Abstract
PURPOSE: To establish the novel prognostic markers for breast cancer, gene expression profile was examined genome-wide. METHODS: We used cDNA microarray consisting of 18,432 human genes to compare genome-wide expression profiles of eight primary breast cancers, four from patients who died of breast cancer within 5 years after surgery (5D group) and four who survived disease-free for more than 5 years (5S group). RESULTS: We identified 21 genes whose expression was greater in tumors from the 5D group than in 5S tumors, and 23 with higher expression in the 5S group than in the 5D group. We established a Prognostic Index (PI) for prediction of postoperative prognosis, based on the aberrant expression profiles of ten of those genes. Among 20 additional cases chosen blindly, ten presented with high prognostic scores (>7, good) according to the PI; the remaining ten cases revealed scores <7 (poor). The PI predicted the actual 5-year clinical outcomes of these 20 cases with 100% accuracy. CONCLUSION: Our PI system is reliable in clinical settings for predicting postoperative risk for breast cancer. The extensive list of genes provides valuable information about progression of breast cancer and suggests potential target molecules for treating this disease.