Background: Risk stratification and management of patients with acute coronary syndromes (ACS) is challenging. Aim of this study was to evaluate the possible role of metabolomics in the prognostic stratification of ACS patients. Methods: 918 patients (345 females, 633 males, median age 74) were enrolled; among these 146 died (negative outcome), whereas 832 showed a positive outcome within 2 years from the cardiovascular event. Patients serum samples were analyzed via high resolution Proton Nuclear Magnetic Resonance and the obtained spectra were used to characterized the metabolic profiles of the two cohorts of patients. Multivariate statistics and a Random Forest classifier were used to create a prognostic model for the prediction of death within 2 years from the cardiovascular event. Results: In the training set (n=80 positive outcomes, n=80 negative outcomes), metabolomics showed significant differential clustering, with a good separation of the two outcomes cohorts. A prognostic risk model predicted death with sensitivity, specificity, and predictive accuracy of 78.5% (95%CI 77.7-79.3%), 69.9% (95%CI 69.2-70.5%) and 74.3% (95%CI 73.6- 74.8%), respectively, and an area under the ROC curve of 0.846. These results were reproduced in an independent test set (n=752 positive outcomes, n=66 negative outcomes), obtaining 67.3% sensitivity, 86.4% specificity and 84.8% predictive accuracy. The known prognostic factors age, sex, previous CABG, previous PCI, heart failure, atrial fibrillation, cerebrovascular disease, diabetes, creatinine concentration, Killip class, and acute coronary syndrome classification were compared with the NOESY1D RF risk score, calculated on the test set, in univariate and multivariate regression analyses. In the univariate analysis many of prognostic factors were statistically associated with the outcomes, but the RF score shows the p-value by far more significant (p=2.65e-12). Moreover, in the multivariate regression only the age, the Killip class and the RF score still remain statistically significant, demonstrating their independence with respect to the other variables.
|Titolo:||Metabolomics by Nuclear Magnetic Resonance identifies patients with high risk of death within two years after a cardiovascular event: The case of the amiflorence II study|
|Anno di registrazione:||2017|
|Autori di Ateneo:|
|Autori:||Tenori, L.; Giusti, B.; Vignoli, A.; Gori, A.M.; Luchinat, C.; Grifoni, E.; Barchielli, A.; Balzi, D.; Marchionni, N.; Marcucci, R.|
|Appare nelle tipologie:||1c - Abstract su rivista|