Four of five different models for profiling gene expression in breast cancers produced concordant results, showing significant agreement in predicting disease outcome even though they tested for evidence of different genes.
Researchers used samples from breast cancers excised from 295 women to compare the results of five different gene-expression profiling studies. The studies assess whether given genes are expressed in the tumors, which enables clinicians to more accurately estimate the tumor's aggressiveness, wrote Dr. Cheng Fan of the University of North Carolina, Chapel Hill, and associates (N. Engl. J. Med. 2006 355:560–9).
The first model uses gene-expression profiles to identify the cancer's subtype, differentiating between tumors that originate from luminal cells or basal cells and determining if the tumor is human epidermal growth factor receptor (HER)-2 positive.
The second model identifies the levels of expression of 70 genes thought to regulate the cell cycle, invasion, metastasis, and angiogenesis. The third model calculates the likelihood of cancer recurrence in estrogen receptor (ER)-positive, node-negative tumors by assessing their expression of 21 genes. The fourth model assesses the expression of wound-response genes, which identifies tumors that are more likely to metastasize because they have activated pathways for matrix remodeling, cell motility, and angiogenesis.
All four of these models were significant predictors of relapse-free survival and overall survival.
The fifth model uses a ratio of the levels of expression of two genes: one encodes homeobox 13; the other encodes the interleukin-17B receptor. This model was not predictive of survival.