Zunyan Dai1, Catalin Barbacioru, Ying Huang, Wolfgang Sadée. 1. Program of Pharmacogenomics, Department of Pharmacology, The Ohio State University, 5078 Graves Hall, 333 West 10th Avenue, Columbus, 43210, USA.
Abstract
PURPOSE: This study develops and evaluates a systematic approach to finding biomarker genes for predicting potency of anticancer drugs against tumor cells, focusing on gene families related to growth factor signaling. METHODS: Cytotoxic potencies of 119 drugs against 60 neoplastic cell lines (NCI-60) were correlated with expression of 343 genes, including 90 growth factors and receptors, 63 metalloproteinases, and 92 ras-like GTPases as downstream signaling factors. Progressively more stringent criteria and predictive models aim at identifying the smallest subset of genes predictive of cytotoxic potency. RESULTS: Comparing gene expression with drug potency across the NCI-60 yielded genes with negative and positive correlations (p < 0.001), indicative of a role in chemoresistance and chemosensitivity, respectively. Of 17 genes with multiple negative correlations, 8 are known chemoresistance factors, validating the approach. Negatively correlated genes clustered into two main groups with distinct expression profiles and drug correlations, represented by EGFR and ERBB2 (Her-2/Neu). Accordingly, no synergism was observed between EGFR and ERBB2 inhibitors. However, combinations with classical anticacer drugs were not correlated with EGFR and ERBB2 expression in four cell lines tested, suggesting complex interactions in combination treatments. Finally, a subset of only 13 genes was found to be sufficient for near optimal prediction of drug potency against the NCI-60. CONCLUSIONS: Our approach using a small subset of genes reveals known and potential biomarkers in cancer chemotherapy, providing a strategy for genome-wide analysis.
PURPOSE: This study develops and evaluates a systematic approach to finding biomarker genes for predicting potency of anticancer drugs against tumor cells, focusing on gene families related to growth factor signaling. METHODS:Cytotoxic potencies of 119 drugs against 60 neoplastic cell lines (NCI-60) were correlated with expression of 343 genes, including 90 growth factors and receptors, 63 metalloproteinases, and 92 ras-like GTPases as downstream signaling factors. Progressively more stringent criteria and predictive models aim at identifying the smallest subset of genes predictive of cytotoxic potency. RESULTS: Comparing gene expression with drug potency across the NCI-60 yielded genes with negative and positive correlations (p < 0.001), indicative of a role in chemoresistance and chemosensitivity, respectively. Of 17 genes with multiple negative correlations, 8 are known chemoresistance factors, validating the approach. Negatively correlated genes clustered into two main groups with distinct expression profiles and drug correlations, represented by EGFR and ERBB2 (Her-2/Neu). Accordingly, no synergism was observed between EGFR and ERBB2 inhibitors. However, combinations with classical anticacer drugs were not correlated with EGFR and ERBB2 expression in four cell lines tested, suggesting complex interactions in combination treatments. Finally, a subset of only 13 genes was found to be sufficient for near optimal prediction of drug potency against the NCI-60. CONCLUSIONS: Our approach using a small subset of genes reveals known and potential biomarkers in cancer chemotherapy, providing a strategy for genome-wide analysis.
Authors: Gretchen A Murphy; Suzanne M Graham; Staeci Morita; Sarah E Reks; Kelley Rogers-Graham; Anne Vojtek; Grant G Kelley; Channing J Der Journal: J Biol Chem Date: 2002-01-11 Impact factor: 5.157
Authors: P Skehan; R Storeng; D Scudiero; A Monks; J McMahon; D Vistica; J T Warren; H Bokesch; S Kenney; M R Boyd Journal: J Natl Cancer Inst Date: 1990-07-04 Impact factor: 13.506