4.1 Introduction

A range of investigations, including TVUS, are currently used in different gynaecology units across Scotland. Given the present limitations of the evidence base, the guideline development group has not specifically recommended the use of one investigation in preference to another. However, users of this guideline should note that the following section on the selection and interpretation of TVUS (section 5) is underpinned by a reliable evidence base.

4.2 The role of investigation in PMB

The principal aim of the investigation of post-menopausal bleeding is to identify or exclude endometrial pathology, most notably endometrial carcinoma. It is also important to ensure that women are sufficiently reassured following normal tests, that symptomatic benign disease is identified, and that the process of investigation is both acceptable and efficient.

4.3 Available techniques

Two meta-analyses judged to be of sufficient rigour and involving one diagnostic technique - transvaginal ultrasonography - were identified.^{28, 29} Most studies evaluating other techniques were small cohort studies.

4.3.1 TRANSVAGINAL ULTRASONOGRAPHY

Transvaginal ultrasonography can reliably assess thickness and morphology of the endometrium and can thus identify a group of women with post-menopausal bleeding who have a thin endometrium and are therefore unlikely to have significant endometrial disease. This group may not require any further investigation unless there is a recurrence of bleeding.^31,32,33,34 The relatively non-invasive nature of TVUS may make it more acceptable than other investigations, especially to elderly women. Evidence level 2+

The endometrial thickness cut-off

It is conventional to measure the double thickness measurement of both endometrial surfaces at the thickest point in the mid-sagittal view. If there is fluid in the cavity separating the two layers of endometrium then the layers are measured individually and summated. Use of the endometrial thickness cut-off assumes that the endometrial morphology is normal. Any abnormal features, e.g. suspicion of a polyp, would require further investigation irrespective of the endometrial thickness.

There are different endometrial thickness thresholds that may be used for recommending further investigation. However there is a trade-off between sensitivity and specificity.^{28, 31,32,33,38} In other words, the lower the threshold of endometrial thickness chosen as a cut-off point, the fewer cases of endometrial cancer will be missed but at the cost of referring for further investigation a much greater number of women without cancer. Setting the threshold of endometrial thickness at a higher level will result in more cases of cancer being missed whilst reducing the number of women without cancer referred for further investigation.

A recent meta-analysis reviewed evaluations of TVUS.²⁹ After lower quality studies had been excluded (those associated with a higher risk of bias that might affect estimates of test accuracy), only four high quality studies remained. These four studies all assessed the 5 mm threshold and, when pooled, showed that a negative TVUS result of 5 mm or less reduced the risk of disease by 84%. Whether this is sufficient to rule out disease depends on the pre-test risk of disease in the relevant patient group. Furthermore the authors caution against relying on the pooled findings of only four studies, for which there is a wide 95% confidence interval for the reduction in risk (54% to 94%) arising from a negative result. A range of factors need to be considered in determining the best use of TVUS and the optimal threshold. Evidence level 2+

See section 5 for recommendations on the use of endometrial thickness cut-offs for estimating risk of cancer .

No endometrial thickness threshold completely excludes possible early endometrial carcinoma. This disease can, of course, be present in women without post-menopausal bleeding.³³ Inter-observer error in measuring endometrial thickness among experienced clinicians is minimal especially when the endometrium is thin. However, this may not hold if individual clinicians have less experience.³⁰ Evidence level 2+

Endometrial thickness and sequential HRT

The mean endometrial thickness in women on sequential hormone replacement therapy with post-menopausal bleeding is greater than in those women with post-menopausal bleeding who are not on sequential HRT. Thus an abnormal endometrial thickness in women with post-menopausal bleeding who are not using sequential HRT represents a greater probability of endometrial disease than in women taking hormone replacement therapy.^{28, 35} Evidence level 2++

In women using sequential HRT, the thickness of the endometrial wall can vary with each phase of the cycle. In order to standardise readings, TVUS measurements should probably take place during the first half of the cycle.

Training requirements

The interpretation of TVUS depends upon the skills of the operator as well as inter-operator variation. Therefore, gynaecological, radiological or radiography staff undertaking TVUS should be sufficiently trained in the technique and work to locally agreed standards.

Where sufficient local skills and capacity exist, transvaginal ultrasound is an appropriate first-line procedure to identify which women with post-menopausal bleeding are at higher risk of endometrial cancer.

In patients on sequential HRT, TVUS measurements should take place during the first half of the cycle where possible.

Staff undertaking TVUS should be trained to ensure a consistent and acceptable level of performance.

4.3.2 TRANSABDOMINAL ULTRASOUND

Ultrasonic assessment of the uterus by a transabdominal approach has now been superseded by transvaginal ultrasound.^{36, 37} The principle advantage of the latter is improved resolution, allowing more accurate endometrial thickness measurement and better assessment of endometrial morphology. Other advantages of TVUS include the lack of dependency upon a full bladder, improved access to the retroverted uterus and more successful assessment of more obese patients. Evidence level 3

Transabdominal ultrasound may be used as a complementary examination if the uterus is significantly enlarged or a wider view of the pelvis or abdomen is required. Transabdominal ultrasound may also be used in the small proportion of women in whom it proves technically impossible to perform a transvaginal ultrasound.

4.3.3 OTHER ULTRASONOGRAPHIC TECHNIQUES

Other methods of endometrial assessment such as transvaginal Doppler ultrasonography,
three-dimensional ultrasonography, saline enhanced transvaginal ultrasonography and endometrial texture and margin analysis have been assessed in a limited manner with a view to improving specificity.^{38,39,40,41,42} There is no evidence to support their introduction into routine clinical practice at present. Evidence level 3

4.3.4 DILATATION AND CURETTAGE (D&C)

In several small case series, patients had a D&C immediately prior to hysterectomy. In these series, endometrial lesions were overlooked in up to 10% of the instances in which D&C was the only procedure used.⁴³ One study evaluating the completeness of endometrial sampling by D&C showed that in 60% of patients less than half the cavity was curetted.⁴⁴ Evidence level 3

D&C should no longer be used as the first-line method of investigating PMB in most cases.

4.3.5 ENDOMETRIAL BIOPSY

All methods of sampling the endometrium will miss some cancers.^{33, 45,46,47,48} In a small proportion of patients, outpatient endometrial sampling is not technically possible.

Outpatient endometrial sampling has a procedure failure rate as well as a tissue-yield failure rate, each of approximately 10%.^13,46,48,49 It should be noted that yield failures are not unexpected in women with atrophic endometrial linings, whereas failure to obtain tissue would be less likely if cancer was present. However, the success of Pipelle^(TM) endometrial sampling, in terms of obtaining a sample, is also related to operator experience, hence training is very important.⁴⁶ Evidence level 2+,4

When outpatient blind endometrial sampling was compared with inpatient hysteroscopy and curettage, it was found to be an effective screening procedure for atypical hyperplasia but did miss benign lesions, especially endometrial polyps.^31,48,49,50 Evidence level 2+

Hysteroscopy and biopsy (curettage) is the preferred diagnostic technique to detect polyps and other benign lesions.

Outpatient sampling may reduce inconvenience, anxiety and risk for the patient and costs for the hospital. ^47,48,51 Evidence level 1+,2+,3

Histological specimens may be obtained either at the same time as inpatient or outpatient hysteroscopy with curettage or using an endometrium sampling device, e.g. Pipelle(TM).

4.3.6 HYSTEROSCOPY

Hysteroscopy allows the operator to visualise directly the endometrial cavity. The procedure can be performed either in the outpatient setting with the patient awake or under general anaesthesia. A biopsy of the endometrium is usually taken following hysteroscopy either with a sampler or by curettage.

Despite general consensus that it is the current gold standard, the evidence base for hysteroscopy is poor.

Hysteroscopy in the outpatient setting appears to have an accuracy and patient acceptability equivalent to inpatient hysteroscopy under general anaesthetic.^47,52 Evidence level 1+

Outpatient techniques for hysteroscopy and suction sampling of the endometrium should be available in all diagnostic units.

Facilities to perform hysteroscopy and curettage under general anaesthetic should be available for when the outpatient procedure is not possible or the patient has a strong preference for a general anaesthetic.

4.4 Sequencing of investigations

It is reasonable to pursue a different approach to investigation if direct access to a 'one stop' service is not available or obtaining TVUS would be inconvenient (e.g. requiring a long distance return trip) or would appreciably delay assessment. Obtaining an initial endometrial sample may be in the patient's interest if it identifies a cancer prior to the ultrasound appointment. Even so, there is no robust evidence that demonstrates that endometrial sampling alone is sufficient to exclude endometrial cancer.

A possible flowchart of investigation is included in the Quick Reference Guide at the back of this guideline and available separately. The exact sequence of investigation will depend upon clinical judgement, local resources, local expertise and patient preference.

4.5 Investigation of women using tamoxifen

In view of the increased risk of endometrial cancer associated with tamoxifen therapy, there is a case for heightened vigilance for post-menopausal bleeding by both the women and the clinician(s) responsible for their care. However, studies that have included women with vaginal bleeding in the assessment of diagnostic tests to detect abnormalities in women receiving tamoxifen have been of insufficient quality or size to enable any robust conclusions to be drawn.^{53, 54}

Although strictly outside the scope of this guideline, the issue of whether to undertake screening investigations in women on tamoxifen is controversial and merits discussion. Current evidence does not justify the use of any investigation (ultrasonography, hysteroscopy, endometrial biopsy or D&C) in post-menopausal women receiving treatment with tamoxifen in the absence of vaginal bleeding.^{55,56,57,58,59,60,61} Unnecessary investigation should be avoided as there are risks associated with further investigation.⁶⁰ Evidence level 1+,2+

Endometrial investigation should only be carried out in post-menopausal women on tamoxifen who experience vaginal bleeding.

Ultrasonography is poor at differentiating potential cancers from other tamoxifen-induced thickening because of the distorted endometrial architecture associated with long term use of tamoxifen.

Endometrial thickening associated with tamoxifen therapy but not with pathology may decrease the specificity (i.e. increase the number of false positives) of ultrasonography.^{62, 63} Ultrasonographic evaluation could employ a higher cut-off point of endometrial thickness (9 mm) to prompt further investigation but further, more rigorous studies are required to test this approach.⁵³ Furthermore, both women and clinicians may not accept any residual uncertainty about the risk of undetected pathology given this cut-off point. Therefore, the use of hysteroscopy and biopsy as first line investigations may be more appropriate and provide less ambiguous results in this high-risk group of women. Evidence level 2+,4

Hysteroscopy with biopsy is preferable as the first line of investigation in women taking tamoxifen who experience post-menopausal bleeding.

4.6 Recurrent post-menopausal bleeding

There is no evidence to recommend when re-investigation should take place following recurrent or persisting post-menopausal bleeding. Clinical judgement is required but consideration of re-investigation must be given in view of the false negative rate associated with all methods of diagnosis.^{32,37,39,43,45} Evidence level 2+,3,4

Re-investigation of recurrent post-menopausal bleeding should be considered after six months.