ABSTRACT
BACKGROUND: The occurrence of liver metastasis significantly affects the prognosis of colorectal cancer (CRC). Existing research indicates that primary tumor location, vascular invasion, lymph node metastasis, and abnormal preoperative tumor markers are risk factors for CRC liver metastasis. Positive expression of programmed cell death ligand 1 (PD-L1) may serve as a favorable prognostic marker for nasopharyngeal and gastric cancers, in which combined positive score (CPS) quantifies the level of PD-L1 expression. This study aimed to explore CPS as a potential risk factor for CRC liver metastasis and integrate other independent risk factors to establish a novel predictive model for CRC liver metastasis. METHODS: A retrospective analysis was conducted on 437 patients with CRC pathologically diagnosed at The Second Xiangya Hospital of Central South University from January 1, 2019, to December 31, 2021. Data were collected, including CPS, age, gender (male and female), primary tumor location, Ki-67 expression, pathologic differentiation, neural invasion, vascular invasion, lymph node metastasis, and preoperative tumor markers. The optimal cutoff point for the continuous variable CPS was determined using the Youden index, and all CPSs were dichotomized into high- and low-risk groups based on this threshold (scores below the threshold were considered high risk, and score above the threshold were considered low risk). Univariate logistic regression analysis was employed to identify risk factors for CRC liver metastasis, followed by multivariate logistic regression analysis to integrate the selected risk factors. The predictive model was validated through the construction of receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). A nomogram was constructed for visualization. RESULTS: The determined cutoff point for PD-L1 CPS was 4.5, with scores below this threshold indicating a high risk of CRC liver metastasis. In addition, primary tumor origin other than the rectum, presence of pericolonic lymph node metastasis, and abnormal levels of tumor markers carcinoembryonic antigen and cancer antigen 19-9 were identified as independent risk factors for CRC liver metastasis. The constructed clinical prediction model demonstrated good predictive ability and accuracy, with an area under the ROC curve of 0.871 (95% CI, 0.838-0.904). CONCLUSION: The exploration and validation of CPS as a novel predictor of CRC liver metastasis were performed. Based on these findings, a new clinical prediction model for CRC liver metastasis was developed by integrating other independent risk factors. The DCA, clinical impact curve, and nomogram graph constructed on the basis of this model have significant clinical implications and guide clinical practice.
