ABSTRACT Ultra‐high‐field 7‐T 1 H‐MRS resolves the bis‐allylic resonance at 2.8 ppm, a marker present only in fatty acids (FAs) with two or more double bonds, enabling distinction between monounsaturated (MUFA) and polyunsaturated fatty acids (PUFAs) in IMCL that is not achievable at 3 T. The objective of this study was to determine the relative proportions of saturated (SFA), MUFA, and PUFA in IMCL in vivo using 7‐T 1 H‐MRS. To further improve the precision of this fitting, we acquired high‐resolution 600‐MHz 1 H‐NMR spectra of representative FA species and incorporated their chemical shifts as prior knowledge in the spectral fitting model. Five predominant FAs—oleic, palmitic, linoleic, and α‐ and γ‐linolenic acids, which together represent the major FA species in human skeletal muscle triglycerides—were analyzed via 600‐MHz NMR to determine precise chemical shifts for methyl (–CH 3 ), allylic (–CH 2 –CH=CH–), and bis‐allylic (=CH–CH 2 –CH=) protons. The NMR spectra revealed distinct shifts: linoleic acid (0.891, 2.044, and 2.782 ppm), oleic acid (0.878 and 2.002 ppm), palmitic acid (0.880 ppm), α‐linolenic acid (0.976, 2.064, and 2.807 ppm), and γ‐linolenic acid (0.889, 2.074, and 2.808 ppm). These values were subsequently fitted to in vivo 7‐T 1 H‐MRS spectra (range of 0.90–5.55 ppm) acquired from the tibialis anterior of 11 healthy volunteers. Using specialized analysis software, IMCL peaks for methyl (0.90 and 1.00 ppm), allylic (2.02 and 2.06 ppm), and bis‐allylic (2.78 and 2.82 ppm) protons were incorporated into the spectral fitting model, enabling improved characterization of FA composition. The mean relative composition of intramyocellular FAs was 36% palmitic acid, 27% oleic acid, and 23% linoleic acid (including α‐ and γ‐linolenic acids). Our findings demonstrate that incorporating NMR‐derived prior knowledge enables chemically specific fitting of intramyocellular FA peaks, allowing distinction between SFA, MUFA, and PUFA components within the intramyocellular compartment—information that cannot be obtained from biopsy‐based homogenate analysis.
Hioki et al. (2026) studied this question.