Validation of Biomarkers in Gene Expression Datasets of Inflammatory Bowel Disease: IL13RA2, PTGS2 and WNT5A as Predictors of Responsiveness to Infliximab Therapy

Abstract

Background: Some patients with inflammatory bowel disease (IBD) do not respond to infliximab (IFX) therapy. Gene expression studies revealed genes that may help to predict non-responding IBD patients. Our purpose was to validate the discriminating power of published genes. Methods: Microarray datasets of IBD patients responding or non-responding to IFX were downloaded from GEO database (‘transcriptomic arm’). Published genes discriminating responding and non-responding patients were identified in PubMed (‘biomarker arm’). Using ROC-analysis, the discriminating performance of genes in ‘biomarker arm’ were re-tested in each datasets of ‘transcriptomic arm’. We also performed an independent analysis of colon biopsy datasets to identify novel discriminating genes. Results: The transcriptomic arm of four GEO datasets (3 and 1 from colon biopsies and blood cells, respectively) included 99 patients (of those, 59 and 40 were IFX responder and non-responder, respectively). Of the 65 candidate genes reported in biopsy specimens 25 genes discriminated significantly (p<0.05) infliximab responders and nonresponders in the three biopsy datasets consistently. Of the 39 candidate genes reported in peripheral blood 9 genes provided significant discrimination after re-testing. Independent analysis of three biopsy datasets identified the top five genes. Three genes (IL13RA2, PTGS2 and WNT5A) were highly effective discriminator in each analysis. Conclusion: This analysis identified IL13RA2, PTGS2 and WNT5A, three genes in colonic tissues of IBD patients as suitable to discrimination between patients responding and non-responding to IFX therapy. These genes encode proteins implicated in intestinal pathology; the high expression in non-responding patients may indicate important targets in IBD therapy.