Background Magnetic resonance imaging offers a direct means of evaluating inflammation disease activity and damage in rheumatoid arthritis joints. However, few studies have evaluated intra-machine MRI.¹

Objectives Evaluate the agreement of intra-machine MRI using the OMERACT 5 Rheumatoid Arthritis MRI Score (Om5 RAMRIS)² and compare MRI reliability to clinical and laboratory measures.

Methods 12 subjects with seropositive RA, median disease duration 6 years, were chosen to reflect spectrum of RA severity. MRI was performed at same time of day at baseline and 48 h. All sources of variability were either controlled or measured. A cannula was inserted into the nondominant arm and serum was collected prior to MR for ESR/CRP. The cannula was left insitu for administration of gadolinium (Gd). MRI of wrist was performed with dominant hand splinted in semipronated position alongside the body. Axial and coronal images were obtained pre-Gd using T1, proton density and T2 (FS) sequences. Gadolinium was administered at 0.1 mmol/kg over 90 secs. Post Gd fat suppressed sequences commenced 2 min post injection (axial then coronal). MRI images were read in random order by a single blinded observer using the Om5 RAMRIS. All images were reread in random order at 48 h by same observer. Additional clinical data at baseline and 48 h: HAQ, SF36, 28 swollen and tender joint count (same observer).

Results Intra-machine MRI (ie at baseline and 48 h) intraclass correlation coefficients (ICCs) were lower for all measures of disease activity (synovitis global, synovial maximal thickness and bone oedema) than for lesions of disease damage (bone erosions and bone defects). ICCs of summated wrist activity score for intra-machine MRI were 0.53 – 0.70, whereas intra-reader ICCs were 0.86 and 0.90. However ICCs of summated wrist damage score for intra-machine MRI were 0.89 – 0.96 and for intra-reader were 0.93 and 0.97. Mean (SD) scores at baseline and 48 hrs. ICCs for clinical and laboratory data were: DAS28 = 0.94, HAQ = 0.96, TJC = 0.89, ESR = 0.76, SJC = 0.72 and CRP = 0.55. Mean differences at baseline and 48 hrs for all clinical, laboratory and MRI measures were not statistically significant. Furthermore, none of the differences consistently reflected either an increase or decrease in disease activity. Smallest detectable difference (SDD) for selected measures were as follows: wrist activity ± 35 (50% of score range (SR)), wrist damage ± 18 (15%SR), DAS28 ±. 98 (23% SR), HAQ ±. 49 (21% SR), SJC ± 5.4 (57%SR), TJC ± 5.2 (35%SR) and CRP ± 10 (71%SR).

Conclusion Intra-reader agreement using the OMERACT 5 Rheumatoid Arthritis MRI Score has excellent intra-reader agreement for the assessment of synovitis, bone oedema, bone erosions and defects. Intra-machine agreement at 48 h was also excellent for bone erosions and defects and the SDD was quite acceptable. Lower ICCs for measures of MRI RA activity probably reflect underlying true variability in disease activity even at 48 h (suggested by variability in CRP), as well as a component of measurement error, (implied by relative stability of DAS28). Thus to disentangle the discrimination (reliability vs responsiveness) of this MRI activity scoring system, it requires further evaluation in a RCT.

References

1. Lassere M, et al. J Rheum, in press

2. Conaghan P, et al. J Rheum, in press