Background/Purpose: Radiographic scoring of erosions and joint space narrowing (JSN) of the hand/wrist is standard to evaluate RA joint damage for regulatory approval of DMARDs; however, MRI has demonstrated its superiority in assessing joint inflammation and damage in clinical trials. MRI field strength impacts feasibility, image quality and cost. Measurement methods using high field (hf) ≥1.0 tesla (T) MRI requires large machines and specialized space. Low field (lf) <1.0T MRI has also demonstrated utility in clinical trials and is easily performed in a clinical office setting. We report the results of an SLR examining Outcome Measures in Rheumatology Clinical Trials (OMERACT) RA MRI scoring (RAMRIS) validation for lf images of RA wrist/hand.

Methods: We searched PubMed with Cochrane hedge for articles from 1970 to Aug 2012 using search terms enabling relevant data extraction. Data included: demographics/RA features in adults, MRI field strength (<1.0 T), wrist/hand images assessing ≥1 joint feature: synovitis, bone marrow edema (BME)/osteitis, erosions, JSN, tenosynovitis, in addition to measurement method and validity evidence. We applied OMERACT validity definitions: face, criterion, content, construct, also assessing reliability, responsiveness, discrimination, and feasibility. Quality was determined by Cochrane Handbook criteria, adapted for imaging research.

Results: 35 lf MRI articles met criteria for extraction. RAMRIS was most often used, as with our previous SLR evaluating ≥1.0T MRI. Altogether, 16 articles using <1.0T/RAMRIS were analyzed, seeking validated MRI measures: 0.2T (13 articles, including 3 RCTs), 0.6T (2 articles), and 0.3T (1 article). 5 articles compared lf to high field (hf) MRI. Table 1 shows lf MRI validation data. There is indirect criterion validity for synovitis and erosions: 2 articles compared lf to hf MRI for synovitis and 2 others for erosions. One study provided good erosion score correlation between lf MRI and computed tomography, and another one with x-ray. Validating data with 0.2T included prediction of x-ray progression, sensitivity to change by 4 weeks, and intra/inter reliability for synovitis, BME, and erosions. Data currently lacking include discrimination for synovitis, BME, erosions, construct discriminant for erosions, and validation for JSN or tenosynovitis.

Conclusion: Using a rigorous PRISMA-compliant SLR to examine low field MRI/RAMRIS of the RA wrist/hand, we found 0.2T MRI is partially validated to measure synovitis, BME/osteitis, and erosions. To finalize validation, evidence for discrimination between treatment groups for synovitis, BME, erosions, and discriminant construct validity for erosions are needed.

     Table 1. Number of articles providing validity data for RAMRIS measurement with lf MRI