Urinary metabolic profiling in inflammatory bowel disease Dr Horace Williams Clinical Research Fellow Imperial College London
Background: Metabolic profiling Metabolic profiling or metabonomics describes the generation of metabolic information from biofluids or tissues NMR spectroscopy (NMRS) simultaneous acquisition of multiple biochemical parameters Urinary metabolic profiling study specific diseases based on underlying metabolic processes no such application to IBD
Rationale for metabolic profiling in IBD Gut microbiota differ between CD, UC and healthy controls Gut microbiota have important influences on specific urinary metabolites: Hippurate Dimethylamine Formate TMAO 4-cresol sulphate
Experimental design: subject groups Crohn s disease Ulcerative colitis Healthy controls Number (Male/Female) 86 (47/39) 60 (30/30) 60 (30/30) Median range] age (years) 33 16-66] 40 17-66] 30 18-61] L1: 18 E1: 14 - Disease location L2: 25 E2: 18 - L3: 43 E3: 28 - Longitudinal Samples 35 14 26 Urinary NMR Spectra acquired using JEOL 500MHz Eclipse+ NMR spectrometer Largest urinary metabonomic study in any disease to date
UC patients CD patients Controls 0.4 p=0.001 p=0.82 p=0.001 0.3 0.2 0.1 FORMATE relative index Controls UC patients CD patients 0 0.0 0.6 0.4 0.2 0.0 p=0.0003 p=0.98 p=0.0007 4-CRESOL SULPHATE relative inde x Hypothesis-driven analysis 4 p<0.0001 p=0.0001 p<0.0001 3 2 1 HIPPURATE relative index Controls UC patients CD patients
Multivariate factor analysis Principal Components Analysis (PCA) Assumes no a priori knowledge Overview of the samples, analysing the whole spectrum Principal components are linear combinations of variables (metabolites) accounting for the greatest variation within the dataset Scores plot: each sample represented in the new coordinate space Loadings plot: combination of metabolites responsible for the scores plot
Multivariate factor analysis A B 0.4 Principal Component 2 0.2 0-0.2 Principal Component 2 0.2 0-0.2-0.4 0 0.2 0.4 0.6 0.8 Principal Component 1-0.2-0.1 0 0.1 0.2 0.3 0.4 Principal Component 1 A: PCA scores plot: 86 CD patients and 60 controls. B: PCA scores plot: 68 CD patients and 60 controls (excluding outliers). CD Control
Multivariate factor analysis Partial least squares discriminant analysis (PLS-DA) NMR spectroscopic variables (metabolites) are related to class membership Scores plots generated Loadings plots and regression vector: identification of spectral regions (metabolites) responsible for separation between groups Orthogonal Signal Correction (OSC) Rigorous validation techniques
Multivariate factor analysis PLS Component 1 0.5 0.0-0.5 CD Control CD 61 7 Control 10 50 Sensitivity: 90% Specificity: 83% Control samples CD samples CD Control
PLS-DA models: CD vs. UC All individuals Individuals on no medication CD: UC L2 CD: UC CD: UC Sensitivity: specificity 86: 82 79: 83 82: 97 Results compare favourably with predictive abilities of ASCA / panca Metabolites primarily responsible for distinguishing CD and UC: HIPPURATE, citrate, methylhistidine, guanidoacetate, 4-cresol sulphate
Hippurate Metabolism Investigating Hippurate Metabolism in IBD 16 CD patients, 16 healthy controls Low benzoate diet Administered 5mg/kg sodium benzoate.
Controls CD patients Controls CD patients 4 3 2 1 0 Controls CD patients Baseline 1 hour 2.5 2.0 1.5 1.0 0.5 0.0 p=0.53 HIPPURATE relative index (to total spectral integral) HIPPURATE excretion (1 hour - baseline) 1000 800 600 400 200 0 p=0.0007 % CHANGE in hippurate excretion (baseline to 1 hour) Results A B C
Discussion Significant differences in IBD for specific urinary metabolites whose concentration is modulated according to make-up of the intestinal microbiota: Hippurate, formate, 4-cresol sulphate Multivariate analysis (PLS-DA modelling): Able to distinguish between cohorts, even colonic CD vs UC Hippurate of major importance Elucidation of hippurate metabolism in IBD Gut microbiota implicated in the lower levels found in CD
Acknowledgements Broad Medical Research Program Collaborators/Researchers Dr. Bernard North Dr. David Walker Prof. Ken Welsh Dr. Hiroe Sato Ms. Vicky Loh Miss. Venisha Patel Dr. Simon Jakobovits
Urine NMRS from male healthy control = 67 = 910 KEY. 1: lactate 2: alanine 3: 4-cresol 4: citrate 5: DMA 6: creatine 7: creatinine 8: TMAO 9: glycine 10: hippurate 11: formate 3 11 10 10 9 78 54 8.9 8.8 8.7 8.6 8.5 8.4 8.3 8.2 8.1 8 7.9 7.8 7.7 7.6 7.5 7.4 7.3 7.2 7.1 9 8 65 43 3 2 1 4.4 4.2 4 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 ppm
Addressing potential confounders Subject selection Caucasian Exclusion criteria Comorbidity Intercurrent illness Antibiotic usage Subject questionnaire Diet Drugs
Subject groups No significant differences between groups in terms of dietary constituents, EtOH intake, exercise, smoking Female subjects matched for reproductive status and use of hormonal therapies
Addressing potential confounders Recent studies: Inter-individual variation > intra-individual variation Good reproducibility between individuals First void urine samples exhibit greatest variability Random urine samples, 13.00 ± 3 hours