Integrated gene expression profile predicts prognosis of breast cancer patients

Abstract

Gene expression data has in recent years demonstrated the superior capacity to predict the prognosis of breast cancer patients unreceiving adjuvant chemotherapy comparing to the information available from traditional clinical and pathological sources. Meanwhile, adjuvant chemotherapy can significantly improve survival of breast cancer. It would be inappropriate to ignore its effect on prognosis. We hypothesized that an integrated gene expression profile can predict the prognosis of breast cancer patients receiving chemotherapy. Therefore, we screened the specific gene markers and constructed an integrated 24-gene signature by low-density microarray including the "poor signature" and genes related to resistance to chemotherapy. The gene signature stratified correctly patients into good prognosis group and poor prognosis group. In addition, the Kaplan-Meier analyses for disease-free survival as a function of the 24-gene signature showed highly significant differences between the two groups (Log Rank test P < 0.0001 = Univariate and multivariate Cox's proportional-hazards regression analyses indicated that the signature represents the strongest independent prognostic factor for breast cancer patients. When compared with single signature, such as Oncotype DX and 70 poor signature, the integrated signature showed more predominant power of predication in breast cancer patients receiving chemotherapy. Such integrated signature will critically aid clinical decision making at the level of individualization for most breast cancer patients receiving chemotherapy.