Abstract PS3-09-06: Comparative Predictive Utility of Oncotype DX and MammaPrint Using a Clinicopathologic Model in Matched HR+/HER2− Early Breast Cancer Cohorts: The BRAIN Study

Abstract Oncotype DX (ODX) and MammaPrint (MMP) are two widely adopted multigene assays used to guide adjuvant chemotherapy decisions in patients with hormone receptor-positive (HR+), HER2-negative early breast cancer. Despite their similar clinical roles, ODX and MMP often produce discordant risk classifications. While head-to-head comparison in the same patient would be ideal, it is rarely feasible in practice due to cost, clinical utility, and reimbursement constraints. Therefore, we designed a pilot study to indirectly compare the predictive performance of ODX and MMP by applying a clinicopathologic model—the Tennessee nomogram—to matched cohorts of patients who received either test. A total of 2,542 patients who underwent ODX (n = 2,011) or MMP (n = 531) testing were retrospectively selected. Patients were matched 1:1 based on age and nodal status to control for key prognostic variables. Risk distribution was compared across both cohorts, followed by application of the Tennessee model to assess its predictive concordance with each assay’s reported risk classification. Model performance was evaluated using AUC, specificity, positive predictive value (PPV), and negative predictive value (NPV). The proportion of patients classified as high-risk differed between the two cohorts (ODX: 12.22% (8.21 - 17.28), MMP: 33.94% (27.72-40.59). The Tennessee model showed an AUC of 85.57 (77.49-93.65) for predicting high-risk ODX results and 70.70 (63.32-78.08) for MMP. Specificity and NPV were 90.21%, 94.59% in the ODX cohort, compared to 89.04% and 73.45% in the MMP cohort. Our findings indicate that the Tennessee model aligns more closely with ODX, suggesting that each assay may capture distinct biological dimensions of tumor behavior. This study highlights the importance of understanding the limitations of multigene tests when used interchangeably and supports the development of cost-effective, data-driven prediction tools that integrate clinicopathologic variables. These pilot results lay the groundwork for future AI-based models capable of refining treatment stratification in early HR+/HER2− breast cancer. Citation Format: J. Kim, S. Lee, S. Lee, I. Lee, J. Ahn, S. Park, S. Gwon, N. Son. Comparative Predictive Utility of Oncotype DX and MammaPrint Using a Clinicopathologic Model in Matched HR+/HER2− Early Breast Cancer Cohorts: The BRAIN Study [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS3-09-06.