Aim: We compared the clinical utility of magnetic resonance angiography (MRA) to catheter cerebral angiography (CA) in the investigation of children with suspected central nervous system (CNS) vasculitis.
Method: Single-centre retrospective review of children with a suspected diagnosis of CNS vasculitis studied with both MRA and CA. MRA and CA-detected abnormalities (location, multiplicity, and morphology) were compared; sensitivity and specificity were calculated on a per lesion and per patient basis for MRA, with CA as the reference standard.
Results: Findings in fourteen patients (median age at presentation of 5 y 10 mo [range 1 y 5 mo-14 y 5 mo]; eight males, six females) relating to sixteen paired studies of MRA and CA were reviewed. CA-detected lesions were commonly bilateral (13/16 studies, p<0.05), and likely to be proximally distributed (15/16 studies, p<0.05).The sensitivity and specificity of MRA for CA lesion detection was 63% (95% confidence interval [CI] 48-78) and 89% (95% CI 81-93), respectively with moderate agreement between the two modalities (kappa=0.51, 95% CI 0.37-0.66). The majority of the false negative observations involved the posterior circulation (9/14). The overall sensitivity for MRA diagnosis of vasculitis per patient was 94% (95% CI 67-99).
Interpretation: MRA failed to identify all lesions detected on CA, particularly those in the posterior circulation. MRA is a reasonable initial modality in the investigation of suspected CNS vasculitis but in cases of abnormal parenchymal MRI and normal MRA, CA should be considered.