Cerebrospinal fluid findings in reversible cerebral vasoconstriction syndrome: a way to differentiate from cerebral vasculitis?

Summary

Reversible cerebral vasoconstriction syndrome (RCVS) is characterized by an acute onset of severe headache and multi‐focal segmental vasoconstriction of cerebral arteries resolving within 12 weeks. Diagnostic criteria include normal or near‐normal findings in cerebrospinal fluid (CSF) analysis, especially leucocyte levels < 10/mm³. Distinguishing RCVS from primary angiitis of the central nervous system (PACNS) is essential to avoid unnecessary and sometimes unfavourable immunosuppressive treatment. We reviewed retrospectively the clinical and diagnostic data of 10 RCVS patients who presented in our neurological department from 1 January 2013 to February 2017. The main purpose was to verify whether CSF leucocyte counts < 10/mm³ serve to discriminate RCVS from PACNS. Five of six patients who underwent lumbar puncture presented with CSF leucocyte levels ≥ 10/mm³. Two patients had a history of misinterpretation of CSF pleocytosis as cerebral vasculitis and of immunosuppressive treatment. A complete restitution of cerebral vasoconstriction was evident in all. No patient had further cerebral strokes or bleedings without immunosuppressive treatment over more than 12 weeks. Despite the established diagnostic criteria, RCVS can manifest with CSF leucocyte levels > 10/mm³. Careful anamnesis and the response of ‘vasculitis‐like angiography’ to nimodipine given as a test during angiography and as oral medication are key to differentiate RCVS from cerebral vasculitis.

Introduction

Reversible cerebral vasoconstriction syndrome (RCVS) is a cerebral angiopathy whose main features are acute onset of severe ‘thunderclap’ headache and multi‐focal segmental cerebral artery vasoconstriction. There is a high risk for stroke and cortical subarachnoid haemorrhage. The diagnostic criteria include reversibility of angiographic abnormalities within 12 weeks and normal or near‐normal findings on cerebrospinal fluid (CSF) analysis, especially normal glucose levels, protein levels < 80mg/dl and leucocyte counts < 10/mm³ 1 (Tables 1 and 2).

Table 1.

Critical elements for the diagnosis of RCVS adapted from Calabrese et al. 2007

Clinical features

Severe, acute ‘thunderclap’ headaches, with or without additional neurological signs due to (watershed) strokes, subarachnoidal haemorrhages

Associated conditions (see Table 2)

Radiological features

Conventional angiography or indirect CTA or MRA with ‘vasculitis‐like’ multi‐focal segmental cerebral artery vasoconstriction

Watershed (strokes)

Subarachnoidal haemorrhages (cortical)

No evidence for aneurysmal subarachnoid haemorrhage

Reversibility of angiographic abnormalities within 12 weeks after onset

Others

Normal or near‐normal cerebrospinal fluid analysis (protein level < 80 mg%, leucocytes < 10 mm3, normal glucose level)

CTA = computed tomographic angiography; MRA = magnetic resonance angiography.

Table 2.

Conditions associated with reversible cerebral vasoconstriction

Pregnancy and postpartum period
Early puerperium, late pregnancy, eclampsia, pre‐eclampsia and delayed postpartum eclampsia
Medication, drugs and blood products
Phenylpropanolamine, pseudoephedrine, ergotamine tartrate, methergine, nicotine patches, bromocryptine, lisuride, selective serotonin reuptake inhibitors, sumatriptan, isometheptine, cocaine, ecstasy, amphetamine derivatives, marijuana, lysergic acid diethylamide, tacrolimus (FK‐506), atovaquone, fingolimod, isoflavones, irreversible, non‐selective monoamine oxidase (MAO) inhibitors cyclophosphamide, erythropoetin, intravenous immune globulin and red blood cell transfusions
Other triggering factors
Hypomagnemia, hypercalcaemia, porphyria, catecholamine‐secreting tumours such as pheochromocytoma and bronchial carcinoid tumour, head trauma, spinal subdural haematoma, post‐carotid endarterectomy and neurosurgical procedures
Associated conditions
No identifiable precipitating factor
Associated with headache disorders, such as migraine, primary thunderclap headache, benign exertional headache, benign sexual headache and primary cough headache
Associated with posterior reversible encephalopathy syndrome cervical artery dissections, unruptured saccular cerebral aneurysma, cerebral arterial dysplasia

Traditionally, CSF analysis is thought to be essential to distinguish RCVS from primary angiitis of the central nervous system (PACNS) to avoid initiation of unnecessary immunosuppressive treatment and inappropriate and ineffective treatment with nimodipine 2. The differential diagnosis is often challenging, sometimes leading to brain biopsies in the workup to achieve diagnostic certainty 3.

In this context, we analysed retrospectively whether an elevated CSF cell count of ≥ 10/mm³ according to the diagnostic criteria confidently allows exclusion of RCVS.

Material and methods

Patients were examined at the Department of Neurology of the Alfried Krupp‐Hospital in Essen (Germany). We reviewed retrospectively the case records of 115 patients from 1 January 2013 to February 2017 who were encoded as ‘Other cerebrovascular diseases’, according to ICD‐10. Twelve patients matched the criteria of acute thunderclap headache and/or multi‐focal reversible abnormalities on catheter, computerized tomography (CT) or magnetic resonance imaging (MRI) angiography and transcranial ultrasound without the evidence for cerebral vasculitis. One patient was excluded because her headache was interpreted as associated with a viral infection. One patient was excluded because the diagnostic workup revealed a persistent stenosis of her cerebral arteries and diagnosis was later changed in fibromuscular dysplasia. The remaining 10 patients were diagnosed as RCVS according to the established diagnostic criteria 1. Clinical and diagnostic data were obtained by retrospective chart review. A standardized data collection form was completed for all cases. All patients had a neurological examination at symptom onset and approximately 12 weeks later. The study was approved by the Ethics Committee of the University of Duisburg‐Essen, Germany.

Results

Ten patients were diagnosed as RCVS (six females, four males, mean age = 43·2 years). All showed multi‐focal vasoconstriction (in duplex ultrasound, magnetic resonance angiography or conventional angiography). Eight of the 10 patients reported severe headache; only one patient was classified as headache intensity ≤ 7 on the NRS. One patient presented with reduced vigilance, so anamnestic data could not be evaluated. Seven of 10 patients were considered to be exposed to RCVS provoking triggers such as postpartal status (two patients) and vasoactive medication or drugs such as fluoxetine, tiotropium bromide or consumption of cannabis. In one patient RCVS occurred in the context of medication with fingolimod, which has been incriminated as a provoking factor elsewhere 4. Strokes were detected in nine of 10 patients; one patient showed cortical subarachnoid haemorrhage in two locations.

Six of 10 patients underwent lumbar puncture in the acute phase of symptom presentation. In five of these six patients the CSF leucocyte counts were ≥  10/mm³ (10/mm³–29/mm³). Oligoclonal bands were positive in two patients; one of these had a comorbid multiple sclerosis 4. Cytological analysis revealed a lymphomonocytic pleocytosis in four cases and granulocytes and lymphocytes in one patient. One patient with leucocyte counts of 23/mm³ and bloody CSF had a protein level of > 1000 mg/l; the remaining protein levels were all normal (Table 3).

Table 3.

Summarized clinical findings and results of examinations

Age	Sex	Associated migraine	Associated pregnancy	Comorbidities	Acute ischaemic changes in MRI	Subarachnoid haemorrhage	Typical angio‐graphic/ MRI‐ angiographic findings (acute)	Proof of reversibility	CSF analysis (acute)
22	M	Yes	–	Tension headache	R cerebellar R PCA	No	MRI	Normal DSA and MRA*	Leucocytes 10/µl Protein level 473 mg/l OCB positive
30	F	No	Three months postpartum	Highly active multiple sclerosis, thalassaemia	L PCA	No	Both	Normal MRA*	Leucocytes 12/µl Protein level 470 mg/l
39	M	No	–	None	Multiple, bilateral	No	Both	Normal DSA, MRA*	Leucocytes 18/µl Protein level 269 mg/l OCB‐negative
44	M	No	–	Hypercholesterolaemia Heterozygous FVL mutation	Multiple, bilateral	No	MRI	Normal DSA, MRA*	–
74	F	No	–	Sigma diverticulosis	R thalamus L PCA	No	Angiographic	Normal DSA, MRA*	–
23	M	Yes	–	Nasal septum surgery	L PCA	No	Duplex	Normal DSA, MRA*	Leucocytes 23/µl Protein level 1180 mg/l OCB not performed
53	F	No	–	Arterial hypertension hyperlipidaemia Obesity	L MCA L ACA R PCA	No	Both	Normal duplex*	Leucocytes 3/µl Protein level 385 mg/l OCB‐negative
59	F	No	–	Hashimoto's thyroiditis	No	No	Both	Reversibility in angiography 20 min after intra‐arterial Nimodipin; normal MRA*	Leucocytes 29/µl Protein level 604 mg/l OCB‐negative
52	F	No	–	Arterial hypertension, COPD, vocal cord carcinoma	–	Yes, cortical in two locations	Angiographic	Reversibility in angiography 20 min after intra‐arterial Nimodipin; normal DSA, MRA*	–
36	F	Unknown	Yes, 1 week postpartum	Arterial hypertension Cerebral salt wasting syndrome	Yes, R MCA (multiple)	No	Angiographic and duplex	Reversibility in angiography 20 min after intra‐arterial Nimodipin;improvement of duplex after 2 days	–

ACA = anterior cerebral artery; DSA = digital subtraction angiography; L = left; MCA = middle cerebral artery; MRA = magnetic resonance angiography; OCB = oligoclonal bands; PCA = posterior cerebral artery; R = right; COPD = chronic obstructive pulmonary disease; MCA = middle cerebral artery; PCA = posterior cerebral artery; FVL = factor V Leiden. *Angiographic or magnetic resonance angiographic control within 12 weeks.

In three patients, intracranial vasospasms improved by intravenous infusion of nimodipine during angiography (Fig. 1). All patients showed a complete remission of vasoconstriction under oral nimodipine treatment evident in angiography, MR angiography or duplex sonography within 12 weeks.

Figure 1.

Intracranial vasospasms (a) in the right middle cerebral artery (MCA) resolving by intravenous nimodipine infusion over 20 min (b, c).

Lumbar puncture was repeated to control the CSF pleocytosis in three patients and revealed normal leucocyte levels within 3–6 months without immunosuppressive treatment. In one patient who initially received glucocorticoids over 3 months, another CSF analysis was performed 5 years later and still revealed normal results.

Discussion

Distinguishing reversible cerebral vasoconstriction syndrome from PACNS is essential in young patients presenting with headaches and/or neurological deficits when angiography reveals stenosis of the intracranial arteries 5 (Table 4).

Table 4.

Similarities and differences between RCVS and primary angiitis of the CNS (PACNS) adapted from Calabrese et al. 2007

Sign	RCVS	PACNS
Headaches	Acute severe ‘thunderclap‘ headache	Dull, chronic
Stroke localization	Watershed	Variable
Intracranial haemorrhages	Often	Very rare
Cortical subarachnoidal haemorrhages	Often	Very rare
Oedema	Often	Absent
Conventional angiopathy	100% ‘sausage on a string’ sign	Only 25% irregular, narrowing, dilation

RCVS = reversible cerebral vasoconstriction syndrome; CNS = central nervous system; PACNS = primary angiitis of the central nervous system.

Catheter angiography itself often fails to distinguish these two entities reliably during the acute course of the disease, while CSF analysis is thought to be of high diagnostic importance, especially as established diagnostic criteria include a threshold of < 10/mm³ CSF leucocytes 1. However, in some cases the course of disease is typical for RCVS despite higher leucocyte counts. We therefore decided to analyse if a CSF cell count of ≥ 10/mm³ consequently allows exclusion of RCVS or if the diagnosis can also be posed in patients with higher CSF cell counts.

We reviewed a cohort of 10 patients fulfilling the diagnostic criteria of RCVS with complete relief of symptoms and angiographic pathology within 12 weeks. Concerning the rate of triggered vasculopathies, our cohort is concordant with approximately two‐thirds of patients in former literature 6. Moreover, demographic data, clinical presentation and angiographic signs were comparable with the literature. However, in contrast to other reports, migraine was not a frequent comorbidity in our patients 7.

In five of six patients with the diagnosis of RCVS who underwent lumbar puncture the CSF analysis revealed leucocyte levels ≥ 10/mm³. According to the diagnostic criteria this finding should have led to exclusion of the diagnosis. Nevertheless, the diagnostic workup showed a complete restitution of cerebral artery vasoconstriction within 4 weeks to 3 months which could be excluded to appear in PACNS without immunosuppressive therapy. Other case–series examining more patients than ours also included patients with cell counts > 10/mm³, with rates of 3% or almost 10% 6, 8. However, a conclusive interpretation is lacking in the literature, as well as the discussion of whether the diagnostic criterion of CSF leucocyte counts <  10/mm³ should be maintained. Singhal et al. 6 concluded that CSF abnormalities may be associated with the co‐existence of stroke. Rates of stroke in RCVS patients range at approximately 40% 6, 8. Generally, watershed infarction and/or vasogenic oedema are thought to be classical MRI findings in RCVS. Currently, there are no data available of whether the localization of infarction, co‐existence of vasogenic oedema or the time–course, i.e. the time CSF analysis takes place, have impacts upon leucocyte counts.

This is the first communication, to our knowledge, reporting the presence of a CSF pleocytosis > 10/mm³ in the majority (83%) of patients with RCVS. Of course, data from a small patient population such as ours does not allow general conclusions to be drawn. Nevertheless, RCVS patients with CSF pleocytosis > 10/mm³ are at risk of misdiagnosis as cerebral vasculitis, leading possibly to inappropriate immunosuppressive treatment, especially in a non‐specialist setting 2. Furthermore, RCVS patients are at risk of neurological worsening if treated with glucocorticoids, which will mainly be the first‐line therapy in suspected cerebral vasculitis 9. Consequently, one could argue whether a more accurate threshold of CSF pleocytosis should be defined or if the exclusion criterion of > 10/mm³ CSF leucocytes should be refuted. Currently, there are not enough data available to answer this question. Ducros et al. 8 found up to 35/mm³ leucocyte levels in RCVS patients; hence, differentiating RCVS from PACNS by a strict leucocyte threshold may be challenging.

How can this dilemma be solved? Despite being considered as the diagnostic gold standard, brain biopsy bears the risk of complications and false‐negative results 2, 10. Enhancement on vessel wall imaging is not specific for cerebral vasculitis 11. It is crucial to consider that the character of headaches is stronger and occurs more abruptly (‘thunderclap’) in RCVS than in vasculitis 2, 10. Singhal et al. 3 suggested a bedside test based on the anamnestic data of thunderclap headache with a 100% positive predictive value for RCVS. Recently, intra‐arterial infusion of nimodipine was shown to be a helpful differential diagnostic tool in selected patients with suspected RCVS 12. This results in a normalization of both the diameter of the main trunks of the cerebral vessels and the calibre of the peripheral vessels quickly after intra‐arterial nimodipine in RCVS 12. According to the data presented, we recommend the intra‐arterial nimodipine test or oral nimodipine treatment and watchful waiting over 12 weeks as a safe option in patients with thunderclap headaches, and angiography with suspected vasculitis and CSF pleocytosis 2. We also recommend not to rule out RCVS in patients with CSF leucocyte levels ≥ 10/mm³ and to keep in mind that this could be a more frequent pitfall than presumed previously 2, 5.

Clinical implications

• Headache is an important presenting symptom of RCVS and cerebral vasculitis

• CSF can fail to differentiate between these two entities

Disclosure

There were no conflicts of interest.

Author contributions

L. K. performed data acquisition and analysis and drafted the manuscript. P. B. revised the manuscript for intellectual content and helped with the interpretation of the data. P. A. revised the manuscript for intellectual content and helped with the interpretation of the data. H.‐P. H. revised the manuscript for intellectual content and helped with the interpretation of the data. M. K. conceived the study, performed data acquisition and analysis and drafted the manuscript.