Nuclear magnetic resonance-based metabolomics predicts exercise-induced ischemia in patients with suspected coronary artery disease

Abstract

The purpose of this study was to develop a (1)H-nuclear magnetic resonance metabolomic approach capable of predicting the occurrence of exercise-induced ischemia in patients with suspected coronary artery disease and to identify the metabolite patterns that contribute most importantly to the prediction. In 31 patients with suspected effort angina and without previous myocardial infarction, serum was obtained just prior to a stress single-photon emission computed tomography. Serum NMR spectra were acquired with pulse-and-acquire and T(2)-edited sequences. The region between 0.50 and 4.25 ppm was used for analysis. Twenty-two patients had reversible myocardial perfusion defects and nine did not. Both groups had similar age and clinical profile, except for more smokers and diabetics in the ischemia group, and attained a similar peak heart rate. The best separation was achieved with long T(2)-edited spectra, 84% of patients being correctly classified based on the partial least square discriminant analysis. The main contributors to discrimination were lactate, glucose, as well as methyl and methylene moieties of lipids and long-chain amino acids. Metabolomic analysis of serum can predict exercise-inducible ischemia in patients with suspected coronary artery disease. This capability could be useful in screening and risk stratification of patients with coronary risk factors.