It is well documented that when a gynecologic-oncologist performs the initial surgery for ovarian cancer staging, there is a higher likelihood of complete staging and superior overall survival (OS). However, the proportion of women with ovarian cancer who undergo surgery by a gynecologic-oncologist is less than 50% largely because it is difficult to identify which subgroup of women with an adnexal mass is most likely to have ovarian cancer. Several prediction models and referral guidelines have been investigated to assist in the diagnostic triage of women with an adnexal mass and determine the need for gynecologic oncology referral, but none has received widespread acceptance. Recently, a diagnostic multiple biomarker test, the multivariate index assay (OVA1) marketed by Vermillion, was developed to assist in the diagnostic triage of women with an adnexal mass. In 2011, a pivotal trial examined the predictive value of OVA1 for ovarian malignancy; it contained a mix of subjects enrolled by both gynecologists and gynecologic-oncologists, and the overall prevalence of ovarian malignancy was 29%.
The aim of the current prospective, multi-institutional trial was to investigate the effectiveness of OVA1 for identifying ovarian malignancy among an intended-use cohort of women undergoing surgery for an adnexal mass after enrollment by nongynecologic oncology providers. The predictive performance of OVA1 was compared with that of clinical assessment and serum CA125-II. Participants were patients from 27 nongynecologic oncology practices scheduled to undergo surgery for a documented adnexal mass within 3 months of imaging before enrollment. Preoperative serum samples and physician assessment of the risk of malignancy were correlated with the final pathological report. The comparative sensitivity of each method for detecting malignancy was stratified according to histology, stage of disease, and menopausal status.
A total of 494 patients were evaluable for OVA1 and CA125-II and clinical impression. A pelvic malignancy was found in 92 (18.6%) of these patients. Primary ovarian cancer was diagnosed in 65 patients (13.2%), with 43.1% having International Federation of Gynecology and Obstetrics stage I disease. When combined with clinical impression, the sensitivity of OVA1 in detecting ovarian malignancy was 95.7% (95% confidence interval [CI], 89.3%–98.3%). OVA1 had higher sensitivity than CA125-II in detecting ovarian malignancy among women with early-stage disease (OVA, 91.4% [95% CI, 77.6%–97.0%] vs CA125-II, 65.7% [95% CI, 49.2%–79.2%]). OVA1 correctly identified 83.3% of malignancies missed by clinical impression and 70.8% of cases missed by CA125-II. The performance of OVA1 was also superior for prediction of the absence of an ovarian malignancy: negative predictive value was 98.1% (95% CI, 95.2%–99.2%). Both clinical impression and CA125-II were more accurate than OVA1 for identifying benign disease. When combined with clinical impression, OVA1 correctly identified benign pathology in 204 of 402 patients (specificity, 50.7%; 95% CI, 45.9%–55.6%).
University of California (R.E.B.), Irvine-Medical Center, Orange, CA; Applied Clinical Intelligence (A.S.), Bala Cynwyd, PA; The Johns Hopkins Medical Institutions (D.W.C.), Baltimore, MD; Vermillion, Inc (G.C., E.T.F.), Mountain View, CA; and Vermillion, Inc (D.G.M.), Austin, TX