Comprehensive Evaluation of Relapse Risk (CERR) Score for Colorectal Liver Metastases: Development and Validation

Abstract

Background: The calculation of the tumor burden score (TBS) is not perfect because the bilobar spread of colorectal liver metastasis (CRLM) is neglected. The identification of an ideal prognostic scoring system for CRLM remains controversial.

Materials and methods: Patients who underwent curative intent liver resection for CRLM from one medical center were enrolled in cohort 1 (787 patients) and cohort 2 (162 patients). Tumor relapse-free survival (RFS) was the main outcome. A Cox regression model was used to identify independent predictors of prognosis. The time-dependent area under the curve, calibration curve, and C-index were employed to validate the predictive ability of the survival model.

Results: Modified TBS (mTBS) was established by a mathematical equation with parameters including CRLM size, CRLM number, and unilobar or bilobar metastasis. Five preoperative predictors of worse RFS were identified in cohort 1 and incorporated into the Comprehensive Evaluation of Relapse Risk (CERR) score: KRAS/NRAS/BRAF-mutated tumor (1 point); node-positive primary (1 point); extrahepatic disease (1 point); carcinoembryonic antigen level > 200 ng/mL or carbohydrate antigen 19-9 (CA19-9) >200 U/mL (1 point); and mTBS between 5 and 11 (1 point) or 12 and over (2 points). Patients in cohort 1 were stratified by their CERR score into risk groups: the high-risk group (CERR score 4 or more), the medium-risk group (CERR score 2-3), and the low-risk group (CERR score 0-1). Importantly, internal validation in cohort 1 and further validation in cohort 2 both showed the superior discriminatory capacity of the CERR score.

Conclusion: mTBS should be promoted. The CERR score is a powerful prognostic tool that can help determine optimal clinical management strategies.

Implications for practice: This work resulted in the successful modification of the tumor burden score and development of a comprehensive and practical prognostic scoring system-the Comprehensive Evaluation of Relapse Risk (CERR) score. The CERR score, with a better prognostic discriminatory ability, outperformed the Fong score. Perhaps more importantly, the CERR score is a powerful prognostic tool because it unified the most consistently reported prognostic factors. Therefore, the CERR score can assist doctors in determining optimal clinical management strategies.

Keywords: Colorectal cancer; Hepatectomy; Metastasis; Prognosis; Relapse.

Figure 1

Establishment of mTBS. (A, B): Pattern diagrams of mTBS. (C): Calculation of mTBS: mTBS² = e^χ × [(maximum colorectal liver metastasis [CRLM] diameter)² + (number of CRLM)²], if unilobar metastasis, χ = 0; if bilobar metastasis, χ = 1; e (natural constant) = 2.71828. (D): Time‐dependent AUCs of mTBS in Cohort 1; mTBS, modified tumor burden score, three categories: score <5, score 5–12, and score >12; TBS, tumor burden score, 3 categories: score <3, score 3–9, and score >9; N1S5, number of CRLM >1 and/or size of CRLM ≥5 cm, 3 categories: number = 1 and size <5 cm, number >1 only or size ≥5 cm only, number >1 and size ≥5 cm.Abbreviations: AUC, area under the curve; mTBS, modified tumor burden score; TBS, tumor burden score

Figure 2

Kaplan‐Meier estimates of relapse‐free survival stratified by score values and risk groups of CERR, Fong, and GAME score in cohort 1. (A–C): Relapse‐free survival curves for different values of the CERR, GAME, and Fong score. (D–F): Kaplan‐Meier estimates of relapse‐free survival stratified by risk groups of CERR, Fong, and GAME score. The shaded part of survival curve represents 95% confidence interval.Abbreviations: CERR, Comprehensive Evaluation of Relapse Risk; GAME, Genetic and Morphological Evaluation.

Figure 3

Comparison and validation of CERR score. (A): Time‐dependent AUCs of the CERR, GAME, and Fong score in cohort 1. (B): Internal validation of the CERR score using the bootstrap sampling method in cohort 1. (C): Time‐dependent AUCs of the CERR, GAME, and Fong score in cohort 2. (D): Further validation of the CERR score using the bootstrap sampling method in cohort 2.Abbreviations: AUC, area under curve; B, bootstrap; CERR, Comprehensive Evaluation of Relapse Risk; GAME, Genetic and Morphological Evaluation; RFS, relapse‐free survival.

Figure 4

Kaplan‐Meier estimates of relapse‐free survival stratified by score values and risk groups of CERR, Fong, and GAME score in cohort 2. (A–C): Relapse‐free survival curves for different values of the CERR, GAME, and Fong score. (D–F): Kaplan‐Meier estimates of relapse‐free survival stratified by risk groups of CERR, Fong, and GAME score. The shaded part of survival curve represents 95% confidence interval.Abbreviations: CERR, Comprehensive Evaluation of Relapse Risk; GAME, Genetic and Morphological Evaluation.