Concordance among Gene-Expression–Based Predictors for Breast Cancer

Background Gene expression profiling studies of primary breast tumors performed by different laboratories have resulted in the identification of many apparently different prognostic profiles/gene sets, which show little overlap in gene identity.

Methods In order to compare the individual sample predictions made by these different gene sets, we applied to a single data set of 295 samples, five different gene expression-based predictors: (1) Intrinsic Subtypes, (2) 70-gene Good vs. Poor, (3) Wound-Response Activated vs. Quiescent, (4) Recurrence Score and (5) the 2-gene ratio profile for tamoxifen-treated patients.

Results There was high concordance in outcome predictions across most of these different predictors when the outcome predictions on individual samples were compared. In particular, patients of the Basal-like, HER2+/ER- and Luminal B Intrinsic Subtypes were almost all 70-gene Poor, Wound-Response Activated, and had a High Recurrence Score. The 70-gene and Recurrence Score predictors, which are beginning to be used in the clinical setting, showed 77-81% agreement.

Conclusions These data show that even though different gene sets are being used for prognostication on breast cancer patients, four of the profiles tested here showed significant agreement in outcome predictions on individual patients and are likely tracking a common set of biological phenotypes.