Ultrasound diagnosis of endometrial cancer by subjective pattern recognition in women with postmenopausal bleeding: prospective inter-rater agreement and reliability study.

Abstract

To assess the inter-rater agreement and reliability of using subjective pattern recognition for diagnosing endometrial cancer (EC) on ultrasound in women with postmenopausal bleeding (PMB). This was a prospective cross-sectional study conducted at a gynecological rapid-access clinic, between October 2016 and December 2017, in which consecutive women with PMB and endometrial thickness of ≥ 4.5 mm on transvaginal ultrasound examination were included. Women on hormone replacement therapy or tamoxifen and those with a history of primary gynecological malignancy were excluded. Two raters independently performed ultrasound examinations, blinded to each other's findings, and classified women as having uniformly thickened endometrium, benign endometrial polyp or EC, using subjective pattern recognition. Inter-rater reliability of ultrasound diagnosis was assessed using Cohen's kappa (κ) statistic. All women subsequently underwent either outpatient endometrial biopsy, hysteroscopy or hysterectomy. Forty women were included in the study, with a median age of 61 (interquartile range (IQR), 57-69) years and a median endometrial thickness of 11.0 (IQR, 6.2-20.3) mm. Final histological analysis confirmed 16 (40%) women with EC, 16 (40%) with benign endometrial polyp, four (10%) with atrophic endometrium, three (8%) with proliferative endometrium and one (3%) with endometrial hyperplasia. Inter-rater agreement for the ultrasound diagnoses of uniformly thickened endometrium, benign endometrial polyp and EC was 14/16 (87.5%), 22/30 (73.3%) and 28/34 (82.4%), respectively; inter-rater reliability was good (κ = 0.69; 95% CI, 0.49-0.88). When the ultrasound diagnoses were grouped as either cancer or no cancer, inter-rater agreement was 85% and inter-rater reliability was good (κ = 0.78; 95% CI, 0.61-0.95). Rater A correctly identified 14/16 cases of EC and Rater B identified 15/16. EC was misdiagnosed as benign polyps on ultrasound in two women by Rater A and in one woman by Rater B. The overall accuracies of Rater A and Rater B in differentiating between benign endometrial pathologies and malignancy were 90% and 90%, respectively. Our results show good inter-rater reliability of subjective pattern recognition in diagnosing uniformly thickened endometrium, benign endometrial polyp and EC on ultrasound in women with PMB. Our findings should facilitate wider use of subjective pattern recognition in routine clinical practice. © 2020 International Society of Ultrasound in Obstetrics and Gynecology.