Reliability of progressive and additive MRI scoring systems for evaluation of haemophilic arthropathy in children: Expert MRI Working Group of the International Prophylaxis Study Group

Effective treatment of haemophilic arthropathy requires a detailed evaluation of joint integrity. Methodological assessment of magnetic resonance imaging (MRI) scores are needed to assure reproducibility of measurements when comparing results of clinical trials conducted in different centres. We compared the reliability of two MRI scoring systems for assessment of haemophilic arthropathy: one progressive system that displays the most severe change and one additive system that depicts osteochondral and soft tissue-related changes. A total of 47 1.5 T MRI examinations of knees (n = 21) and ankles (n = 26) of 42 haemophilic boys, age range, 22 months to 18 years, performed at different centres (Toronto, n = 20, Europe, n = 12 and Denver, n = 15) were independently reviewed by four radiologists at two occasions. Twenty-two examinations were from children < 9 years and 25 from children ≥ 9. Sagittal and coronal gradient-echo (MPGR, 3D FLASH with fat saturation, GRASS) images were obtained. The MRI examinations of the ankle and knee studies presented with osteochondral abnormalities in 38.5% and 23.8% of the cases respectively. The two scoring systems demonstrated an excellent inter-reader [progressive, 0.88; additive (A, e, s and h components), 0.86] and intra-reader [progressive, 0.92; additive (A, e, s and h components), 0.93] reliability using intraclass correlation coefficients (ICCs). Although ICCs were slightly higher for knees when compared with ankles, and for older children when compared with younger children, all values fell within excellent inter- and intra-reader reliability categories. The two MRI scoring systems demonstrated a comparable reliability. This result constitutes the basis for further development of a combined MRI scoring system for assessment of haemophilic arthropathy, which incorporates progressive and additive components.