Objective: There is no established ultrasonographic endometrial cutoff value for the diagnosis of endometrial pathologies in asymptomatic postmenopausal tamoxifen (TAM)-treated patients. We attempted to seek the most accurate cutoff value.
Materials and methods: Multiple ultrasonographic cutoff points were evaluated by logistic regression analysis based on 279 patients who had transvaginal ultrasonographic examinations followed by endometrial histopathological analysis. Performance characteristics were calculated with correlation to the endometrial histological findings. We also calculated how many endometrial pathologies could have been left undiagnosed and the number of endometrial samplings, with different cutoff values, which could have been avoided.
Results: There was a gradual increase in specificity and a gradual decrease in sensitivity of the ultrasonographic studies with the increase of endometrial thickness. More overall and more various endometrial pathologies were identified with the increase in cutoff values. The best cutoff value appeared to be 15 mm (sensitivity 37.9%, specificity 87.2%, positive predictive value 63.0%, and negative predictive value 70.2%). However, by avoiding performance endometrial sampling up to this cutoff value, 62.2% endometrial pathologies including 48 endometrial polyps, one case of endometrial hyperplasia with atypia, and one case of endometrial cancer may have been left undiagnosed. At the same time, endometrial sampling in 78.5% of cases may have been avoided.
Conclusion: In asymptomatic postmenopausal breast cancer tamoxifen-treated patients, the use of wider ultrasonographic endometrial cutoff values could be associated not only with the performance of fewer endometrial samplings, but also with a higher possibility of endometrial pathologies, including endometrial cancers, being left undiagnosed.