Development and external validation of new ultrasound-based mathematical models for preoperative prediction of high-risk endometrial cancer.

OBJECTIVES: To develop and validate strategies, using new ultrasound-based mathematical models, for the prediction of high-risk endometrial cancer and compare them with strategies using previously developed models or the use of preoperative grading only. METHODS: Women with endometrial cancer were prospectively examined using two-dimensional (2D) and three-dimensional (3D) gray-scale and color Doppler ultrasound imaging. More than 25 ultrasound, demographic and histological variables were analyzed. Two logistic regression models were developed: one 'objective' model using mainly objective variables; and one 'subjective' model including subjective variables (i.e. subjective impression of myometrial and cervical invasion, preoperative grade and demographic variables). The following strategies were validated: a one-step strategy using only preoperative grading and two-step strategies using preoperative grading as the first step and one of the new models, subjective assessment or previously developed models as a second step. RESULTS: One-hundred and twenty-five patients were included in the development set and 211 were included in the validation set. The 'objective' model retained preoperative grade and minimal tumor-free myometrium as variables. The 'subjective' model retained preoperative grade and subjective assessment of myometrial invasion. On external validation, the performance of the new models was similar to that on the development set. Sensitivity for the two-step strategy with the 'objective' model was 78% (95% CI, 69-84%) at a cut-off of 0.50, 82% (95% CI, 74-88%) for the strategy with the 'subjective' model and 83% (95% CI, 75-88%) for that with subjective assessment. Specificity was 68% (95% CI, 58-77%), 72% (95% CI, 62-80%) and 71% (95% CI, 61-79%) respectively. The two-step strategies detected up to twice as many high-risk cases as preoperative grading only. The new models had a significantly higher sensitivity than did previously developed models, at the same specificity. CONCLUSION: Two-step strategies with 'new' ultrasound-based models predict high-risk endometrial cancers with good accuracy and do this better than do previously developed models.

Imaging in gynaecological disease: clinical and ultrasound characteristics of mucocele of the appendix. A pictorial essay.

The aim of this pictorial essay is to describe the ultrasound characteristics of the mucocele of the appendix (MA) by transvaginal ultrasound. The databases of the pathology and gynaecology departments of two Belgian hospitals were retrospectively searched for all cases of MA. Only patients who received a pre-operative transvaginal ultrasound scan by an experienced ultrasound examiner following the IOTA protocol were included. Ultrasound -characteristics were described and pattern recognition was applied in order to find morphological characteristics typical for MA. Thirty-one cases were found of which five underwent a transvaginal ultrasound. In three out of these five cases a mucocele was described. All three masses had a similar morphology, i.e. a para-ovarian uni-locular tubular mass with a color score of 1. A typical characteristic was the aspect of the cyst content, which resembled whipped cream. Although MA is seldom diagnosed by gynaecologists, transvaginal ultrasound is able to make an accurate diagnosis. MA is a -para-ovarian, unilocular tubular mass that typically has a cyst content that resembles whipped cream.

Ovarian cancer arising in endometrioid cysts: ultrasound findings.

OBJECTIVES: To describe sonographic characteristics of malignant transformation in endometrioid cysts. METHODS: Women with a histological diagnosis of ovarian endometrioid cysts, borderline tumors arising in endometrioid cysts and carcinoma arising in endometrioid cysts, preoperatively examined sonographically, were included in this retrospective study. Gray-scale and Doppler ultrasound characteristics of the endometrioid cysts were compared with those of the borderline tumors and primary cancers arising in endometrioid cysts. The performance of an experienced examiner in classifying the masses was also assessed. RESULTS: Of 324 cases collected for the study, 309 (95.3%) lesions were classified as endometrioid cysts, four (1.2%) as borderline tumors arising in endometrioid cysts and 11 (3.4%) as carcinoma arising in endometrioid cysts. Women with malignant findings (borderline ovarian tumors and cancers) were older (median age 52 (range, 28-79) years) than those with benign endometrioid cysts (median age 34 (range, 18-76) years) (P<0.0001), and the prevalence of postmenopausal status was significantly higher in malignant cases. All (15/15) malignant tumors vs. 16% (50/309) of benign tumors were characterized by the presence of solid tissue (P<0.0001). The prevalence of solid tissue with positive Doppler signals was higher in malignant tumors (100%) than in benign cysts (7.8%) (P<0.0001). Papillary projections were a more frequent sonographic feature among malignant lesions (86.7%) than among benign endometrioid cysts (11.3%) (P<0.0001); power Doppler signals were detected within the projections in 92.3% and 37.1% of malignant and benign lesions, respectively. The examiner correctly diagnosed 94.8% (293/309) of benign lesions as benign and 93.3% (14/15) of malignant lesions as malignant. The risk estimation of the examiner was 'uncertain' in three (20%) and 'probably/certainly malignant' in 12 (80%) of 15 malignant cases. CONCLUSION: Borderline tumors and carcinomas arising in endometrioid cysts show a vascularized solid component at ultrasound examination.