J. Yu1, G. Zhou1, S. Liu1, R. Sanchez1, R. Damoiseaux2, F. C. Brunicardi1 1University Of California At Los Angeles,Department Of Surgery, David Geffen School Of Medicine,Los Angeles, CA, USA 2University Of California At Los Angeles,Department Of Physiology, David Geffen School Of Medicine,Los Angeles, CA, USA
Introduction:
While each type of malignancy has its own genotype and phenotype, there are commonalities among all cancers as a disease. In the era of precision medicine, it is essential to determine the genomic signature of each type of cancer, which could reveal commonalities among cancer. Microarray and next-generation sequencing techniques have revealed a series of somatic mutations and differentially expressed genes (DEG) associated with multiple types of cancers. Our objective was to identify a set of potentially actionable genes for nine cancers using a novel combination of systematic genomic analysis and published cancer microarray databases and to determine whether there exist overlapping actionable genes among these cancers.
Methods:
A total of 12 gene expression microarray datasets containing 9 different solid cancer types (n=1016) were downloaded from the Gene Expression Omnibus, including 104 breast cancer, 117 brain tumor, 36 colon cancer, 108 gastric cancer, 155 liver cancer, 72 pancreatic cancer, 72 renal cancer, 6 prostate cancer and 346 matching non-tumor control tissue samples. A weighted gene co-expression network analysis (WGCNA) was used to compute gene expression network and to determine the connectivity and significance of genes for each cancer type and whether there were common genes among the nine cancer types.
Results:
WGCNA of a total of 1016 gene expression data revealed specific gene modules for each cancer type. Gene co-expression networks were constructed and actionable genes for the nine cancer types were identified. Importantly, one particular module contained differentially overexpressed genes across all nine cancer types versus their matching non-tumor controls. Of these, the actionable genes BIRC5, TPX2, CDK1, and MKI67, well-known cancer genes, were significantly enriched in cell cycle and cell proliferation pathways. In addition, a DEG-survival correlation matrix was constructed and resulted in a list of actionable genes that were significantly associated with the overall patient survival across these cancer types. Strikingly, the solute carrier family proteins SLC6A13, SLC13A1 and SLC5A12 mediating ion/glucose transport led the gene list, suggesting a strong correlation between ion/glucose transport and cancer patient survival.
Conclusion:
The systematic genomic analysis utilizing a large collection of cancer gene expression microarray datasets and WGCNA reveals potentially actionable cancer genes unique for each cancer type. A shared gene module containing actionable genes common to the nine cancer types involving cell cycle and cell proliferation pathways was identified in support that cancer types may have a shared core molecular pathway. Specific gene modules for each type of the cancers may provide better understanding of molecular mechanisms for these cancers, and provide potential therapeutic targets for precision medicine.