Abstract

Introduction
Thunderclap headache is a term first used by Day & Raskin to describe hyperacute onset of severe headache (1). These headaches may herald subarachnoid haemorrhage, cerebral venous thrombosis or pituitary apoplexy, or may occur without underlying intracranial pathology. Diagnostic criteria for thunderclap headache proposed by Slivka & Philbrook (2) and extended by Dodick et al. (3) suggest that the headache may recur for up to 7 days and that in the absence of an underlying intracranial disorder neurological symptoms and signs are transient and residual deficits minimal. We report a case of recurrent thunderclap headache associated with angiographically documented reversible segmental intra-cerebral vasospasm occurring without underlying intracranial pathology. Severe intermittent headaches recurred for 14 days prior to the development of persisting cortical blindness, dysphasia and visuospatial deficits accompanied by bilateral watershed cerebral infarction on neuroimaging. This case illustrates that thunderclap headache associated with reversible cerebral vasospasm may not be as benign a condition as previously suspected and can result in disabling stroke.
Case report
RC is a 58-year-old woman who presented with loss of vision over 2 days on a background of 2 weeks of explosive episodic headaches. Her past medical history included hypertension and hyperlipidaemia. She was a lifelong smoker and imbibed less than 20 g of alcohol daily. She had no personal or family history of migraine or stroke. Her medications were atorvastatin and enalapril. On day 1 of her illness hyperacute onset of severe bioccipital headache occurred 1 h after a thyroid uptake scan using intravenous technetium pertechnetate (subsequent investigation of a non-functioning thyroid nodule proved it to be benign). She was investigated in the Emergency Department with a computed tomography (CT) scan which was normal and cerebrovascular fluid (CSF) examination showing no cells, normal protein and glucose and no xanthochromia. She was discharged with a diagnosis of migraine.
Paroxysms of severe occipital headache continued to occur on average twice daily at approximately 12-h intervals and lasting 1–2 h, and in between headaches she was completely well. On day 6 she was admitted to hospital and despite treatment with regular intravenous 6-h chlorpromazine and oral amitriptyline, intermittent morphine was required when the headaches recurred. Magnetic resonance imaging (MRI) and magnetic resonance angiography (MRA) on day 12 were unremarkable at the time, although in retrospect a mild degree of diffuse arterial spasm of the intracranial vessels was probably present. On day 11 prednisolone 50 mg daily was added. She was discharged home on day 14 as the headaches had improved. She described blurred vision on day 15 and on day 16 woke unable to see. Examination was normal except for complete loss of vision. MRI brain showed bilateral occipital hyperintensities on T2 images consistent with bioccipital infarction and MRA showed decreased flow in distal middle and posterior cerebral arteries suggestive of vasospasm or vasculitis (Fig. 1). Full blood count, erythrocyte sedimentation rate, electrolytes, liver function tests, anti-nuclear and anti-neutrophil cytoplasmic antibodies and a hypercoagulability screen were all normal or negative. Treatment with intravenous heparin and high-dose dexamethasone was commenced.

Magnetic resonance angiography showing diffuse segmental intracerebral vasospasm with improvement from day 16 (top) to day 21 (bottom).
Over the next 2 days her condition worsened with development of disorientation, perseveration, dysphasia and visuospatial difficulties. On day 19 repeat CSF was normal and digital subtraction angiography revealed widespread narrowing and beading of small vessels throughout both cerebral hemispheres. Intravenous nimodipine was added to the high-dose steroids. A brain biopsy was planned but refused by the patient's family. Vision and orientation improved over subsequent days and a repeat MRA on day 21 showed considerable improvement in vessel calibre (Fig. 1). The clinical course fluctuated over the next few weeks, with improvement in language and frontal lobe function. Vision improved steadily but prominent visual agnosia and visuospatial problems persisted. Treatment was with intravenous nimodipine for 3 days, then orally for 3 weeks. Steroids were weaned over 5 weeks total and heparin was given for 7 days. The patient was discharged to live with her daughter after a further 3 months intensive rehabilitation. A repeat MRI on day 79 showed watershed infarction between MCA and PCA territories extending into MCA territories bilaterally (Fig. 2) and MRA was entirely normal.

Magnetic resonance imaging (MRI) brain scan performed on day 79 showing bilateral chronic watershed territory infarction.
Discussion
Our patient suffered bilateral watershed strokes due to reversible cerebral vasospasm resulting in cortical blindness and significant language, spatial and executive functioning problems. We believe the underlying diagnosis is reversible cerebral segmental vasospasm presenting as recurrent thunderclap headache. The headaches were extremely severe and the onset on each occasion was explosive. Migrainous infarction is an uncommon cause of stroke and cases with evidence of vasospasm are found in the literature (4), but our patient had no previous or family history of migraine and the headaches were self-limited over a 2-week period. Vasculitis may be difficult to exclude in comparable cases as it can be associated with similar angiographic appearances (3). Patients who continue to deteriorate may require brain biopsy to aid differentiation. In this case against the diagnosis of vasculitis is the lack of aneurysm formation on angiography; the rapid improvement in vessel calibre between days 19 and 21; the watershed pattern of infarction rather than multiple infarcts with a deep white matter predominance (5); and the lack of systemic signs or inflammatory markers. Leptomeningitis can produce similar appearances but two normal CSF examinations exclude this. Subarachnoid haemorrhage is a common precipitant for vaso-spasm (6) but was excluded with neuroimaging and CSF examination. Other causes of reversible cerebral vasoconstriction include sympathomimetics (7), eclampsia (8), phaeochromocytoma (9), intravenous immunoglobulin in Guillian–Barre syndrome (10), closed head injury (11), porphyric encephalopathy (12), and after surgical manipulation (13); none of which is implicated in this case. Our case most resembles the four idiopathic cases of reversible cerebral segmental vasoconstriction described by Call et al. (14). These patients all had an illness with no obvious precipitant characterized by severe headache, fluctuating neurological signs, and vasospasm around the circle of Willis that subsequently resolved.
The onset of headache 1 h after a thyroid uptake scan raises the possibility of a reaction to technetium pertechnotate or an associated carrier compound, although no such occurrence with intravenous technetium has been described. Of interest is the dissociation between the severity of headache and the degree of vasospasm seen on angiography. The headache was improving prior to the onset of visual loss and did not return following this. The improvement in headache is unlikely to be due to resolving ischaemia as the angiographic appearance of intracranial vasospasm worsened significantly at this time. This indicates that headache severity is not a reliable indicator of degree of intracranial vasospasm and therefore monitoring with serial MRA is required.
Prednisone at a dose of 50 mg daily is often advocated as a treatment for recurrent or cluster type headaches, but it did not protect against further vasospasm and neurological damage in this case, perhaps indicating that higher doses are required. The optimum treatment for malignant vasospasm in this setting is unknown. Intravenous nimodipine has been shown to be useful in improving neurological outcome when cerebral vasospasm occurs following subarachnoid haemorrhage (15). We used a similar protocol. The improvement in neurological status began once this treatment was initiated and the vasospasm on MRA 2 days later had significantly resolved. Whether this represents efficacy of treatment or natural disease course is uncertain. Further studies to determine efficacy of nimodipine in vasospasm unrelated to subarachnoid haemorrhage are required.
In this case there was a prodrome of 2 weeks of recurrent explosive headache prior to the development of persisting neurological deficits. We conclude that recurrent thunderclap headache is not necessarily a benign entity and can be associated with disabling stroke.
