Methods This study enrolled 532 CRLM patients at West China Hospital (WCH) from January 2009 to December 2019. Prognostic factors were identified from the training cohort to construct a WCH‐nomogram and evaluating accuracy in the validation cohort. Receiver operating characteristic (ROC) curve analysis was used to compare the prediction accuracy with other existing prediction tools. Results From the analysis of the training cohort, four independent prognostic risk factors, namely tumor marker score, KRAS mutation, primary lymph node metastasis, and tumor burden score were identified on which a WCH‐nomogram was constructed. The C‐index of the two cohorts were 0.674 (95% CI: 0.634–0.713) and 0.655 (95% CI: 0.586–0.723), respectively, which was better than the previously reported predication scores (CRS, m‐CS and GAME score). ROC curves showed AUCs for predicting 1‐, 3‐, and 5‐year overall survival (OS) of 0.758, 0.709, and 0.717 in the training cohort, and 0.860, 0.669, and 0.692 in the validation cohort, respectively. A cutoff value of 114.5 points was obtained for the WCH‐nomogram total score based on the maximum Youden index of the ROC curve of 5‐year OS. Risk stratification showed significantly better prognosis in the low‐risk group, however, the high‐risk group was more likely to benefit from neoadjuvant chemotherapy. Conclusions The WCH‐nomogram demonstrates superior prognostic stratification compared to prior scoring systems, effectively identifying CRLM patients who may benefit the most from neoadjuvant chemotherapy.