Background There is compelling evidence that early treatment of rheumatoid arthritis (RA) associates with better disease outcome, underlying the need of early identification of RA. Magnetic resonance imaging (MRI) has been proven sensitive in detecting joint erosion and inflammation. Although it has been suggested that extremity MRI can aid in differentiating RA patients from undifferentiated arthritis (UA) patients, the value of MRI in diagnosing RA is still unclear.
Objectives To evaluate if the addition of extremity MRI findings to the 2010 RA classification criteria is beneficial to accurately diagnose RA in UA patients.
Methods 205 early arthritis patients, included in the Leiden early arthritis clinic between August 2010 and August 2013, that did not fulfill the 1987 RA criteria or did not have other rheumatic diagnoses were studied. Patients underwent unilateral 1.5T MRI of the MCP and wrist joints at inclusion. MRIs were made and scored following the RA-MRI-scoring system (RAMRIS) by one experienced reader (intra-observer intraclass correlation 0.925). The outcome measures were fulfilling the 1987 RA criteria or the start of disease modifying anti rheumatic drugs (DMARDS) within the first year. The diagnostic accuracy of the 2010 criteria alone and the criteria combined with MRI findings (presence of erosions, bone marrow edema and synovitis) were compared. An MRI-feature was scored positive if the score of that RAMRIS-feature was ≥1; sensitivity analyses were also run on a cut-off of ≥2.
Results Patients had a mean age of 55, 61% were women, the median number of swollen joints was 3, median symptom duration was 10.7 weeks and 22% was ACPA-positive. During the first year, 47 patients (23%) fulfilled the 1987 RA criteria and 96 patients (47%) were prescribed DMARDs. When studying the start of DMARDS as outcome and applying the 2010 criteria the sensitivity was 40% (95%CI 33% to 46%), specificity 88% (95%CI 84% to 93%), accuracy 65% (95%CI 59% to 72%) and the AUC 0.64 (95%CI 0.56 to 0.72). When the presence of synovitis on MRI was combined with the 2010 criteria, the sensitivity increased to 91% (95%CI 87% to 95%)) and the specificity decreased to 28% (95% CI 21% to 34%)); the accuracy (57% (95%CI 50% to 64%)) and AUC (0.59 (95%CI 0.54 to 0.64)) did not differ significantly. Similar data were found for the addition of bone marrow edema or MRI detected erosions to the 2010-critieria. Similar findings were obtained when evaluating fulfillment of the 1987 criteria after 1-year as outcome. When a MRI was defined positive for a finding with a score ≥2, data were also comparable.
Conclusions Adding MRI findings to the 2010 criteria for RA increased the sensitivity compared to the criteria alone, but at the cost of a considerable decrease in specificity and the accuracy and AUC did not improve. More thorough evaluations are needed to determine which MRI findings are specific for RA in UA patients. It also needs to be determined if the severity and location of inflammation needs to be taken into account. Based on the present data we conclude that adding information on the presence of any MRI-detected synovitis, BME or erosions to the 2010 criteria does not improve the diagnostic process of RA in UA patients.
Disclosure of Interest None declared