Introduction/Background To further extend and improve the utility of two FDA-cleared multivariate index assays for preoperative malignancy risk assessment of adnexal masses, we developed a third generation multivariate index assay MIA3G that stratifies patients with an adnexal mass into three risk groups. A small high-risk (HR) rule-in group captures a majority of the total cancer cases with a clinically actionable positive predictive value (PPV). A low-risk rule-out group includes a significant portion of the remaining test population with a very high negative predictive value (NPV).
Methodology Seven serum protein analytes: apolipoprotein A-1, beta-2 microglobulin, CA125, follicle stimulating hormone, HE4, transferrin (TRF), and transthyretin (TT) were analyzed on samples from four IRB-approved independent multicenter study cohorts (table 1). The OVA1 Study set was used for training. Two algorithms and corresponding cutoffs were derived for pre- and post-menopausal patients separately. They were then independently validated on the remaining three study sample sets. In addition to sensitivity/specificity, PPVs/NPVs were estimated after adjusting the pre-test cancer prevalence of the validation sets to 5% for premenopausal and 10% for postmenopausal patients.
Results The results are summarized in table 2 and figure 1. They were relatively consistent across the 3 validation sets. As an example, for the premenopausal patients in the validation sets, with an assumed prevalence of 5%, the HR group identified <10% of the test population yet captured 52% - 81% of the cancer cases, resulting in a PPV>33%. On the other hand, >64% of the premenopausal samples were assigned to the LR group at an NPV of 99%. MIA3G performed well for stage I/II invasive cancer and multiple histologic subtypes (data not shown).
Conclusion With further validation in a prospective studies, MIA3G could be potentially used in patients with suspicious adnexal masses as part of the decision of surgery.
Disclosure RG Bullock and HA Fritsche are employees of Vermillion Inc. which funded this work. Z Zhang is a consultant for Vermillion Inc. and contributed to this work independent of his affiliation with Johns Hopkins University
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.