Sensitivity of transvaginal ultrasound screening for endometrial cancer in postmenopausal women: a case-control study within the UKCTOCS cohort

Background The increase in the worldwide incidence of endometrial cancer relates to rising obesity, falling fertility, and the ageing of the population. Transvaginal ultrasound (TVS) is a possible screening test, but there have been no large-scale studies. We report the performance of TVS screening in a large cohort. Methods We did a nested case-control study of postmenopausal women who underwent TVS in the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) following recruitment between April 17, 2001, and Sept 29, 2005. Endometrial thickness and endometrial abnormalities were recorded, and follow-up, through national registries and a postal questionnaire, documented the diagnosis of endometrial cancer. Our primary outcome measure was endometrial cancer and atypical endometrial hyperplasia (AEH). Performance characteristics of endometrial thickness and abnormalities for detection of endometrial cancer within 1 year of TVS were calculated. Epidemiological variables were used to develop a logistic regression model and assess a screening strategy for women at higher risk. Our study is registered with ClinicalTrials.gov, number NCT00058032, and with the International Standard Randomised Controlled Trial register, number ISRCTN22488978. Findings 48 230 women underwent TVS in the UKCTOCS prevalence screen. 9078 women were ineligible because they had undergone a hysterectomy and 2271 because their endometrial thickness had not been recorded; however, 157 of these women had an endometrial abnormality on TVS and were included in the analysis. Median follow-up was 5.11 years (IQR 4.05-5.95). 136 women with endometrial cancer or AEH within 1 year of TVS were included in our primary analysis. The optimum endometrial thickness cutoff for endometrial cancer or AEH was 5.15 mm, with sensitivity of 80.5% (95% CI 72.7-86.8) and specificity of 86.2% (85.8-86.6). Sensitivity and specificity at a 5 mm or greater cutoff were 80.5% (72.7-86.8) and 85.7% (85.4-86.2); for women with a 5 mm or greater cutoff plus endometrial abnormalities, the sensitivity and specificity were 85.3% (78.2-90.8) and 80.4% (80.0-80.8), respectively. For a cutoff of 10 mm or greater, sensitivity and specificity were 54.1% (45.3-62.8) and 97.2% (97.0-97.4). When our analysis was restricted to the 96 women with endometrial cancer or AEH who reported no symptoms of postmenopausal bleeding at the UKCTOCS scan before diagnosis and had an endometrial thickness measurement available, a cutoff of 5 mm achieved a sensitivity of 77.1% (67.8-84.3) and specificity of 85.8% (85.7-85.9). The logistic regression model identified 25% of the population as at high risk and 39.5% of endometrial cancer or AEH cases were identified within this high risk group. In this high-risk population, a cutoff at 6.75 mm achieved sensitivity of 84.3% (71.4-93.0) and specificity of 89.9% (89.3-90.5). Interpretation Our findings show that TVS screening for endometrial cancer has good sensitivity in postmenopausal women. The burden of diagnostic procedures and false-positive results can be reduced by limiting screening to a higher-risk group. The role of population screening for endometrial cancer remains uncertain, but our findings are of immediate value in the management of increased endometrial thickness in postmenopausal women undergoing pelvic scans for reasons other than vaginal bleeding.