Clustering of 37 circulating biomarkers by exploratory factor analysis in patients following complicated acute myocardial infarction

Background: The objective of this study was to elucidate the complex interactions between families of circulating biomarkers representing different biochemical responses to the pathophysiology following complicated acute myocardial infarction (AMI). Methods: Blood samples, drawn at a median of 3 days post AMI were obtained from 236 patients with complicated AMI and evidence of heart failure or left ventricular dysfunction. Using exploratory factor analysis, 37 biomarkers were grouped according to their collinearity to each other into clusters. The clusters were used as a model to elucidate interdependencies between individual biomarkers. Each cluster defines a specific pathophysiological process, called factor. These factors were used as covariates in multivariable Cox-proportional hazard regression analyses for prediction of all-cause death and the combined endpoint of cardiovascular death and re-infarction. Results: Exploratory factor analysis grouped the biomarkers under 5 factors. The composition of these groups was partially unexpected but biological plausible. In multivariable analysis, only 1 factor proved to be an independent predictor of outcome. Major contributions (factor loadings>0.50) in this cluster came from: mid-regional pro-adrenomedullin, tumor necrosis factor receptor, pro-endothelin-1, growth differentiation factor 15, C-terminal pro arginine vasopressin, uric acid, chromogranin A and procollagen type III N-terminal. Conclusion: Clustering of multiple biomarkers by exploratory factor analysis might prove useful in exploring the biological interactions between different biomarkers in cardiovascular disease and thus increase our understanding of the complicated orchestral interplay at the molecular level.