To investigate distinguishing primary central nervous system (CNS) vasculitis compared to non-inflammatory intracranial vasculopathy using features in clinical presentation and blood and cerebrospinal fluid (CSF) markers.

We evaluated all cases enrolled in the prospective registry of CNS vasculopathy at a tertiary medical center. Final diagnoses were reviewed and patients with incomplete workup were excluded. Infectious vasculopathy, reversible cerebral vasoconstriction syndrome, autoimmune or inflammatory disease, and cardio-embolic disease were excluded since they may be more readily distinguished from primary CNS vasculitis. Several clinical features on first presentation such as gender, ethnicity, clinical presentation, past medical history, and medications in addition to serum and CSF findings were investigated for their ability to predict the final diagnosis. Data was randomly divided to discovery and internal validation cohorts with 7:3 ratio.

Overall, 208 patients were identified who had completed the workup with a final diagnosis. Thirty-six patients with primary CNS vasculitis and 26 patients with non-inflammatory vasculopathies (large vessel atherosclerotic disease, small vessel disease, intracranial dissection, and amyloid vasculopathy) were included. A logistic regression model was evaluated, and using recursive feature elimination top 3 features were selected to optimize the accuracy including presence of seizure (OR: 11.2, 95% CI = 1.1-119.4), past medical history of transient ischemic attack or hyperlipidemia (OR: 0.3, 95% CI = 0.1-0.8), and CSF pleocytosis (OR: 4.3, 95% CI = 1.2-15.7). The model predicted primary CNS vasculitis in validation cohort with sensitivity of 83%, specificity of 100%, and accuracy of 89%.

Primary CNS vasculitis might be differentiated from non-inflammatory intracranial vasculopathy based on clinical presentation and CSF findings.