Search

Identifying malignant ovarian masses preoperatively in young women can prove difficult due to their nonspecific presentation in this population (1). To describe (i) the clinicopathological predictors of malignancy in surgically removed ovarian masses in women ≤ 21 years of age; (ii)the ability of three risk models to distinguish malignant (including borderline) from benign tumours. A 30-year retrospective analysis of surgically removed ovarian masses in females ≤ 21 years at the Royal Children's Hospital and the Royal Women's Hospital Melbourne. External validation of risk of malignancy scores were undertaken. These included RMI2 (Risk of Malignancy Index 2, [menopausal status, CA125 and ultrasound features]); the Papic-method (tumour markers, maximum diameter of lesion and solid component) and the PRMI (Paediatric Risk of Malignancy Index, sex hormone related symptoms, maximum diameter of largest solid component, enhancement or flow in a septum or solid papillary projection). For each model the positive predictive value (PPV), negative predictive value (NPV), sensitivity, specificity and area under the curve (AUC) were calculated. We examined 460 masses of which 383 were benign, 20 were borderline and 57 frankly malignant. Factors significantly associated with frank malignancy (compared to benign masses) were the presence of a palpable mass (73.9% versus 24.2%), precocious puberty, a larger mean diameter of the mass (15.1 cm versus 8.3 cm), presence of a solid component (91.3% versus 44.4%), multilocularity (66.7% versus 40.4%), ascites (42.9% versus 2.4%) respectively (all p < 0.05). Those with borderline tumours had an older median age compared to frankly malignant tumours (14.4 vs 18.5 years respectively) and were less likely to have a solid component or ascites (p < 0.005). Tumour markers were undertaken in 52 borderline and frankly malignant masses, and were positive in 92.0% (23/25) of germ cell malignancies, 42.8% (3/7) of sex cord stromal tumours, 80% of frankly malignant epithelial tumours (4/5) and 30.8% (4/13) of borderline tumours. The risk models RMI2 and the Papic-method had good discriminatory value in differentiating benign from malignant disease (area under the curve [AUC] 0.83 and 0.84 respectively). However, the PRMI was less discriminatory (AUC 0.62). The RMI2 model with a cut off score of 112 provided the highest predictive odds for malignancy in this population out of all three models (OR 25.3 [95% C.I. 9.0-71.1]). Pre-existing risk of malignancy indices (RMI2) and Papic show good reliability for detection of malignant ovarian disease in young females. [ABSTRACT FROM AUTHOR]