Background Juvenile Idiopathic Arthritis (JIA) is the main cause of chronic arthritis in children, and it can lead to joint damage and disability. Detecting early articular involvement in JIA is therefore of crucial importance to prevent cartilage and bone damage in these young patients. Since 2004 we have undertaken a prospective study of children with newly diagnosed JIA with knee involvement, and we noted that, not infrequently, knees deemed clinically normal had appreciable effusion on ultrasound (US).

Objectives We wished to prospectively compare agreement between clinical, US and MRI assessments of the knee joints in children with JIA, and to develop an MRI scoring system that allows assessment of disease activity and progression in children with JIA and compare it with US examination.

Methods Three hundred and thirty one knees from 48 children, affected by JIA with knee arthritis, were assessed clinically and ultrasonographically on the same day, using a semi-quantitative scoring system from 0 to 3 for swelling and effusion, respectively. A subgroup of these children (25) with a total of 40 knees had matching MRI scans obtained within 0 to 14 days from clinical and US examinations. For those, US and MRI scans (T2 weighted images) were scored 0-3 for effusion, synovial hypertrophy, bone oedema and bone erosions, using our newly developed knee MRI scoring system. An open 0.2 Tesla scanner without administration of gadolinium was used.

Results A moderate agreement for effusion was found between the 331 knees assessed clinically and ultrasonographically (K 0.54). Out of the 260 knees without swelling, 30 (11.5%) had mild to moderate effusion on US (Table). In the subgroup of 40 knees that had matching US and MRI scans it was demonstrated a good agreement for effusion (K 0.66) and a moderate agreement for synovial hypertrophy (K 0.47) between the two methods of assessments. The inter-observer reliability for US was very good for effusion (K 0.87), and good for synovial hypertrophy (K 0.68). The intra-observer reliability for MRI was good for effusion (K 0.73) and very good for synovial hypertrophy (K 0.85). Moreover, US tends to be more sensitive than MRI at detecting joint effusion (US scored on average 0.08 higher on the scoring system than MRI), maybe due to the lower sensitivity of MRI as T2 weighted images rather than contrast with gadolinium was used.

Conclusions A significant number of knee joint effusions are missed on clinical examination. Therefore US of this joint should be used as an adjunct to clinical examination to avoid under-diagnosis, especially when joint injections are being considered, when clinical examination is negative and symptoms are equivocal for active arthritis, at follow-up to assess treatment’s efficacy. We have developed a reproducible scoring system for MRI of the knee. US appeared to be more sensitive than MRI at detecting effusions, however this is probably due to the absence of using gadolinium in MRI. It does however highlight the benefits of MSUS in paediatric rheumatology units where scans can be obtained at the same time as clinical assessment.

Disclosure of Interest None Declared