Differential Methylation as a Biomarker of Response to Etanercept in Patients With Rheumatoid Arthritis

Abstract

Objective: Biologic drug therapies represent a huge advance in the treatment of rheumatoid arthritis (RA). However, very good disease control is achieved in only 30% of patients, making identification of biomarkers of response a research priority. We undertook this study to test our hypothesis that differential DNA methylation patterns may provide biomarkers predictive of response to tumor necrosis factor inhibitor (TNFi) therapy in patients with RA.

Methods: An epigenome-wide association study was performed on pretreatment whole blood DNA from patients with RA. Patients who displayed good response (n = 36) or no response (n = 36) to etanercept therapy at 3 months were selected. Differentially methylated positions were identified using linear regression. Variance of methylation at differentially methylated positions was assessed for correlation with cis-acting single-nucleotide polymorphisms (SNPs). A replication experiment for prioritized SNPs was performed in an independent cohort of 1,204 RA patients.

Results: Five positions that were differentially methylated between responder groups were identified, with a false discovery rate of <5%. The top 2 differentially methylated positions mapped to exon 7 of the LRPAP1 gene on chromosome 4 (cg04857395, P = 1.39 × 10(-8) and cg26401028, P = 1.69 × 10(-8) ). The A allele of the SNP rs3468 was correlated with higher levels of methylation for both of the top 2 differentially methylated positions (P = 2.63 × 10(-7) and P = 1.05 × 10(-6) , respectively). Furthermore, the A allele of rs3468 was correlated with European League Against Rheumatism nonresponse in the discovery cohort (P = 0.03; n = 56) and in the independent replication cohort (P = 0.003; n = 1,204).

Conclusion: We identify DNA methylation as a potential biomarker of response to TNFi therapy, and we report the association between response and the LRPAP1 gene, which encodes a chaperone of low-density lipoprotein receptor-related protein 1. Additional replication experiments in independent sample collections are now needed.

Figure 1

Volcano plot of −log₁₀ P value versus mean difference in methylation. CpG positions that are more methylated in nonresponders compared to good responders are plotted to the right.

Figure 2

Plot of CpG methylation values for the top 5 differentially methylated positions in both good responders and poor responders (i.e., nonresponders). Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/journal/doi/10.1002/art.39590/abstract.

Figure 3

Locus view of the LRPAP1 region on chromosome 4. Top, Plot of −log₁₀ P value (y‐axis) versus position (x‐axis). The red line indicates the value of 5 × 10⁻⁸ (genome‐wide significance threshold). Middle, Schematic drawing of gene and position of CpG islands. Bottom, Correlation structure of DNA methylation at the CpG sites in the genomic region.