Ultrasound assessment and risk prediction in women with endometrial cancer

Abstract

Background: Endometrial cancer is the most common gynecological cancer in the industrialized world, constituting 4.5% of all cancer in Swedish women. Standard surgery is hysterectomy and bilateral salpingo-oophorectomy, with or without lymphadenectomy, depending on the estimated risk of lymph node metastases. Adjuvant therapy is given, depending on the estimated risk of adverse outcome. Numerous risk stratification systems guide the choice on lymphadenectomy, but none is associated with a high accuracy. Lymph node metastases are found in 11—22% of women. If they could be accurately predicted, many lymphadenectomies could be avoided. The aim of this thesis was to improve preoperative risk assessment in women with endometrial cancer, with regard to predicting deep (≥ 50%) myometrial invasion (MI), cervical stromal invasion (CSI), lymph node metastases and recurrence or progression.

Methods: All study cohorts originate from the prospective, international, multicenter IETA (International Endometrial Tumor Analysis) 4 ultrasound study on women with endometrial cancer. In Study I agreement to histopathology and interobserver reproducibility of subjective ultrasound assessment of MI and CSI among ultrasound experts and gynecologists were compared by off-line evaluation of videoclips from 53 women from a single center cohort. In Study II sonographic features and accuracy of ultrasound assessment of MI ≥ 50% were compared in tumors with and without the microcystic elongated and fragmented (MELF) pattern of myometrial invasion and the relationship of the MELF pattern to more advanced stage (≥ IB) and lymph node metastases was assessed in 850 women with endometrioid endometrial cancer from a multicenter cohort. In Study III a risk prediction model was developed on 1501 women from a multicenter cohort, to estimate the individual risk of lymph node metastases before surgery. In Study IV demographic, sonographic and Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) variables and their ability to predict recurrence or progression was assessed in 339 women from a single center cohort.

Results: Gynecologists and ultrasound experts assessed MI ≥ 50% with comparable diagnostic accuracy and interobserver reproducibility, while ultrasound experts assessed CSI with greater diagnostic accuracy and interobserver reproducibility than gynecologists. Tumors with the MELF pattern were slightly larger, the color score was higher and the multiple multifocal vascular pattern was more common, compared to tumors without the MELF pattern. The MELF pattern did not affect the diagnostic accuracy of MI assessment, however it was associated with ≥ 50% MI, CSI, higher stage and lymph node metastases. A risk model with variables from endometrial biopsy results (histotype), clinical (age and bleeding duration) and ultrasound characteristics (tumor extension and tumor size) could reliably predict the risk of lymph node metastases before surgery, and had higher clinical utility than risk stratification by combined endometrial biopsy and ultrasound. Demographic (age ≥ 65 years and waist circumference ≥ 88 cm), sonographic (ultrasound tumor extension and ultrasound AP diameter ≥ 2 cm) and ProMisE variables combined had higher ability to predict recurrence or progression than the ESMO (European Society for Medical Oncology) classification. Ultrasound tumor size < 2 cm and non-p53 abnormal status identified a large group of women (48%) with a very low risk of tumor recurrence or progression (1.8%).

Conclusion: Preoperative ultrasound staging should be performed by ultrasound experts and is not negatively affected by presence of the MELF status. A risk model with variables from endometrial biopsy, clinical and ultrasound characteristics improves preoperative risk prediction of lymph node metastases. Demographic, sonographic and ProMisE variables show the potential to predict tumor recurrence or progression already before surgery.