The value of new sonographic modalities, demographic, biometric and proliferative variables for the risk prediction in endometrial cancer

Sammanfattning: Background: Endometrial cancer is the sixth most common type of cancer among women worldwide and constitutes 4.8% of all cancer in women. The incidence is highest in highincome countries, given how age and obesity are the two most important risk factors. The majority of cases present at an early stage, with a favourable prognosis. In those patients, added treatment after surgery does not improve survival rates. In advanced stages, the prognosis is generally poor, and treatment options are few with a modest effect at best. Therefore, correct staging and triaging of patients to low- and high-risk groups are the cornerstone of endometrial cancer treatment. Today, risk stratification is based on imaging with ultrasound and/or magnetic resonance imaging and histologic assessment. These are all based on human interpretation and coding and can thus be more or less reliable. Improving reliability and diagnostic accuracy of our instruments, or finding more objective risk factors to complement or replace the current standard, is thus necessary to further tailor treatment to each patient. This thesis aimed to assess DNA ploidy and S-phase fraction as prognostic markers, measure the degree of rater dependency in ultrasound staging, investigate the possible benefit of adding dynamic contrast-enhanced ultrasound and compare diagnostic performance in assessing local tumour extension with ultrasound as well as the prevalence of known high-risk ultrasound features. Methods: Study I used a population-based, consecutive cohort of 1140 women with FIGO stage I endometrioid endometrial cancer. Cox regression was used, including age, degree of differentiation, myometrial invasion, DNA ploidy, S-phase fraction and adjuvant treatment as covariates, with endometrial cancer death being the end-point. In study II, fifteen ultrasound experts assessed off-line 2D video clips and 3D volumes from 58 patients with endometrial cancer for deep myometrial invasion and cervical stromal involvement. Kappa statistics and diagnostic performance were calculated, and rater accuracy was correlated to rater experience measures. In study III, the added benefit to the routine ultrasound of dynamic contrastenhanced ultrasound in diagnosing local tumour extension of endometrial cancer was assessed by comparing the results in a prospectively enrolled study cohort (n=93) to a matched control cohort (n=279). In study IV, pre- and postmenopausal women from the prospective IETA-4 multicentre cohort (n=1538) were compared concerning the prevalence of high-risk sonographic features and the diagnostic performance in assessing local tumour extension with ultrasound. Results: A high S-phase fraction, but not DNA aneuploidy, was an independent prognostic factor for endometrial cancer death in FIGO stage I endometrioid cancer. Interrater reliability was higher with 2D video clips than with 3D volumes and diagnostic performance was also higher. Diagnostic performance was correlated to the number of cases each rater assessed annually. Dynamic contrast-enhanced ultrasound improved sensitivity in diagnosing deep myometrial invasion, without lowering specificity. Premenopausal women more often had low-risk cancer but still had high-risk sonographic features related to the vascularity of the endometrium, likely related to physiological features of a cycling endometrium. An intact endometrial-myometrial border suggested low-risk disease in both pre- and postmenopausal women. Local tumour extension was more accurately assessed in pre- compared to postmenopausal women. Conclusion: The utility and optimal cut-off value for S-phase fraction have to be sought in a new study before it can be introduced into clinical practice. Ultrasound is a reliable imaging modality but should be centralised to high-throughput centres to increase rater experience. Dynamic contrast-enhanced ultrasound is safe and does not require any special preparations and can thus be used as a complement in tricky cases to rule out myometrial invasion. Tumour vascularity in endometrial cancer assessment has to be interpreted in light of menopausal status.