Human Brown Adipose Tissue Estimated With Magnetic Resonance Imaging Undergoes Changes in Composition After Cold Exposure: An in vivo MRI Study in Healthy Volunteers

Abstract

Aim: Magnetic resonance imaging (MRI) is increasingly being used to evaluate brown adipose tissue (BAT) function. Reports on the extent and direction of cold-induced changes in MRI fat fraction and estimated BAT volume vary between studies. Here, we aimed to explore the effect of different fat fraction threshold ranges on outcomes measured by MRI. Moreover, we aimed to investigate the effect of cold exposure on estimated BAT mass and energy content. Methods: The effects of cold exposure at different fat fraction thresholding levels were analyzed in the supraclavicular adipose depot of nine adult males. MRI data were reconstructed, co-registered and analyzed in two ways. First, we analyzed cold-induced changes in fat fraction, T2* relaxation time, volume, mass, and energy of the entire supraclavicular adipose depot at different fat fraction threshold levels. As a control, we assessed fat fraction differences of deltoid subcutaneous adipose tissue (SAT). Second, a local analysis was performed to study changes in fat fraction and T2* on a voxel-level. Thermoneutral and post-cooling data were compared using paired-sample t-tests (p < 0.05). Results: Global analysis unveiled that the largest cold-induced change in fat fraction occurred within a thermoneutral fat fraction range of 30-100% (-3.5 ± 1.9%), without changing the estimated BAT volume. However, the largest cold-induced changes in estimated BAT volume were observed when applying a thermoneutral fat fraction range of 70-100% (-3.8 ± 2.6%). No changes were observed for the deltoid SAT fat fractions. Tissue energy content was reduced from 126 ± 33 to 121 ± 30 kcal, when using a 30-100% fat fraction range, and also depended on different fat fraction thresholds. Voxel-wise analysis showed that while cold exposure changed the fat fraction across nearly all thermoneutral fat fractions, decreases were most pronounced at high thermoneutral fat fractions. Conclusion: Cold-induced changes in fat fraction occurred over the entire range of thermoneutral fat fractions, and were especially found in lipid-rich regions of the supraclavicular adipose depot. Due to the variability in response between lipid-rich and lipid-poor regions, care should be taken when applying fat fraction thresholds for MRI BAT analysis.

Keywords: brown adipose tissue; cold exposure; fat fraction; lipid metabolism; magnetic resonance imaging; thermogenesis.

Figure 1

Example of image registration and a reconstructed fat fraction map before and after cooling. The first column shows thermoneutral and post-cooling images (one slice from the first echo in the acquisition). In the second column, the overlay of the same images before (top) and after registration (bottom) is shown. The images are colored orange (thermoneutral) and blue (post-cooling) for better visualization of differences between the scans. The third column shows the thermoneutral and post-cooling fat fraction maps of the supraclavicular adipose depot, overlaid on the corresponding images. Lipid content in the supraclavicular region is color-mapped over a 30–100% fat fraction range.

Figure 2

Estimated volumetric BAT analysis. Thermoneutral and post-cooling volume histograms as a function of fat fraction with bin size 0.5%: thermoneutral volumes are shown in red and post-cooling volumes in blue (A). Cold-induced volume changes plotted as a function of fat fractions (10% FF interval) (B). Data are represented as mean ± SEM for n = 9. In (B), a paired sample t-test was used to analyze the changes in volume after cold exposure. *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 3

Effect of FF thresholds on estimated BAT volume differences. Heatmap of the effect of different FF segmentation thresholds on estimated BAT volume differences after cooling. The color (second y-axis) depicts the estimated BAT volume difference for each lower (x-axis) and upper left (y-axis) threshold. The largest decrease in estimated BAT volume is present with a lower threshold of 72% and no upper threshold. The triangle in the lower right corner indicates invalid FF threshold options, as we implemented a minimum FF threshold of 30%. (A) Cold-induced volume changes analyzed using the paired sample t-test (*p < 0.05, **p < 0.01) at different threshold ranges: 30–100% (B), 50–100% (C), and 70–100% (D). Data is represented as mean ± SEM for all participants (n = 9).

Figure 4

Effect of different FF thresholds on global supraclavicular adipose tissue FF and T2*. Cold-induced FF and T2* changes analyzed using the paired t-test at different threshold ranges: 30–100% (A,D), 50–100% (B,E), and 70–100% (C,F). Data are represented as mean ± SEM for n = 9. The paired sample t-test was used to analyze the changes in volume after cold exposure (**p < 0.01 and ***p < 0.001).

Figure 5

Distinction between lean and lipid masses within supraclavicular adipose tissue. Lean and lipid masses were estimated as described in the “Methods” section and represented as a function of their specific fat fractions (A). Cold exposure decreased both lean and fat masses to in the upper fat fractions (above 70%) and slightly increased these in the lower fat fractions (B). (C) Correlation between total estimated BAT volume and lipid or lean mass analyzed using linear regression (R² is reported). Change in total lipid and lean mass after cold exposure, analyzed with the paired sample t-test (D). Data in (A,B,D) represent mean ± SEM for n = 9 volunteers. *p < 0.05.

Figure 6

Metabolizable energy content in the supraclavicular adipose depot. Representation of energy content in the supraclavicular depot at thermoneutrality, with specific values attributed to lean tissue or lipids (A). Changes in energy content attributed to lean or fat masses, represented over different fat fraction ranges (B). Total energy storages (kcal) before and after cold exposure analyzed, by using the paired t-test (C). Heatmap of the effect of different FF segmentation thresholds on estimated energy content differences after cooling. The color (second y-axis) depicts the estimated energy content difference for each lower (x-axis) and upper left (y-axis) threshold. The largest decrease in estimated energy is present with a lower threshold of 70% and no upper threshold. The triangle in the lower right corner indicates invalid FF threshold options, as we implemented a minimum FF threshold of 30% (D). Data represent mean ± SEM of all participants (n = 9). *p < 0.05.

Figure 7

Structural heterogeneity of brown adipose tissue in the supraclavicular region during cold exposure. Example of a reconstructed fat fraction map with merged z-slices before and after cooling (A,B) and cold-induced change (post-minus pre) (C) for n = 1. The 2D joint voxel histogram representing variation in change in lipid content of each voxel in relation to its thermoneutral FF from the voxel-wise analysis, wherein the colors represent the number of voxels belonging to each combination (D) for all participants (n = 9). Cold colors indicate decreases in fat fraction and warm colors indicate increases in fat fraction.

Figure 8

Voxel histograms representing the relation between thermoneutral values and cold-induced changes in T2* and FF. Thermoneutral measurements of T2* against thermoneutral fat fractions (A). Relation between the cold-induced changes in T2* and thermoneutral fat fractions (B). The association between cold-induced changes in both T2* and FF (C). Data is presented as the mean of all participants (n = 9).