Abstract 1553: Comparative Gene Expression Analysis of Blood and Heart Identifies Genomic Predictors of Myocardial Injury Following Cardiopulmonary Bypass and Cardioplegic Arrest in the Rat

Abstract

Background: Perioperative myocardial injury (PMI) complicating cardiac surgery remains poorly predicted by clinical risk factors. In a novel rat model of cardiopulmonary bypass (CPB) and cardioplegic arrest (CA) we tested the hypothesis that microarray expression profiles in peripheral blood leukocytes (PBL) can accurately predict degree of PMI, and compared genomic signatures in left ventricular myocardium (LVM) and PBL to identify PMI-associated genes that generalize across tissues. Methods: Male rats subjected to CPB only (75 min), CPB with CA (30 min), and sham surgery (n=5 each) had plasma heart fatty acid binding protein (HFABP) and cardiac troponin I (cTnI) measured 1 hour post CPB and microarray profiling of total RNA from LVM and PBL. PMI genomic classification models were constructed using shotgun stochastic search approach and Bayesian model averaging and their accuracy tested by five-fold cross validation. Results: A spectrum of PMI was observed in the experimental groups as well as robust deregulation of gene expression in a stimulus and tissue-specific manner (Figure ). Estimates of PMI classification accuracy from PBL and LVM are presented in Table . Conclusions: Cardiac surgery induces changes in leukocyte gene expression that predict the spectrum of PMI, with potential applications in perioperative risk stratification and selection of cardioprotective strategies. Predictive accuracy of tissue-specific myocardial injury genomic classifiers