Expression profiling in lung adenocarcinomas with or without epidermal growth factor receptor (EGFR) gene mutation at exons 18 - 21 reveals expression signatures related to the EGFR pathway

Abstract

Background: Lung cancer is a leading cause of cancer morbidity and mortality worldwide related to advanced staging at the time of presentation or diagnosis. Epidermal Growth Factor Receptor (EGFR) with activating mutations is a potential new molecular therapeutic target responsive to tyrosine kinase inhibitor therapy. Our hypothesis is that there are differential expression gene signatures between lung adenocarcinomas with or without EGFR mutation Methods: Total RNA from 49 resected primary lung adenocarcinomas were analyzed individually for their expression profiles by Affymetrix GeneChip HG-U133 set according to standard protocols. EGFR mutations at exons 18-21 were evaluated by sequencing of cDNA. Expression data were analyzed with publicly available software packages for microarray analysis. Results: 29 tumors showed EGFR mutations (23/32 women, 6/17 men). The mutant cases were significantly associated with life time never smokers (26/32) compared to current or previous smokers. After data normalization across all cases and exclusion of genes showing significant proportion of absent call and genes showing low variance, differential gene expression was compared between tumors with or without EGFR mutations using log transformed signal intensities and two sample t-tests, yielding 318 genes that showed significantly (p<=0.01) higher or lower expression by two fold difference between groups. Hierarchical clustering aiming at detection of co-regulated or co-expressed genes showed significant sample clustering including groups that contained predominantly tumors with or without EGFR mutation. Molecular classification was done with Support Vector Machine with Leave-one out analysis and the Machine Learning Performance was 96% sensitivity and 80% specificity in the prediction of tumours with EGFR mutation. Some examples of differentially expressed genes and molecular classifiers included EGFR, FGFR3, EREG, WIF3, INHBA and TNFSF10. Conclusion: We have identified differentially-expressed and class-predictive genes in primary lung adenocarcinomas with respect to EGFR gene mutation status at exons 18-21. These genes are potentially involved in the EGFR oncogenic pathway and serve as putative diagnostic or therapeutic targets in the management of lung cancer. (Supported by HKSAR RGC grants 7310/01M, 7486/03M, 7468/04M)