Background Early diagnosis and treatment of rheumatoid arthritis (RA) is advocated. However, in part of the RA-patients a classifiable diagnosis cannot be made at first presentation; these patients present with unclassified arthritis (UA). The use of MRI for early diagnosis of RA is recommended, yet the evidence is limited.

Objectives To assess the performance of hand and foot MRI for early diagnosis of RA.

Methods Unilateral contrast-enhanced 1.5T MRI of the hand and foot was performed in 589 early arthritis patients included in the Leiden Early Arthritis Clinic, of whom 229 presented with RA, 159 with other arthritides and 201 with UA. MRI-findings observed in symptom-free controls served as reference to define an abnormal MRI. In preliminary investigations, patients that presented with RA were compared with symptom-free controls and with patients that presented with other arthritides. Thereafter, the accuracy of MRI was determined in UA-patients that were followed for 1-year on fulfilling the 1987-RA-criteria (primary outcome); the secondary outcome was start of disease-modifying drugs.

Results The results of the preliminary investigations were promising and showed that MRI-detected tenosynovitis was more discriminative than other types of MRI-inflammation. Of all UA-patients, 29 (14%) developed RA and 75 (37%) started disease-modifying treatment. MRI-detected tenosynovitis was associated with RA-development independent of other types of MRI-inflammation (odds ratio (OR) 7.5 95%CI 2.4–23) and also independent of age and commonly used measures of inflammation (number of swollen joints and C-reactive protein) (OR 4.2 95%CI 1.4–12.9). Within UA-patients, the negative predictive value (NPV) of abnormal tenosynovitis was 95% and the positive predictive value (PPV) 25%. The performance was best in the subgroup UA-patients presenting with oligo-arthritis (18% developed RA): PPV was 36%, NPV 98%, sensitivity 93%, and specificity 63%. Decision curve analysis revealed a higher net benefit for a model including MRI-detected tenosynovitis.

Conclusions MRI-detected inflammation, MRI-detected tenosynovitis in particular, contributes to identification of the UA-patient that will develop RA.

Disclosure of Interest None declared