Abstract 2183: Serum Cytokine Profiles Identify a Subset of Patients with Early Inflammatory Arthritis Responsive to Early DMARD Therapy
Authors
A. Brar, K. Wong, N. Knowlton
Purpose:
Currently selection of DMARDs by physicians for treatment of patients with RA are empiric, as there are no biomarkers that predict drug response. This study utilized serum cytokine profiles at baseline (pretreatment) to predict responsiveness to DMARD treatment.
Methods:
80 patients with early (< 1 yr) untreated inflammatory arthritis (43 RA, 22 undifferentiated arthritis; 17 other arthritis) were identified and followed for > 1 yr. Serum levels of 23 cytokines were measured using a multiplex bead based immunofluorecence assay. Levels were normalized by z transformation and classified using hierarchical cluster analysis and pearson correlation co-efficients. Cytokine levels were correlated with treatment response and remission as defined by EULAR criteria.
Results:
Cytokine levels did not distinguish patients with RA from UA, nor those with erosions from those without erosions at baseline, nor those who were RF or anti-CCP positive. However, patients could be clustered by cytokine profiles. Discriminate functional analysis suggested MCP1, RANTES, GMCSF and IFNγ were the main determinants of cluster grouping. In fact, INFγ most accurately predicted classification. Cluster group 1 had a higher prevalence of CCP positivity (60% vs 37% in cluster group 2, p<0.03), but outcomes at the last visit (prevalence of erosions and % achieving remission) were not different for the two groups. Although there were no overall differences in response to treatment (MTX or SSZ), subjects in cluster 2 were more likely to be in remission at the last clinic visit than similarly treated subjects in cluster 1 (15/21 71% vs 3/12 25% p=0.01). Subjects in cluster 1 were also treated with more DMARDs during the study period than cluster 2.
Conclusion:
Cytokine profiles may identify a subset of early inflammatory arthritis patients with potentially less response to MTX or SSZ, suggestive of more aggressive disease. This suggests that cytokine profiles may assist in treatment decisions.
Editorial Comment:
The need for biomarkers that identify patients most likely to respond to a particular therapy is greatly needed. Most likely a single biomarker will not be adequate for this purpose, and an array of proteins, autoantibodies, or RNA will likely prove to be more informative. This study attempts to identify a cytokine “signature” that might differentiate treatment responders from nonresponders. Interestingly the patients clustered into two clear groups and the groups were best distinguished by IFNγ expression. This is somewhat surprising, as the IFNγ signature has been considered to be a marker of SLE more than RA. Also of interest was the inability to find a clinical feature at baseline that could explain the clustering of patients into the two groups. What is it about Group 1 that expressed high levels of IFN relative to group 2? It is not clear. With regards to treatment, however, cluster 1 may have had a somewhat less robust response to standard nonbiologic DMARDs. These results will clearly have to be studied in a larger population. And, perhaps of even greater interest will be the identity of the feature that distinguishes the differential cytokine expression in the two groups.
