Search in the Abstract Database

Abstracts Search 2017

* = Presenting author

P018 Complete metabonomic and microbiota profiling identifies biomarkers for anti-TNF therapy response

Ding N.S.*1,2,3, Sarafian M.3,4, Perdones-Montero A.5, Misra R.4, Hendy P.4, Penez L.4, Holmes E.3, Hart A.3,4

1St Vincent's Hospital, Inflammatory Bowel Disease, Melbourne, Australia 2St Mark's Hospital, Inflammatory Bowel Disease Unit, London, United Kingdom 3Imperial College London, Department of Surgery and Cancer, London, United Kingdom 4St Mark's Hospital, IBD, London, United Kingdom 5Imperial College, Surgery and Cancer, London, United Kingdom

Background

Anti-TNF therapy forms the backbone for treatment in moderate to severe Crohn's disease (CD). Metabonomic approaches to profiling Crohn's disease has led to numerous discoveries in disease pathogenesis. We aim to use metabonomic and microbiome profiling to identify predictive biomarkers of anti-TNF response.

Methods

CD patients commencing anti-TNF had 3 monthly visits for 12 months with collection of biofluids (urine, faeces and serum) and disease assessment with biochemistry and faecal calprotectin or mucosal healing. A response index combining biochemistry (decrease in FC or CRP)and mucosal healing was used to define therapeutic response in the presence of adequate drug level.

We collected 179 urine, 210 serum and 168 faecal samples from 68 anti-TNF naive CD patients (luminal phenotype undergoing anti-TNF therapy without surgical resections) and 20 healthy controls. Liquid-Chromotography Mass Spectroscopy using Waters® instruments with lipid, bile acid (BA) and polar molecule (HILIC) profiling of metabolites and multivariate analysis compared to response index on SIMCA software was undertaken. 16SrRNA extraction using Powerlyzerkit®, sequencing with MiseQ illumina® and processing using Mothur was performed.

Results

There were 18 non-responders and 9 responders to anti-TNF therapy according to our strict criteria for response. Multiple biomarkers were identified across assays to be significant for predicting anti-TNF response to therapy across all visits (Fig. 1).

Figure 1. OPLS-DA models created for response vs. non-response (A) with biomarkers from serum bile acid profiling (B) and urine HILIC profiling (C).

The strongest models were from serum bile acid (R2X 0.29, Q2Y 0.41, p=4.97×10–7) and urinary HILIC (R2X 0.14, Q2Y 0.30, p=1.28×10–4) (Fig. 1A). Serum BA profiling analysis identified 2 conjugated and 1 unconjugated BAs (Fig. 1B) while urinary HILIC profiling identified cysteine as biomarkers (Fig. 1C) creating a model allowing prediction of anti-TNF response, with levels being significantly different between non-responders and responders (Fig. 1).

On 16SrRNA, lactobacillis is higher in responders while clostridiales were lower in abundance for non-responders. The quantities of species did not alter significantly over time nor with therapy. Lactobacilles is known to synthesise cysteine from serine and for expansion in gut microbiota.

Conclusion

This prospective, longitudinal cohort study of microbiome and metabonomic analysis demonstrates that there are predictive biomarkers involved with bile acid and inflammatory pathways. The microbiome of patients with Crohn's disease does not alter significantly despite anti-TNF therapy response which allows for prediction of therapeutic outcome.