Objective  This study aims to evaluate the performance of the predictive model risk of ovarian malignancy algorithm (ROMA) using serum human epididymis protein 4 (HE4) and CA125 for discriminating benign from malignant pelvic masses.

  Methods  The electrochemical luminescence method was carried out before and after the menopause prediction model (ROMA) to detect serum HE4 and CA125 levels of patients admitted in hospital for surgery of pelvic masses or ovarian cyst. Patients were classified into high- and low-risk groups to evaluate the value of the applied prediction model.

  Results  Of the 1683 patients evaluated, 1448 had pelvic benign disease and 235 had pelvic malignancy, including 106 EOC. In the pathological diagnosis of patients with benign pelvic masses, 1356 cases were classified into the low-risk group with a specificity of 93.6%. A total of 121 cases of pelvic malignant tumors with sensitivity of 80.7% and 20 cases of border ovarian epithelial tumor with sensitivity of 28.2% were categorized in the high-risk group. EOC cases (93) were also classified in the high-risk group, with sensitivity of 87.7%. The cases that were not classified in the high-risk group included 2 of mucous adenocarcinoma and 11 of clear cell carcinoma. Five cases of non-epithelial ovarian malignant tumor patients were classified into the high-risk group with a sensitivity of 38.5%. Thirty-five cases of ovarian malignant tumor patients with a sensitivity of 85.3% and one case of metastatic ovarian cancer patient with a sensitivity of 25.0% were categorized in the high-risk group.

  Conclusions  ROMA can be successfully applied in pelvic malignant tumor patients to classify them into high-risk group. Most EOC patients belong to the high-risk group. ROMA can be used in patients with malignant EOC, especially during preliminary screening.