Page 144 - The diagnostic work-up of women with postmenopausal bleeding

Chapter 7
could only identify four more studies compared to the three existing meta-analyses on this subject.7-9
Another weakness is that, because only a small number of studies is available and most studies are based on small samples, we had to draw conclusions based on a limited number of patients. Apar t from the limited power, the relatively small number of studies and variability in methods also did not allow for more standard statistical analyses recommended for diagnostic test accuracy reviews, such as pooling sensitivity and specificity using the bivariate model or estimating summary ROCs.
The three existing meta-analyses focused on the diagnostic accuracy of endometrial sampling in a mixed population of pre- and postmenopausal women.7-9 As the diagnostic accuracy of a test is strongly dependent on the prevalence (or pre-test probability) of a diagnosis, and the prevalence of endometrial cancer and atypical hyperplasia is much lower in pre- versus postmenopausal women, we think it is important to study this subject in a selected population of women with PMB.Therefore, we searched specifically for articles on the diagnostic accuracy of endometrial sampling in women with PMB and included only these studies, which reported data on postmenopausal bleeding separately.
Endometrial sampling fails in 42% of cases (either technical failure or insuffient material) and in 7% of these cases a (pre)cancer is found.This finding is in accordance with findings in other studies, which describe a failure or inconclusive rate of 16% to 50 % and in 5 to 20 % of these cases significant endometrial pathology is found.25-27 Therefore, a case of a failed or inconclusive sample, should lead to further diagnostic work-up. Also, a benign result of endometrial sampling is not completely reassuring, as sensitivities are lower than anticipated based on previous literature. In the three existing meta-analyses (blind) D&C has been used as a reference standard, which is worrisome as D&C is known to miss 50-85% of focal intracavitary pathology.28,29 As D&C could miss focal pathology,it could also possibly miss endometrial (pre) cancer in an endometrial polyp.Therefore, nowadays, D&C is almost completely replaced by hysteroscopy as a reference standard, both in clinical as well as in research settings.10 It suggests that endometrial sampling, which is performed as a mini-curettage, as well misses a significant number of focal pathologies and therefore possibly also focal (pre) cancers. Because in women with atypical hyperplasia (which is regarded as endometrial (pre) cancer in 17-52% an underlying cancer is found at hysterectomy,30 it is important to diagnose not only endometrial cancer but also atypical hyperplasia. Given the above findings,
142