Inter and Intra-observer Variability of Epicardial Adipose Tissue T1 Mapping

Background: Epicardial adipose tissue (EAT) is increasingly recognized as a metabolically active visceral fat depot with both local and systemic effects on cardiovascular function. Quantitative assessment of EAT volume and thickness has been associated with adverse cardiac outcomes. More recently, qualitative assessment of EAT by cardiac magnetic resonance (CMR) T1 mapping has been proposed as a potential assessment of EAT quality, potentially offering superior risk stratification compared with EAT volume. However, the reproducibility and utility of EAT T1 mapping as a clinical tool remains undefined. We aimed to evaluate the reproducibility of T1 measurements of the various thoracic adipose tissue depots.

Methods: We retrospectively identified 25 patients who had undergone clinically indicated CMR and had short and long axis T1 mapping performed using a MOLLI 5(3)3 acquisition. Two independent readers performed measurements of the T1 values of EAT, paracardiac adipose tissue (PAT), visceral adipose tissue (VAT), and subcutaneous adipose tissue (SAT). For each depot, the largest freehand ROIs possible were drawn on separate slices while avoiding areas of artifact, myocardium, or coronary vessels. The ROI for each fat depot from the three slices was then averaged to create the final T1 value. Intra- and inter-observer reproducibility was assessed using intraclass correlation coefficients (ICC).

Results: EAT T1 mapping demonstrated poor inter-reader reproducibility, with ICC of 0.27 (95% CI -0.1. to 0.59, p=0.079). These results were similar among the other adipose depots with VAT T1 (ICC 0.23 (95% CI -0.18 to 0.57, p=0.135)), PAT T1 (ICC 0.39 (95% CI 0.01 to 0.68, p=0.024)), and SAT (ICC 0.27 (CI -0.09 to 0.58, p=0.073)). Inter-observer reproducibility was also poor (ICC 0.39, p=0.024). In summary, none of the adipose depots demonstrated good inter or intra-observer reproducibility for T1 mapping measurements.

Conclusion: EAT T1 mapping, as currently implemented, demonstrates poor reproducibility and is not yet a reliable method for assessing EAT quality. The high artifact burden, poor spatial resolution for identifying small structures, and small depots of adipose tissue all contribute to this poor reproducibility. Technical refinements, including improved acquisition protocols and automated segmentation approaches, may help enhance the utility of this as a measure in the future.