Abstract
Primary thunderclap headache (TCH) is sometimes associated with cerebral vasospasm. However, the role of vasospasm in relation to the development of reversible or irreversible posterior leukoencephalopathy among patients with primary TCH has never been fully addressed. This paper includes a report on a 51-year-old woman with primary TCH complicated with posterior leukoencephalopathy and a literature review of 16 further patients with the same illness. Their magnetic resonance or conventional angiographic findings were clearly described. Our review found that all these 17 patients showed evidence of cerebral vasospasm. Eleven (65%) of them developed permanent ischaemic infarctions, almost exclusively located at the watershed zones. We suggest that the presence of vasospasm might be requisite for posterior leukoencephalopathy as well as for permanent infarctions in these patients. Therefore, searching for any clue of vasospasm is mandatory in treatment of patients with primary TCH. Absence of an accompanying vasospasm might predict a good outcome.
Introduction
Thunderclap headache (TCH), first coined by Day and Raskin (1), describes an unanticipated, severe headache coming to peaks in seconds. Diagnostic criteria for TCH proposed by Slivka and Philbrook (2) and extended by Dodick et al. (3) assigned TCH into three subtypes: (i) TCH without neurological signs or symptoms; (ii) TCH with neurological signs or symptoms; and (iii) TCH associated with intracranial disorders, such as subarachnoid haemorrhage, cerebral venous sinus thrombosis, or pituitary apoplexy, etc. Those without associated intracranial disorders may have angiographic features of normal cerebral vasculature or reversible segmental cerebral vasoconstriction (RSCV); some of the latter had been denoted as ‘Call–Fleming syndrome’ (4). The new edition of the International Classification of Headache Disorders, 2nd edition (ICHD-II) first coded this primary headache and proposed diagnostic criteria (Table 1) (5).
Diagnostic criteria of primary thunderclap headache in International Classification of Headache Disorders, 2nd edition
A newly described disease entity, reversible posterior leukoencephalopathy syndrome (RPLS) (6), which is characterized by reversible posterior-predominant white and grey matter lesions on brain magnetic resonance imaging (MRI), can sometimes present as abrupt and severe headache (TCH-like), nausea, vomiting, visual disturbance, focal neurological deficits and seizures. The clinical manifestation and imaging findings of RPLS bear strong resemblance to Call–Fleming syndrome, suggesting a spectrum of overlapping syndromes. In severe cases of these two syndromes, permanent ischaemic insults can occur. The nomenclature of RPLS has been disputed since the lesions need not always be reversible, with restriction persisting in the posterior hemisphere and involving only white matter (7). Some other terms have been coined, but have not been widely accepted (7).
In this report, we present a 51-year-old woman experiencing recurrent primary TCH with vasospasm, who developed posterior leukoencephalopathy and permanent ischaemic infarctions later. Herein, we raise this question: is the existence of vasospasm mandatory for posterior leukoencephalopathy or, furthermore, ischaemic infarctions in patients with primary TCH? Therefore, after the case report, we undertook a literature review (to date) by searching MEDLINE and BIOSIS databases to provide clues, i.e. the angiographic findings in patients with primary TCH complicated with either reversible or irreversible posterior leukoencephalopathy.
Case report
A 51-year-old otherwise healthy woman was hospitalized for ‘the worst headache that she had ever experienced’. She had been an infrequent migraineur since adolescence, but her migraines bore quite different characteristics from the new headache, for which she had called for emergency management five times within 1 week prior to this admission. The first headache occurred while she was straining during a bowel movement (day 1). The headache intensity reached its maximum in seconds. It was pulsatile and located at the occipital regions and accompanied by nausea and vomiting. Brain computed tomography (CT) showed negative results at a local hospital, where the headache resolved 3 h later after some parenteral analgesics. She reported complete freedom from pain between headache intervals.
The results of physical and neurological examination were normal on admission. A brain CT taken within 6 h after the latest headache (day 8) was still negative. Spinal tap performed immediately after the brain CT disclosed an opening pressure of 110 mmH2O, white cells 4/μl, red cells 32/μl, protein 42 mg/dl and glucose 61 mg/dl. Culture of the cerebrospinal fluid was non-yielding. Brain MRI and MR angiography (MRA) performed on day 10 disclosed a segmental narrowing at the right middle cerebral artery (MCA). There were no demonstrable intraparenchymal lesions or cerebral aneurysms. The patient had high blood pressure (220/110 mmHg) during the headache attacks, but she denied a prior history of hypertension or use of any sympathomimetic or serotonergic agents; and in addition, blood pressure measured during her headache-free intervals was normal. Surveys for paroxysmal hypertension were all normal, including 24-h urine and plasma catecholamines, serum cortisol and renin levels.
After admission, the patient's TCH occurred daily and almost always in the morning and tended to last for hours despite medications. From day 16, her headache did not recur without specific management. However, she began to complain of blurred vision, mild weakness in both left limbs and unsteadiness starting the second day (day 17). Brain MRI performed on day 20 showed multiple hyperintense lesions bilaterally at the parieto-occipital regions, bilaterally on the cerebellar hemispheres, right centrum semiovale and left aspect of the splenium of corpus callosum on T2-weighted images (Fig. 1). High signal change on the diffusion-weighted images (DWI) and low apparent diffusion coefficient (ADC) values suggested cytotoxic oedema in the corresponding lesions. The calcarine and paramedian regions of the occipital lobes were spared, making the diagnosis of simultaneous infarction of bilateral posterior cerebral artery territories less likely (6). A diagnosis of posterior leukoencephalopathy was considered. Three-dimensional time-of-flight (3D-TOF) MRA demonstrated multiple segmental narrowing of the middle and posterior cerebral arteries, suggesting the possibility of vasospasm, vasculitis or atherosclerosis of the involved vessels. Surveys for vasculitis and hypercoagulability were all negative. Electrocardiography and a transthoracic echocardiogram also showed normal findings.
Brain magnetic resonance imaging and magnetic resonance angiography (MRA) on day 20 (a–d) showed hyperintense lesions at bilateral occipital regions on T2-weighted image (a); high signal change on diffusion-weighted image (b) and low apparent diffusion coefficient (ADC) values on ADC mapping (c) suggested cytotoxic oedema in the corresponding regions. The three-dimensional time-of-flight MRA showed multiple segmental narrowing (arrows) in bilateral middle and posterior cerebral arteries (d). Brain MRA performed on day 30 (e) demonstrated reversal of the previous narrowing of intracranial vessels.
Oral nimodipine and aspirin were administered beginning on day 20 based on the suspicion of vasospasm. Discrete but low-intensity headaches with vomiting still occurred on days 21, 23 and 27. Transcranial Doppler sonography (TCD) on day 26, digital subtracted angiography on day 27 and MRA on day 30 showed resolution of the segmental narrowing of the involved vessels (Fig. 1e).The rapid reversibility favoured the diagnosis of vasospasm and made vasculitis and atherosclerosis unlikely. No other significant vascular abnormalities such as aneurysm or arterial dissection were noted. The patient complained of no more headaches beginning on day 28. A repeated brain MRI performed on day 30 showed partial resolution of the previous diffusion-restriction lesions. MRI 5 months later showed encephalomalacia bilaterally over the occipital regions and the right corona radiata. Sequential MRA studies (5 months and 1 year later) revealed normal intracranial vasculature. She reported no further attacks in the following 4 years. Her latest neurological examination 3.5 years after the disease onset demonstrated residual right lower quadrantanopsia and minimal left hemiparesis.
Literature review
To identify studies which contained angiographic findings in cases with posterior leukoencephalopathy accompanying primary TCH for this review, a database search of the literature was conducted using MEDLINE (1966 to January 2005) and BIOSIS (1985 to March 2005) using English language and human study limits. The keywords ‘thunderclap headache.tw.’, ‘posterior leukoencephalopathy.tw.’, ‘reversible segmental cerebral vasoconstriction.tw.’ and ‘reversible cerebral segmental vasoconstriction.tw.’ were employed as subject titles for the article searches. In order to find other potential cases not obtained through the above search, a further search of the reference lists of all of the identified articles was also performed.
The inclusion criteria of articles for this review were as follows. Studies should contain case reports and have pertinent information including: (i) ‘brain CT or MRI findings of posterior leukoencephalopathy’ or ‘clinical evidence of focal neurological deficits’, (ii) results of conventional angiography, CT angiography (CTA), MRA or TCD, and (iii) patients’ age ≥18 years. Studies were excluded if: (i) they contained no related information of a case description, (ii) the causes of TCH were ‘evident’, i.e. ‘not’ primary TCH, or (iii) the content of case reports was irrelevant to the study subject. We defined infarction as ‘the persistent lesions on the last MRI’ or ‘restricted diffusion on DWI/ADC mapping’ or ‘clinical evidence of residual neurological deficits’ in this review. The localization of infarct territories was determined by the description within the original papers or direct interpretation of the published figures according to present knowledge of arterial territories (8). Border zone or watershed infarcts were defined as the result of a critically reduced cerebral perfusion pressure in far-downstream brain arteries that led to a critically reduced cerebral blood flow and oxygen supply in certain vulnerable brain areas (9, 10). Supratentorial watershed infarcts could further be divided into anterior watershed infarcts, posterior watershed infarcts, and internal watershed infarcts, according to the monograph of Momjian-Mayor (10).
A single reviewer (S-P.C.) made the review of all searched articles. The full texts of all articles were evaluated in conformity with the inclusion and exclusion criteria.
Results
The MEDLINE and BIOSIS search with foregoing keywords and limitations found 148 articles in total (MEDLINE 122, BIOSIS 92, overlapping 66). Fifty-seven articles were excluded because they contained no case description or the content was unrelated to the review subject; 62 articles were excluded because they had no pertinent information on cranial vasculature evaluation (seven of them had only MR venography); and another six articles were not eligible because their cases were all children. Seven articles were put aside because they did not mention the headache characteristics or imaging evidence of posterior leukoencephalopathy. Another eight articles containing cases with TCH (or severe headaches), posterior leukoencephalopathy and neuroradiological assessments of cranial vasculature were not taken into the final analysis because the aetiologies of their headaches or posterior leukoencephalopathy were evident, such as hypertensive encephalopathy, eclampsia, postpartum vasculitis, use of chemotherapeutic or immunosuppressive agents, etc. All the possible aetiologies of RPLS with cerebral vasospasm identified through the present review (2, 4, 11–26) were tabulated (Table 2). Only eight eligible articles (2, 4, 21–24, 26) were identified after censoring with the inclusion and exclusion criteria. Two more articles (20, 25) were obtained from the reference lists of the identified articles. In total, there were totally 10 eligible articles for review.
Possible aetiologies of RPLS with vasospasm through the current literature review
RPLS, Reversible posterior leukoencephalopathy syndrome.
In the end, 17 cases (2, 4, 20–26) including ours with primary recurrent TCH were identified for analysis (Table 3). Imaging evidence for posterior leukoencephalopathy was unequivocally identified in 11 cases, while in another three cases (2, 4) the imaging findings were not certain for posterior leukoencephalopathy because of atypical locations (4), the neuroimaging was not presented, or the authors failed to provide detailed descriptions about the lesions (2) (see Table 3). The other three cases without imaging evidence of posterior leukoencephalopathy were also included because they showed focal neurological deficits (2, 4).
Identified cases ∗ with recurrent primary thunderclap headache, posterior leukoencephalopathy and assessments for cranial vasculatures
Original articles might contain more cases, but the patients listed here were only eligible ones.
Possible association with vasoconstrictive drugs.
Our case.
Controversial image findings. aRight basis pontis. bSeveral white matter lesions and a small signal abnormality in the right thalamus. cSmall areas of high signal abnormalities in the periventricular areas bilaterally.
Right lateral cerebellum, border zone of anterior inferior cerebellar artery and posterior inferior cerebellar artery.
s, focal deficit presenting as seizure; AWS, anterior watershed; CT-PL, CT evidence of posterior leukoencephalopathy; FD, focal neurological deficits; HTN, hypertension; IWS, internal watershed; MRI-I, MR evidence of infarction; MRI-PL, MR evidence of posterior leukoencephalopathy; N, absence; NA, no available prudential information; PWS, posterior watershed; RD, residual neurological deficits; TCH, thunderclap headaches; Y, presence.
Most of reported patients were women (76%), with a mean age of 45.4 years (range 19–63). Using the definition of systolic blood pressure >140 mmHg or diastolic blood pressure >90 mmHg as hypertension, less than one-third had hypertension (29%). All 17 cases showed diffuse segmental vasospasm either by conventional angiography (2, 4, 20, 21, 23–25) (
Discussion
Based on the literature review and our case report, we found, among patients with primary TCH, either reversible posterior leukoencephalopathies (
In our review, we found most patients with primary TCH were middle-aged women. All patients, including ours, experienced recurrent high-intensity headaches of abrupt onset and afterwards developed posterior leukoencephalopathy or permanent ischaemic infarctions. Their headache profile fulfilled the diagnostic criteria of primary TCH proposed by the ICHD-II (Table 1) (5), except for a more prolonged course of recurrent headaches (>10 days) in some (4, 21–23). In fact, the patient reported by Strum et al. (21) was very similar to our own, i.e. bi-occipital infarctions occurred after 2-week recurrent primary TCH. Of note, both our and Strum et al.'s patients had demonstrated vasospasm several days prior to the appearance of subsequent posterior leukoencephalopathy and infarctions. This sequence might suggest a link between vasospasm and posterior leukoencephalopathy or infarctions.
The exact causes of posterior leukoencephalopathies are still unknown and might be multifactorial. Some of them are proposed to be similar to the pathogenesis of hypertensive encephalopathies, the central mechanism of which is presumably ‘autonomic breakthrough’ (27). The endothelial control of vascular tone is overwhelmed; which progresses into a vicious cycle of homeostatic failure, leading to progressive increase of vascular resistance, which, in turn, further worsens endothelial dysfunction. Increased vascular permeability, attributed to endothelial dysfunction, may contribute partly to characteristic imaging findings of RPLS, i.e. vasogenic oedema. In contrast, the progressive increase of vascular resistance phenomenologically presents itself as diffuse vasospasm, and when severe, may lead to irreversible ischaemic change, especially over the watershed regions. It is unknown if ‘autonomic breakthrough’ also plays a role in the pathogenesis of posterior leukoencephalopathy among patients with primary TCH. In fact, only one-third of our reviewed patients had hypertension. Since DWI and ADC mapping were not routinely performed in our reviewed patients, we could not determine whether the posterior lesions were predominantly due to vasogenic or cytotoxic oedema or first due to vasogenic oedema, then cytotoxic oedema. The posterior lesions in our patient were compatible with cytotoxic oedema based on the results of DWI and ADC mapping. Therefore, at least in our patient, diffuse and high vascular resistance (vasospasm) of a protracted course might result in haemodynamic compromise, and, in turn, cause watershed infarctions (9, 10). Actually, posterior lesions with low ADC value suggesting cytotoxic oedema has been reported in several cases with RPLS (28–30). These lesions might have a greater propensity of evolving to permanent infarctions, although the underlying pathomechanism has not been elucidated completely (28).
There are several limitations to this study. First, we did not perform extensive investigations to exclude cardioembolic stroke in our patient. The presence of patent foramen ovale or atrial septal aneurysm could not be completely ruled out. However, the impressive clinical presentation and characteristic imaging findings of this patient precluded the diagnosis of cardioembolic stroke. Second, some articles containing pertinent information might have been missed using current databases and search limitations; for example, non-English written articles have been excluded from the present review. Third, publication bias should also be considered, i.e. patients without a finding of vasospasm might not be reported in the literature. This might overestimate the true percentage of vasospasm in patients with TCH in this review. Fourth, though bearing high sensitivity, specificity, and accuracy (up to 92%, 98% and 96%, respectively) in evaluating vasospasm (31), MRA has not been validated as conventional angiography as a gold standard. Nevertheless, the reversibility of the segmental narrowing of cerebral vasculature made other diagnoses less likely. Finally, and most importantly, based on a single observation and a retrospective review of the literature of 16 additional patients, our study could not provide solid and scientific evidence to delineate the causal relationship of posterior leukoencephalopathy and vasospasm in patients with primary TCH. All the aforementioned mechanisms were only putative.
Despite these limitations, the data reviewed here do have clinical implications. We suggest that angiographic studies should be performed to look for not only aneurysms but also for vasospasm among patients with primary TCH. If there is evidence of cerebral vasospasm, the chances of posterior leukoencephalopathy or permanent ischaemic events are markedly increased. In contrast, the absence of vasospasm may indicate a good prognosis. In addition, more data are mandatory to investigate whether nimodipine can prevent vascular complications in primary TCH patients with vasospasm.
Footnotes
Acknowledgements
This study was supported in part by a grant from Veterans General Hospital-Taipei (V94-295).