Objective The objective of this study was to assess the effectiveness of human epididymis protein 4 (HE4) and cancer antigen 125 (CA125), and risk of ovarian malignancy algorithm (ROMA) in identifying type I and type II epithelial ovarian cancers (EOCs).

Methods A cross-sectional diagnostic study was conducted of 499 Thai women older than 18 years who had clinically diagnosed pelvic masses and underwent elective surgery at our hospital between July 2012 and July 2014. Preoperative serum HE4 and CA125 levels were measured, and postoperative pathologic slides were reviewed.

Results Of 499 Thai women enrolled in this study, 357 were noncancerous (NC), 79 had type I EOCs (EOCs-I) and 63 had type II (EOCs-II). Risk of ovarian malignancy algorithm revealed significantly better performance than HE4 and CA125 in discriminating between NC and EOCs-I (receiver operating characteristic curve and area under the curve [ROC-AUC]: ROMA, 0.86; HE4, 0.80; and CA125, 0.77), and NC and EOCs-II (ROC-AUC: ROMA, 0.97; HE4, 0.95; and CA125, 0.93). In differentiation between EOCs-I and EOCs-II (setting EOC-II as reference), ROMA yielded a better performance than HE4 and CA125 (ROC-AUC: ROMA, 0.83; HE4, 0.82; and CA125, 0.77); however, CA125 and HE4 showed higher sensitivity (CA125, 77.8%; HE4, 76.2%), whereas ROMA had the highest specificity (79.8%).

Conclusions In women who present with pelvic masses, ROMA performed very well in distinguishing between noncancer lesions and EOCs but the combined HE4 and CA125 test was more effective in predicting EOC types.