5.1 Introduction

Transvaginal ultrasound is useful in the investigation of women with PMB because it helps to identify those at higher risk of endometrial cancer who require further investigation. TVUS is also an effective means of excluding endometrial cancer. But the interpretation of test results depends upon several factors, including the cut-off point used to define a positive result and how large or small the risk of cancer is for different groups of women.

This section outlines:

• Estimates of cancer risk for different groups of women presenting with PMB
• In which groups of women with PMB, TVUS is most likely to be useful
• How to interpret TVUS results for groups of women with varying risks of cancer

Traditional parameters essential to the evaluation of test performance, such as sensitivity and specificity, are neither necessarily intuitive nor readily transferable to clinical situations. However, the understanding and application of information about test performance to clinical practice is vital to the successful selection and implementation of the diagnostic test. This section sets out a complementary approach to judging how useful an investigation is, using parameters derived from available data on sensitivity and specificity.^{64, 65} Further technical details on the derivation of recommendations in this section may be found on the SIGN website.

For departments where TVUS is available, clinicians may wish to review their criteria for referral and further investigation in light of these recommendations. Where no TVUS is available, clinicians and managers interested in developing a service should consider the guideline development group's conclusions alongside resource and training implications.

5.2 To which patients does this section apply?

The recommendations in this chapter apply to women referred to gynaecologists for investigation of PMB. It does not apply to women directly referred for TVUS by general practitioners, as most research is based upon secondary care populations.

5.3 Estimating the probability of endometrial cancer in women presenting with PMB

An important issue in the initial clinical assessment concerns the probability that an individual woman with PMB will have endometrial cancer. This probability will influence the choice and interpretation of any investigation.

Approximately 10% of women up to 60 years of age presenting and referred with post-menopausal bleeding have endometrial cancer.^{6, 28} Among women over 60 years presenting with post-menopausal bleeding, extrapolated from age-specific incidence rates, 13% are assumed to have endometrial cancer.⁶ Evidence level 2++

Data on the proportion of women on combined HRT presenting with PMB or unscheduled bleeding who are found to have endometrial cancer are not available. The estimated proportions of such women developing endometrial cancer per annum range from 0.02% to 0.05%.^{66, 67} A cautious baseline estimate of 1% for the risk of endometrial cancer in women on combined HRT presenting with PMB or unscheduled bleeding is assumed. The risk for women under 50 years is assumed to be 0.1%, and women over 60 years 1.5%. The risks for women on continuous combined HRT regimens is assumed to be similar to those not on HRT.

The guideline does not attempt to quantify the probability of a woman on tamoxifen with PMB having endometrial cancer. However, it is likely to be substantially higher than 10%.

5.4 Interpretation of TVUS

As stated earlier, it is in general true that the thicker the endometrium of a post-menopausal woman the higher the likelihood of endometrial cancer being present.³⁰ Therefore the measurement of a patient's endometrial thickness by TVUS aims to distinguish between a low or significant probability of endometrial cancer in her case. The threshold value for deciding if there is a significant risk of cancer will depend, in part, on the 'normal' endometrial thickness for the relevant patient group. For example, women on sequential HRT tend to have thicker endometria, as a matter of course, and so a higher threshold needs to be used for identifying those with a significant probability of cancer.

The interpretation of a TVUS test result, for a patient presenting with PMB, depends both on what probability of cancer is considered acceptable (requiring no further investigation) and on the pre-test probability of that patient having cancer. The pre-test probability depends on her history of HRT (see Figure 1) , and on her age (see section 5.3 and additional material on the SIGN website). Consequently, differing cut-off points can be applied to specific groups of women with PMB or unscheduled bleeding. These cut-off points also need to be chosen in light of what is agreed as an acceptable remaining risk of cancer. In current practice the commonly accepted cut-off of endometrial thickness of 5 mm does not reduce the probability of endometrial cancer below 1% for women who have never taken HRT or have been on continuous combined HRT. For these groups of women with a pre-test probability of endometrial cancer of around 10%, given a TVUS measurement of 5 mm or less, the probability of cancer is 1.7%. Achieving agreement upon an acceptable level of probability of cancer after testing is problematic but for the purposes of this guideline, it is less than 1% (or less than 1 in 100).

According to the most recent meta-analysis, using an endometrial thickness of over 3 mm to define an abnormal result would represent a more sensitive approach.²⁹ Based upon a pre-test probability of cancer of 10%, the post-test probability following a negative test result is 0.4%. Unfortunately, the evidence base for a 3 mm threshold is less reliable and more prone to bias than that for the 5 mm threshold. Poorer quality studies tend to overestimate the accuracy of diagnostic tests.⁶⁸ The guideline development group faced the dilemma of whether to recommend a cut-off at a threshold (5 mm) which would probably be of insufficient use as a diagnostic test in most women, or one (3 mm) based on evidence overestimating its accuracy. It was decided to make approximate adjustments to the likelihood ratio to adjust for bias. Therefore, measured by TVUS, an endometrial thickness of 3 mm or less gives an approximate post-test probability of cancer of 0.6% to 0.8% in the following groups of women: Evidence level 1+,2++

• Post-menopausal women who have never been on HRT
• Post-menopausal women who have not been on any form of HRT for a year or more
• Post-menopausal women on continuous combined HRT

For post-menopausal women who present with unscheduled bleeding on sequential HRT, an endometrial thickness of 5 mm or less gives an approximate post-test probability of 0.2%. Therefore use of the 5 mm threshold in this group provides equivalent (or greater) reassurance that cancer is not present, if the TVUS result is negative.

A cut-off threshold of 3 mm or less should be used for TVUS in women with PMB or unscheduled bleeding who: have never used HRT have not used any form of HRT for a year or more are using continuous combined HRT.

If the clinician and the woman judge that the level of reassurance and reduced risk are acceptable following a TVUS measurement of 3 mm or less, no further action need be taken. Further investigations should be carried out if symptoms recur.

If the clinician, the patient or both are not satisfied with this level of reassurance, further investigation is justified. This should include an endometrial biopsy to obtain a histological assessment.

For women on sequential combined HRT presenting with unscheduled bleeding, TVUS using a cut-off point of 5 mm or less should be used to exclude endometrial cancer.

Women presenting with PMB and taking tamoxifen have a higher probability of malignancy (substantially greater than 10%). In these cases the ultrasound image is more difficult to interpret. Therefore it is advisable to sample the endometrium initially and examine the cavity hysteroscopically.

As indicated in the previous section, other considerations may influence the selection of TVUS as an initial investigation. TVUS may be appropriate as a non-invasive procedure in more elderly or frail women. Other investigations may be appropriate as a first line approach in certain circumstances, where there is a risk of substantial delay in waiting for TVUS, or as a second line if TVUS cannot be performed for technical reasons.

5.5 Next step after identifying a thickened endometrium

The finding of an endometrial thickness above the relevant cut-offs (3 mm or 5 mm) indicates that there is a risk of abnormality significant enough to warrant further investigation. The exact procedures to be followed will depend on clinical circumstances and local service provision, but should include a pathological assessment of the endometrium. As the false negative rate of endometrial sampling or biopsy is significant, current advice is that this should be combined with hysteroscopy. This should be carried out prior to endometrial sampling. If local facilities and organisation allow, hysteroscopy and endometrial sampling could be performed in a one stop service during the same session as the transvaginal ultrasound, although patient circumstances may not always be suitable for this.

If the local service is such that there would be a delay prior to performing hysteroscopy, there is merit in performing an endometrial sampling technique in the outpatient clinic. This may lead to earlier recognition of a significant lesion such as malignancy.

The hysteroscopic evaluation of the endometrial cavity reassures the clinician whether or not visible abnormality exists, indicates whether material for histological assessment is obtainable and may demonstrate the best method to achieve it. When endometrial sampling gives no yield, a negative hysteroscopy confirms that this outcome is acceptable.

Figure 1 : Transvaginal ultrasonographic (TVUS) evaluation of women with post-menopausal bleeding