An evaluation and application of publically available microarray data repositories in colorectal cancergeneration and enrichment of a consensus profile differentiating colorectal primary and metastatic tumours

Introduction: Publically available microarray repositories (PAMR) continue to expand. These could represent significant opportunities (through pooling of large data-sets) for high-throughout bioinformatic appraisals of colorectal metastasis formation. This study aimed to (1) assess data accessible in PAMR relating to colorectal metastasis formation, (2) develop a mechanism for cross-comparison of different experiments (i.e. generating a consensus transcriptomic profile) and (3) perform an hypergeometric enrichment analysis using Ingenuity, a technique which mines literature to identify significant linkages with signalling pathways and therapies. Methods: All PAMR repositories were searched for experiments relating primarily to colorectal primary and lymphatic or hepatic metastases. Following filtration to exclude non-MIAME compliant experiments, seven suitable studies were identified from which 332 microarrays were imported into multiple experimental viewer (MeV) for analysis. Differential gene expression profiles (GEPs) were generated from each experiment then cross compared to identify overlaps. Overlaps were characterised in terms of frequency and concordance (i.e. genetic changes in the same direction in relation to up or down-regulation) thereby establishing a consensus profile (CP) of genes. CP Genes underwent hypergeometric enrichment using Ingenuity. Results: Primary analysis generated a155 genetic profile differentiating primary and metastatic tumours. Enrichment identified two significantly associated signaling pathways (a) SMAD2, BMP2, CSDA, ITCH and ZMYND1 and (b) SCLS1, LYN, PRKCD and WNT5A. 33 genes of this geneset were linked to colorectal cancer in scientific literature (p = 2.52 9 1014). Both pathways were retained in the consensus profile differentiating primary and metastatic tumours. Comparison of primary tumours and lymphatic metastases generated a separate consensus profile containing genes encoding FYN, LYN and MAPK9. 17 of these genes from this geneset were shown to be associated with colorectal cancer in the scientific literature (p = 2.76 9 103). Further enrichment analysis identified Dasatinib, an FYN inhibitor, and thus potential inhibitor of lymphatic metastasis formation. Conclusions: The identification of overlaps in gene expression profiles permits generation of consensus genetic profiles. Using this approach consensus expression profiles were established that differentiate primary colorectal tumours from lymphatic and hepatic metastases. Enrichment analysis identified novel signaling targets and corresponding therapeutic agents. Conflict of interest: None Disclosures: None