Plasma Levels of Propionylcarnitine Improved Prediction of Heart Failure and All-Cause Mortality in Patients with Stable Coronary Artery Disease

Abstract

Background: Plasma metabolites could be suitable as predictive biomarkers for cardiovascular pathologies or death, thereby improving the prediction of protein biomarkers. The release of acylcarnitines may be altered after coronary artery disease (CAD) in subjects with recurrent clinical outcomes, and this could be used as a prognosis tool.

Methods: Patients with stable coronary artery disease (SCAD) who had suffered an acute coronary syndrome 6-9 months before were followed for up to 4.3 years for adverse events. Soluble pro-inflammatory/fibrotic proteins, and a panel of 13 amino acids and 13 acylcarnitines, were evaluated by ELISA and metabolomics analyses as potential predictors of a primary outcome [heart failure (HF) or death].

Results: Among 139 patients (67.0 years old, BMI = 28.6 kg/m², and 71.2% male), 25 developed the primary outcome after a mean follow-up of 2.2 years. These patients showed increased plasma levels of NT-proBNP (1300 vs. 250 pg/mL; p < 0.001), pro-inflammatory/fibrotic MCP-1 (1.7 vs. 1.4 × 10² pg/mL; p = 0.043), Gal-3 (12.7 vs. 7.9 ng/mL; p < 0.001), and NGAL (2.7 vs. 1.6 × 10² ng/mL; p < 0.001), and lower acetyl- and propionylcarnitines (0.59 vs. 0.99 µM, p = 0.007, and 3.22 vs. 6.49 × 10^-2 µM, p < 0.001, respectively). Instead, plasma amino acids were not significantly changed. Through a multivariable logistic regression analysis, a combined model of age, Gal-3, and the NGAL/propionylcarnitine ratio showed the highest prediction for HF or death (AUC = 0.88, sensitivity = 0.8, and specificity = 0.81; p < 0.001).

Conclusions: Patients with SCAD led to recurrent HF or all-cause death. Interestingly, increased levels of plasma NGAL and Gal-3, and a reduction in propionylcarnitine, could predict the occurrence of these events.

Keywords: acylcarnitine; biomarkers; coronary artery disease; metabolomics.

Figure 1

Risk factors, comorbidities, and treatments. A bar chart representing the relative frequencies of risk factors and comorbidities (a), and pharmacological treatments (b), in each group of patients with or without the primary outcome. The p-values were obtained from the Chi-square or Fisher’s exact tests, supplemented by the Z-test for proportions. * p-value < 0.05, ** p-value < 0.01, and *** p-value < 0.001. LMCA, left main coronary artery; PAD, peripheral artery disease; LVEF, left ventricular ejection fraction; PTCA, percutaneous transluminal coronary angioplasty; MRA, mineralocorticoid receptor antagonists; ARBs, angiotensin II receptor blocker; CCBs, calcium channel blockers; ACEIs, angiotensin-converting enzyme inhibitors.

Figure 2

Simple logistic regression analysis of risk factors for HF or all-cause death. A binary logistic regression followed by a receiver operating characteristic (ROC) analysis was applied. The forest plot displays the odds ratios (ORs) and 95% confidence intervals (CIs) for each univariable model. eGFR, estimated glomerular filtration rate; C2:0, acetylcarnitine; C3:0, propionylcarnitine. NT-proBNP, N-terminal pro-brain natriuretic peptide; MCP-1, monocyte chemotactic protein 1; Gal-3, galectin-3; NGAL, neutrophil gelatinase-associated lipocalin. Those values of the area under the ROC curve (AUC) greater than 0.7 (in bold) were used for further analysis. * p = 0.05 vs. C3:0 (by DeLong’s test).

Figure 3

Multivariable logistic regression and ROC analysis for the predictive factors. Top, Model “A”, represents the initial model obtained by using the forward stepwise method (NGAL/C3:0); Model “B” reflects the incorporation of Gal-3 into the previous model, and Model “C” shows the final model achieved by the stepwise procedure. The odds ratio (OR) with 95% confidence interval (CI), p-value, and area under the ROC curve (AUC) with 95%CI are exposed for each model. Bottom, ROC curves (AUC) obtained by using the leave-one-out (LOO) cross-validation method. C3:0, propionylcarnitine; Gal-3, galectin-3; NGAL, neutrophil gelatinase-associated lipocalin.

Figure 4

Potential evolution of SCAD to HF or all-cause death. After the occlusion of the coronary artery, the cardiac cell may induce mitochondrial adaptations (i.e., switch of energetic substrate and the subsequent reduction in SCAC) followed by pro-inflammatory (i.e., NGAL) and pro-fibrotic (i.e., Gal-3) overexpression. However, some patients might enforce and prolong this cardiac remodeling, which could lead to HF or worse evolution SCAC, short-chain acylcarnitines.