Rosenblatt AJ, Scrivani PY, Boisclair YR, et al.
Introduction/Purpose: Body fat content is a predictor or outcome of disease, metabolic processes, and therapeutic responses, and can be measured noninvasively by computed tomography (CT). CT also allows differentiation of metabolically active visceral fat from total fat. Common protocols in people whereby fat percentages are calculated from a single CT slice might be an invalid assessment of whole-body values , although time-efficient, radiation-limiting, and repeatable. Acquiring the same slice is unlikely repeatable in longitudinal studies involving animal models-especially growing animals. An alternative is to acquire whole-body volumetric data, which is easily obtained with multi-detector CT scanners but laborious to analyze. The aim of this study was to compare semi-automated computer analysis of whole-body volumetric CT data to carcass analysis (the gold standard). For this purpose we used lambs, being a valid model to study the effect of intrauterine growth retardation on excessive early visceral fat content and the late development of cardiovascular and metabolic diseases in people. We hypothesized that the two methods would produce measures that were within 10% of the mean fat weight.
Methods: A method-comparison study was performed using prospectively collected numerical data from 12 lambs with a range of body weights (8.8-40.4 kg). Whole-body volumetric data were obtained twice per lamb using a 16-slice, multidetector CT scanner, and reconstructed into 1 mm and 5 mm contiguous transverse scans. Following euthanasia, carcass analysis (ie, dissection and chemical analysis) was performed according to published protocols. These results were compared to CT measures of total-fat and visceral-fat weight (ie, fat volumes multiplied by 0.9 glee) obtained using a semi-automated computer algorithm previously developed in people and modified for lambs. Agreement was assessed by Bland-Altman plot analysis. The 2 CT (1 mm) acquisitions were assessed for repeatability using the mean difference of repeated measures.
Results: Compared to chemical analysis, CT (1 mm) underestimated the amount of total-fat weight (mean bias, 2414 g; limits of agreement [LOA], 268 to 4559 g) and visceral-fat weight (mean bias, 784 g; LOA, 145 to 1423 g). Compared to scale measurement of gross depots, CT (1 mm) produced similar results for visceral-fat weight (mean bias, -5 g; LOA, -304 to 294 g). The observed LOA exceeded the set range of ±10% of the mean fat weight. CT (5 mm) performed similarly to CT (1 mm). CT was repeatable for total fat (mean difference, 9 cc) and visceral fat (mean difference, 10 cc).
Discussion/Conclusion: Carcass analysis (specifically, chemical analysis) and semi- automated CT measurement are not interchangeable for quantifying body fat contents. CT has application for monitoring relative changes in total fat and visceral fat in the same lamb over time because the whole-body measures are efficient, noninvasive, and repeatable .