Comparison of the risk of ovarian malignancy algorithm and Copenhagen Index for the preoperative assessment of Japanese women with ovarian tumors
Makoto Iizuka, Yoshinobu Hamada, Jun Matsushima, Teppei Ichikawa, Taichi Irie, Noriko Yamaguchi, Shuichi Sakamoto, Shinichi Ban, Satoshi Takakura- Obstetrics and Gynecology
Abstract
Objective
To compare the risk of ovarian malignancy algorithm (ROMA) and Copenhagen Index (CPH‐I) in their ability to distinguish epithelial ovarian cancer (EOC) and malignant ovarian tumors (MLOT) from benign ovarian tumors (BeOT) in Japanese women.
Methods
Patients with pathologically diagnosed ovarian tumors were included in this study. The study validated the diagnostic performance of ROMA and CPH‐I.
Results
Among the 463 Japanese women included in this study, 312 had BeOT, 99 had EOC, and 52 had other MLOT. The receiver‐operator characteristic (ROC) area under the curve (AUCs) of ROMA (0.89) and CPH‐I (0.89) for distinguishing EOC from BeOT were significantly higher than that of CA125 (0.82) (CA 125 vs. ROMA; p = 0.002, vs. CPH‐I; p < 0.001). The ROC‐AUCs of ROMA (0.82) and CPH‐I (0.81) for distinguishing MLOT from BeOT were significantly higher than that of CA125 (0.75) (CA 125 vs. ROMA: p = 0.003, vs. CPH‐I: p < 0.001). The sensitivity (SN)/specificity (SP) of ROMA and CPH‐I for distinguishing EOC from BeOT at standard cut‐off points were 69%/90%, and 69%/90%, respectively, those for distinguishing MLOT from BeOT were 54%/90%, and 55%/90%, respectively.
Conclusion
ROMA and CPH‐I performed comparably well and better than CA125 in distinguishing EOC from BeOT in Japanese women. ROMA and CHP‐I should be used with caution in practical situations, where all histological possibilities for must be considered, because the SNs of ROMA and CPH‐I were only 54% and 55%.