Metabolomic analysis with 1H-NMR for non-invasive diagnosis of hepatic fibrosis degree in patients with chronic hepatitis C

Abstract Background The assessment of fibrosis degree in liver diseases is based on several non-invasive techniques, but none has been accurate. Aim This study employed proton nuclear magnetic resonance spectroscopy to identify metabolic profiles in serum and urine, specific for different fibrosis degree in chronic hepatitis C patients. Method 71 plasma, 73 serum, and 578 urine samples were collected. All samples were analyzed using 1 H-NMR spectroscopy technique and three different NMR spectra were acquired for each serum/plasma sample. The data analyses were performed by partial least square regression, principal component analysis, and Monte Carlo cross-validation in a supervised methodology. Results The cross-validation test correctly assigned each sample to its specific donor with 98.44% accuracy for urine samples and 65% for serum/plasma samples. Advanced fibrosis and cirrhosis were recognized with 71% sensitivity for CPMG plasma spectra and 69% specificity for NOESY serum spectra. Accuracy for NOESY serum spectra was 68%. Noesy spectra recognized advanced fibrosis and cirrhosis with 71% sensitivity, 30% specificity, and 50% accuracy in urine samples. Conclusion Metabolomic analysis of urine spectra using 1 H-NMR spectroscopy can recognize a specific individual profile in all patients with chronic hepatitis C. However, this method cannot yet differentiate with sufficient accuracy, patients with advanced fibrosis from patients with milder disease.