Evaluation of models to predict lymph node metastasis in endometrial cancer: A multicentre study

International audienceBackground Several models (preoperative and postoperative) have been developed to predict lymph node metastasis (LNM) in patients with endometrial cancer. The purpose of our investigation was to compare available models in a multicentre study. Methods In a cohort of 519 patients with endometrial cancer who had undergone primary hysterectomy and at least a pelvic lymphadenectomy, we compared the areas under the receiver-operating characteristic curves (AUCs), calibrations, rates of false negatives (FN), and the number of patients at low-risk for LNM using ten different models (three preoperative and seven postoperative). Results In all, 17.5% of patients among the study population (91 in 519) had LNM. Only one of the three preoperative models and three of the seven postoperative models had an AUC >0.75. Six models were well calibrated. Eight models yielded an FN rate of 0.75, to yield an FN rate of