Composition of adipose tissue and marrow fat in humans by 1H NMR at 7 Tesla

Abstract

Proton NMR spectroscopy at 7 Tesla (7T) was evaluated as a new method to quantify human fat composition noninvasively. In validation experiments, the composition of a known mixture of triolein, tristearin, and trilinolein agreed well with measurements by (1)H NMR spectroscopy. Triglycerides in calf subcutaneous tissue and tibial bone marrow were examined in 20 healthy subjects by (1)H spectroscopy. Ten well-resolved proton resonances from triglycerides were detected using stimulated echo acquisition mode sequence and small voxel ( approximately 0.1 ml), and T(1) and T(2) were measured. Triglyceride composition was not different between calf subcutaneous adipose tissue and tibial marrow for a given subject, and its variation among subjects, as a result of diet and genetic differences, fell in a narrow range. After correction for differential relaxation effects, the marrow fat composition was 29.1 +/- 3.5% saturated, 46.4 +/- 4.8% monounsaturated, and 24.5 +/- 3.1% diunsaturated, compared with adipose fat composition, 27.1 +/- 4.2% saturated, 49.6 +/- 5.7% monounsaturated, and 23.4 +/- 3.9% diunsaturated. Proton spectroscopy at 7T offers a simple, fast, noninvasive, and painless method for obtaining detailed information about lipid composition in humans, and the sensitivity and resolution of the method may facilitate longitudinal monitoring of changes in lipid composition in response to diet, exercise, and disease.

Fig. 1.

¹H NMR spectra of subcutaneous fat (left) and tibial bone marrow (right) from a 26 year-old healthy male at 7 Tesla (7T). Ten resonances can be resolved (A–J). The bottom trace shows the acquired spectrum, and the upper trace shows the fitted spectrum. A water signal is seen in the spectrum of subcutaneous fat but not bone marrow. A T₂-weighted image shows the position of the voxel in the subcutaneous fat tissue and tibial bone marrow (5 × 5 × 5 mm³) from which the spectra were acquired.

Fig. 2.

A: Correlation of the known, prepared fraction of diunsaturated triglyceride with MR-measured fraction of diunsaturated triglyceride. Phantoms were prepared by mixing three C18 triglycerides (number of double bonds), tristearin (0), triolein (1), and trilinolein (2) in the following ratios (tristearin:triolein:trilinolein): 50:0:50, 50:10:40, 50:20:30, 50:30:20, 50:40:10, 50:50:0, and 25:50:25. ¹H NMR spectra of the phantoms were analyzed using Equations 1 and 2. B: Correlation of the known, prepared fraction of C18 triglyceride with MR-measured value from area (B/E). Phantoms were prepared by mixing C16 tripalmitin and C18 tristearin with the following ratios (C16:C18): 100:0, 80:20, 75:25, 50:50, 25:75, 20:80, and 0:100. The data were analyzed using f_18C = 0.5 * area (B/E) − 12.

Fig. 3.

Inversion-recovery measurement of T₁ (left panel) and echo time (TE)-dependence measurement of T₂ (right panel) from subcutaneous fat tissues of a 34 year-old healthy male at 7T. The inversion bandwidth (BW) was set to span resonances A and B (middle stack), or resonances C, D, E, and F (left-most stack). The inversion delay times for the given spectra are 10, 20, 50, 100, 200, 500, 1,000, 2,000, and 3,000 ms with a constant repetition time (TR) of 7 s and TE of 40 ms (left panel). The echo times for the T₂ measurements are 20, 40, 60, 80, 100, 140, and 180 ms (right panel).

Fig. 4.

T₂ measurement of tibia bone marrow (left panel) and calf subcutaneous fat (right panel) from a 25 year-old healthy female by varying TEs at 7T. Note that the voxel (5 × 5 × 5 mm³) fits well in the single fat cell of the subcutaneous tissue and the collected ¹H spectrum is water-free. All spectra are vertically scaled to equal magnitude of the methylene resonance (B), and as a result, with TE increase, the methyl resonance A with longer T₂ than B shows signal rising, whereas the shorter T₂ resonances such as C and D show signal decaying relative to resonance B. To avoid overcrowding, the fitting of the “A” and “C” peaks is not shown. Other parameters: TR 8 s, number of acquisitions, 8; number of points, 4 k; BW 4 kHz.

Fig. 5.

Correlation of measured area (F/E) between tibial bone marrow and subcutaneous fat for the 20 healthy adult subjects studied, showing that the diunsaturated fatty acid is similar for these two adipose sites and that the fat composition variation among subjects is detectable by ¹H magnetic resonance spectroscopy.