Multivariate index assay MIA3G vs other assessment tools for the ovarian cancer risk assessment of indeterminate masses.

Abstract

e17609 Background: Estimated 10% of women with adnexal masses will undergo preventive surgery for ovary removal to mitigate malignancy risk. This is significant, compared to the 15% rate of ovarian cancer (OC) confirmed by pathology. On the other hand, a women’s lifetime risk of getting OC is estimated at 1 in 78, with death at 1 in 108. This predicament motivates a surveillance paradigm for assessment of malignancy risk. ROMA (Roche) is a multi-marker assay intended to assess the cancer risk of adnexal masses in women already scheduled for surgery. We recently characterized the clinical performance of MIA3G[1], to assess OC risk for women with adnexal mass clinically characterized as benign or indeterminate. Here we present the comparative analysis for MIA3G and ROMA in a multi-site study. Methods: MIA3G -a 7-biomarker assay with a DNN-based risk prediction algorithm – was tested for clinical management of women (median: 51yo, IQR:41-61) across 11 centers. Data is presented for 206 patients with adnexal mass (286 unique samples), who follow an admix of surgical and clinical management. The continuous MIA3G probabilistic scores and dichotomized risk classification were used for the analyses. The primary objective was to compare MIA3G with ROMA assays. The secondary objective was to validate – via a pair-wise approach – the correlation of the biomarkers, age and menopausal status(MENO) - with the predicted risk stratification. Results: MIA3G stratified 13% [CI:0.096-0.174] and 87% [CI:82.6-90.4] of the 286 samples as indeterminate and low risk of malignancy respectively (p-value<2.2e-16, Welch’s t-test). In comparison, ROMA stratified 11.3% with a high likelihood of finding a malignancy on surgery. In the pre-menopausal sub-cohort MIAG3 risk-stratified only 7% as indeterminates, whereas for ROMA and CA125 the analogous estimates increased to 15% and 20% respectively. For the post-menopausal population, the converse trend was observed. In MIAG3 with increasing risk, the inhibitive effect of APO was signaled (R2:-24%, p:0.0006) and potential implication of iron metabolism was observed in TFR (R2:-31%,p:9.76e-5). B2M indicated inflammation with increasing risk (R2:28%, p:0.052. In addition to being a clinical attribute, MENO status was confirmed with FSH and age (R2 >80%,p<2.2e-16). A strong correlation of risk with HE4 (R2:38%) and CA125 (R2:32%) was observed and a HE4-CA125 pairwise-correlation of >85% confirmed. Additionally, the positive association of PRE with APO( R2 :33%) and negative association with CA125 and HE4 (R2~:-20%) rounded out the comprehensive nature of the MIA3G assay for characterizing the nuanced evolution of the masses. Conclusions: This real-world study established the efficacy of MIA3G as the most clinically validated tool for the risk assessment of ovarian cancer of indeterminate masses. 1: DOI 10.3389/fmed.2023.1102437.