This is a further contribution to the important debate in TOG. Bakour and Ball have revised their conclusion that ‘asymptomatic thickening of the endometrium/polyps does not routinely justify further investigations’. They state that their modified recommendation is based on a recent meta-analysis superseding the conclusions from the preliminary data of the UKCTOCS trial. This meta-analysis, undertaken in January 2011, selected 20 observational studies (out of 95 eligible) including 6974 women with sample sizes of 47–1926 and the authors acknowledge the possible bias resulting from this selection process. Importantly, it did not include the data from UKCTOCS trial which was just being published in January 2011. The latter is a single prospective study with fairly robust data collection and verification, which analysed endometrial thickness (ET) measurements in 37 038 women with 136 (99 with no bleeding) developing endometrial cancer (EC) or atypical hyperplasia (AEH) within 1 year. Hence, it could be argued that the much larger data from UKCTOCS trial remains relevant with some limitations.
A major and repeated emphasis has been placed on describing the ‘sensitivity’ and ‘specificity’ of the various cut-offs of ET in diagnosing EC.[1-3] For example, Jacobs et al. mention the optimum ET cut-off for EC/AEH to be 5.15 mm with a high specificity of 86%. This can give a false impression that increased ET has high correlation to EC/AEH. However, this high sensitivity and specificity is a function of very low incidence of EC in the population. In fact, the vast majority i.e. 94 out of 100 asymptomatic women with ET >10 mm will not have EC/AEH. Hence, the ‘positive and negative predictive value’ of the ET cut-offs are more important considerations to draw meaningful conclusions for clinical practice. An in-depth look at the UKCTOCS data reveals that in the absence of bleeding, the ‘positive predictive value’ of ET >5 mm is 1.4% and >10 mm is 4.5% (hence much less than 1.4% when ET is between 5 and 10 mm). Would a policy of ‘no further investigation’ in the presence of a cancer risk of 4.5% and more (5.9%) be justified? This risk may be a bit of an overestimate because UKCTOCS data did not include postmenopausal women below 60 years of age. However, it seems a more scientific approach to perform an ‘outpatient endometrial biopsy’ (EB) when significant asymptomatic endometrial thickening is discovered which would bring the post-test probability of EC below 1%. Provided the endometrial biopsy is benign (or even insufficient tissue), then hysteroscopy would generally be unnecessary, unless there is hyperplasia, high suspicion or very significant high risk factors.
It is generally accepted that most (if not all) endometrial polyps of less than 2 cm are benign. However, the confirmation of polyps has generally been based on hysteroscopy and then polyps could be twisted off at the same time under paracervical anaesthetic block. It seems difficult to make a case for ‘under vision’ removal of polyps with expensive disposable equipment. The ‘morphology’ of endometrial echo merits further attention and discussion. Jacobs et al pooled all ‘abnormal endometrial morphology’ together. A presumptive sonographic diagnosis (when possible) of well demarcated endometrial polyps with cystic areas (even in the presence of ET >10 mm) would reduce the risk of cancer below 1%, but an amorphous irregular endometrial echo even of lesser thickness would increase the possibility of cancer. Hence, given that the transvaginal scans are performed by sonographers and clinicians with variable experience, at least the asymptomatic women with ET >10mm should have an EB irrespective of risk factors. We use hysteroscopy quite selectively. The results of the meta-analysis are relevant in that ET measurements have no place in screening of asymptomatic women and ET <10 mm without bleeding in low risk women (and non-suspicious endometrial morphology) may not need further investigation. Some women may prefer to have endometrial polyps removed (ET >10 mm) for reassurance but this would be optional with discussion of pros and cons and economic considerations as further evidence emerges.