Diagnostic accuracy of ultrasound versus cranial MRI for giant cell arteritis: dependence on cardiovascular risk for ultrasound only

OBJECTIVES: To compare measures of diagnostic accuracy of cranial, contrast-enhanced 3 T 2D T1-fatsat black-blood (T1fs-BB) MRI and ultrasound of the temporal arteries (TA) for suspected giant cell arteritis (GCA). To evaluate the influence of cardiovascular risk (CVR). METHODS: Retrospective, single centre study with patients ≥50 years, evaluated for GCA with both MRI and ultrasound. Diagnostic reference: expert diagnosis ≥6 months. Patients were categorized into two CVR-categories. TA segments were re-rated on a scale from 0 to 3 for MRI. Ultrasound grading used halo/compression signs and intima media thickness (IMT) cut-offs using the compressed method. RESULTS: A total of 44 patients included: 74 (51.4%) with GCA and 70 (48.6%) without. In total, 54 (37.5%) had high to very high CVR. Binary agreement between MRI and ultrasound: 79.2% to 83.3% for patient level and 84.7% to 85.0% for segment level. MRI had a sensitivity of 87.8% (78.5-93.5%) and a specificity of 87.1% (77.3-93.1%). Ultrasound IMT cut-offs had a sensitivity of 86.5% (76.9-92.5%) and a specificity of 74.3% (63.0-83.1%); OMERACT signs 87.8% (78.5-93.5%) and 72.9% (61.5-81.9%). While in the high to very high CVR subgroup, MRI preserves its diagnostic accuracy, specificity drops considerably for ultrasound. CONCLUSION: For the total population T1fs-BB-MRI outperformed ultrasound due to its higher specificity. For patients without high to very high CVR, both modalities perform comparably. In patients with high to very high CVR, MRI has better diagnostic accuracy. Different IMT cut-offs and liberal use of confirmatory tests seem justified for patients with high CVR.