Research ArticleNon-alcoholic fatty liver disease: Spectral patterns observed from an in vivo phosphorus magnetic resonance spectroscopy study

Background & AimsLiver biopsy is the gold standard for diagnosing non-alcoholic fatty liver disease (NAFLD) but with practical constraints. Phosphorus magnetic resonance spectroscopy (31P-MRS) allows in vivo assessment of hepatocellular metabolism and has shown potential for biochemical differentiation in diffuse liver disease. Our aims were to describe spectroscopic signatures in biopsy-proven NAFLD and to determine diagnostic performance of 31P-MRS for non-alcoholic steatohepatitis (NASH).Methods31P-MRS was performed in 151 subjects, comprised of healthy controls (n = 19) and NAFLD patients with non-NASH (n = 37) and NASH (n = 95). Signal intensity ratios for phosphomonoesters (PME) including phosphoethanolamine (PE), phosphodiesters (PDE) including glycerophosphocholine (GPC), total nucleotide triphosphate (NTP) including α-NTP, and inorganic phosphate (Pi), expressed relative to total phosphate (TP) or [PME+PDE] and converted to percentage, were obtained.ResultsCompared to controls, both NAFLD groups had increased PDE/TP (p <0.001) and decreased Pi/TP (p = 0.011). Non-NASH patients showed decreased PE/[PME+PDE] (p = 0.048), increased GPC/[PME+PDE] (p <0.001), and normal NTP/TP and α-NTP/TP. Whereas, NASH patients had normal PE/[PME+PDE] and GPC/[PME+PDE], but decreased NTP/TP (p = 0.004) and α-NTP/TP (p <0.001). The latter was significantly different between non-NASH and NASH (p = 0.047) and selected as discriminating parameter, with area under the receiver-operating characteristics curve of 0.71 (95% confidence interval, 0.62-0.79). An α-NTP/TP cutoff of 16.36% gave 91% sensitivity and cutoff of 10.57% gave 91% specificity for NASH.Conclusions31P-MRS shows distinct biochemical changes in different NAFLD states, and has fair diagnostic accuracy for NASH.