Abstract
PURPOSE: To investigate whether the cost-effective, pretreatment tumor markers carcinoembryonic antigen (CEA) and carbohydrate antigen-125 (CA-125) can be used to predict lymph node metastasis (LNM) in endometrioid-type endometrial cancer (EC) and to develop a predictive model. METHODS: This was a single-center retrospective study of patients with endometrioid-type EC who underwent complete staging surgery between January 2015 and June 2022. We identified the optimal cut-off values of CEA and CA-125 for predicting LNM using receiver operating characteristic (ROC) curves. Stepwise multivariate logistic regression analysis was used to identify independent predictors. A nomogram for predicting LNM was constructed and validated by bootstrap resampling. RESULTS: The optimal cut-off values of CEA and CA-125 were 1.4 ng/mL (area under the ROC curve (AUC) 0.62) and 40 U/mL (AUC 0.75), respectively. Multivariate analysis showed that CEA (odds ratio (OR) 1.94; 95% confidence interval (CI) 1.01-3.74) and CA-125 (OR 8.75; 95% CI 4.42-17.31) were independent predictors of LNM. Our nomogram showed adequate discrimination with a concordance index of 0.78. Calibration curves for the probability of LNM showed optimal agreement between the predicted and actual probabilities. The risk of LNM for markers below the cut-offs was 3.6%. The negative predictive value and negative likelihood ratio were 96.6% and 0.26, respectively, with moderate ability to rule out the possibility of LNM. CONCLUSION: We report a cost-effective method of using pretreatment CEA and CA-125 levels to identify patients with endometrioid-type EC who are at a low risk for LNM, which may guide decision-making regarding aborting lymphadenectomy.