This thesis has evaluated aspects of ultrasound-based assessment and management of women with postmenopausal bleeding and endometrial polyps.

The efficacy of transrectal ultrasound scan (TRS) was assessed in 103 consecutive postmenopausal women with an axial uterus. TRS was accepted by two-thirds of the women and the proportion of satisfactory endometrial assessments was significantly higher on TRS compared to transvaginal scan (TVS), 91% (95% CI 84-98) vs 62% (95% CI 50-74), respectively. In the subgroup of 50 women with postmenopausal bleeding and an axial uterus, the endometrial thickness measured significantly thinner on TRS by a median of 1.2mm (IQR 0.4-3) compared to TVS. Furthermore, subjective pattern recognition for endometrial cancer was less accurate on TVS compared to TRS when the uterus is in an axial position.

The interrater reliability of ultrasound subjective pattern recognition for endometrial cancer was prospectively assessed in 40 women with postmenopausal bleeding and a thickened endometrium (≥4.5mm); a good level of agreement (κ = 0.78, 95% CI 0.61-0.95) was found between an expert and an average operator.

The diagnostic accuracy of ultrasound subjective pattern recognition for endometrial cancer was assessed in 240 consecutive women with postmenopausal bleeding and a thickened endometrium (≥4.5mm) and available histology. It performed well with a sensitivity and specificity of 88% (95% CI 77-95) and 97% (95% CI 94-99), respectively. The presence of focal malignancy within endometrial polyps was the most common cause of a false-negative diagnosis of endometrial cancer.

Endometrial cancer was diagnosed on ultrasound by subjective pattern recognition and simultaneously assessed for the presence of deep myometrial invasion and cervical stromal invasion in 51 women. We found that the accuracy of ultrasound in the preoperative staging of endometrial cancer was comparable to MRI (sensitivity and specificity, 86% vs 77% and 66% vs 76%, respectively).

A clinical model was presented to estimate the risk (low, intermediate, or high) of pre-malignancy or malignancy in postmenopausal endometrial polyps. The model included polyp size, the presence or absence of intralesional cystic spaces and the patient’s BMI as clinical variables. Accordingly, approximately one-third of postmenopausal polyps would be categorised as high- or intermediate-risk and they would account for over 90% of all premalignant/malignant polyps, while the remaining polyps would be categorised as low-risk with a 1/18 risk of pre-malignancy or malignancy. The overall accuracy of the model in predicting premalignant or malignant postmenopausal polyps was 92% (95% CI 86.0-97.4).

The natural history of expectantly managed endometrial polyps was assessed retrospectively in 112 polyps over a median follow-up of 22.5 months (range 6-136). We found that polyps’ growth rates varied, and it was not possible to predict an individual polyp’s growth based on the patient’s clinical characteristics or polyp’s morphological features. Polyp’s growth rate was not associated with the risk of developing abnormal uterine bleeding (AUB). Some polyps underwent spontaneous regression (7/112, 6%) and this occurred more frequently among premenopausal women and those who were symptomatic of AUB.