Combined use of biomarkers for detection of ovarian cancer in high-risk women

Abstract

The aim of this study was to examine the negative predictive value for a panel of serum markers in women at high risk for developing ovarian cancer. A total of 201 serum samples were collected and analyzed from 102 women at "high risk" for ovarian cancer: 26 with primary ovarian cancer, 31 with recurrent ovarian cancer, 28 with benign gynecologic diseases, and 14 with other cancers. Samples were tested for cancer antigen (CA) 125 II, CA19-9, CA72-4, CA15-3, and macrophage colony-stimulating factor, OVX1, and the marker values were further used as input to be evaluated by a previously trained artificial neural network (ANN). CA125 alone identified 72% of the primary ovarian cancers at a specificity of 95%. If either CA125 or CA72-4 were elevated, sensitivity rose to 80%. Adding macrophage colony-stimulating factor-improved sensitivity to 84% and when CA15-3 was included, a sensitivity of 88% was achieved. Specificity of the four marker panel was, however, reduced to 82.5%. By contrast, at the same sensitivity of 88%, the ANN exhibited a much higher specificity at 92.5% (p = 0.0105). Our data suggest that the combined use of multiple biomarkers improve sensitivity in women at high risk for ovarian cancer. In contrast to the simple "or" combination rule, the ANN was able to achieve a higher sensitivity without significant loss in specificity.