Preoperative evaluation and a nomogram prediction model for pelvic lymph node metastasis in endometrial cancer.

ABSTRACT

OBJECTIVE: The primary objective of this study is to explore the preoperative risk factors of pelvic lymph node metastasis (PLNM) in endometrial cancer patients, and construct a nomogram prediction model. MATERIALS AND METHODS: We retrospectively collected various preoperative clinical characteristics of patients and analyzed their relationship with PLNM. Logistic regression analysis was used to screen for independent risk factors for PLNM of endometrial cancer. A nomogram prediction model was constructed, the receiver operating characteristic (ROC), calibration curve and decision curve analysis (DCA) were constructed and used to assess discrimination, calibration, and net benefit. RESULTS: Out of the 276 patients, 74 (26.81%) with postoperative pathological confirmation of PLNM. Multivariate logistic regressive analysis demonstrated that preoperative depth of myometrial invasion (DIM) ≥50% determined by Magnetic Resonance Imaging (MRI) (p = 0.003), carbohydrate antigen 125 (CA125) (p = 0.030), carbohydrate antigen 19-9 (CA 19-9) (p = 0.044), and platelet/lymphocyte ratio (PLR) (p = 0.025) could serve as independent risk factors for PLNM. A risk factors-based nomogram prediction model was constructed, which showed good discrimination (AUC = 0.841, p < 0.001) and good efficacy (C-index = 0.842) and good calibration (mean absolute error = 0.046). DCA showed that the model can provide clinical benefits. CONCLUSIONS: Preoperative DIM ≥50% determined by MRI, serum CA 19-9, CA125 and PLR could be utilized to predict PLNM in endometrial cancer patients. This nomogram prediction model can provide preoperative help for evaluation and identification of patients with endometrial cancer, and provide a theoretical basis for clinical intervention.