The Predictive Value of the Cholesterol-to-Natural Killer Cell Ratio in Colorectal Cancer

Purpose: We constructed a novel biomarker cholesterol (C)-to-natural killer (NK) cell ratio (CNR) to reflect the synergistic effect of cholesterol metabolism and inflammation on colorectal cancer (CRC) outcomes. This study aimed to investigate the clinical significance and predictive value of CNR in CRC and develop a simple and reliable prognostic model for predicting OS in CRC patients. Methods: We retrospectively collected the hematology data and medical records of 213 patients with CRC at Renji hospital and the histological data and medical records of 94 patients with CRC included in a tissue microarray. The association between tumor biomarkers and survival was evaluated using the log-rank test. The diagnostic efficacy of CNR was assessed using receiver operating characteristic curves. The overall survival (OS) rates were estimated using the Kaplan-Meier method. Cox proportional hazards regression was employed in both univariate and multivariate analyses to identify independent prognostic factors, which were subsequently utilized to develop a predictive model for OS. The performance of the model was evaluated using the concordance index (C-index) and calibration plots. The patients were stratified based on the total risk scores calculated from the model. The differences in OS among these groups were assessed using the Kaplan-Meier method. The relationship between cholesterol and NK cells was analyzed by investigating the colon cancer datasets TCGA and GSE39582. Results: The TNM stage III-IV CRC group had significantly higher blood levels of cholesterol, triglycerides, low-density lipoprotein cholesterol (LDL-C), CNR, and carcinoembryonic antigen (CEA), and shorter progression-free survival (PFS) than the TNM stage I-II CRC group (all, P < 0.05). The blood CNR correlated negatively with PFS (P < 0.001). Elevated tissue CNR levels were an independent risk factor for CRC, where low-tissue CNR patients demonstrated significantly prolonged survival (P < 0.05). The area under the curve for blood CNR was 0.743. The multivariate analyses indicated that tumor location (P = 0.004), TNM stage (P = 0.004), and tissue CNR (P = 0.033) were independent prognostic factors of OS in CRC. The nomogram model was based on these variables and demonstrated good calibration and predictive performance, achieving an excellent C-index of 0.737 [95% confidence interval (CI), 0.674-0.779]. The expression of the key cholesterol biosynthesis players HMGCR, SQLE, and SREBP2 was not significantly associated with NK cell-mediated cytotoxicity-related gene signatures. HMGCR and SQLE were negatively associated with CD56, a phenotypic NK cell marker (P < 0.001). Conclusion: This is the first study to explore the predictive value of CNR in CRC, which was a promising predictor of CRC progression. The developed nomogram model may serve as a reliable tool for predicting survival in patients with CRC, which may complement the TNM staging system.