Objective        To investigate the metabonomics of serum samples from patients with mild hepatic encephalopathy (MHE) and to establish a diagnostic model for MHE.       Methods        Nuclear magnetic resonance spectroscopy (NMR) was applied to analyze serum metabonomics from 25 patients with simple liver cirrhosis (cirrhosis group)

25 cirrhotic patients with mild hepatic encephalopathy (MHE group) and 30 healthy volunteers (control group). Principal component analysis (PCA) an

d orthogonal partial least-squares-discriminant analysis (OPLS-DA) were employed to establish a pattern recognition analysis of NMR spectra. Clinical samples were used to evaluate the performance of the diagnostic model for MHE.       Results      The OPLS-DA model could distinguish the cirrhotic patients from the healthy controls with R

2

X=35.1% and Q

2

Y =0.902

and could also distinguish the patients with and without MHE with R

2

X =30.4% and Q

2

Y =0.817. In comparison with the healthy volunteers

the patients with liver cirrhosis and with MHE were characterized by increased serum levels of glucose

succinate

citrate and decreased levels of low-density lipoprotein (LDL)

very low-density lipoprotein (VLDL) and choline. Compared with the cirrhotic patients without MHE

the MHE patients showed increased serum levels of glutamine

phenylalanine and acetone and decreased serum levels of valine and isoleucine. The accuracy ratio of OPLS-DA model for MHE was 80% (8/10).       Conclusion      Metabonomics spectra based on

1

H NMR combined with pattern recognition analysis can detect various metabolites in serum of cirrhotic patients with and without MHE at the same time

which offers a new way for studying the pathogenesis of liver cirrhosis and MHE

and for the diagnosis of MHE in the future.