Session Information
Title: Rheumatoid Arthritis - Clinical Aspects III: Predictors of Disease Course in Rheumatoid Arthritis
Session Type: Abstract Submissions (ACR)
Background/Purpose: C reactive protein (CRP) levels have been used extensively to model disease activity in rheumatoid arthritis (RA). However, TNF antagonist therapy (and likely therapy with other biologics) reduce CRP levels, even without associated reductions in RA disease activity. Serum amyloid A (SAA) also models RA disease activity and has been suggested as a better biomarker for RA disease activity. We previously showed that TNF antagonist therapy lowered both SAA and CRP levels to a greater extent than oral disease modifying anti-rheumatic drugs (DMARDs) after adjusting for disease activity. However, TNF antagonists, compared to oral DMARDs, decreased CRP levels more than SAA levels. Our aim was to compare the effect of etanercept (ETN) and/or oral DMARD therapy on serum SAA versus CRP levels, and to determine whether SAA levels model RA disease activity better than CRP levels, especially during treatment with a TNF antagonist.
Methods: Samples were analyzed from RA subjects (n = 755) in the Treatment of Early RA (TEAR) study, a randomized, double-blind, comparative effectiveness trial. In the TEAR trial, early RA subjects with less than 2 years of disease duration were randomized to receive either combination ETN/methotrexate (MTX) therapy (n = 244), MTX/hydrochloroquine/sulfasalazine (triple oral) therapy (n = 132) or MTX monotherapy (n = 379). MTX monotherapy subjects with a DAS28-ESR > 3.2 after 6 months were stepped up to either ETN/MTX (n = 205) or triple oral therapy (n = 93). Serum samples and clinical data were collected when treatment was initiated and at 24, 48 and 102 week follow up visits. We used Spearman correlation coefficients (rho) to determine the overall relationship between SAA and CRP levels, and mixed effects models to determine the fit between 1) SAA levels and DAS28-ESR and 2) CRP levels and DAS28-ESR, for subjects treated with ETN/MTX and subjects treated with oral DMARDs, while correcting for baseline SAA and CRP levels, respectively. Akaike information criterion (AIC) and parameter estimates were used to determine model fit, with ΔAIC > 10 used as strong evidence for one model fitting better than another.
Results: At the baseline visit, SAA levels were only moderately correlated with CRP levels (rho = 0.42), and the SAA and CRP correlation was similar regardless of treatment or study visit (rho range 0.35 to 0.62). Across all subjects and time points, models of the DAS28-ESR using SAA levels were better than models using CRP; the ΔAIC between the models was 305. The model of DAS28-ESR using SAA was associated with a ~6 fold better fit vs. the CRP model for subjects treated with ETN/MTX, and a ~5 fold better fit vs. the CRP model for subjects treated with oral DMARDs (ΔAIC = 159 vs. ΔAIC = 137, respectively).
Conclusion: The lack of a strong correlation between SAA and CRP levels coupled with their different modeling of RA disease activity suggests that SAA levels may be a better biomarker for RA disease activity than CRP, especially during treatment with TNF antagonists.
Disclosure:
I. D. Metes,
None;
D. W. Chew,
None;
A. M. Patel,
None;
G. K. Balasubramani,
None;
J. R. Curtis,
None;
S. L. Bridges Jr.,
None;
S. R. Wisniewski,
None;
L. W. Moreland,
None;
M. C. Levesque,
None.
« Back to 2013 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/serum-amyloid-a-levels-model-rheumatoid-arthritis-disease-activity-better-than-c-reactive-protein-levels-especially-during-treatment-with-anti-tumor-necrosis-factor-%ce%b1-therapy-etanercept/