Performance of transvaginal ultrasound in evaluation of endometrioid versus non-endometrioid cancers.

401 Background: Measurement of endometrial thickness by transvaginal ultrasound (TVUS) is the standard of care for evaluation of patients with postmenopausal bleeding, the most common presenting symptom of endometrial cancer. However, the sensitivity of TVUS has not been clearly established in the setting of non-endometrioid histology, which is more aggressive and does not arise from endometrial hyperplasia. This study aims to evaluate the endometrial stripe (EMS) detection rate and sensitivity of TVUS in non-endometrioid compared to endometrioid endometrial cancers. Methods: In this retrospective cohort study, we included endometrial cancer cases diagnosed at a single institution from 01/01/2019 to 04/01/2024 who were enrolled in the Discovery and Evaluation of Testing for Endometrial Cancer in Tampons (DETECT) study who had TVUS during their initial clinical evaluation. Demographic characteristics and results from TVUS reports were obtained from the electronic medical record. Endometrial stripe thickness of less than 5mm is considered a negative result, whereas endometrial thickness of 5mm or greater (5mm+) is considered positive. The primary outcomes were the ability to measure EMS and the sensitivity of TVUS at a cutoff of 5mm+ for endometrial cancer detection, overall and by race and histology. We assessed differences using chi-squared tests. Results: Of the 349 patients included in our analysis, 247 (71%) had endometrioid histology, and 102 (21%) had non-endometrioid histology. There were 98 (28%) Black patients (49% non-endometrioid) and 242 (69%) White patients (22% non-endometrioid). The EMS was adequately measured in 98% of patients with endometrioid histology compared to 91% of patients with non-endometrioid histology (p=0.001). When stratified by race, EMS was visualized in 100% of Black patients with endometrioid histology but 89% of Black patients with non-endometrioid histology (p=0.02). This difference was seen but not significant in White patients with 98% measurable endometrial stripes in endometrioid cancer and 92% in non-endometrioid cancer (p=0.05). The ability to measure EMS was not different when compared by race alone. The sensitivity of TVUS (EMS of 5mm+) was higher in patients with endometrioid histology (99%) than patients with non-endometrioid histology (94%; p=0.03). Conclusions: In one of the largest and most diverse studies of TVUS performance in endometrial cancer patients to date, TVUS was more effective at visualizing the EMS and had higher sensitivity in patients with endometrioid compared to non-endometrioid endometrial cancers. While performance did not vary by race, the prevalence of non-endometrioid cancers was twice as high among Black compared to White patients. Our findings suggest that TVUS may be less accurate for diagnosing non-endometrioid endometrial cancer, which disproportionately affects Black women.