Abnormalities in metabolic pathways in celiac disease investigated by the metabolic profiling of small intestinal mucosa, blood plasma and urine by NMR spectroscopy.

Celiac disease (CeD) is an autoimmune enteropathy caused by gluten intake in genetically predisposed individuals. We investigated the metabolism of CeD by metabolic profiling of intestinal mucosa, blood plasma and urine using NMR spectroscopy and multivariate analysis. The metabolic profile of the small intestinal mucosa was compared between patients with CeD (n = 64) and disease controls (DCs, n = 30). The blood plasma and urinary metabolomes of CeD patients were compared with healthy controls (HCs, n = 39). Twelve metabolites (proline (Pro), arginine (Arg), glycine (Gly), histidine (His), glutamate (Glu), aspartate, tryptophan (Trp), fumarate, formate, succinate (Succ), glycerophosphocholine (GPC) and allantoin (Alln)) of intestinal mucosa differentiated CeD from controls. The metabolome of blood plasma with 18 metabolites (Pro, Arg, Gly, alanine, Glu, glutamine, glucose (Glc), lactate (Lac), acetate (Ace), acetoacetate (AcAc), β-hydroxybutyrate (β-OHB), pyruvate (Pyr), Succ, citrate (Cit), choline (Cho), creatine (Cr), phosphocreatine (PCr) and creatinine) and 9 metabolites of urine (Pro, Trp, β-OHB, Pyr, Succ, N-methylnicotinamide (NMN), aminohippurate (AHA), indoxyl sulfate (IS) and Alln) distinguished CeD from HCs. Our data demonstrated changes in nine metabolic pathways. The altered metabolites were associated with increased oxidative stress (Alln), impaired healing and repair mechanisms (Pro, Arg), compromised anti-inflammatory and cytoprotective processes (Gly, His, NMN), altered energy metabolism (Glc, Lac, β-OHB, Ace, AcAc, Pyr, Succ, Cit, Cho, Cr and PCr), impaired membrane metabolism (GPC and Cho) and intestinal dysbiosis (AHA and IS). An orthogonal partial least square discriminant analysis model provided clear differentiation between patients with CeD and controls in all three specimens. A classification model built by combining the distinguishing metabolites of blood plasma and urine samples gave an AUC of 0.99 with 97.7% sensitivity, 93.3% specificity and a predictive accuracy of 95.1%, which was higher than for the models built separately using small intestinal mucosa, blood plasma and urine. In conclusion, a panel of metabolic biomarkers in intestinal biopsies, plasma and urine samples has potential to differentiate CeD from controls and may complement traditional tests to improve the diagnosis of CeD.