Standardized
MRI
‐Based Quantification of Epicardial Adipose Tissue Volume: Reproducibility and Agreement With Cardiac Computed Tomography
Judith Gronwald, Svante S. Gersch, Zoé Böttiger, Clara Hagedorn, Torben Lange, Sören J. Backhaus, Sebastian Kelle, Constanze Schmidt, Karl Toischer, Andreas Schuster, Alexander Schulz ABSTRACT
Background
Epicardial adipose tissue volume (EATV) is increasingly recognized as a cardiometabolic risk marker associated with adverse outcomes. The most established approach for EATV quantification is cardiac computed tomography (CT). MRI offers a radiation‐free alternative allowing simultaneous assessment of myocardial function and tissue characteristics; however, standardization and validation are limited.
Purpose
To evaluate a standardized MRI‐based method for EATV quantification and determine its agreement with CT.
Study Type
Retrospective.
Population
127 patients with aortic stenosis (AS) (78 ± 6 years; 38% female) who underwent paired CT and MRI and 11 volunteers (74 ± 7 years, 45% female) who underwent repeat MRI after ≥ 6 weeks.
Field Strength/Sequence
Short‐axis balanced steady‐state free precession cine sequence at 3T (AS patients) and 1.5T (volunteers).
Assessment
On CT, EATV was quantified by manual delineation of the visceral pericardium and voxel‐thresholding (−190 to −30 HU). MRI‐based EATV quantification used manual volumetry with delineation of the visceral pericardium and epicardium on end‐diastolic short‐axis cine stacks.
Statistical Tests
Inter‐modality agreement was assessed by Spearman correlation and Bland–Altman analysis. Reproducibility was evaluated in 20 patients using intraclass correlation coefficient (ICC) and coefficient of variation (CoV). Scan‐rescan reproducibility for MRI‐derived EATV quantification was assessed using ICC and linear regression. p < 0.05 was considered significant.
Results
Median EATV was significantly higher on MRI than CT (47 vs. 38 mL/m 2 ), with a significant moderate correlation ( ρ = 0.627) between measures. Inter‐ and intra‐observer analyses showed excellent reproducibility for both modalities (CT: intra‐observer ICC: 0.983, inter‐observer ICC: 0.994; MRI: intra‐observer ICC: 0.955, inter‐observer ICC: 0.970). MRI‐derived EATV quantification also showed excellent scan‐rescan reproducibility (ICC: 0.985).
Data Conclusion
The standardized MRI‐based approach enabled highly reproducible EATV measurements with excellent repeatability. Agreement with CT was moderate, with systematically higher values on MRI, limiting direct comparability.
Evidence Level
3.
Stage of Technical Efficacy
2.