Introduction: The diagnostic process for possible PACNS is complex and deals with an array of diseases including neurologic, infectious, neoplastic and inflammatory disorders with which no single specialty is highly familiar. We have reviewed our experience with such referrals and have been impressed with the diagnostic diversity of suspected cases.

Methods: We have reviewed 80 consecutive consultations (1986-1998) performed by one of us (LHC) including only those of sufficient pre-test probability to warrant angiography and/or CNS biopsy and report our clinical and laboratory findings.

Results: Thirty-six were ultimately classified as definite PACNS, 8 as possible PACNS, 22 as non-PACNS and 14 remained unclassified. Among the non-PACNS cases: 3 were infections, 5 neoplasms, 1 demyelinating disease, 3 inflammatory processes and 10 various diseases. Among the non-PACNS cases, 12 were initially treated as vasculitis but had this therapy discontinued after evaluation. Of these, 75% were females, 91.7% had an acute onset of presentation, and the most frequent complaints were headache (75%) and/or focal neurologic symptoms (83.3%). Among the 22 non-PACNS patients the initial misdiagnosis was most frequently made based on an abnormal MRI (81.8%) and/or a low probability angiogram (61.1%). Despite a vigorous diagnostic process 14 were not classified definitively either as PACNS cases or as non-cases. Eight of these underwent empirical treatment plans for PACNS.

Conclusions: 1) Even among cases with high pre-test probability for PACNS most are found to have another nosologic diagnosis, 2) Not all cases can be definitively categorized despite a vigorous diagnostic approach and 3) Based on such complexity most suspected cases warrant a multidisciplinary approach.

 Purpose: To determine whether a typical magnetic resonance (MR) imaging appearance exists for central nervous system vasculitis (CNSV), and whether MR findings predict the presence of vascular abnormalities on angiography.

Methods: Of 261 angiograms read at our hospital between 1986 and 1997, 32 were consistent with CNSV. Twenty-two of these studies were performed in patients with rheumatologic disorders, and 18 of the 22 had both angiograms and MR studies available for review. We characterized the MR lesions by type (bland infarct versus hemorrhage), size, number and location, and correlated the MR findings with angiography. We used Scheffe's procedure to compare MR and angiography in terms of the frequency of lesion occurrence in each vascular distribution.

Results: Causes of CNSV included primary angiitis of the CNS (PACNS), systemic lupus erythematosis (SLE), polyarteritis nodosa and giant cell arteritis. On average, each patient had 4.1 lesions on MR. The involved brain regions included subcortical white matter (20 lesions), cortex (16 lesions), deep gray matter (16 lesions), deep white matter (9 lesions) and cerebellum (9 lesions). There was a predilection for lesions occurring in the distribution of the middle cerebral artery (compared with the anterior or posterior cerebral artery distributions and the posterior fossa). Angiography and MR both detected a total of 74 lesions, but correlation between the two imaging modalities was poor for individual lesions. Of the 74 arterial lesions detected by angiography, 33 (45%) had corresponding abnormalities on MR. Conversely, of the 74 lesions detected by MR, 49 (66%) were evident on angiography.