A comparative analysis of magnetic resonance imaging and high-resolution peripheral quantitative computed tomography of the hand for the detection of erosion repair in rheumatoid arthritis

To investigate whether MRI allows the detection of osteosclerosis as a sign of repair of bone erosions compared with high-resolution peripheral quantitative computed tomography (HR-pQCT) as a reference and whether the presence of osteosclerosis on HR-pQCT is linked to synovitis and osteitis on MRI. A total of 103 RA patients underwent HR-pQCT and MRI of the dominant hand. The presence and size of erosions and the presence and extent (grades 0-2) of osteosclerosis were assessed by both imaging modalities, focusing on MCP 2 and 3 and wrist joints. By MRI, the presence and grading of osteitis and synovitis were assessed according to the Rheumatoid Arthritis MRI Score (RAMRIS). Parallel evaluation was feasible by both modalities on 126 bone erosions. Signs of osteosclerosis were found on 87 erosions by HR-pQCT and on 22 by MRI. False-positive results (MRI(+)CT(-)) accounted for 3%, while false-negative results (MRI(-)CT(+)) accounted for 76%. MRI sensitivity for the detection of osteosclerosis was 24% and specificity was 97%. The semi-quantitative scoring of osteosclerosis was reliable between MRI and HR-pQCT [intraclass correlation coefficient 0.917 (95% CI 0.884, 0.941), P < 0.001]. The presence of osteosclerosis on HR-pQCT showed a trend towards an inverse relationship to the occurrence and extent of osteitis on MRI [χ(2)(1) = 3.285; ϕ coefficient = -0.124; P = 0.070] but not to synovitis [χ(2)(1) = 0.039; ϕ coefficient = -0.14; P = 0.844]. MRI can only rarely detect osteosclerosis associated with bone erosions in RA. Indeed, the sensitivity compared with HR-pQCT is limited, while the specificity is high. The presence of osteitis makes osteosclerosis more unlikely, whereas the presence of synovitis is not related to osteosclerosis.