Objective To describe patterns of diagnostic findings, and identify subgroups of primary angiitis of the central nervous system (PACNS).

Methods We retrospectively analysed 31 patients with PACNS. Cases were selected by predetermined diagnostic criteria and stratified into biopsy-proven and imaging-based PACNS. We compared clinical characteristics, cerebrospinal fluid (CSF) findings and imaging results including high-resolution vessel wall MRI between groups.

Results There were 31 cases of PACNS (mean age 45.6 years, 58.1% female), of whom 17 (55%) were biopsy-proven, 14 (45%) were based on imaging findings. Patients with a positive biopsy had fewer infarcts (29.4% vs 85.7%, p=0.003), were more likely to have meningeal and parenchymal contrast enhancement (76.5% vs 28.6%, p=0.012), were less likely to have abnormal MR angiography (11.8% vs 100%, p<0.001) and did not show vessel wall enhancement at the time of diagnosis (0% vs 76.9%, p<0.001). In contrast, patients with imaging-based diagnosis showed more frequently multiple infarcts and vessel abnormalities, with vessel wall enhancement in most of the cases. Clinical characteristics and CSF analysis did not reveal marked differences between groups.

Interpretation Multi-parametric MRI distinguishes two subtypes of PACNS that most likely differ concerning the affected vessel size. Biopsy-proven PACNS primarily involves smaller vessels beyond the resolution of vascular imaging, while imaging-based PACNS affects predominantly medium-sized vessels leading to false-negative biopsy results. Using distinct MRI patterns may be helpful for selecting patients for appropriate invasive diagnostic modalities.