Identifying resistance biomarkers against five clinically approved tyrosine kinase inhibitors in 45 cell lines.

Abstract

e21005 Background: Clinical studies show low overall response rates of 10-47% to targeted cancer therapeutics. Our aim was to identify new biomarkers for predicting sensitivity against five already approved tyrosine kinase inhibitors. Methods: Sensitivity to sunitinib, erlotinib, lapatinib, sorafenib and gefitinib were tested in 45 cancer cell lines, and a resistance index was calculated for each cell line. The gene expression profiles (data were obtained by interrogating the raw microarray data of the caArray database) in the subset of resistant vs. sensitive cell lines were compared for each drug. Feature selection was carried out using significance analysis of microarrays and rank products. The results were validated by qPCR in the cell lines and - in case of four sunitinib resistance associated genes - in clinical samples by immunohistochemistry. Results: A set of 63 genes was identified as associated with resistance against the examined drugs. Overall classification accuracy of the prediction was 92.8% in a leave-one-out cross validation using prediction analysis of microarrays. The expression of the genes was validated by qPCR in the cell lines. The results confirmed 45/63 of the microarray-based resistance associated genes and 7/32 of literature based genes. All together 48 sunitinib-treated metastatic renal cell carcinomas were collected. The immunohistochemical analysis in these confirmed the correlation of the expression of RAB17, LGALS8, and EPCAM with overall survival. Conclusions: We identified new predictive biomarker candidates for five tyrosine kinase inhibitors, and validated a set of sunitinib resistance associated genes in an independent patient cohort.