Introduction/Background Predicting borderline ovarian tumors can be challenging. Despite a favorable prognosis, these patients should not be treated as benign disease as comprehensive surgical staging is needed. In Denmark, RMI is the gold standard for predicting malignancy and referral to PET/CT. We evaluated ultrasound features in accordance to IOTA terminology and risk of malignancy using the ADNEX model.

Methodology Patients ≥18 years with ovarian lesions were prospectively included at Dept. of Gynecology, Rigshospitalet, Denmark. Gynecologists described lesions using IOTA terminology in a template (EPIC). Clinical decisions were not based on IOTA scores.

Results N=47 patients with histologically verified borderline ovarian tumors were included (89.4% stage I, 10.6% stage II-III). Median age was 54 years (range 21 – 82). RMI was >200 in 29 (61.7%) and <200 in 18 (38.3%). PET/CT was performed in 36 (79.6%) and concluded malignancy suspicion in 18 (FDG-uptake in 15, suspicious CT in 3). Thus, malignancy was suspected in 18 (38.3%) and benign disease in 29 (61.7%) women preoperatively. A total of 10 (21.3%) women underwent secondary staging surgery. The majority were classified multilocular solid (53.2%) or multilocular (23.4%), and less often unilocular solid (21.3%) and unilocular (2.1%). Papillary projections were present in 59.6%, and 38.3% had ≥4. The largest diameter of lesion was >100 mm in 57.4%. Cystic content was anechoic in 46.8%, low level in 32.0%, ground glass in 10.6%, and mixed in 10.6%. Color score >1 was seen in 55.3%. A total of 41/47 (87.2%) had a malignancy risk >10% using the ADNEX model. All 6/47 (10.6%) with malignancy risk <10% were uni-/multilocular lesions (<10 locules), 2 with diameter >100 mm.

Conclusion Accurate diagnosis of borderline is essential for planning appropriate management. Ultrasound pattern recognition is a valuable clinical observation. The ADNEX model identified a malignancy risk above 10% in almost 90% of the population.