Prediction of doxorubicin sensitivity in breast tumors based on gene expression profiles of drug-resistant cell lines correlates with patient survival

Up to date clinical tests for predicting cancer chemotherapy response are not available and individual markers have shown little predictive value. We hypothesized that gene expression patterns attributable to chemotherapy-resistant cells can predict response and cancer prognosis. We contrasted the expression profiles of 13 different human tumor cell lines of gastric (EPG85-257), pancreatic (EPP85-181), colon (HT29) and breast (MCF7 and MDA-MB-231) origin and their counterparts resistant to the topoisomerase inhibitors daunorubicin, doxorubicin or mitoxantrone. We interrogated cDNA arrays with 43 000 cDNA clones (similar to 30 000 unique genes) to study the expression pattern of these cell lines. We divided gene expression profiles into two sets: we compared the expression patterns of the daunorubicin/doxorubicin- resistant cell lines and the mitoxantrone-resistant cell lines independently to the parental cell lines. For identifying predictive genes, the Prediction Analysis for Mircorarrays algorithm was used. The analysis revealed 79 genes best correlated with doxorubicin resistance and 70 genes with mitoxantrone resistance. In an independent classification experiment, we applied our model of resistance for predicting the sensitivity of 44 previously characterized breast cancer samples. The patient group characterized by the gene expression pro. le similar to those of doxorubicin-sensitive cell lines exhibited longer survival (49.7 +/- 26.1 months, n = 21, P = 0.034) than the resistant group (32.9 +/- 18.7 months, n = 23). The application of gene expression signatures derived from doxorubicin-resistant and -sensitive cell lines allowed to predict effectively clinical survival after doxorubicin monotherapy. Our approach demonstrates the significance of in vitro experiments in the development of new strategies for cancer response prediction.