A Predictive Model Using Histopathologic Characteristics of Early-Stage Type 1 Endometrial Cancer to Identify Patients at High Risk for Lymph Node Metastasis

This study aimed to develop a predictive model using histopathologic characteristics of early-stage type 1 endometrial cancer (EC) to identify patients at high risk for lymph node (LN) metastases. The data of 523 patients who received primary surgical treatment between January 2001 and December 2012 were abstracted from a prospective multicenter database (training set). A multivariate logistic regression analysis of selected prognostic features was performed to develop a nomogram predicting LN metastases. To assess its accuracy, an internal validation technique with a bootstrap approach was adopted. The optimal threshold in terms of clinical utility, sensitivity, specificity, negative predictive values (NPVs), and positive predictive values (PPVs) was evaluated by the receiver-operating characteristics (ROC) curve area and the Youden Index. Overall, the LN metastasis rate was 12.4 % (65/523). Lymph node metastases were associated with histologic grade, tumor diameter, depth of myometrial invasion, and lymphovascular space involvement status. These variables were included in the nomogram. Discrimination of the model was 0.83 [95 % confidence interval (CI) 0.80-0.85] in the training set. The area under the curve ROC for predicting LN metastases after internal validation was 0.82 (95 % CI 0.80-0.84). The Youden Index provided a value of 0.2, corresponding to a cutoff of 140 points (total score in the algorithm). At this threshold, the model had a sensitivity of 0.73 (95 % CI 0.62-0.83), a specificity of 0.84 (95 % CI 0.82-0.85), a PPV of 0.40 (95 % CI 0.34-0.45), and an NPV of 0.95 (95 % CI 0.94-0.97). The results show that the risk of LN metastases can be predicted correctly so that patients at high risk can benefit from adapted surgical treatment.