Predicting OncoDx recurrence scores with immunohistochemical markers

Recent reports suggest that immunohistochemistry (IHC) markers can be used to give prognostic information in breast cancer that is similar to that contained in the Genomic Health Inc. OncoDx recurrence score (Onco-RS). The goal of this study is to examine the potential prognostic value of a score derived from results of a simple set of IHC tests in the prediction of the Onco-RS. A score (IHC-RS) was derived to predict the Onco-RS using IHC-based quantitative and semiquantitative results from a subset of markers selected from those used in the generation of an Onco-RS score. The patient population consists of a retrospectively identified cohort of 158 women with ER-positive, HER2neu-negative breast cancer who completed OncoDx testing. A predictive model was developed to generate the IHC-RS using stepwise multiple regression incorporating Ki67 percentage and semiquantitative ER and PR scores. Using only these 3 IHC markers, the IHC-RS predicted 62% of the Onco-RS variability (adjusted R=0.624, P=0.004). In addition, analysis of outliers in the correlation between the IHC-RS and the Onco-RS reveals the possibility of sampling error as a drawback of the Onco-RS. This is contrasted against potential interlab and intralab variability and preanalytic issues that may negatively impact the implementation of an IHC-RS.