Non-contrast cardiac CT-based quantitative evaluation of epicardial and intra-thoracic fat in healthy, recently menopausal women: Reproducibility data from the Kronos Early Estrogen Prevention Study
Authors: Eranthi Jayawardena, Dong Lia, Rine Nakanishi, Damini Dey, Christopher Dailing, Assad Qureshia, Brooke Dickens, Nicolai Hathiramani, Michael Kim, Ferdinand Flores, Ann E. Kearns, Li-Yung Lui, Dennis Black, Matthew J. Budoff
May 4th, 2019
Cardiac fat is emerging as an important parameter for cardiovascular risk stratification. Accurate and reproducible volumetric measurements can facilitate in the serial assessment of cardiac fat by computed tomography (CT). We assessed the intra- and inter-observer variability of cardiac fat volumetric measurements using a semi-automated CT software.
We used non-contrast coronary calcium CT scans to quantify epicardial and intra-thoracic fat volumes. Two expert readers analyzed baseline and follow up CT scans of 45 subjects by using a semi-automated CT software (QFAT 2.0, Cedars Sinai-Medical Center). Correlation and Bland-Altman analysis was performed for both intra- and inter-observer comparisons for each cardiac fat type.
The intra-observer correlation coefficients ranged between 0.86 to 0.99 and 0.87 to 0.99 for epicardial (median fat per reader (cm3) 20.9 to 25.7) and intra-thoracic (median fat per reader (cm3) 27.1 to 31.6) fat volumes respectively, with no significant differences between individual data points (all p > 0.38). The inter-observer correlation coefficient was 0.99 (p < 0.0001 for correlation) for both epicardial and intra-thoracic fat. By Bland-Altman analysis for epicardial fat measurements, mean difference of intra-observer was 0.90 cm3 with 95% confidence intervals (0.22,1.7) and −1.8 cm3 for inter-observer, with 95% CI (−2.9, −0.69). Bland-Altman plots for intra-thoracic fat measurements were similarly impressive for both inter- and intra-observer reads.
Our data showed that measuring epicardial and intra-thoracic fat volumes by CT using a semi-automated software has excellent intra-observer and inter-observer reliability. Cardiac fat volumes can be obtained easily and reproducibly from routine calcium scoring scans and may help in assessing cardiovascular risk.