931 GENOMIC SIGNATURE CORRELATING SENSITIVITY OF BLADDER CANCER CELL LINES TO EPIDERMAL GROWTH FACTOR RECEPTOR INHIBITORS

Abstract

INTRODUCTION AND OBJECTIVES: Inhibition of epidermal growth factor receptor (EGFR) has demonstrated efficacy in treatment in multiple late stage cancers. Predictive markers of EGFR inhibitor response vary across tumor types and a significant need exists for diagnostic tests capable of predicting response of tumors to EGFR directed therapies. To determine if a gene expression signature correlates with EGFR inhibitor sensitivity in bladder cancer we sought to correlate cell line gene expression patterns with sensitivity to EGFR inhibition. METHODS: 13 cell lines were screened for their sensitivity to the EGFR inhibitors lapatinib, neratinib, afatinib and canertinib. Supervised hierarchal cluster analysis was performed on gene expression data obtained from Affymetrix U133 arrays collected as part of the Cancer Cell Line Encyclopedia database. RESULTS: Of the lines tested, 4 (31%) demonstrated submicromolar sensitivity to at least one or more EGFR inhibitor. Those cell lines most sensitive include: HT1376, HT1197, Scaber and UMUC7. Interestingly, in those lines with EGFR resistance, gene expression was enriched in MAPKAPK2, AR, and ITGA1 consistent with alternative growth signaling pathways driving these tumors. Gene clustering demonstrated that a set of genes trended towards enrichment in sensitive cell lines with sensitive lines being more similar to each other than other lines in hierarchal cluster analysis. Interestingly, HT1376 and HT1197 while most similar by hierarchal cluster analysis differed by the presence of a PIK3CA mutation in the HT1197 line, perhaps explaining the difference in sensitivity to EGFR inhibition (0.1 vs 0.9 uM GI50). CONCLUSIONS: Future research expanding analysis to an additional 17 bladder cancer cell lines is expected to strengthen the association of a gene expression signature with sensitivity to EGFR inhibitors. Clearly, an unmet need exists for molecular diagnostics capable of predicting therapeutic response. Gene expression profiling as demonstrated here shows promise as an approach to predict patient response. These findings provide a rationale for designing prospective clinical trials with treatments decisions guided by molecular profiling.