Reply:

Abstract

We read with interest the recent letter by Cobbold and colleagues regarding our study published in HEPATOLOGY. Using a clinical 1.5 T whole-body magnetic resonance (MR) scanner, we validated purported indices of lipid saturation (SI) and polyunsaturation (PUI) using oil phantoms of known composition. Application of this technique to measure lipid signals in human liver in vivo produced results that were consistent with human biopsy-based approaches demonstrating that, when compared with healthy lean individuals (lean), relative hepatic lipid saturation increases and purported “polyunsaturation” decreases with obesity (OB) and obesity-related hepatic steatosis (OB+HS).1 Based on in vitro high field strength (11.7 T) MR studies of liver tissue, Cobbold et al. have suggested that the peak assigned as the “diallylic” lipid resonance upon which our PUI index was based reflects total choline-containing compounds (tCho). We point out the comparative lack of spectral resolution of peaks this size in in vivo studies performed at clinical field strength (1.5 T), compared with high-resolution in vitro spectroscopy of rodent2 and human3 liver tissue. In our in vivo spectra, the peak in question is broad (≈2.9–3.3 ppm). The recommended use of chemical shift from an internal reference for peak nomination ameliorates but does not ablate this issue. Given the high concentration of polyunsaturated fatty acids in healthy human liver tissue samples4 and the amplitude of this peak in rodent2 and human3 MR spectroscopy, one would expect to observe a signal at ≈2.85 ppm. The separation of this peak from that of choline in vivo may well be limited at 1.5 T. This issue may have been clarified by comparing these findings with those from liver biopsy samples. However, this was not indicated in our healthy cohort. Cobbold and colleagues have raised important issues and highlight the difficulties of obtaining well-resolved human in vivo spectra at a lower field strength. The precise nature of this peak in vivo at 1.5 T warrants clarification, because the index we employed (PUI) shows significant power in discriminating individuals on the basis of obesity and level of hepatic steatosis. Our study provides an important first step in applying clinically available MR spectroscopy scanning to the measurement of hepatic lipid composition in vivo. Interpretation of our data on the basis of the suggestion of Cobbold et al. that this resonance be assigned as tCho and omitting the “diallylic” resonance from our SI index still shows that hepatic lipid SI is significantly higher in OB and OB+HS versus lean individuals (P < 0.01), and further that obesity-related hepatic steatosis is characterized by saturation of hepatic lipids (OB+HS versus OB, P < 0.05). Thus, 1H-MR spectroscopy provides a rapid method to qualitatively assess hepatic lipid composition. Nathan A. Johnson* , Toos Sachinwalla , Jacob George?, * Discipline of Exercise and Sport Science, The University of Sydney, Sydney, Australia, Institute of Obesity, Nutrition and Exercise, The University of Sydney, Sydney, Australia, Department of Magnetic Resonance, Rayscan Imaging Liverpool, Australia, ? Storr Liver Unit, Westmead Millennium Institute and Westmead Hospital, The University of Sydney, Westmead, Australia.