Histopathology of women with non-uniform endometrial echogenicity and risk factors for atypical endometrial hyperplasia and carcinoma

Objective: In sonography, homogeneous endometrium is defined as uniform endometrial echogenicity and heterogeneous, asymmetrical or cystic endometrium is defined as non-uniform. However, the relationship between the non-uniform endometrial echogenicity and the presence or absence of pathology is not known. A retrospective study of the patients with ultrasound non-uniform endometrium who underwent hysteroscopy-directed biopsy was performed to explore its clinical meaning in the diagnosis of endometrial lesions. Materials and methods: Patients with non-uniform endometrial echogenicity who underwent hysteroscopy-directed biopsy were enrolled in the Obstetrics and Gynecology Hospital of Fudan University from January 2015 to May 2018 as the primary cohort. In total, 692 patients with non-uniform endometrial echogenicity were diagnosed and underwent hysteroscopy-directed biopsy. Characteristics were assessed using univariate logistic regression between patients with and without atypical endometrial hyperplasia and carcinoma (atypical EH+). Multivariate analyses were used to develop the predicting model. We incorporated statistically significant variables and presented with nomogram. Internal validation was assessed. An independent validation cohort consisted of 237 consecutive patients from June 2018 to February 2019. Results: Hysteroscopy-directed biopsy showed that 55.20% (382/692) of the patients with non-uniform endometrium had normal endometrium, while 44.80% (310/692) had endometrial lesions, including 39.31% (272/692) benign lesions and 5.49% (38/692) atypical EH+. Univariate logistic analysis showed that older age (P=0.027), abnormal uterine bleeding (AUB) before menopause (P=0.011), postmenopausal bleeding (P<0.001) and endometrial thickness >= 7 mm (P=0.013) were statistically significant for atypical EH+. Multivariate logistic regression analysis showed that age >= 50 years old (OR: 3.97, 95% CI: 1.17-13.43, P=0.027), endometrial thickness >= 7 mm (OR: 8.08, 95% CI: 1.86-35.08, P=0.005) and postmenopausal bleeding (OR: 8.98, 95% CI: 3.26-24.76, P<0.001) were risk factors for atypical EH+. Predictors in the individualized predicted nomogram included age >= 50 years old, AUB before menopause, postmenopausal bleeding and endometrial thickness >= 7 mm. The model showed good discrimination with area under curve (AUC) of 77.09%. With cutoff value of 0.0089267, the recall of atypical EH+ is 100% with precision 6.52% and 6.22% in both primary and validation cohort, respectively. Conclusion Non-uniform endometrial echogenicity is clinically meaningful in assessment of atypical EH+ with risk factors of age >= 50 years old, postmenopausal bleeding and endometrial thickness >= 7 mm. The model can help clinician to predicate the probability of atypical EH+ and make clinical decision.