Development and validation of therapeutically relevant multi-gene biomarker classifiers.

Abstract

In June 2004 Ma et al. ( 1 ) described a two-gene expression ratio that they claimed accurately predicted clinical outcome of early-stage breast cancer patients treated with adjuvant tamoxifen monotherapy. In the current issue of this journal, Reid et al. ( 2 ) reported their failure to confirm the usefulness of the two-gene expression ratio on independent data. I will attempt to try to provide possible explanations for the inconsistency of results of the two studies and to draw some general conclusions.