PhRAT Profile

Pharmacogenomics of Rheumatoid Arthritis Therapy

Abstract (Updated, April 2011)

Long-term outcome in patients with rheumatoid arthritis (RA) is highly dependent upon aggressive pharmacological control of inflammation early in the disease course. Despite the importance of selecting the optimal medication soon after disease onset, there is no clinical or biomarker predictor of drug treatment response. A genetic biomarker would be particularly useful for drugs that block the inflammatory cytokine TNF-alpha (TNF), as these drugs are first-line biological disease modifying anti-rheumatic drugs DMARDs, yet induce remission in only ~30% of patients. A biomarker would also provide insight into biological mechanism of treatment response. In this application, our central hypothesis is that common genetic variants of modest effect size predict response to anti-TNF therapy. To test this hypothesis, we propose to expand upon our established multi-center collaboration and available GWAS data to develop (1) new statistical methods for conducting GWAS (estimating variance explained by common single nucleotide polymorphisms, SNPs), and (2) new informatics methods for defining treatment response in the EMR (which will allow us to collect many more samples for GWAS).

Aim 1: Analyze GWAS data on ~2,500 RA patients to search for common variants that predict response to anti-TNF therapy. We will conduct a GWAS to search for single SNPs that are associated with treatment response. We will also develop multi-SNP "polygenic models" of response to anti-TNF therapy using GWAS data.

Aim 2: Use electronic medical records (EMR) at Partners HealthCare, Vanderbilt and Northwestern to define treatment response, and conduct a GWAS on ~1,200 additional RA patients treated with anti-TNF therapy. We will first define cohorts of RA patients at each medical center. We will then develop semi-automated algorithms using natural language processing (NLP) and codified EMR data to define disease activity, temporal patterns of medication usage, and response to treatment. Once defined and samples are collected, we will conduct a GWAS to replicate single SNPs and the polygenic model developed in Aim 1. The approach will establish an informatics approach to define treatment response at other institutions using EMR data.

PhRAT Team

Brigham and Women's Hospital, Partners Health Care Team

Harvard School of Public Health Team

Vanderbilt University Team

Northwestern University Team

Broad Institute Team