Introduction Advances in protein profiling using proximity extension assays (PEA) have transformed our ability to compare concentrations of multiple proteins across biological samples. PEA utilises the specificity of antibody proximity and the sensitivity of polymerase chain reaction (qPCR) to detect proteins of interest. As part of the EC funded IBD Character initiative to discover biomarkers for clinical use using multiomic technologies, we performed high-throughput prospective case-control serum profiling to identify protein biomarkers that can predict Inflammatory Bowel Disease (IBD).
Method Utilising Proseek Multiplex panels (Olink Bioscience, Sweden), serum profiling was performed in patients with a new diagnosis of IBD. Our control group consisted of symptomatic individuals. Phenotypic data was captured including age, sex, diagnosis and IBD medications. Statistical analysis was performed using R. Data were normalised and then batch corrected using ComBAT. Linear models were created for each protein including age and sex as covariates. After quality control, data from 186 proteins was available for analysis.
Results A total of 245 patient serum samples (n = 153) newly diagnosed IBD, n = 92 symptomatic controls) from 4 IBD centres from Norway, United Kingdom and Sweden were included in the study from December 2012 to June 2014. 70 had Crohn’s disease (CD), 70 ulcerative colitis (UC) and 13 Inflammatory Bowel Disease Unclassified (IBDU). The mean age of the entire cohort was 33 years (range 0–79 years) and 54% were female. Multivariable analysis identified a set of 48 protein markers that were significantly associated with IBD. The 5 most significant protein markers were MMP-12 (Holm-adjusted p = 3.3 × 10–13), OSM (p = 2.4 × 10–12), CXCL9 (p = 1.7 × 10–9), MMP10 (p = 1.7 × 10–9) and EGFR (p = 1.8 × 10–9). Of these five, all except EGFR were upregulated in IBD. The top 2 markers, MMP-12 and OSM were able to discriminate IBD from controls with an area under the receiver operator characteristics curve of 0.81 and 0.75 respectively. Using linear discriminant analysis, a combined biomarker consisting of MMP-12 and OSM was able to discriminate IBD from controls with a sensitivity and specificity of 80% and 72% respectively.
Conclusion We have identified serum biomarkers that can predict IBD. These data demonstrate a potential for a PEA multiplex technology in IBD diagnostics and its ability to identify novel proteins that may be relevant in disease pathogenesis.
Disclosure of interest R. Kalla Grant/Research Support from: IBD Character, N. Kennedy Grant/Research Support from: Wellcome Trust, Conflict with: Abbvie, MSD, Warner Chilcott, Ferring speaker fees, F. Hjelm: None Declared, E. Modig: None Declared, M. Sundell: None Declared, J. Söderholm: None Declared, B. Andreassen: None Declared, D. Bergemalm: None Declared, N. Ventham Grant/Research Support from: IBD BIOM, H. Hjortswang: None Declared, R. Petr: None Declared, M. Vatn: None Declared, J. Halfvarson: None Declared, M. Gullberg: None Declared, J. Satsangi Grant/Research Support from: EC Grants, Wellcome, CSO, MRC, Consultant for: Takeda, Conflict with: MSD speaker fees.