Background/Purpose:

The focus of current research for new therapeutic approaches in Rheumatoid arthritis (RA) is shifting from mere suppression to induction and maintenance of tolerance. Adaptive immunity, specifically the balance and functions of effector and regulatory CD4⁺ T cells, is central in this context. We have recently developed a protocol of oral tolerance induction to an epitope (dnaJP1) derived from a heat shock protein (HSP), a key component of the mechanism of perpetuation and amplification of chronic autoimmune inflammation. Phase II trial results showed clinical efficacy and an intriguing immune deviation of T cell immunity. In order to dissect further the mechanism of action and also to identify biomarkers predictive of susceptibility to treatment, we employed here whole genome DNA methylation analysis, which can be correlated to gene expression and is relatively more stable compared to mRNA, the basis of the well-known whole transcriptome analysis.

Methods:

Total CD4⁺ T cells from 12 patients taken at baseline, equally segregated in placebo responders/non-responders (R/NR) and HSP peptide R/NR, were sorted from PBMCs. DNA was extracted and bisulfite converted to analyze the cytosine methylation pattern. Converted DNA was analyzed using Illumina Infinium HumanMethylation450 BeadChip, yielding the methylation percentage of more than 485,000 CpG sites across the genome.

Results:

A nearly perfect segregation of R and NR was observed when Principal Component Analysis (PCA) was applied to the whole dataset. When sites selected for discriminating R from NR (based on p-value and fold change) were used, PCA was also able to discriminate, based on average DAS-28 scores, weak responders from strong responders. Weak responders were more frequent among placebo treated subjects, while HSP peptide-treated subjects comprised stronger responders. In order to enrich for sites discriminating R and NR with high confidence without losing power due to the high number of CpG sites, pre-filtering of sites with low variance was performed. With this method, we found 51 sites differentially methylated between R and NR, clearly  supporting the potential of the whole genome methylation analysis for discovery of new candidate biomarkers. As a control, mock comparisons, such as between all placebo against all HSP peptide-treated subjects irrespective of clinical outcome (which is not meaningful at baseline due to patient randomization) were run and consistently yielded fewer hits. 

Conclusion:

We found significant differences in the general pattern of DNA methylation between RA patients responding or not responding to an epitope-specific therapy. We have also identified a set of CpG sites with high predictive and prognostic power for discriminating the responsiveness to treatment, which may be developed as a screening tool. The functions of the genes found to be differentially methylated is currently under investigation.

---

« Back to 2012 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/epigenetic-features-as-predictive-markers-of-responsiveness-to-epitope-specific-therapy-in-rheumatoid-arthritis/