Abstract
We consecutively recruited 21 patients (all women, mean 54 ± 8 years) with bath-related thunderclap headache (BRTH). Thirteen of them were in menopause, two had just ceased hormonal therapy, and one was at 3 months postpartum. Bathing was the initial trigger for thunderclap headaches in nine patients (43%). Many patients (n = 15, 71%) had other non-bath-related attacks. Most patients (n = 18, 86%) reported that the headache occurred immediately when water was sprayed over their body, with warm water (52%) as the most common. During the disease course [mean 14 days (6-34)], the mean number of BRTH was 5.1 ± 3.6 attacks. Nineteen patients (90%) changed bathing habits to prevent attacks. Thirteen patients (62%) had magnetic resonance angiography vasoconstrictions, and two of them (15%) developed reversible posterior encephalopathy. None of the patients without vasoconstrictions had this complication. Nimodipine was effective in stopping further attacks in 84% (16/19) treated patients. No relapse was reported at a mean follow-up of 30 months. BRTH occurred exclusively in women and predominantly in middle age. Deficiency or fluctuation of female sex hormones may play a role. About 60% patients showed cerebral vasospasms, fulfilling the diagnosis of reversible cerebral vasoconstriction syndrome and indicating a risk of posterior encephalopathy.
Keywords
Introduction
Negoro et al. (1) first reported ‘benign hot bath-related headache’ in three middle-aged Japanese women in whom simply pouring hot water over themselves or soaking in a hot bath triggered a split-second onset of severe headache. Despite the explosive presentation, the headache was considered benign because it subsided spontaneously after a period of 2 weeks to 3 months (1). In 2003, we first related this headache disorder to primary thunderclap headache, which refers to a severe and explosive headache attack that reaches maximal intensity within a minute (2). We suggest using ‘bathing headache’ instead because one of our four patients developed intracranial arterial vasospasm and ischaemic complications (i.e. not always benign) and this headache syndrome was not always induced by ‘hot’ bath. In addition, we found nimodipine effective in preventing painful attacks (2–4). After our publication, seven more cases were reported, including six Asians (5) and one European (6), all women, which downplayed the ethnic factor in this headache disorder.
Thunderclap headache was once synonymous with symptomatic headaches (7–10), until recent studies showed that thunderclap headache can be primary, without identifiable pathology except for the occasional finding of reversible segmental cerebral vasoconstriction (11, 12). The International Classification of Headache Disorders, 2nd edn (ICHD-2) (13), first adopted ‘primary thunderclap headache (code 4.6)’ as a primary headache disorder. On the other hand, thunderclap headache with evidence of intracranial vasoconstrictions is treated as a secondary headache disorder and coded as ‘headache attributed to benign (or reversible) angiopathy of the central nervous system (CNS) (code 6.7.3)’. Recently, reversible cerebral vasoconstriction syndrome (RCVS) has been proposed as a unifying term which describes thunderclap headache associated with reversible vasoconstriction due to different aetiologies, including benign CNS angiopathy (8). As such, bath-related thunderclap headache (BRTH) can be diagnosed as primary thunderclap headache or RCVS based on the presence or absence of cerebral arterial vasospasm.
Since previous studies on BRTH were all case reports, this unique headache disorder has never been systematically described. With that in mind, over several years, we collected 21 patients with BRTH. In this study we aimed to characterize this rare but potentially stroke-prone headache disorder.
Patients and methods
We recruited consecutive patients with BRTH diagnosed at our headache clinic at Taipei-Veterans General Hospital from October 2000 to February 2007. The headache clinic has been operating since 1997. All patients presenting at this headache clinic fill out a detailed headache intake form, have their medical and headache history taken, and complete clinical investigations and neurological examinations. They are asked to keep a headache diary to provide a guide for diagnosis and treatment.
In this study, BRTH was diagnosed if a patient's headache fulfilled the criteria of primary thunderclap headache or benign CNS angiopathy based on the ICHD-2 criteria except for attack duration (Table 1); moreover, bathing was related to at least two attacks of thunderclap headache, i.e. multiple related attacks. The demographics and headache profile, especially in relation to bathing, were obtained by chart review. The results of computed tomography (CT) of brain were recorded if available. All patients underwent brain magnetic resonance imaging (MRI), magnetic resonance angiography (MRA) and MR venography (4). Spinal tap with cerebrospinal fluid (CSF) analysis was performed if patients agreed. Proper laboratory tests were conducted to exclude thyrotoxicosis and pheochromocytoma. Except for those reporting spontaneous headache resolution before presenting to our clinic, all patients were given oral nimodipine 30–60 mg every 4 h. When oral nimodipine was not effective or when MRA revealed marked cerebral vasospasm, i.v. nimodipine 0.5–1 mg/h was administered via a central venous line with blood pressure monitoring every 2–4 h (3, 4). The efficacy of the nimodipine treatment was recorded as ‘effective’ if no further thunderclap headaches, bath-related or non-bath-related, occurred in the 48 h after the last dose escalation (4). All patients were followed up until no further attacks recurred after discontinuation of nimodipine. In May 2007, we called each patient to determine whether or not they had a relapse of BRTH. This study protocol was approved by the Institutional Review Board of the hospital.
Current diagnostic criteria of primary thunderclap headache and benign (or reversible) angiopathy of the CNS in the ICHD-2
CNS, central nervous system; ICHD-2, The International Classification of Headache Disorders, 2nd edn.
Results
During the study period, 26 consecutive patients with thunderclap headache with attacks related to bathing were seen among 5338 patients visiting our headache clinic. Of these, 21 (0.4%) fulfilled our diagnostic criteria of BRTH, i.e. at least two attacks were related to bathing. Five patients with only one bath-related attack were excluded because it might have been a coincidence rather than bath-related. All our patients were women with a mean age of 54 ± 8 years (32–76 years). Most patients (n = 18, 86%) were between 40 and 60 years old. Some patients have been reported elsewhere (2–4). Thirteen patients (62%) were in menopause and one was just at 3 months postpartum. Five patients were currently on hormonal therapy, two of whom had just stopped hormones for 9 days and 2 months, respectively. The incidence of BRTH was not related to any season of the year. Eight patients (38%) had a prior history of hypertension and six (29%) were recorded to have a blood pressure surge (systolic blood pressure > 160 mmHg) during the headache attacks. Neurological examinations were normal and none of our patients had neck rigidity.
Past history
Two patients (10%) had had similar BRTH in the past, one 7 years ago and the other 8 years ago. Six patients (29%) also had a history of migraine and three of episodic tension-type headache.
Laboratory and CSF studies
All laboratory studies were unrevealing. Two patients agreed to perform lumbar puncture. Their CSF was clear and colourless, and the cell counts and levels of protein and sugar were normal.
Headaches in relation to bathing
Before the headache disorder, all patients took a bath every day. Nineteen patients always took a shower (90%), whereas the other two usually took a shower, but occasionally used a tub bath. In this study, all the first BRTH occurred in relation to a standing shower. If all thunderclap headaches were included, bathing was the initial trigger in nine patients (43%), but was the only trigger in six (29%). Most patients (n = 18, 86%) reported that headaches occurred immediately when water was sprayed over their body. Two patients (10%) reported headache occurred in the middle of bathing, and only one after bathing. Warm water was the most common (n = 11, 52%) water temperature to induce headache, followed by hot water (n = 5) and cold water (n = 2). Three patients did not find any difference in different water temperatures. The most sensitive body part to trigger headaches was the chest (n = 11, 52%), followed by hair (n = 6, 29%) and face (n = 2, 10%). In order to prevent BRTH, 19 patients (90%) changed bathing habits for a mean duration of 14 ± 11 days (3–46 days). Of these, nine took a scrubbing bath, three stopped bathing, four shortened the bathing time and three only washed the inguinal areas. All of them considered that their BRTH could either be prevented or became less severe when they changed bathing habits. Fifteen patients (71%) reported other thunderclap headache-related triggers, including exertion (n = 9, 43%), straining at stools (n = 9, 43%), coughing (n = 5, 24%), anger (n = 3, 14%), sexual intercourse (n = 2, 10%) and singing (n = 1, 5%).
Headache profile
The BRTHs were most often described as explosive (n = 18, 86%) or pulsatile (n = 14, 67%). They were most often bilateral (n = 13, 62%), especially bilateral occipital regions (n = 8, 38%), followed by over the midline, such as the vertex (n = 7, 33%). The median duration of BRTH was 2 h (range 30 min to 30 h). The headaches were sometimes accompanied by nausea (n = 6, 29%), vomiting (n = 5, 24%), photophobia (n = 3, 14%) or phonophobia (n = 3, 14%). In addition to thunderclap headaches, all patients reported persistent mild to moderate background headaches. Overall, the mean number of all thunderclap headaches was 8.5 ± 4.9 (range 2–20), whereas that of BRTH was 5.1 ± 3.6 (range 2–14). BRTH accounted for an average of 68% (15–100%) of all thunderclap headaches.
CT scan of brain
Fifteen patients (58%) underwent CT of brain studies. None of them showed evidence of subarachnoid haemorrhage (SAH) or other abnormalities.
MR imaging studies
The initial MRA studies were done an average of 16 ± 16 days (range 2–74 days) after headache onset. The studies did not show any evidence of aneurysms, but showed multiple segmental arterial constrictions (MRA vasospasm) in 13 patients (62%): middle cerebral artery (MCA) (n = 12, 57%), posterior cerebral artery (n = 9, 43%) and anterior cerebral artery (n = 4, 19%). All these vasoconstrictions recovered or greatly improved at follow-up MRA studies in an average of 77 ± 47 days (range 21–160 days) after headache onset. The timing of initial MRA studies did not differ between those with and without vasospasm (16 ± 9 vs. 16 ± 24 days after headache onset, P = 0.985). Compared with those without MRA vasospasm (n = 8, 38%), i.e. primary thunderclap headache, patients with MRA vasospasm, i.e. RCVS, were more likely to be triggered by straining at stool (62% vs. 13%, P = 0.03). The other possible clinical profile variables did not differ statistically between these two groups, including age, menopause, headache location, headache pattern, accompanying symptoms, numbers of total or bath-related thunderclap headaches, disease duration and different triggers such as cough, sex, laughter, anger, singing or exertion.
The brain MRI studies did not show evidence of SAH in any patient. However, two patients in the MRA vasospasm group (15%) showed findings compatible with posterior encephalopathy [fluid-attenuated inversion recovery (FLAIR)]: high-signal lesions in right posterior temporal and bilateral occipital regions on day 6 in one patient (2) and right parieto-occipital and left occipital regions on day 14 in the other (Fig. 1). The former patient was asymptomatic, whereas the latter complained of blurred vision. These lesions disappeared in the follow-up MRI studies; however, the former patient developed a small, asymptomatic right cerebellar infarct (2). Two patients had a history of hypertension, but only one of them was found to have blood pressure surge during the disease course.
A 41-year-old woman with bath-related thunderclap headache showed reversible posterior encephalopathy 14 days after headache onset. Magnetic resonance angiography shows multiple segmental cerebral vasoconstrictions (arrowheads, a) and magnetic resonance imaging fluid-attenuated inversion recovery (FLAIR) reveals a small high-signal-intensity lesion over the right parieto-occipital region (arrow, b). Both lesions recovered 1 month later (c, d).
Response to treatment
Nineteen patients (90%) were treated with nimodipine, of whom 13 received only oral treatment and six received i.v. treatment (mean 7.2 days, range 2–11 days) followed by oral treatment. Nimodipine was rated as effective in 84% (16/19) of the treated patients. No hypotensive events or other adverse events were reported. The duration of oral nimodipine treatment was 3–4 weeks with gradual tapering. No further attacks developed after discontinuation of nimodipine.
Clinical course and follow-up
Overall, BRTH resolved within an average of 14 days (range 6–34 days). Telephone interview of all patients did not find any relapse in any patient at a mean follow-up of 30 ± 23 months (3–79 months).
Discussion
This large series has shown that BRTH mainly affected middle-aged women. Their female sex hormones were either deficient or fluctuated in most of them. Based on the absence or presence of MRA vasoconstrictions, about 40% of patients fulfilled the diagnosis of ICHD-2 primary thunderclap headache, and the remaining 60% RCVS. This headache disorder usually ran a benign course and subsided within 2 weeks on average; however, two patients in the latter group (15%) developed reversible posterior encephalopathy. The great resemblance of the symptoms between those with and without vasoconstrictions suggests that the two disorders may represent two points on a spectrum (4). The so-called primary thunderclap headache may represent a milder form of RCVS with vasoconstriction not visible on MRA studies.
Our patients easily related their headaches to bathing because most patients (86%) developed severe headaches once the water was poured over their body. In addition, bath-related attacks accounted for 68% of all thunderclap headache attacks, even though most patients (71%) also had non-bath-related attacks. Of note, BRTH is not a new headache disorder, because bathing, like sexual activity, straining, coughing or singing, is just a trigger factor of thunderclap headaches in patients with RCVS. Our study has confirmed that BRTH exists and often points to RCVS; however, the cause of headaches is the underlying RCVS, not bathing. Unlike the first Japanese report (1), our study has shown warm rather than hot water was the most common triggering temperature. In addition, chest and hair were the most common body parts to be triggered. However, since our patients took a standing shower, upper torso and hair were usually the first body parts to be showered. It is of interest that the majority of patients could avoid bath-related headache attacks by changing bathing habits or stopping bathing.
This study has shown that among patients with BRTH, the diagnosis of RCVS, i.e. presence of cerebral vasoconstrictions, may indicate a high-risk group, since two patients developed posterior encephalopathy (15%). This is in line with our previous review that vasospasm is requisite for posterior encephalopathy among patients with thunderclap headaches (14). However, we can not exclude the possibility of the contribution of blood pressure surge (50%, 1/2) to this complication, because our case number was still small. Therefore, when treating patients with BRTH, it is mandatory to determine whether they had cerebral arterial vasospasm. We suggest MRA examination can be used to provide this prognostic information because (i) vasoconstrictions are usually multiple and commonly involve large vessels, e.g. MCA, accounting for 92% of involved vessels in those with RCVS as in our study, and (ii) catheter angiography may by itself induce vasospasm and cause other complications (15). However, the timing of the MRA has not been determined and the vasospasm in branches beyond the circle of Willis might be missed.
We empirically treated patients with BRTH with oral or i.v. nimodipine because some patients showed vasospasm (3, 4). The treatment was considered effective because patients could resume bathing 1 or 2 days later, although we did not adopt a double-blind controlled trial design. In contrast, nimodipine did not show efficacy in a previous case report of BRTH (5). The exact mechanism of action of nimodipine on BRTH remains unclear.
The pathophysiology of primary thunderclap headache or RCVS is poorly understood, not to mention BRTH. Dodick has proposed that an underlying neurogenic mechanism, i.e. the sympathetic system around the vessel, is responsible for the abrupt onset of both thunderclap headache and vasospasm (16, 17). This can explain why thunderclap headache attacks are frequently preceded by events that increase the sympathetic tone, such as rage, exertion or sex. Standing shower, a popular bathing habit in Taiwan, was adopted by all our patients during their first BRTH. Moreover, except for two Japanese patients, all published cases developed BRTH during a shower (1, 2, 5, 6). Standing shower can induce a greater increase in heart rate and blood pressure than basin or sink baths or sitting shower (18, 19), i.e. increasing sympathetic tone, which may more easily trigger thunderclap headache attacks. Furthermore, all the reported cases, including ours, were female, which implies that gender and sex hormones should not be overlooked. In fact, many of our patients were in a state of oestrogen deficiency or fluctuation, such as menopause, withdrawal of oestrogen replacement or postpartum. Of note, a recent study has revealed that postmenopausal women showed a decline in cardiac vagal modulation with a shift towards a state of sympathetic hyperactivity (20).
Our study has limitations. First, MRA has not replaced the gold standard of catheter angiography in evaluating cranial vasculature. In addition, vasoconstrictions at distal cerebral arterial branches may not be detected by MRA, which may lead to underestimation of the incidence of vasospasm. Second, lumbar puncture is considered to be part of the basic evaluations for thunderclap headache; therefore, physicians should try to persuade patients to submit to this important study. However, the low rate of spinal tapping in this study was due to the negative view of this examination in our society. In addition, without CSF spectrophotometry analysis, the chance of SAH could not be completely ruled out in our patients. Nevertheless, the chance of SAH should be low because none of the studies of CT, MRI, MRA and FLAIR showed SAH or aneurysms. Future studies should evaluate the chance of the diagnosis of SAH after lumbar puncture in patients with repetitive thunderclap headache attacks and their MR studies showing multiple segmental vasoconstrictions and absence of SAH or aneurysms.
In our opinion, BRTH has important clinical implications, because it points to an acute, severe, stroke-prone but potentially treatable headache disorder. Physicians should be alert to this headache syndrome and look not only for SAH or aneurysm but also for cerebral vasoconstrictions.
Footnotes
Acknowledgements
This study was supported in part by a grant from the National Science Council, Taiwan under Contract No. NSC 93-2314-B-010-044- and NSC 95-2314-B-010-005.