Abstract P273: Association of Gene Expression Signatures with Small Molecule Metabolic Intermediates that Predict Cardiovascular Mortality in CATHGEN
Diagnostic gene expression (GEx) signatures for cardiovascular disease (CVD) have been developed and are now in clinical use. However, a gene expression signature predicting CVD events has yet to be described and the development of which would be considered a major advance in diagnostic and mechanistic science. Over the last several years we have identified blood-borne small metabolic intermediates that are associated with major adverse coronary events, including mortality alone. Specifically, short chain dicarboxylacylcarnitines (SCDC), long chain dicarboxylacylcarnitines (LCDC) and long neutral amino acids (LNAA) have been the strongest and most consistent diagnostic and predictive markers. We reasoned that examining the relation of peripheral blood GEx signatures to these metabolites might provide insight into GEx signatures independently related to disease. CATHGEN is a sample repository of over 9300 individuals that underwent coronary artery catheterization in the Duke laboratory 2001-2011 and for which vital and clinical disease status is updated yearly. Here, we examined samples collected 2004-2008. Metabolite factor scores were derived from previous work, wherein metabolic intermediates were grouped into metabolic factors using principal components analysis. We had available 1283 peripheral blood gene expression profiles, derived from sequentially ascertained subjects irrespective of disease status, obtained with the Illumina platform and surviving a rigorous QC process. P-values for GEx candidates were corrected for false discovery rate (FDR) and only those with FDR p-values<0.05 were further considered. GEx tags were identified that were significantly related to the SCDC (n=42), LNDC (n=2) and LNAA (n=515) factors using regression analysis and controlling for age and gender. We consider here only the SCDC-associated genes. For these SCDC-associated genes, the relation to mortality was considered, again controlling for FDR. For all 42 probes, the uncorrected p-values ranged from 4.1E-8 to 9.5E-5 and when correcting for FDR, 4.7E-5 to 2.8E-3. Identified genes were over-represented in tyrosine, adipocytokine, glycerol phospholipid, and urea cycle metabolism KEGG pathways using a gene set enrichment analysis. This represents, to our knowledge, the first description of predictive GEx signatures for CVD mortality and could add in development of new diagnostics or therapeutic targets.
- © 2013 by American Heart Association, Inc.