Objectives Sentinel lymph node (SLN) is considered the standard of care in early-stage endometrial cancer (EC) patients. In case of SLN failure, a side-specific lymphadenectomy of the no mapping hemipelvis is recommended. Nevertheless, most hemipelvis lymphadenectomies showed no nodal involvement. Previously, we published a preoperative predictive score of nodal involvement. In case of a negative score (value 3–4), the risk of nodal metastases was extremely low. The present multicentre study aims to validate the predictive score of nodal involvement in patients undergoing nodal assessment.

Methods EC patients undergoing surgical treatment with nodal staging were included in the analysis. A preoperative predictive score of nodal involvement was calculated for all patients before surgery was performed. The score included myometrial infiltration, tumor grading (G), tumor diameter, and Ca125 assessment. STARD (standards for Reporting Diagnostic accuracy studies) guidelines were followed for the score accuracy.

Results 1038 patients were included in the analysis and 155 (14.9%) nodal metastases were detected. The score was negative (3 and 4) in 475 patients and positive (5–7) in 563 cases. The score showed 83.2% sensitivity, 50.8% specificity, 94.5% negative predictive value, and 55.7% diagnostic accuracy. The area under the curve (AUC) was 0.75. The logistic regression between negative score and absent nodal metastases showed OR 5.133, 95% CI (3.30–7.98), p <0.001.

Conclusions The nodal preoperative predictive score is a fair diagnostic test. The risk of nodal metastasis is extremely low in case of negative score. In SLN failure, the application of the present score associated with SLN algorithm could avoid unnecessary lymphadenectomies.