DOP88 Understanding the molecular mechanisms of anti-TNF treatment failure: Whole blood DNA methylation changes associated with primary non-response to anti-TNF treatment in patients with Crohn’s disease

Abstract

Abstract Background Anti-TNF treatment failure in patients with IBD is common and frequently related to low drug concentrations. DNA methylation sites from whole blood have been identified as effective biomarkers predicting treatment response to methotrexate and etanercept in patients with rheumatoid arthritis. We sought to define differences in DNA methylation associated with primary non-response (PNR) to anti-TNF treatment in patients with Crohn’s disease. Methods The Personalised Anti-TNF Therapy in Crohn’s disease (PANTS) study is a prospective UK-wide study investigating anti-TNF treatment failure in patients with active luminal Crohn’s disease. DNA methylation from whole blood was assessed using the Illumina EPIC Beadchip at baseline, weeks 14, 30 and 52. We compared DNA methylation profiles in a subset of adalimumab- and infliximab-treated patients who experienced PNR at week 14 and were not in remission at week 54 (infliximab = 99, adalimumab = 94) with patients who responded at week 14 and were in remission at week 54 (infliximab = 99, adalimumab = 93). Low anti-TNF drug concentration was defined as an infliximab level <3mg/L and adalimumab level <5mg/L. Epigenome-wide association (EWAS) analysis was performed using linear mixed models, where p values < 9 x 10-8 were considered significant. Because we observed significant changes in CD4 and CD8 T cells, B cells, NK cells, monocytes and granulocytes following anti-TNF treatment we adjusted our models for derived cell proportions (Figure 1). Results Overall, between baseline and week 14, we observed 4999 differentially methylated probes (DMPs) annotated to 2376 genes. Pathway analysis identified 108 significant gene ontology terms enriched in biological processes related to immune system processes and responses (Figure 2). EWAS analysis at baseline identified 48 DMPs annotated to 36 genes associated with PNR (Figure 3). The majority of DMPs (40/48) were hypermethylated (p < 0.001): of those annotated to genes, 25 were located in the gene body, 5 in the 5’ untranslated regions (UTR), 4 in transcription start sites and the remaining in 3’UTR. When EWAS analysis at baseline was stratified by anti-TNF drug concentration, 27 DMPs annotated to 17 genes were associated with PNR and low drug concentration. Of these, 11 DMPs were independently associated with low anti-TNF drug concentration at week 14 (Figure 4). We did not identify any DMPs associated with PNR with adequate drug concentration. Conclusion Our data suggests that baseline DNA methylation profiles may be used as a predictor for anti-TNF drug concentration at week 14.