INTRODUCTION: While continuous efforts are being made for efficient diagnosis of coronary artery disease (CAD), research for appraisal of myocardial ischemia in unstable angina (UA) is still needed because UA lacks tissue damage in contrast to myocardial infarction. Therefore, we applied NMR-derived metabolomic of serum samples obtained from patients presenting with UA and healthy volunteers (HC) with no history of angina. Our aims were two-fold: First, is NMR derived metabolomics sufficiently robust to identify perturbations in circulating metabolites and detect an ischemic episode in UA patients. Second, we wanted to employ this least-invasive metabolomic strategy to discriminate UA patients from HC in order to predict acuteness of the disease by preparing a statistical model of prediction using multivariate statistics.

MATERIALS AND METHODS: A total of 65 patients presenting with symptoms of acute UA (Braunwald Classification-II and III), and prolonged chest pain (even at rest), (a negative value of hs-cardiac Troponin-T, [<0.014ng/ml]) were enrolled in the study. Blood samples were collected within 4 hours of onset of angina. These patients were confirmed of having CAD by coronary angiography. The study comprised 62 age-comparable healthy controls (HC) with no prior history of angina/CAD. Serum was separated from blood (2ml) of each subjects using standard protocol. Each serum samples were passed through 3kDa cutoff centrifugal filters to remove abundant amount of several types of proteins and lipoproteins, and NMR experiments were performed on collected filtrates. A Bruker Avance III 800 MHz spectrometer was used to perform NMR experiments using 400μL of each filtered serum sample in a 5-mm NMR tube. Trimethylsilyl propionic acid sodium salt (TSP, 0.84 mmol/l) deuterated at CH2 groups was used for the deuterium lock, reference, and standard peak for the quantitation of metabolites. For all the specimens, one dimensional 1H NMR experiments were performed at 22oC by suppression of water resonance by pre-saturation. The parameters used were as follows: spectral width, 16666 Hz; time domain points, 65k; relaxation delay, 10s; pulse angle, 90o; number of scans, 128; and line broadening, 0.3 Hz. Multivariate chemometric analysis was applied on the NMR data with the help of ‘Unscrambler X’ software. The data were subjected for principal component analysis (PCA) followed by orthogonal partial least square discriminant analysis (OPLS-DA). To avoid over-fitting of the OPLS-DA model, an internal cross-validation (ICV) was also applied using 60% of the data as training set and 40% of the data as test set. A prediction model was constructed to detect potential biomarkers related to the discrimination between UA and HC cohorts. To evaluate the clinical utility of biomarkers derived from PLS-DA model, ROC analysis was also performed.

RESULTS: Figure 1 shows 1D 1H NMR spectra of filtered serum of HC and UA with an overview of complete metabolic profile and chemical shift assignments. PCA and OPLS-DA reveals that total eleven variables—creatinine, valine, alanine, glutamine, malonate, cis-aconitate, lactate, uracil, inosine, hippurate, and adenine—were playing major role to differentiate UA from HC. Subsequent spectral analysis reveals that valine, cis-aconitate, uracil, inosine, hippurate, and adenine were down regulated and creatinine, alanine, glutamine, malonate, and lactate were up regulated in UA with compared to HC. ROC analysis with leave-one-out approach reveals that mainly five variables—valine, alanine, glutamine, inosine and adenine—were playing major role to achieve utmost AUC of ROC (0.99) for the 95% differentiate of UA from HC with 96% sensitivity and 95% specificity.DISCUSSION: The main finding of this study is that the NMR-derived metabolomics technique is sufficiently robust and accurate to identify perturbations in the circulating small metabolites during an ischemic event in UA patients. Valine has been demonstrated to reverse the electrophysiologic changes¹ caused by metabolic inhibition and offers cardioprotection during acute ischemia^2-3. In order to maintain the ATP levels during ischemic episode, probably an augmented glutamine and alanine release leads to metabolic remodeling, as observed in present and previous observations^4-6. Inosine exhibited inhibitory action against adenosine diphosphate (ADP)-induced platelet activation⁷. However, platelet activation plays an important role in pathology of UA and the decreased level of both adenine and inosine correlates with platelet aggregation⁸. The correlation between platelet aggregation and CAD has well been documented in various studies^9-10. Apart from the above narrated description of most significant metabolites, six metabolites—lactate, malonate, creatinine, cis-aconitate, hippurate, and uracil— were also found to be considerably perturbed in UA compared to HC metabolic profile. In essence, our findings provide evidence in support of metabolomic strategies that can be efficiently used to describe and relate the metabolic remodeling during ischemia in acute UA milieu.