Background The “Rheumatoid Arthritis MRI Score” (RAMRIS) is a standardized method for evaluating MRI images of the hand and wrist of rheumatoid arthritis (RA) patients. Although widely used in research, it is associated with a long learning curve and considered time consuming. In addition, its good reproducibility may be limited to experienced users.
Objectives We evaluated the proficiency of RAMRIS readings by two inexperienced readers before and after a brief training period.
Methods Two sets of 12 MRIs each of hand and wrist of RA patients were used for the study. Patients of the two sets were matched for MRI disease severity. Pre- and post gadolinium Turbo 3D T1 sequences in the coronal plane with reconstruction in the axial plane and STIR sequences in the coronal and axial planes were obtained on a extremity-dedicated 0.2 T machine (Artoscan, Esaote, Genova, Italy). The RAMRIS reference (R) was the score given in agreement by two experienced readers (FB, MAC). Two rheumatology trainees graduated from 2 years and without direct experience in RAMRIS reading (A, B) were explained how to perform it and independently evaluated the first set of images using as reference the RAMRIS atlas. After one week, the MRIs of 6 of the patients of the first set and of 6 other patients were jointly evaluated and commented by all readers. Finally, A and B evaluated the second set of MRI images. The absolute differences in scores between R, A, and B were calculated for individual bones (erosions and bone marrow oedema, BME) in hand and wrist and for synovitis locations in the wrist only.
Results The median per patient absolute differences with R was 3 (range 0-30) for A and 4 (range 0-30) for B, considering all 24 patients (p=0.019). Differences with R decreased between the first and second readings (A= from 4 (range 0-21) to 2 (range 0-30), p=0.06; B= from 7.5 (range 0-30) to 3 (range 0-21), p=0.0003). κ values compared with R are shown in table 1 for the individual lesions and total RAMRIS. The mean time necessary for reading the first and second set was 54.6±14.5 min and 48.8±8.3 min for A and 32.0±8.5 min and 29.6±1.4 min for B, respectively (both ns changes).
Conclusions Ability in RAMRIS reading varied between observers according to the different lesion scored. A one week intensive training with readings of 12 MRIs of RA patients significantly increased the proficiency of inexperienced RAMRIS readers.
Disclosure of Interest None declared