Search in the Abstract Database

Abstracts Search 2017

* = Presenting author

P253 Effects of time on urinary metabolic signatures in inflammatory bowel disease

Hicks L.*1, Powles S.1, Swann J.2, Chong L.1, Holmes E.2, Williams H.1, Orchard T.1

1Imperial College Healthcare NHS Trust St. Mary's Hospital, Department of Gastroenterology, London, United Kingdom 2Imperial College London, Computational and Systems Medicine, London, United Kingdom

Background

Metabolic profiling (metabonomics) has been proposed as a novel clinical tool in IBD to predict development of complex disease, or for longitudinal non-invasive monitoring of activity and/or response to drug treatment. Urinary metabonomics can distinguish IBD from healthy controls [1] but no studies to date have assessed the stability of these discriminatory profiles over time. Studies in healthy adults show metabolic signatures are largely unchanged over periods of up to 3 years [2], but signals are influenced by multiple external factors including medication and surgery, so how these change in IBD is unknown. The aim of this study was to compare baseline urinary metabolic profiles of IBD patients with a repeated sample several years later to assess similarity, and also to test if any clinical outcomes could be retrospectively predicted from the baseline sample.

Methods

Two urine samples from 39 IBD patients (22 Crohn's disease (CD) and 17 ulcerative colitis (UC)) were collected - one at baseline and one several years later (range 7–9 yrs). These were analysed by 1H NMR spectroscopy. Disease progression was defined as initiation of immunosuppression or biologics, progression of disease location or phenotype, or surgery. Principal components analysis was used to visualise the variance between the two time-points within the cohort. Orthogonal partial least squares discriminant analysis (OPLSDA) was used to establish if the metabolic signatures could be used to predict adverse clinical outcomes in the patients studied.

Results

There was a diverse clinical outcome across the groups; 57% of CD patients and 17% of UC patients had clinical progression at follow up sampling. PCA showed clustering of sample pairs from the baseline and several years later in most individuals, suggesting intra-individual similarity across time. OPLSDA showed no statistical models could be built to predict combined poor outcome based on the initial urinary metabolic profile (p=0.26). However, the small subgroup who went on to require surgical intervention could be separated from the cohort in a model (Q2=0.015; p=0.03) constructed on their baseline profiles.

Conclusion

The metabolic profile of IBD in an individual appears relatively stable over a significant time period despite a variety of clinical outcomes and interventions. Variations in longitudinal measurements appear to be subtle, and therefore application of this technique for disease monitoring and risk stratification could prove difficult. These results may suggest that metabolic profiling could be exploited to predict a higher risk of requiring future surgery. Large prospective studies are required to further investigate this.

References:

[1] Williams H.R.T. et al. (2009), Characterization of inflammatory bowel disease with urinary metabolic profiling, Am J Gastroenterol

[2] Bernini, P., et al. (2009), Individual human phenotypes in metabolic space and time, J Proteome Res