Endometrial Thickness as Diagnostic Triage for Endometrial Cancer Among Black Individuals

Abstract

Poor performance of the transvaginal ultrasonography triage strategy has been suggested as a contributor to racial disparity between Black individuals and White individuals in endometrial cancer (EC) stage at diagnosis in population-level simulation analyses. To examine the false-negative probability using ultrasonography-measured endometrial thickness (ET) thresholds as triage for EC diagnosis among Black individuals and assess whether known risk factors of EC modify ET triage performance. This retrospective diagnostic study of merged abstracted electronic health record data and secondary administrative data (January 1, 2014, to December 31, 2020) from the Guidelines for Transvaginal Ultrasound in the Detection of Early Endometrial Cancer sample assessed Black individuals who underwent hysterectomy in a 10-hospital academic-affiliated health care system and affiliated outpatient practices. Data analysis was performed from January 31, 2023, to November 30, 2023. Pelvic ultrasonography within 24 months before hysterectomy. Ultrasonography performed before hysterectomy as well as demographic and clinical data on symptom presentation, endometrial characterization, and final EC diagnosis were abstracted. Endometrial thickness thresholds were examined for accuracy in ruling out EC diagnosis by using sensitivity, specificity, and negative predictive value. False-negative probability was defined as 1 - sensitivity. Accuracy measures were stratified by risk factors for EC and by factors hypothesized to influence ET measurement quality. A total of 1494 individuals with a uterus (median [IQR] age, 46.1 [41.1-54.0] years) comprised the sample, and 210 had EC. Fibroids (1167 [78.1%]), vaginal bleeding (1067 [71.4%]), and pelvic pain (857 [57.4%]) were the most common presenting diagnoses within 30 days of ultrasonography. Applying the less than 5-mm ET threshold, there was an 11.4% probability that someone with EC would be classified as not having EC (n = 24). At the 4-mm (cumulative) threshold, the probability was 9.5%, and at 3 mm, it was 3.8%. False-negative probability at the 5-mm threshold was similar among EC risk factor groups: postmenopausal bleeding (12.4%; 95% CI, 7.8%-18.5%), body mass index greater than 40 (9.3%; 95% CI, 3.1%-20.3%); and age 50 years or older (12.8%; 95% CI, 8.4%-18.5%). False-negative probability was also similar among those with fibroids on ultrasonography (11.8%; 95% CI, 6.9%-18.4%) but higher in the setting of reported partial ET visibility (26.1%; 95% CI, 10.2%-48.4%) and pelvic pain (14.5%; 95% CI, 7.7%-23.9%). These findings suggest that the transvaginal ultrasonography triage strategy is not reliable among Black adults at risk for EC. In the presence of postmenopausal bleeding, tissue sampling is strongly recommended.