Comparison of Magnetic Resonance Analysis of Myocardial Scarring With Biomarker Release Following S-T Elevation Myocardial Infarction.

Abstract

Background: Late gadolinium enhancement (LGE) with cardiac magnetic resonance (CMR) is commonly assumed to represent myocardial fibrosis; however, comparative human histological data are limited, and there is no consensus on the most accurate method for LGE quantitation. We evaluated the relationship between CMR assessment of regional fibrosis and infarct size assessment using serial biomarkers after ST elevation acute myocardial infarction (STEMI). Method(s): Ninety-three patients treated for STEMI (59 +/- 10 years, 86% male) underwent CMR 6 months after infarction. Infarct size was quantified by CMR-LGE using manual and range of semi-automated thresholds (range: 2-10 standard deviations [SD]) above reference myocardium and the full width-half maximum (FWHM) technique, and compared with the rise in serum biomarkers. The agreement between CMR and biomarker in the identification of large infarcts based on peak troponin (TnI) levels was also analysed. Result(s): Quantification methods had a strong influence on the infarct size assessment with CMR-LGE. Significant correlations were observed between LGE and biomarkers across all of the signal intensity thresholds. Whilst there was a wide variation with respect to the estimation of total LGE size (from 6.8 +/- 7.7 to 32.1 +/- 11.3 grams), the variation in the correlation with peak troponin level was much smaller (r-values ranging from 0.670 to 0.876). There was good agreement between CMR-LGE and biomarker assessment of infarct size; the best agreement between CMR-LGE and large infarction using a threshold of 8SD for peak TnI > 50 ng/mL (Cohen's kappa (kappa) = 0.722), and a threshold of 4SD for peak TnI > 95 ng/mL (kappa = 0.761). Conclusion(s): The correlation between CMR-LGE quantification of infarct size and biomarker release following STEMI at a range of semi-automated thresholds was consistently strong, with good agreement between measures across a range of thresholds.Copyright © 2018