EarlyR: A Robust Gene Expression Signature for Predicting Outcomes of Estrogen Receptor–Positive Breast Cancer

Abstract

IntroductionEarly stage estrogen receptor (ER)-positive breast cancer may be treated with chemotherapy in addition to hormone therapy. Currently available molecular signatures assess the risk of recurrence and the benefit of chemotherapy; however, these tests may have large intermediate risk groups, limiting their usefulness. MethodsThe EarlyR prognostic score was developed using integrative analysis of microarray data sets and formalin-fixed, paraffin-embedded–based quantitative real-time PCR assay and validated in Affymetrix data sets and METABRIC cohort using Cox proportional hazards models and Kaplan-Meier survival analysis. Concordance index was used to measure the probability of prognostic score agreement with outcome. ResultsThe EarlyR score and categorical risk strata (EarlyR-Low, EarlyR-Int, EarlyR-High) derived from expression of ESPL1, MKI67, SPAG5, PLK1 and PGR was prognostic of 8-year distant recurrence-free interval in Affymetrix (categorical P = 3.5 × 10−14; continuous P = 8.8 × 10−15) and METABRIC (categorical P < 2.2 × 10−16; continuous P < 10−16) data sets of ER+ breast cancer. Similar results were observed for the breast cancer–free interval end point. At most 13% of patients were intermediate risk and at least 66% patients were low risk in both ER+ cohorts. The EarlyR score was significantly prognostic (distant recurrence-free interval; P < .001) in both lymph node–negative and lymph node–positive patients and was independent from clinical factors. EarlyR and surrogates of current molecular signatures were comparable in prognostic significance by concordance index. ConclusionThe 5-gene EarlyR score is a robust prognostic assay that identified significantly fewer patients as intermediate risk and more as low risk than currently available assays. Further validation of the assay in clinical trial–derived cohorts is ongoing.