Background Radiographs (hands ± feet) measured with total Sharp scores (TSS) are considered the reference standard for quantifying structural damage in RA clinical trials. Magnetic resonance imaging (MRI) is increasingly utilized due to its greater sensitivity to detect erosive damage, but responsiveness and discrimination of MRI measures for joint damage and inflammation are not fully characterized. In a systematic literature review (SLR), we identified the OMERACT RA hand/wrist MRI Score (RAMRIS) as the only validated MRI tool.1
Objectives To evaluate responsiveness and discrimination of RAMRIS synovitis, osteitis/bone marrow edema (BME), and erosions, we conducted a meta-analysis of published RCT data.
Methods RCTs utilizing 1.5T MRI and measuring at least one RAMRIS feature were identified. Level of evidence (LoE) was assessed using modified Cochrane Handbook criteria. Change data for RAMRIS measures were converted to weighted mean changes ± SDs, and analyzed per treatment type and time point. ANOVA, weighted effect size, and estimated sample sizes were calculated. Relative weighted effect sizes (wES, using sample size as weighting factor) for biologic versus DMARD+placebo and for Biologic+DMARD versus DMARD+placebo at each time point for each RAMRIS measure were estimated using the difference between weighted mean of mean changes between groups, divided by the mean of weighted standard deviation across treatment groups.
Results Ten RCTs provided resp/discrim data: nine with moderate (5/9) to good (4/9) LoE reported sufficient data to contribute to meta-analysis; one had inadequate LoE. Sample size ranged from 11 to 146 RA patients. Interventions studied were 7 biologics, 2 novel therapies +/- DMARDs in active RA patients. The table shows weighted effect size and sample size estimates for synovitis, osteitis and erosions across time. Compared to placebo, synovitis improved (p<0.05) at 4/6, 12/14/18, and 24 weeks; osteitis improved (p<0.05) at 12/18 and 24 wks. Erosion score showed very small changes with high variability and no statistical significance at these time points.
Conclusions RAMRIS measures with 1.5T MRI provide consistent responsiveness and discrimination for synovitis and osteitis/BME, but not for erosions.
Woodworth TG et al. Arthritis Rheum 2012.64:1050
Acknowledgements UCLA research librarians Rikke Ogawa and Lisa Federer, and Bristol Myers Squibb for an unconditional research grant.
Disclosure of Interest T. Woodworth: None declared, O. Morgacheva: None declared, L. Duan: None declared, O. Pimienta: None declared, O. Troum Grant/research support: Bristol Myers Squibb, president International Society for Musculoskeletal Imaging in Rheumatology (ISEMIR), N. Barroso: None declared, V. Ranganath Grant/research support: Bristol Myers Squibb, D. Elashoff: None declared, D. Furst Grant/research support: Abbvie, Actelion Pharmaceuticals US, Amgen, Bristol Myers Squibb, Gilead, GlaxoSmithKline, Novartis, Pfizer, Roche/Genentech, UCBPharma, Janssen Pharmaceutical Products