Diagnostic Yield and Safety of Brain Biopsy for Suspected Primary Central Nervous System Angiitis
Background and Purpose—The utility and safety of brain biopsy for suspected primary angiitis of the central nervous system (PACNS) are uncertain. Factors predictive of a positive biopsy have not been well described. Our aim was to evaluate the diagnostic yield and safety of brain biopsy in suspected PACNS and determine whether any prebiopsy variables are associated with a positive biopsy.
Methods—This is a retrospective study of consecutive patients who underwent diagnostic brain biopsy for PACNS at a single institution. The relationship between biopsy yield and patient demographics, surgical technique, laboratory testing, neuroimaging, biopsy characteristics, and prebiopsy immunosuppressive therapy were examined.
Results—PACNS was confirmed in 9 of 79 patients (11%). Biopsy identified alternative diagnoses in 24 patients (30%), with cerebral amyloid angiopathy (8 patients), encephalitis (5 patients), demyelination (3 patients), and CNS lymphoma (3 patients) most commonly found. There was no correlation between a positive biopsy and cerebrospinal fluid results, neuroimaging, surgical technique, biopsy characteristics, or preoperative immunosuppressive therapy. Smaller biopsies (P=0.02) and closed procedures (P=0.013) were less likely to yield a diagnosis. Postoperative complications occurred in 13 patients (16%), 3 (4%) of which were serious.
Conclusions—Brain biopsy leads to pathological confirmation of vasculitis in a minority of suspected PACNS cases but alternative diagnoses are often identified. Importantly, rare but meaningful complications may occur.
Primary angiitis of the central nervous system (PACNS) is a rare inflammatory disorder of brain vessels and leptomeninges causing progressive neurological dysfunction.1 Diagnosis of PACNS is challenging because it often presents with headache, altered mental status, seizures, and focal neurological deficits that can be seen with other conditions, including infection and malignancy.2 Diagnostic criteria are varied and controversial but many clinicians consider vasculitis as definite if there is confirmation on biopsy and probable if, in the absence of a tissue sample, there are findings consistent with angiitis on ancillary testing, including angiography (vessel beading), magnetic resonance imaging (infarct, hemorrhage, contrast enhancement), and cerebrospinal fluid (CSF) analysis (elevated white blood cell count and protein). It is considered PACNS angiitis if there is no evidence of systemic vasculitis or other conditions that could elicit the same clinical features.1 Because the treatment for PACNS usually involves strong immunosuppressants with significant side effects,3,4 brain biopsy is often recommended to confirm the diagnosis.1,5,6 Additionally, biopsy may identify alternative diagnoses that can lead to changes in treatment or, at least, avoidance of immunosuppressants and their attendant risks. Currently, there is limited data analyzing the yield and safety of brain biopsy for suspected PACNS.7–9 Our aim was to evaluate the diagnostic yield and safety of diagnostic brain biopsy for suspected PACNS and to determine whether any prebiopsy variables are associated with a positive biopsy.
We searched the electronic medical record at the Hospital of the University of Pennsylvania for patients with procedure codes for open and closed brain biopsies between January 1, 2005, and December 31, 2013. We cross-referenced these results with independent lists of biopsies from the Pathology and Neurosurgery Departments. We reviewed patients’ charts to identify patients who underwent biopsy for suspected PACNS. We obtained data regarding (1) demographics; (2) CSF results; (3) magnetic resonance imaging and conventional angiogram findings; (4) surgical technique (open versus closed, right frontal versus other approaches); (5) biopsy diagnosis, size, and presence of leptomeninges; (6) pre- and postoperative treatment with steroids and immunosuppressants; and (7) postoperative complications. A board-certified neuropathologist confirmed pathological vasculitis, defined as presence of transmural inflammation of the vessel walls with or without wall necrosis. Study procedures were approved by our Institutional Review Board. Chi-square and Wilcoxon rank-sum analyses were performed as appropriate.
Of 503 biopsies performed between 2005 and 2013, 79 were for suspected PACNS. Forty-four (56%) of these patients were female, 45 (57%) white, 15 (19%) black, and 4 (5%) Hispanic (Table). Forty-two percent were open, 58% closed, 63% performed via a right frontal approach, and 73% contained both brain tissue and leptomeninges (Table). Nine biopsies (11%) had pathological findings diagnostic of PACNS. Of the 70 that were not, 14 (18%) had perivascular inflammation without wall infiltration and did not meet pathological criteria for vasculitis. Alternative diagnoses were found in 24 patients (30%). Of these, 8 (10%) had cerebral amyloid angiopathy, 5 (6%) viral meningoencephalitis, 3 (4%) CNS lymphoma, 3 (4%) demyelination, and 5 (6%) other diagnoses: Alzheimer’s disease, tauopathy, posterior reversible encephalopathy syndrome, small vessel vasculopathy, and progressive multifocal leukoencephalopathy.
We did not find any significant correlation between patient characteristics and a positive yield on biopsy. This included CSF results, magnetic resonance imaging, and conventional angiogram findings, surgical approach, and the use of preoperative steroids or immunosuppressants (Table). Vasculitis patients had numerically but not statistically significantly higher CSF white blood cell counts (median of 5.5 cells/mm3 versus 2.5 cells/mm3; P=0.39) and higher protein (median 92 mg/dL versus 65 mg/dL; P=0.31). Yield was similar for open and closed biopsies (12% versus 11%; P=0.99). No significant differences in yield were seen with respect to biopsy location (8% right frontal versus 24% other locations, typically targeted areas of damage seen on imaging; P=0.21), inclusion of leptomeninges in the biopsy sample (12% when present versus 19% when absent; P=0.67), or larger biopsy size (median 0.20 cm3 in cases with no vasculitis versus 0.90 cm3 in cases diagnostic for vasculitis; P=0.59). Biopsies that failed to provide any useful data were significantly smaller in size than diagnostic biopsies (median 0.15 cm3 versus 0.4 cm3; P=0.02) and more likely with closed procedures (59% versus 41%; P=0.013).
Overall, 13 patients (16%) experienced postbiopsy complications: 6 (8%) had intracerebral hemorrhage, with 2 patients requiring evacuation; 2 (3%) had seizures; 3 (4%) experienced transient altered mental status; 1 patient had an infarct; and 1 had a CSF leak. Seven complications occurred with closed biopsies and 6 with open biopsies. In those with intracerebral hemorrhage, 2 (33%) were diagnosed with cerebral amyloid angiopathy and 4 (66.7%) had microhemorrhages on preoperative magnetic resonance imaging. Bleeding complications occurred more often with closed than with open biopsies (5 versus 1; P=0.39), although this was not statistically significant. There was no association between hemorrhage and biopsy size. None of the patients ultimately diagnosed with PACNS had complications.
In our series of patients undergoing brain biopsy for suspected PACNS, 11% had pathological confirmation of vasculitis and 18% were found to have perivascular inflammation but did not meet full criteria for diagnosis of PACNS. We did not find any baseline clinical, surgical, or imaging variables which predicted a positive biopsy. However, this study was not adequately powered to identify smaller but potentially clinically meaningful associations. Importantly, brain biopsy yielded alternative diagnoses in 24 patients (30%), helping to avoid use of potentially harmful immunosuppressants in 11 of these patients who may have been treated for presumed PACNS. Closed procedures and smaller biopsy size were associated with a greater likelihood of nondiagnostic findings, suggesting that open biopsies with adequate tissue sampling might be preferred.
Both our rate of biopsy-proven vasculitis and overall diagnostic yield (identification of vasculitis or an alternative diagnosis) are lower than prior reports looking specifically at patients suspected of having PACNS. These ranged from 23% to 36% and 53% to 83%, respectively.7–10 Although our study was the largest to date, these differences likely reflect the fact that each of these studies, including ours, has been a single-center, retrospective study, and the threshold for recommending biopsy likely varies from institution to institution and from clinician to clinician.
Complications occurred in 16% of patients, which is consistent with prior series.7–10 Of our complications, 3 (4%) were serious, requiring surgical intervention or causing permanent disability. The decision to perform a brain biopsy must, therefore, weigh the potential risks of the biopsy with the risks of immunosuppressive therapy, which can at times be worse than the biopsies themselves.3,8 Because of this, brain biopsy remains an important diagnostic tool. However, further studies are necessary to better establish which clinical variables are associated with a positive yield and which surgical techniques are most likely to provide diagnostic results with low risk of complications.
Dr Torres reports study design, gathering of data, and authorship of article. Dr Loomis reports study design, gathering of data, and article review. Dr Cucchiara reports study design, gathering of data, and authorship of article. Dr Smith reports provision of data and article review. Dr Messé reports study design, analysis of data, and authorship of article.
- Received April 26, 2016.
- Revision received April 26, 2016.
- Accepted May 17, 2016.
- © 2016 American Heart Association, Inc.