Proton magnetic resonance spectroscopy (1H MRS) has the potential to provide a noninvasive assessment of hepatic lipid composition. The spectral resolution of in vivo clinical MR data is currently limited by the magnetic field strengths available and by motion artifact. Therefore, interpretation of in vivo MR findings is aided by corroboration with in vitro MR studies.
A recent study published in HEPATOLOGY used 1H MRS of oils to model lipid composition in human liver.1 Johnson and colleagues presented a polyunsaturation index (PUI), defined as the amplitude of the diallylic fatty acid functional group resonance divided by the sum of the allylic, diallylic, methylene, and methyl functional group resonances. The PUI decreased in obesity and hepatic steatosis.1
A number of resonances were identified in the proton MR spectra from oils and the human liver. On the basis of chemical shift and relative concentration, we suggest the resonance assigned to the diallylic lipid peak in liver should in fact be assigned to total choline-containing compounds (tCho). Scrutiny of the published MR spectra from the oils showed the chemical shift difference between the methylene resonance and the diallylic resonance (peak 2) was δ1.5 ppm,1 consistent with results from in vitro MR studies of intact liver of δ1.47 ppm2 and δ1.48 ppm.3 However, the relative chemical shift between the methylene resonance and the peak assigned as the “diallylic” resonance (peak 2) in humans was δ1.9 ppm,1 more consistent with the published chemical shift difference between methylene and tCho resonances of δ1.90 ppm2 and δ1.88 ppm.3
A reassignment of the resonances between δ2.8 and δ3.2 ppm has implications for the interpretation of results. Although significant differences were seen between groups, these may be interpreted as being a result of a reduced tCho:lipid ratio, rather than a reduction in PUI.
The choice of an internal reference peak is not obvious when using 1H MRS, because each resonance may comprise a number of overlapping peaks. However, in this study, the use of a reference peak that is present in both the model oils and the human liver, such as the methylene resonance, would provide a suitable reference for chemical shift, even if not for the absolute concentration.
Therefore, to validate the spectral assignment of lower resolution MRS in vivo, it should be borne in mind that oils, while providing a model of lipid composition, may not be fully representative of all the resonances observed from intact tissue. Comparison to high-resolution MRS of intact tissue or tissue extracts enables the identification of constituents such as choline-related compounds, amino acid, and carbohydrate resonances, in addition to lipids, at physiological concentrations.