OP07.10: Artificial intelligence weights the importance of clinical and sonographic factors predicting nodal metastasis in endometrial cancer

Abstract

To determine whether artificial intelligence might be useful in weighting the importance of clinical and sonographic (US) preoperative variables predicting the risk of lymph node metastases in patients (pts) undergoing surgery for endometrial cancer. Retrospective study evaluating the last 100 consecutive endometrial cancer pts undergoing US examination before surgery between 2017 and 2019 at National Cancer Institute of Milan. Using artificial neuronal network (ANN) analysis we estimated the importance of different variables predicting the risk of nodal involvement. Clinical, histological (at pre-operative endometrial biopsy) and US variables were evaluated. ANN simulates a biological neuronal system and similarly to neurons, ANN acquires knowledge through a learning-phase process allowing weighting the importance of covariates, thus establishing how much a variable influences a multifactorial phenomenon. The prevalence of pts diagnosed with nodal disease at surgical staging was 16%. Using ANN we observed that the three main US factors predicting nodal metastasis were: myometrial invasion at US (importance: 0.212), echogenicity of the tumour at US defined as uniform or not-uniform (importance: 0.131) and histology at pre-operative biopsy defined as endometrioid or not (importance: 0.099). Overall, 79% (79/100) of pts had endometrial endometrioid tumour detected at pre-operative biopsy; among this subgroup of pts the main factors predicting nodal spread were: myometrial invasion at US (importance 0.218), echogenicity of the tumour (importance 0.170) and cervical invasion at US (importance 0.147). According to our results, myometrial invasion at US and echogenicity of the tumour should be considered the most important factors predicting nodal involvement in endometrial cancer patients. These data are confirmed in the subgroup analysis of endometrioid tumours only. Further studies are needed to estimate the clinical utility of AI in providing help in decision-making processes. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.