Development and clinical evaluation of an ultrasound-based predictive model for early diagnosis of endometrial cancer.

Early-stage endometrial cancer frequently shares overlapping clinical presentations and sonographic features with benign endometrial disorders, making early detection challenging. This retrospective study evaluated clinical characteristics, symptom profiles, and ultrasonographic parameters in patients with endometrial lesions who underwent standardized transvaginal ultrasound and histopathological confirmation. A total of 257 patients were included, of whom 189 had benign endometrial lesions and 68 were diagnosed with early-stage endometrial cancer. Univariate analyses demonstrated significant differences between groups in age, body mass index, hemoglobin levels, reproductive status, symptom patterns, and multiple ultrasound features, particularly endometrial thickness, endometrial-myometrial junction integrity, vascular density, and vascular distribution. Variables showing statistical significance were incorporated into a multivariate logistic regression model, which identified age, postmenopausal status, abnormal uterine bleeding, endometrial thickness ≥12 mm, disruption of the endometrial-myometrial junction, moderate to marked vascularity, and central or mixed vascular distribution as independent predictors of early-stage malignancy. The resulting predictive model demonstrated excellent discrimination, yielding an area under the receiver operating characteristic curve of 0.902 (95% confidence interval: 0.860-0.944), with 84.3% sensitivity and 82.0% specificity at the optimal cutoff. These findings highlight the value of integrating clinical and ultrasound-derived indicators to improve early risk stratification and support timely diagnostic decision-making in patients with suspected endometrial cancer.