Magnetic resonance imaging (MRI) for detection of active sacroiliitis--a prospective study comparing conventional radiography, scintigraphy, and contrast enhanced MRI.

OBJECTIVE: Sacroiliitis is often difficult to diagnose in the absence of radiographic alterations. For the diagnosis of active sacroiliitis, plain radiography, scintigraphy, and contrast enhanced magnetic resonance imaging (MRI) were evaluated in a prospective study. METHODS: In 44 consecutive patients with complete clinical and laboratory evaluation, plain radiographs, quantitative sacroiliac (SI) scintigraphy, and MRI were performed to evaluate the contribution of these imaging techniques to the diagnosis of active sacroiliitis. Scintiscanning and MRI were done in 20 control subjects to define the normal range of imaging findings. We determined the sensitivity and specificity for each imaging method using a reference standard based on clinical symptoms of inflammatory low back pain with or without laboratory signs, and on clinical and radiographic followup during 1.5-2.5 years to confirm diagnosis. RESULTS: MRI was most sensitive (95%) and superior to quantitative SI scintigraphy (48%) or conventional radiography (19%) for the detection and confirmation of active sacroiliitis. For the assessment of inflammatory signs, MRI had higher specificity (100%) than scintigraphy (97%) or plain radiography (47%). At repeat MRI after 2-30 months, there was persistent pathologic signal intensity in the subchondral bone area despite clinically successful antiinflammatory drug therapy. CONCLUSION: For the assessment of active changes in the synovial portion and the subchondral bone marrow, contrast enhanced MRI is superior to quantitative SI scintigraphy or conventional radiography. MRI picks up an additional 75% of early cases not diagnosed by plain radiography. Scintigraphy is only of limited value. Persistent pathologic signal intensity in the subchondral bone marrow seems to be closely associated with previous inflammatory episodes, thus limiting specificity of MRI for active sacroiliitis. Based on our findings we suggest an algorithm for the evaluation of patients with suspected active sacroiliitis.