The University of Kentucky Model for Selecting Breast Cancer Patients for Oncotype DX Testing
Steven Frame, Meredith Burge, Nicole Miller, Yolanda Brill, Rahul Matnani, Patrick McGrath, Marie-Louise Fjallskog, Luis M Samayoa. University of Kentucky, Lexington, KY; VAMC, Lexington, KY; Uppsala University, Uppsala, Sweden
Background: Currently, the decision on treatment of Estrogen Receptor (ER) (+), Her2-neu (-), N0-1a patients is largely supported by molecular tests such as Oncotype DX. While the benefits of this test are indisputable, the guidelines for ordering it are non-specific and may lead to over-utilization at a significant cost. This study presents a model for Oncotype DX testing, based on common morphologic (H&E) and Immunohistochemical (IHC) variables already established in our daily practices.
Design: Recurrence Scores (RS) from 72 randomly selected N0 patients were compared to levels (%) of ER, Progesterone Receptors (PR), Ki – 67, Cyclin A, Mitotic Index (MI) and Tumor Grade (TG) using univariate regression analyses. Although no one variable showed a significant R value, the ones that correlated the most were selected and given a numerical score according to cut-off levels either previously described (MI and TG) or encountered while performing the analyses (ER and PR) (see table 1). Subsequently, all patients were scored accordingly and paired with their corresponding RS.
|ER expression %||>90||<90|
|PR expression %||90 - 100||70 - 90||30 - 70||1 - 30||0|