Relation of gene expression-based tumor subclasses to clinical phenotypes

Full text

Gene expression analysis by cDNA microarrays is a powerful tool for characterizing the variation in transcriptional programs in cells and tissues. We have analysed surgical specimens from 40 human breast tumors using cDNA microarrays representing 8000 human genes. From 20 of the tumors, pairs of biopsies were obtained both before and after a 16-week course of doxorubicin chemotherapy. Two of the tumors were paired with lymph node metastases. Expression patterns reflecting specific features of physiological variation and cellular composition of the tumors provided unique molecular portraits of each tumor. Different clusters of co-expressed genes, like the estrogen receptor, the ERBB2 gene and basal cell specific genes, were used to further subclassify the tumors. The different subgroups, identified by unique expression patterns, were associated with clinical parameters such as overall survival. Also, the TP53 mutation status in the tumors was associated with disease outcome within these subclasses. Although the sample size in any of the subgroups is too small to support any statistically robust tests, our preliminary results show the potential of gene expression-based subtyping of breast cancer in predicting clinical behaviour and therapy response.