Article Text

Download PDFPDF

Primary angiitis of the central nervous system: avoiding misdiagnosis and missed diagnosis of a rare disease
  1. Shamik Bhattacharyya,
  2. Aaron L Berkowitz
  1. Department of Neurology, Brigham and Women's Hospital, Boston, Massachusetts, USA
  1. Corresponding to Dr Shamik Bhattacharyya, Department of Neurology, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115, USA; sbhattacharyya3{at}


Primary angiitis of the central nervous system (PACNS) is an extremely rare condition, defined as a vasculitis limited to the CNS with no identifiable cause. Its presentation is non-specific and includes headache, cognitive dysfunction and focal neurological signs. Laboratory studies, neuroimaging and angiography are neither sufficiently sensitive nor specific for diagnosis; a definitive diagnosis requires brain biopsy. As a result, PACNS is commonly misdiagnosed. Here, we review its clinical, laboratory and radiological features, and focus on avoiding common diagnostic pitfalls.


Statistics from


Isolated vasculitis of the central nervous system (CNS) with no identified cause has many names, including primary angiitis of the CNS (PACNS), primary CNS vasculitis, cerebral granulomatous angiitis, isolated angiitis of the CNS and others. The components of these many names highlight its important and defining features:

  • CNS vasculitis/angiitis: PACNS is caused by inflammation of blood vessels of the CNS. CNS vasculitis is one of many causes of CNS vasculopathy; vasculopathy refers broadly to any blood vessel pathology, including vasospasm and atherosclerosis.

  • Primary: Thorough clinical and laboratory examination in PACNS does not identify another disorder causing the vascular inflammation. This contrasts with the many secondary causes of CNS vasculitis, such as systemic vasculitis with secondary CNS involvement, or vasculitis of the CNS due to infection (eg, varicella zoster virus-induced vasculitis or angioinvasive fungal infection, as can occur with aspergillosis).

  • Granulomatous: PACNS can only be definitively diagnosed by histopathological examination of CNS blood vessels. The characteristic features of granulomatous vasculitis—the most common histopathological subtype (58%) of PACNS1—are mononuclear infiltrate with granulomas and multinucleated cells.

In our experience, there are many patients who have been incorrectly diagnosed with PACNS and inappropriately treated with aggressive immunosuppression. Diagnostic errors usually arise from a misunderstanding of the three principles above: angiography showed a vasculopathy that was inappropriately assumed to be a vasculitis, a secondary cause of suspected vasculitis was not adequately sought or there was no pathological diagnosis. These types of errors can lead to misdiagnosis and incorrect treatment.

Here, we review the clinical, radiological and laboratory features of PACNS and its differential diagnosis. We highlight the sensitivity and specificity of investigations for PACNS so that practitioners can avoid common diagnostic pitfalls.

Clinical presentation

The clinical features of PACNS are highly variable and non-specific. The most common symptoms are headache (50–60%) and altered cognition (50–70%);2 ,3 clearly, the absence of these symptoms does not exclude the diagnosis. Most patients also have focal features,3 such as hemiparesis, ataxia, cranial neuropathy, visual symptoms, extrapyramidal signs, seizures and/or myelopathy, though none of these individual features occur in more than 25–30% of patients.2 ,4 The diversity of symptoms reflects the variable regions of the CNS affected in any individual patient. These themes of variability and lack of specificity apply not only to clinical features but also to neuroimaging and laboratory features (see below). Patients with PACNS usually do not have systemic symptoms, such as fever, chills and weight loss; their presence in a patient with CNS vasculitis suggests a systemic disorder with secondary vasculitis instead.

PACNS usually arises in the fourth or fifth decade of life, with half of cases beginning between ages 37–59 years.2 ,3 It can affect men and women though most series show a male predominance.2 Symptom onset is typically subacute or chronic: about 40% of patients present after more than 3 months of the first symptoms, while another 40% present within 4 weeks of symptoms.3 Scolding and colleagues described three broad clinical phenotypes: (1) a relapsing-remitting course (‘atypical multiple sclerosis’ phenotype) distinguished from multiple sclerosis by atypical features such as seizures, severe headaches or acute events; (2) an intracranial mass lesion; (3) an acute or subacute encephalopathy.5 As with the individual clinical features, these categories are non-specific and occur in many neurological disorders.

Because PACNS has such varied and non-specific symptoms, it might appear among the differential diagnoses of nearly any unexplained neurological disorder (alongside sarcoidosis), raising the risk of overdiagnosis. On the other hand, PACNS is so rare (estimated incidence of 2.4 cases per million person-years) that it may not be considered in an atypical clinical or radiological presentation.2


Structural neuroimaging

PACNS is generally first considered after cranial imaging shows one or more unexplained abnormalities. The MR scan of brain is abnormal in over 90% of cases, so a normal MRI makes PACNS exceedingly unlikely.6 CT scan of head may be abnormal but is generally not sensitive for PACNS lesions.4 The types of MRI abnormalities can be as highly variable as the clinical manifestations, and can include: T2 or fluid attenuated inversion recovery (FLAIR) sequence hyperintensities involving the cortex, subcortical white matter and/or deep grey matter; acute, subacute and or chronic infarctions in large and/or small vessel distributions;7 mass lesions that can resemble tumour; or enhancement of parenchymal lesions and the leptomeninges (figure 1). These findings may be interpreted by radiologists as suggesting cardioembolic stroke(s), small-vessel vascular disease (leucoaraiosis), demyelinating disease, malignancy, infection, or, if there are extensive white matter changes, cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL). Indeed all of these are diagnostic considerations in a patient with subacute onset of generalised or focal neurological deficits with an abnormal MRI, and most of them are far more common than PACNS.

Figure 1

Primary angiitis of the central nervous system (PACNS) has varied imaging characteristics, as shown here in two patients, each found to have PACNS on brain biopsy. MRI FLAIR sequences of an isolated focal left frontal lesion (A) and bi-insular and right periventricular lesions (B).

Vascular neuroimaging

Finding a stroke or stroke-like lesions on neuroimaging generally leads to vascular imaging with CT or MR angiography. The greatest diagnostic pitfalls in the misdiagnosis—or missed diagnosis—of PACNS arise from misinterpretation of vascular abnormalities (or lack thereof). In Calabrese and Mallek's 1988 original proposed diagnostic criteria for PACNS, angiography that showed alternating stenosis and ectasia—the so-called ‘string of beads’—was accepted as surrogate evidence for PACNS (figure 2).4 However, the ‘beaded’ appearance of vessels on cerebral angiography (CT angiography, MR angiography or digital subtraction angiography) signifies only that there is a vasculopathy and does not indicate a vasculitis. Vascular irregularities on angiography are therefore non-specific.

Figure 2

Anterior-posterior projection of cerebral angiogram with alternating areas of stenosis and dilatation (arrow) showing a ‘beading pattern’ (biopsy-proven primary angiitis of the central nervous system (PACNS)).

In studies with biopsy-confirmed PACNS, the specificity of the ‘string of beads’ appearance on cerebral angiography is only around 30%.2 ,8 This signifies that the large majority of patients with a vasculopathy discovered on angiography do not have PACNS. The differential diagnosis of vessel luminal irregularities on cranial angiography is broad and includes: reversible cerebral vasoconstrictive syndrome, intracranial atherosclerosis, infective vasculitis (eg, due to varicella zoster virus), antiphospholipid syndrome, fibromuscular dysplasia, intravascular lymphoma or imaging artefact in reconstructed angiographic images. The finding of vasculopathy on MR angiography, CT angiography or digital subtraction angiography requires a rigorous evaluation for an underlying cause.

Not only is it a diagnostic pitfall to believe that PACNS can be diagnosed by vessel appearance on angiography, but it is also a potential pitfall to believe that PACNS can be excluded with normal angiography. Series of pathologically proven PACNS show that only 25–43% of cases have vessel changes on angiography.2 ,3 Angiography has insufficient sensitivity to detect vessel changes in PACNS because the disease often exclusively affects small vessels,7 and angiography has poor resolution for vessels smaller than 500 μm in diameter. A normal angiogram therefore by no means excludes PACNS; in fact, most patients with PACNS have a normal angiogram.

Laboratory studies

If angiography shows vessel abnormalities, cerebrospinal fluid (CSF) analysis may help to distinguish vasculopathy from vasculitis. CSF is abnormal in about 90% of cases with histologically proven vasculitis, usually with elevated protein,3 mild lymphocytic pleocytosis2 ,9 and normal glucose. As is the theme in diagnosis and misdiagnosis of PACNS, these findings are non-specific and can occur in many CNS inflammatory conditions. However, a completely normal CSF examination makes PACNS unlikely. Inflammatory CSF in the setting of angiographic abnormalities of cerebral blood vessels raises the possibility of a CNS vasculitis, but patients need additional testing to look for possible secondary causes of CNS vasculitis, such as varicella zoster virus, syphilis, polyarteritis nodosa, granulomatosis with polyangiitis or others (see table 1). Laboratory studies that may help to identify an underlying systemic vasculitis include acute phase reactants (erythrocyte sedimentation rate/serum C reactive protein), antinuclear antibodies, antineutrophil cytoplasmic antibodies, antiphospholipid antibodies, rheumatoid factor, cryoglobulins and hepatitis B/C serologies. Whereas the sedimentation rate is typically elevated in systemic causes of secondary CNS vasculitis, it is normal in over 90% of PACNS cases.2

Table 1

Differential diagnosis of intracranial vasculopathy

Pathological diagnosis

As is clear from the above discussion, there is no combination of clinical, radiological and/or laboratory features that can definitively confirm the diagnosis of PACNS. If the search for secondary causes of CNS vasculitis is unrevealing, the next step should be a brain biopsy. A brain biopsy is required for the definitive diagnosis of PACNS but, when performed for possible PACNS, it commonly reveals an alternative diagnosis. For example, in a series of 30 patients with clinical diagnosis of PACNS undergoing brain biopsy for confirmation, only nine had a final pathological diagnosis of PACNS, while the others had a variety of diagnoses including hypertensive changes (most common), amyloid angiopathy, infarcts, vascular malformation, arboviral infection and others.10 Another series evaluating all brain biopsies performed for consideration of CNS vasculitis within an 8-year window at a major US academic hospital tells a cautionary tale: none of the 14 patients with clinical and angiographic features thought to be diagnostic for PACNS had evidence of vasculitis on biopsy.11 Eight biopsies in this series were non-diagnostic, but alternative diagnoses were reached in six other cases, including CADASIL, astrocytosis, Alzheimer's disease, degenerative changes, infarction and non-inflammatory vasculopathy.

Many clinicians fear the risks of brain biopsy. However, the risks are generally low in experienced centres. In a series of 61 patients undergoing biopsy for consideration of PACNS, none had permanent neurological injury and only three had small haemorrhage at the biopsy site without permanent neurological deficit.12 In a large series examining the safety of stereotactic brain biopsy in over 7000 patients, the mortality rate was less than 1%; the morbidity rate was 3.5% but very few had permanent disability.13 Patients with severe cardiovascular disease, however, may be at increased risk of adverse effects from anaesthesia.

Unfortunately, even brain biopsy does not have perfect sensitivity for the diagnosis of PACNS. In comparing biopsy with gold standard of final clinical diagnosis (combination of history, laboratory, imaging and treatment effect), the sensitivity of brain biopsy is 53–74%.3 ,8 Histopathology is highly specific, however, and can also identify alternative and potentially treatable diagnoses. Overall, about 75% of patients with presumed vasculitis receive a specific diagnosis after biopsy.12 To increase the yield for the diagnosis of potential PACNS, biopsy should include areas of imaging abnormality and contain the leptomeninges and the cortex (sensitivity increases to over 80% when cortex and leptomeninges are examined).3

The histopathological diagnosis of vasculitis generally requires transmural inflammation with injury to the vessel wall. There is usually an angiocentric inflammatory infiltrate. PACNS is a pathologically heterogeneous disease, the most common subtypes being granulomatous (most common, 58%), purely lymphocytic and acute necrotising vasculitis.14 In the granulomatous subtype in older patients, pathology can show extensive beta amyloid deposition in vessel walls with associated inflammation—called amyloid-β related angiitis.15 Some of these patients also have parenchymal amyloid-β deposition while others do not. How these subtypes translate to different clinical courses and prognosis remains unknown due to the rarity of the disease.

Illustrative case

A 61-year-old man developed bilateral paraparesis, visual blurring in the right eye, word-finding difficulty and a mild headache over the course of 5 months. MR scan of brain showed multiple small T2/FLAIR hyperintensities distributed bilaterally with some areas showing diffusion restriction consistent with infarction, and others showed contrast enhancement (figure 3A, B). CT angiography showed areas of possible cerebral arterial stenosis with distal dilatation (figure 3C). Evaluation for the causes of stroke was normal, including echocardiogram, cardiac monitoring and serum lipid studies. CSF analysis showed inflammation with 18 white cells per µL (≤5) (80% lymphocytes), elevated protein 1.08 g/L (0.15–0.45) and normal glucose. Based on his clinical, laboratory and neuroimaging data, he was given a diagnosis of primary CNS vasculitis and started treatment with cyclophosphamide. His cognition continued to worsen over subsequent months and further imaging showed progression of the subcortical lesions. A brain biopsy was performed: this showed intravascular B cell lymphoma. Although this condition has a poor prognosis, the patient missed several months of potentially disease course-altering, life-extending treatment due to premature diagnostic closure without tissue diagnosis.

Figure 3

Bifrontal lesions on FLAIR sequence (A). Some of the lesions have areas of diffusion restriction (B) on diffusion-weighted imaging sequence. CT angiography shows a beading appearance (C, arrow) in middle cerebral artery.


Because PACNS is so rare, there are no randomised trials of its treatment. Experts recommend an initial regimen of corticosteroids (such as intravenous methylprednisolone pulse or oral prednisone at 1 mg/kg/day) and cyclophosphamide, given either as a daily oral dose (2 mg/kg/day) or by monthly intravenous pulse dose (starting, eg, at 750 mg/m2).16 About 80% of patients respond favourably to this regimen; a lack of response should prompt reconsideration of the diagnosis.2 After about 6 months, patients typically transition to an oral corticosteroid-sparing immunosuppressive agent such as azathioprine, mycophenolate mofetil or methotrexate. Because of the risk of serious adverse effects including infection, cystitis, bladder cancer and gonadal toxicity, cyclophosphamide should generally be prescribed in collaboration with a rheumatologist or other physicians experienced in using this medication and monitoring for adverse effects. Cyclophosphamide is teratogenic and contraindicated during pregnancy.

For antineutrophil cytoplasmic antibody-associated vasculitis (such as granulomatosis with polyangiitis and microscopic polyangiitis), rituximab has recently been shown to be equivalent to cyclophosphamide for induction of remission.17 We do not yet know if the same holds true for PACNS.


In summary, PACNS is a rare disorder with highly variable and non-specific clinical and neuroimaging features. Its rare and protean nature can lead to missed diagnosis and to misdiagnosis, each of which can lead to inappropriate treatment. The clinical, radiological and laboratory features are not sufficiently specific to make a firm diagnosis but can suggest alternative diagnoses. In patients with CNS vasculopathy (or other unexplained CNS lesions) with inflammatory CSF, clinicians should pursue an exhaustive search for secondary causes of vasculopathy or vasculitis. Patients with no identifiable cause should be referred for brain biopsy; the diagnostic yield of brain biopsy is greater—and the risks to the patient lower—than generally believed, when performed by skilled practitioners. PACNS often responds to treatment, as do many conditions in the differential diagnosis of the disorder. As Dr Leonard Calabrese, one of the foremost experts this field, once commented, “Both science and art are often needed in securing the diagnosis of various forms of systemic vasculitis but I believe never more so then when applied to vasculitis of the central nervous system”.18

Key points

  • Primary angiitis of the central nervous system (PACNS) most commonly presents with headache and altered cognition; however, no clinical sign or symptom is specific.

  • MRI is sensitive but not specific for PACNS; a normal MRI makes PACNS exceedingly unlikely, but there is no pathognomonic MRI signature.

  • Vascular luminal irregularity on angiography is neither sensitive nor specific for PACNS and should not be used to diagnose or exclude the disease.

  • CSF generally shows pleocytosis and/or elevated protein; again, these findings are sensitive but not specific for PACNS.

  • Brain biopsy is the only way to diagnose PACNS definitively; however, in most cases where PACNS is considered, the biopsy finds an alternative diagnosis.


The authors thank Dr Ivana Vodopivec for help with providing representative images.


View Abstract


  • Contributors Both authors contributed to the conception, drafting and revision of the manuscript, and both authors approved the final version for submission.

  • Competing interests ALB receives royalties from Clinical Pathophysiology Made Ridiculously Simple (Medmaster) and The Improvising Mind (Oxford University Press).

  • Provenance and peer review Commissioned; externally peer reviewed. This paper was reviewed by Fady Joseph, Cardiff, UK.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Editors' commentary
    Phillip E M Smith Geraint N Fuller

Other content recommended for you