563P - Comprehensive evaluation of recurrence risk (CERR) score for colorectal liver metastases: Development and validation

Abstract

BackgroundFor colorectal liver metastases (CRLM), some key prognostic factors— KRAS/NRAS/BRAF, location of primary tumor, and CA19-9—were identified for recurrence and survival but were rarely included in prognostic scoring system analysis. Besides, tumor burden score (TBS) is a prognostic indicator capturing the cumulative impact CRLM size and CRLM number, but TBS does not take spatial factor—unilobar or bilobar metastasis—into account. Methods787 patients undergoing hepatic resection of CRLM were included and were divided into training and validation groups. Modified TBS (mTBS) was established by a mathematical equation (parameters were CRLM size, CRLM number, and unilobar/bilobar metastasis). In the training group, the Cox proportional hazards model was used to identify independent predictors of prognosis; these factors were combined into the Comprehensive Evaluation of Recurrence Risk (CERR) score. The score was compared with Fong score and “Genetic and Morphological Evaluation” (GAME) score and validated in the validation group. Some indices (including C-index, iAUC, Akaike information criterion, net reclassification index, and integrated discrimination improvement) were calculated to compare the discriminatory capacities of three prognostic scoring systems. ResultsmTBS (AUC 0.617) out-performed TBS (AUC 0.568) in predicting recurrence-free survival (RFS) (P=0.006). Five preoperative predictors of worse RFS were identified and were incorporated into CERR score: KRAS/NRAS/BRAF mutated tumor (1 point); node-positive primary (1 point); extrahepatic disease (1 point); CEA >200ng/ml or CA19-9 >200U/mL (1 point); mTBS between 5 and 11 (1 point) or 12 and over (2 points). Patients undergoing hepatectomy for CRLM were stratified by CERR score into risk groups: high-risk group (CERR score 4 or more) had a 3-year RFS rate of 9.77%; medium-risk group (CERR score 2-3) had a 3-year RFS rate of 21.96%; low-risk group (CERR score 0-1) had a 3-year RFS rate of 39.90%. The validation group showed that the discriminatory capacity of the CERR score was superior to that of the Fong score and the GAME score. ConclusionsThe CERR score is a prognostic tool that can be used to determine optimal clinical management strategies. Legal entity responsible for the studyZhongshan Hospital, Fudan University, Shanghai, China. FundingHas not received any funding. DisclosureAll authors have declared no conflicts of interest.