Purpose: The primary hypothesis to be tested in this study was that the diagnostic performance (as assessed by the area under the receiver operator characteristic curve, AUC) of a multianalyte panel to correctly identify women with ovarian cancer was significantly greater than that for CA-125 alone. Methods: A retrospective, case-control study (phase II biomarker trial) was conducted that involved 362 plasma samples obtained from women with ovarian cancer (n = 150) and healthy controls (n = 212). A multivariate classification model was developed that incorporated five biomarkers of ovarian cancer, CA-125; C-reactive protein (CRP); serum amyloid A (SAA); interleukin 6 (IL-6); and interleukin 8 (IL-8) from a modelling cohort (n = 179). The performance of the model was evaluated using an independent validation cohort (n = 183) and compared with of CA-125 alone. Results: The AUC for the biomarker panel was significantly greater than the AUC for CA-125 alone for a validation cohort (p < 0.01) and an early stage disease cohort (i.e. Stages I and II; p < 0.01). At a threshold of 0.3, the sensitivity and specificity of the multianalyte panel were 94.1 and 91.3%, respectively, for the validation cohort and 92.3 and 91.3%, respectively for an early stage disease cohort. Conclusions: The use of a panel of plasma biomarkers for the identification of women with ovarian cancer delivers a significant increase in diagnostic performance when compared to the performance of CA-125 alone.
- Multivariate classification
- Ovarian cancer