Noninvasive Predictor for Premalignant and Cancerous Lesions in Endometrial Polyps Diagnosed by Ultrasound

Background There was no consensus for management of asymptomatic endometrial polyps (EPs) up to date. Objective The aim of present study was to determine the risk factors of malignant lesions in EPs diagnosed by ultrasound and establish a noninvasive predictor to decrease unnecessary hysteroscopy for EPs. Study Design We reviewed the records of all consecutive patients who underwent hysteroscopy for EPs in the Women’s Hospital School of Medicine Zhejiang University between January 1, 2001 and December 31, 2018. The patients with histological diagnoses of atypical hyperplasia or cancer were defined as malignancy, while the patients with histological diagnoses of benign lesions were randomly selected as benign group according to the ratio of 1:4 (malignancy:benign), matching by age and year of hospitalization. Logistic regression analysis was used to analyze the clinical parameters for predicting malignancy of EPs. A Chi-squared automatic interaction detection (CHAID) decision tree analysis was performed to find a noninvasive predictor. The sensitivity, specificity, and the receiver operating characteristic curve (ROC) were used for assess the efficacy of the noninvasive predictor. New diagnosed EPs patients received in 2019 were used for verifying the accuracy of the noninvasive predictor. Results The age in 15,790 cases of benign lesions was significantly younger than that in 230 malignancy cases (41.97 ± 11.53 year vs 53.31 ± 11.61 years, p <0.001). AUB (OR 7.306, 95%CI 4.927–10.835), large EPs (OR 2.595, 95%CI 1.662–4.052), and blood flow signal in EPs (OR 2.690, 95%CI 1.872–3.866) were independent predictive factors of malignancy in all enrolled patients. A noninvasive predictor for malignancy of EPs was established, through combining with AUB, large polyps and blood flow signal. This predictor presented excellent sensitivity and NPV (91.3 and 95.8%), with acceptable specificity and AUC (0.801). Further validation in new diagnosed EPs also suggested excellent sensitivity and reasonable specificity (100 and 58.5%) of the predictor. Factors such as thickened endometrial thickness, menopause shorter than 10 years, hypertension, obesity and nulliparous were also validated as independent predictors of malignancy in different subgroup analysis. Conclusions The noninvasive predictor combined with other risk factors from subgroup analysis would be reliable to distinguish the benign lesions from malignancy for EPs diagnosed by ultrasound.