Hemorrhagic Manifestations of Reversible Cerebral Vasoconstriction Syndrome
Frequency, Features, and Risk Factors
Background and Purpose—Reversible cerebral vasoconstriction syndrome (RCVS), characterized by severe headaches and reversible constriction of cerebral arteries, may be associated with ischemic and hemorrhagic strokes. The aim of this study was to describe the frequency, patterns, and risk factors of intracranial hemorrhages in RCVS.
Methods—We analyzed prospective data on 89 consecutive patients with RCVS, of which 8 were postpartum and 46 used vasoactive substances. Standard bivariate and multivariate statistical tests were applied to compare patients with and without hemorrhage.
Results—Thirty patients (34%), of which 5 were postpartum and 12 used vasoactive substances, developed at least 1 type of intracranial hemorrhage, including cortical subarachnoid (n=27), intracerebral (n=11), and subdural hemorrhage (n=2). Patients with hemorrhage had an older age (46.6 versus 41.6 years, P=0.049) and were more frequently females (90% versus 51%, P=0.0017) or were migrainers (43% versus 19%, P=0.022) than those without hemorrhage. Multivariate testing identified 2 independent risk factors of hemorrhage in RCVS: female gender (OR, 4.05; 95% CI, 1.46 to 11.2) and migraine (OR, 2.34; 95% CI, 1.06 to 5.18). Patients with hemorrhage had a greater risk of persistent focal deficits (30% versus 2%, P=0.0002), cerebral infarction (13% versus 2%, P=0.039), posterior reversible encephalopathy syndrome (17% versus 3%, P=0.041) at the acute stage, and of inability to resume normal activities at 6 months (27% versus 0%, P<0.0001).
Conclusion—In RCVS, women and migrainers seem to be at higher risk of intracranial hemorrhage. Overall, intracranial hemorrhages are frequent in RCVS and are associated with a more severe clinical spectrum.
- female gender
- intracerebral hemorrhage
- reversible cerebral vasoconstriction syndrome
- subarachnoid hemorrhage
- subdural hematoma
Reversible cerebral vasoconstriction syndrome (RCVS) is characterized by severe headaches, often thunderclap headaches, with or without focal deficits and seizures, and a multifocal constriction of cerebral arteries, which resolve spontaneously within 3 months.1 This syndrome has a female preponderance and a mean age of onset of approximately 42 years.1–3 RCVS presumably involves a transient disturbance in the control of cerebral arterial tone leading to segmental multifocal constrictions with alternating segmental dilatations giving the characteristic “sausage on a string” angiographic aspect.1 Although RCVS may be spontaneous, more than half of the cases occur in special circumstances such as postpartum or after an exposure to sympathomimetic or serotoninergic substances.1,3–6 Once considered rare,7 RCVS has been increasingly reported during the last 10 years,1 presumably due to the routine use of CT or MR angiography in patients presenting with acute headache2 and/or stroke and to the greater awareness of this syndrome.1–3 However, RCVS remains a poorly understood syndrome and is viewed completely differently by headache specialists who consider it as a painful but benign syndrome and by stroke specialists who view it mostly as a rare cause of stroke. Indeed, stroke is the most severe manifestation1–3,7 and may lead to permanent sequelae and even death.5,6 Recent reports and case series have suggested that intracranial hemorrhages may be frequent in RCVS3,5,8–12 and harbor different patterns, including cortical subarachnoid hemorrhages (cSAHs), intracerebral hemorrhages (ICHs), and subdural hemorrhages (SDHs), but detailed description of these hemorrhagic manifestations and of their risk factors is lacking so far.
We analyzed the frequency, clinical and radiological features, and risk factors of intracranial hemorrhages in a large prospective cohort of 89 patients with RCVS.
Subjects and Methods
In a predefined prospective protocol,3 we recruited consecutive patients with a diagnosis of RCVS based on the following criteria: (1) unusual, recent, severe headaches of progressive or sudden onset with or without focal neurological deficit and/or seizures; (2) imaging evidence of cerebral vasoconstriction with at least 2 narrowings per artery on 2 different arteries, assessed by MR angiography (MRA), CT angiography, and/or transfemoral angiography (TfA); and (3) disappearance of arterial abnormalities assessed by a control angiography in <3 months.
From January 2004 to January 2008, 89 patients (61 women) fulfilling all 3 study criteria were recruited including 30 patients through our stroke unit admitting approximately 1000 patients per year and 59 through our emergency headache clinic receiving approximately 7500 outpatients per year coming from the great Paris area (10 million inhabitants). Patients with RCVS represented approximately 0.26% (95% CI, 0.21% to 0.31%) of our overall recruitment. Approximately 115 patients per year present to the clinic with thunderclap headaches as their chief complaint and all undergo an extensive workup, including cerebral and vascular imaging.
All patients orally agreed to participate in a descriptive follow-up study. We followed the French legislation stipulating that observational studies do not require formal review by the institutional ethics committee but that any investigation or study requires that patients give their informed consent. This consent may be oral, like here. Detailed medical history, clinical, radiological, and biological data were collected as previously described.3 When present, migraine was diagnosed according to the International Classification of Headache Disorders.13 The first 67 patients were the basis of a previous publication.3 The 89 patients included 8 postpartum women and 46 vasoactive substance users. Clinical symptoms included severe headaches (89), transient focal deficits (14), persistent focal deficits (10), seizures (4), and blood pressure (BP) surges (29). Acute investigations included cerebral CT scan (86); 1.5-T MRI with diffusion-weighted imaging, fluid-attenuated inversion recovery, T1, and gradient-echo weighted images (89); 2-dimensional time-of-flight cerebral MRA (88); cervical and transcranial Doppler (86); TfA (57); complete blood workup (89), urine analysis for drugs (62); and cerebrospinal fluid analysis (CSF; 78). All 89 patients had an angiogram showing “string and beads.” CSF was abnormal in 47 patients; 22 had an elevated white blood cell count (mean, 11/mm); 30 elevated red blood cell count (mean, 2377/mm); and 30 elevated protein levels (mean, 59 mg/L).
Symptomatic analgesic treatment (mostly paracetamol) was used in all patients without standard protocol. Nimodipine was used in 82 patients with intravenous infusion (1 to 2 mg/h adapted to BP) followed by oral administration in 15 patients and direct oral administration in 67 patients (30 to 60 mg every 4 hours, adapted to BP). Duration of treatment ranged from 4 to 8 weeks.
Clinical follow-up visits were performed within 3 to 6 weeks after hospital discharge and then at 3 and 6 months after headache onset. Follow-up angiogram (73 MRA, 16 TfA) demonstrated the reversibility of arterial narrowings in all 89 patients.
Results are expressed as mean and SD or counts and percent. Marginal association between patient characteristics and hemorrhage was assessed by a Student t test for quantitative variables and a Fisher exact test for qualitative variables. Multiple logistic regression was used to determine variables independently associated with each outcome. Variables associated with hemorrhage at a 0.15 level were considered in the multiple models. A backward stepwise procedure was used for variable selection with s probability value cutoff at 0.05. First-order interactions between selected variables were then tested. Validity of the logistic regression model was checked using the goodness-of-fit test.14 Internal validation was performed by bootstrapping,15,16 with random generation of 200 samples from the original data (drawn with replacement) on which the whole variable selection procedure was applied and the performance measures of the derived model calculated. Discriminative ability of the models was evaluated by the C index (identical to the area under the receiver operating characteristics curve)17 and calibration by the calibration slope.18 The models estimated in each bootstrap sample were then evaluated in the original sample, and the differences between the performance on the bootstrap sample and the original sample were taken as a measure of overoptimism. The final model performances were corrected by this overoptimism and regression coefficients estimates multiplied by the calibration slope. All tests were 2-sided. Analyses were carried out using R statistical software (The R Foundation for Statistical Computing, Vienna, Austria).
Frequency and Subtypes of Hemorrhagic Complications
Among 89 patients with RCVS, 30 had an intracranial hemorrhage (34%) and 5 had a cerebral infarction (6%). Stroke occurred in 14 of the 89 patients (16%). One of the 59 patients without hemorrhage had an infarction. Thirteen of the 30 patients with hemorrhagic RCVS had a stroke: ICH with infarction (2), ICH without infarction (9), and infarction with cSAH (2).
All 30 patients with a hemorrhage underwent cerebral CT and MRI scans with MRA, and 27 also had TfA. By definition, all 30 patients had angiographically proven multifocal vasoconstriction, which resolved within 12 weeks as assessed by MRA (28) and/or TfA (8). Angiograms, MRI, and blood tests showed no other cause of hemorrhage.
Three varieties of hemorrhages were observed: cSAH in 27 patients (30%), ICH in 11 (12%), and SDH in 2 (2%). Nine patients had overlapping bleeding locations (Table 1). MRI T2* showed no microbleeds.
Among the 27 cSAHs (Figure 1), 24 were restricted to a few hemispheric sulci, unilaterally (10) or bilaterally (14). One SAH was visible in some sulci of the right cerebellar lobe. The last 2 SAH were more diffuse, lying over both hemispheres and in the perimesencephalic cisterns. SAH was visible on both CT and MRI in 14 patients and only on MRI in 13 patients.
Ten patients had a single ICH and 1 had bilateral frontal hematomas. All 12 ICHs were visible on both CT and MRI. Sites of bleeding included various lobar regions in 6 patients for 7 ICH, basal ganglia in 4 patients, and thalamus in 1 (Figure 2).
The 2 SDHs were acute and associated with other bleeding types (Figure 2).
Demographic Characteristics and Risk Factors in Patients With Hemorrhagic RCVS
Table 2 presents the main demographic characteristics, vascular risk factors, and potential precipitating factors for RCVS in patients with and without hemorrhage. None had a history of recent head trauma. Bivariate analysis suggested that hemorrhagic manifestations were more frequent in older patients, females, and migrainers. This association entirely relied on migraine without aura, but only very few patients had migraine with aura (Table 1). Multivariable analysis showed only 2 independent factors associated with a higher risk of bleeding in RCVS: female gender (OR, 4.05; 95% CI, 1.46 to 11.2) and a history of migraine (OR, 2.34; 95% CI, 1.06 to 5.18). No significant interaction was found between gender and migraine (P=0.20).
Clinical Features and Associated Lesions in Patients With Hemorrhagic RCVS
Acute very severe headache was the presenting symptom in all 30 patients, of whom 24 had the typical RCVS pattern of multiple recurrent thunderclap headaches (Tables 1 and 3⇓). Eighteen subjects reported headache triggers, the most frequent being bending down followed by sexual activity, physical exercise, coughing, sneezing, defecation, urination, and bathing.
Two patients had focal seizures with secondary generalization in 1 case. Six patients had a total of 14 episodes of transient focal deficits, consistent with transient ischemic attacks in 4 patients (11 episodes) and with positive aura-like visual symptoms in 2 patients. Nine patients had a persistent focal deficit lasting >24 hours, which revealed an ICH in 7 and a cerebral infarction in 2. Among the 11 patients with an ICH, 4 had isolated headaches, 4 a single sudden headache concomitant to a focal deficit, and 3 initially had isolated headaches and developed a few days later a persistent focal deficit.
Among the 9 subjects who had a systolic BP ≥160 mm Hg and/or a diastolic BP ≥90 mm Hg during the acute stage of RCVS, 7 patients had BP surges only during the peak of thunderclap headaches, and 2 had permanently elevated BP.
Five patients (17%) had MRI fluid-attenuated inversion recovery sequence hyperintensities consistent with a posterior reversible encephalopathy syndrome (PRES; Table 1). Four patients (13%), including 1 with PRES, developed a cerebral infarction. Four patients, including 2 with PRES and 1 with infarction, had a cervical artery dissection (unilateral vertebral artery dissection in 3 subjects and 4 vessels dissection in 1).
Patients with hemorrhage had a significantly higher frequency of persistent focal deficits (30% versus 2%), cerebral infarction (13% versus 2%), and PRES (17% versus 3%) than patients without hemorrhage.
Temporal Profile of Vasoconstriction and Intracranial Hemorrhage
Table 4 presents the average delay of hemorrhages, ischemic events, and diagnosis of vasoconstriction after headache onset. Vasoconstriction was diagnosed concomitantly to hemorrhage in 15 patients and after hemorrhage in 14 patients. In 4 of these 14 patients, angiogram was available only a few days later. In the remaining 10 patients, early MRA (6), TfA (1), or both MRA and TfA (3) were normal, and vasoconstriction was visible on a second angiogram (5 MRA, 5 TfA). Vasoconstriction was diagnosed before hemorrhage in 1 patient with normal CT and CSF at Day 1, vasoconstriction on TfA at Day 3, and cSAH on MRI at Day 4.
Hemorrhages, PRES, and seizures were mainly diagnosed within the first week, whereas ischemic events occurred mainly during the second week, often after the cessation of thunderclap headaches (Figure 3).
Five patients (17%) with hemorrhage had a normal initial brain imaging. One woman had both normal MRI and CSF the day after her first thunderclap headache, but she subsequently developed cSAH diagnosed on a repeat MRI at Day 2, after 9 additional thunderclap headaches (Figure 1C). Another woman had both normal CT and CSF at Day 2 after 3 thunderclap headaches but later developed bilateral cortical and perimesencephalic SAH diagnosed by CT and MRI at Day 5, after a fourth thunderclap headache. Three patients had initial normal CT scans performed respectively at Days 0, 1, and 3 after headache onset but later developed an ICH diagnosed by a repeat imaging at Day 3 for the first 2 patients and at Day 8 for the third patient (Figure 2D).
Clinical Outcome in Patients With Hemorrhagic RCVS
At 3 months of follow-up, the median modified Rankin Scale score was 0 and the mean modified Rankin Scale was 1, 2 patients remained severely disabled (modified Rankin Scale 3 and 4), and 12 patients (40%) were unable to return to prior professional activities. There were no deaths.
At 6 months of follow-up, 8 patients with hemorrhagic RCVS were still unable to resume work because of persistent deficits (5) or marked asthenia (3), which was significantly higher than in patients without hemorrhage (27% versus 0%, P<0.0001).
Hemorrhagic manifestations were present in 34% of our patients with RCVS. Stroke, with both infarction and ICH, has long been known as the most severe manifestation of RCVS.1,5–7 However, its estimated incidence ranges from 6% to 7% in prospective studies recruiting acute headache patients2 to 54% in a retrospective series including in-hospital patients,1 with ICH in 0% to 25% and infarctions in 6% to 31%.1–3,7,19 This discrepancy presumably reflects recruitment biases and lack of standardized diagnostic workup. In our cohort, recruitment bias is likely to be less marked because of our dual recruitment based on both a stroke unit and an emergency headache center. Stroke occurred in 16% of our 89 patients and ICH was twice as frequent as infarction (12% versus 6%). Moreover, our data further establish that cSAH (30%) is the most frequent type of hemorrhagic manifestation of RCVS,3,4,8,10,11 whereas SDH seems less frequent.10,12
Despite the increasing number of published patients with hemorrhagic RCVS,3–5,7–12 some are still reluctant to attribute an intracranial hemorrhage to RCVS and rather consider the arterial narrowings as a consequence of hemorrhage. This seems very unlikely in the present series for several reasons: first, the absence of other causes of hemorrhage at an extensive and repeated workup; second, the unusual prevailing pattern of bleeding, that is, mild localized cSAH, rarely seen in other conditions; third, the clinical presentation with recurrent thunderclap headaches in >80% of patients; fourth, the angiographic pattern of widespread segmental vasoconstriction and vasodilatation involving brain arteries well remote from the site of bleeding; and fifth, the similarity of angiographic changes in RCVS with and without intracranial hemorrhage.
The exact mechanisms of bleeding in RCVS remain unknown, but our data further illustrate the dynamic nature of this syndrome with headache and hemorrhages usually preceding ischemic complications.3 Our finding that up to 17% of patients with hemorrhagic RCVS initially presented with isolated headaches and normal brain imaging, and only subsequently developed cSAH, ICH, and/or SDH after a few days of recurrent severe headaches, suggests that the abnormal vascular process starts before hemorrhage. The co-occurrence in 17% of our patients of PRES, a transient vasogenic cerebral edema related to small vessel dysfunction with acute disruption of the blood–brain barrier,4 and cSAH suggests that the abnormal process initially affects very small cortical arteries. We previously hypothesized that arterial abnormalities first involve small distal arteries and then progress toward medium- and large-sized vessels, which could explain the high rate of normal early angiograms (up to 33%) in RCVS.3 Serial MRA and transcranial Doppler studies in a large cohort of RCVS cases showed that vasoconstriction affecting first segments of large arteries was maximal 18 to 22 days after headache onset, similar to the timing of headache resolution.19,20 Moreover, marked vasoconstriction could persist weeks after headache resolution, suggesting that vasoconstriction is not directly causing headache.19 We now suggest that segmental vasodilatation could play an important role at the initial stage of RCVS, triggering thunderclap headaches by abrupt stretching of vessel walls and causing hemorrhages by small vessel rupture or reperfusion injuries, whereas small vessel segmental constriction remains asymptomatic (no or rare small vessel infarction). In a second stage, vasoconstriction of second and first segments of major cerebral arteries becomes the major problem causing mainly watershed infarction.1–3,6
Risk factors for hemorrhagic RCVS have not been investigated before. We found that female gender and a history of migraine were 2 independent predictors of intracranial hemorrhages during RCVS. By contrast, a history of arterial hypertension and BP surges during RCVS were not associated with the risk of hemorrhage. The female preponderance of RCVS has long been described,1–3,7 and we have previously shown that RCVS was more severe in females than in males.3 Identifying migraine, especially without aura, as an independent risk factor favoring hemorrhages is somewhat surprising given the fact that prior studies did not single out migraine as a risk factor for RCVS overall.2,3 Larger studies are needed to re-evaluate migraine as a risk factor for RCVS, for hemorrhages during RCVS, or for both.
Our study, conducted in a single tertiary care center, is not free of bias. Detailed clinical and imaging information was prospectively collected, but timing of interviews and investigations did not follow a standardized protocol. Observers of the neuroimaging data were not blinded to the clinical status of the patients or to the hypotheses of the study. Despite these constraints, our study shows that hemorrhagic RCVS is not benign; patients seem to have a higher risk of subsequent infarction than those without hemorrhage, emphasizing the need for close monitoring. Furthermore, one fourth are still unable to work at 6 months of follow-up. There is no established treatment of RCVS. Discontinuation of vasoactive medications in secondary forms seems logical.1,3 We used nimodipine in most cases without any proof that it improved the course of the syndrome more than simple bed rest.
The proportion of RCVS among intracranial hemorrhages in the general population is as yet unknown but may be easily underestimated. Prospective studies are needed to address this issue using a standardized workup in patients with cSAH, ICH, or SDH without other obvious cause and in patients with thunderclap headaches with normal findings on initial brain imaging and spinal tap.
Our results indicate that intracranial hemorrhages affect one third of all RCVS cases and are far more frequent than ischemic events. RCVS should be considered as a differential diagnosis in patients with any type of spontaneous intracranial hemorrhage and especially with localized cortical SAH. The diagnosis of RCVS may be difficult when initial brain and vascular imaging are normal, requiring repeated investigations. Women and migrainers may be at particularly high risk for hemorrhagic manifestations. Studies of cerebral blood flow at the acute stage and of cerebrovascular reactivity at a distance of RCVS could help to understand the mechanisms underlying this poorly understood syndrome.
- Received November 7, 2009.
- Revision received June 17, 2010.
- Accepted June 17, 2010.
Chen SP, Fuh JL, Lirng JF, Chang FC, Wang SJ. Recurrent primary thunderclap headache and benign CNS angiopathy: spectra of the same disorder? Neurology. 2006; 67: 2164–2169.
Ducros A, Boukobza M, Porcher R, Sarov M, Valade D, Bousser MG. The clinical and radiological spectrum of reversible cerebral vasoconstriction syndrome. A prospective series of 67 patients. Brain. 2007; 130: 3091–3101.
Moustafa RR, Allen CM, Baron JC. Call-Fleming syndrome associated with subarachnoid haemorrhage: three new cases. J Neurol Neurosurg Psychiatry. 2008; 79: 602–605.
Wong SH, Dougan C, Chatterjee K, Fletcher NA, White RP. Recurrent thunderclap headaches and multilobar intracerebral haemorrhages: two cases of reversible cerebral vasoconstriction syndrome (RCVS). Cephalalgia. 2009; 29: 791–795.
Headache Classification Subcommittee. The international classification of headache disorders. Cephalalgia. 2004; 24: 1–160.
Steyerberg EW, Eijkemans MJ, Harrell FE Jr, Habbema JD. Prognostic modeling with logistic regression analysis: in search of a sensible strategy in small data sets. Med Decis Making. 2001; 21: 45–56.