gms | German Medical Science

Background: A classification of axial spondyloarthritis (axSpA) by the imaging arm of the ASAS criteria relies partly on the detection of a bone marrow edema (BME) in the magnetic resonance imaging (MRI) of the sacroiliac joints (SIJ) suspicious of SpA. We here evaluate different types of MRI changes possibly relevant for a diagnosis of axSpA as judged by radiologists taking the rheumatologist’s diagnosis as gold-standard.

Methods: Consecutive patients <45 years were included if they presented in a specialized rheumatologic center with chronic low back pain (duration >3 months). Patients underwent a complete diagnostic workup including MRI of the SIJ. All clinical and laboratory information including images but no radiological reports was available for experienced rheumatologists to make a diagnosis of axSpA or non-axSpA. In parallel, two experienced musculoskeletal radiologists, blinded to patients’ demographics and symptoms (except for back pain) evaluated all MR images without knowledge of the rheumatologist’s diagnosis, by quantification of BME, fat metaplasia, erosions, sclerosis and ankylosis based on the Berlin SIJ score. The radiologists also stated whether the patient is likely to have axSpA or not, solely based on MRI findings.

Results: A total of 100 patients were recruited. The rheumatologist diagnosed axSpA in 54 patients (mean age 31.5±8.0 years, 77.8% HLA-B27+, mean symptom duration 36.4±42.0 months), while 46 patients were diagnosed as non-specific back pain (age 33.6±7.1 years, 17.4% HLA-B27+, mean symptom duration 25.5±31.6 months). According to the radiologists, 38 patients were identified as axSpA, 34 of which were also diagnosed as axSpA by the rheumatologist (overall agreement with the clinical diagnosis: 63%), and 4 patients were thought to have axSpA by the radiologist but not by the rheumatologist (disagreement with the clinical diagnosis: 8.7%). Similarly, the quantification of MRIs showed higher scores in patients diagnosed as axSpA by the rheumatologist (Table 1 [Tab. 1]). Only few patients had sclerosis or ankylosis.

From the radiologist’s perspective, the calculated odds ratio (OR) for identification of axSpA by MRI only was 3.1 (95% CI:1.4-7.1) for the presence of BME, 3.5 (95% CI:1.4-9.0) for fat metaplasia, 2.8 (95% CI:1.1-7.0) for erosions, 2.0 (95% CI:0.7-5.5). For the combination of BME and any structural change, the OR was 3.7 (95% CI:1.6-8.5).

Conclusion: This study reveals a discrepancy between the rheumatologist’s and the radiologist’s identification of axSpA, confirming that a diagnosis of axSpA in daily practice should not rely on imaging findings only. Nevertheless, the overall specificity of the radiologists was acceptable, although the sensitivity was relatively low. These data suggest also that not only BME but also fat metaplasia and erosions are of value to diagnose axSpA, beyond classification. The combination of MRI changes seems to enhance the discriminative diagnostic performance. Finally, it will be important to define clinically relevant cut offs for the MRI scores.