Abstract
Cluster headache is a stereotypic, primary headache disorder that is marked by repeated short-lasting attacks of severe, unilateral head pain and associated autonomic symptoms. Cluster headache is probably due to an abnormality in the circadian hypothalamic generator with subsequent trigeminovascular activation. We have reviewed the clinical manifestations, pathophysiology, gender differences, and treatment options in cluster headache.
Introduction
Cluster headache, the most painful of the primary headaches, has been nicknamed ‘suicide headache’ because sufferers have contemplated suicide in fear of another cluster attack or have even taken their own lives during an attack. Although the term cluster headache is now the generally accepted name for this severe, unilateral headache of brief duration, it is still sometimes referred to as migrainous neuralgia (1), especially in Europe. In 1939, Horton et al. (2) described the features of cluster headache and their experience in treating this ‘new syndrome of vascular headache’ with histamine. In 1952, the condition became known as Horton's headache or histaminic cephalalgia (3).
Cluster headache was once considered to be a variant of migraine, but this is no longer a tenable view. In 1988, the Headache Classification Committee of the International Headache Society (IHS) (4) classified cluster headache as a distinct headache disorder and separated it from migraine. The IHS criteria for cluster are as follows:
A.At least five attacks fulfilling B–D.
B.Severe unilateral orbital, supraorbital and/or temporal pain lasting 15–180 min untreated.
C.Headache is associated with at least one of the following signs that have to be present on the pain-side: 1.Conjunctival injection. 2.Lacrimation. 3.Nasal congestion. 4.Rhinorrhoea. 5.Forehead and facial sweating. 6.Miosis. 7.Ptosis. 8.Eye lid oedema.
D.Frequency of attacks: from one every other day to eight a day.
The individual attack, which lasts an average of 60–90 min, is called a cluster headache or cluster attack. The time during which recurrent attacks are occurring, usually weeks but at times months or years, is referred to as the cluster period or bout, or sometimes, especially by patients, simply as a cluster (of attacks). When attacks cease the individual is in remission. This may last from a few days to years.
Episodic cluster headache infers that substantial remissions occur. The IHS criteria for episodic cluster headache state that attacks ‘occur in periods lasting 7 days to 1 year separated by pain-free periods lasting 14 days or more’. Approximately 85% of individuals afflicted by cluster headache have the episodic form (5).
Chronic cluster headache consists of attacks that occur for more than 1 year without remission or with remissions that last less than 14 days. The chronic form of the disease can evolve from the episodic form (secondary chronic form) (6) or may develop de novo as primary chronic cluster headache. The rarest variety is the secondary episodic pattern, which begins as the chronic form and then becomes episodic. Chronic cluster headache occurs in approximately 15% of sufferers, is unremitting from onset (primary chronic cluster) in 10%, and evolves from the episodic form in 5%.
Clinical manifestations
In its commonest form, episodic cluster headache occurs at least once every 24 h for weeks at a time. A period of freedom from attacks (a remission) usually follows and may last from weeks to years (7, 8). A common pattern, especially in the first few years of cluster headache, is exacerbations (periods at risk) that occur seasonally, for example, every spring or every fall (9). This periodicity generally becomes less evident after a few years, and periods of cluster activity become much less predictable, occurring at almost any time of the year. Kudrow (10, 11) studied cluster periodicity in a large series of patients and noted that the most likely times for a cluster to start were apparently associated with the number of hours of daylight; more exacerbations occurred within the 2 weeks following the summer and winter solstices and fewer exacerbations within 2 weeks of the onset and offset of daylight saving time.
On average, a cluster period lasts 6–12 weeks while remissions last for 12 months (7, 8). Considerable variations, both between patients and in individuals, are characteristic, especially after the disorder has persisted for a few years. Despite this variability many individuals continue to have the onset of their yearly cluster at the same time each year. They are pleasantly surprised by the non-appearance of cluster in any particular year and unpleasantly surprised by the unexplained appearance of a cluster at a different time of year.
When exacerbations lengthen, remissions shorten, or more clusters occur than usual, the sufferer may be switching from the episodic to the chronic form. Once chronic cluster headache has developed, whether de novo or by transformation from the episodic form, it tends to persist for years, even into old age. Long-term follow up has shown that as many as 50% of affected individuals eventually revert or switch to an episodic form (12, 13). Individual attacks of headache occur daily or almost daily during a cluster period or in the chronic phase. When only one attack occurs in 24 h, it is not uncommon for it to occur at the same time each day or each night for days or weeks on end. Nocturnal attacks of cluster headache are more frequent than daytime attacks, unlike migraine or trigeminal neuralgia. The first attack often occurs about 90 min after falling asleep and has been related to the onset of rapid eye movement sleep (14, 15). Sleep deprivation, which can result from frequent nocturnal attacks, often results in early-onset rapid eye movement sleep, which may trigger further attacks. As this vicious cycle persists, even a daytime nap can induce rapid eye movement sleep and a further attack of headache.
The individual attack
The cluster headache attack lasts, on average, 45–90 min (shorter and longer attacks contribute to a bell-shaped curve of duration distribution). Attacks are strictly unilateral, almost without exception, in any cluster period and may remain on the same side throughout the individual's history. Less frequently, the pain may switch to the opposite side of the head and face in a subsequent cluster (15%), and even less frequently, attacks will switch sides from one attack to another (5%).
Cluster headache is usually described as coming on without warning. Some observers have described a vague premonitory warning sensation prior to an attack (16), but the onset of pain may be unheralded, rapidly escalating to pain of great intensity. Until recently, cluster headache was not considered to be associated with aura symptoms (as seen in migraine), but Silberstein et al. (17) reported on six cluster headache patients who had stereotyped aura (visual aura in five of six patients) preceding their individual cluster attacks.
Gastrointestinal symptoms are not felt to be typical of cluster headache attacks; vomiting is rare, but nausea occurs in up to 40% of patients. In some patients, nausea may be secondary to drug ingestion. The reported frequency of photophobia in patients with cluster headache varies from 5% to 72%, while phonophobia was reported only occasionally, in 12–39% of cluster headache cases (7, 18). However, recent quantitative data suggest that cluster headache patients have as much sensitivity to light and sound as migraine patients (who are markedly more sensitive than controls) (19). The recognized associated symptoms of cluster came from observation of male cluster patients. Female cluster sufferers appear to have a different cluster symptom profile than men (see below).
Quality and intensity of the pain
The pain of cluster headache is often described in such graphic terms as boring, tearing, or burning, and with such descriptive analogies as ‘a hot poker in the eye’ or as if ‘the eye is being pushed out’. The intensity is arguably the most severe of the primary headache syndromes, comparable to the pain intensity associated with trigeminal neuralgia and short-lasting unilateral neuralgiform pains with conjunctival and tearing syndrome (SUNCT).
Behaviour during an attack
In contrast to migraineurs, cluster patients are restless and occasionally even violent during an attack. Most are unwilling to lie down but prefer to pace about or sit and rock back and forth. Some will exert pressure on the painful area with a hand or place either an icepack or a hotpack over the affected eye and temple. Many will isolate themselves from family members or leave the house to get into cold or fresh air for the duration of the attack. Violent, destructive behaviour that may even result in self-inflicted injuries may occur, but this is rare. Some contemplate suicide during an attack, while others have been known to beg a family member to end their suffering.
As attacks appear to be triggered by rapid eye movement sleep, desperate individuals will attempt to remain awake for as long as possible. The resulting sleep deprivation shortens the rapid eye movement latency time so that when sleep inevitably occurs, an attack occurs shortly thereafter. The resulting vicious cycle of pain and lack of sleep frequently demoralizes the sufferer and can lead to depression and suicidal ideation.
Location
The pain of cluster headache is typically maximal around the eye and orbit of the affected side. It may radiate into the ipsilateral temple, forehead, cheek, and even the jaw. Ekbom (7) described an upper and a lower syndrome based on the distribution of the pain radiating out from the affected eye/orbit. In the upper syndrome, the pain, which is maximal around the eye, radiates to the forehead, temple, and parietal region, in any combination. In the lower syndrome, the pain radiates ipsilaterally to the upper and lower teeth and jaw and even into the neck.
Temporal profile
Cluster headache pain starts without warning (or after a mild pain in the temple) and rapidly worsens, reaching peak intensity within 5–10 min. It may stay at maximal intensity for 45–90 min, but can fluctuate slightly before decreasing in a stepwise manner. In some patients a period of repetitive peaks of pain separated by brief valleys of lesser pain characterizes the attack profile. The end of the attack usually comes quite suddenly, with rapidly decreasing intensity and finally freedom from pain.
Autonomic features
Table 1 lists the autonomic signs and symptoms that may accompany the pain of cluster headache attack.
Autonomic features of cluster headache
All the autonomic features are transient, lasting for the duration of the attack, with the exception of a partial Horner's syndrome (occurs in 57–69% of patients), with ptosis or miosis or both, which rarely may persist after the attack (20). The most common local signs of autonomic involvement are lacrimation and conjunctival injection, each present in more than 80% of patients. Nasal stuffiness or rhinorrhoea occurs in 68–76% of patients during attacks and is usually ipsilateral to the pain, but may, on rare occasions, occur bilaterally. Forehead sweating, facial flushing, and oedema are rare. Fluctuations in heart rate, blood pressure, and cardiac rhythm (including premature ventricular beats, transient episodes of atrial fibrillation, and first-degree atrioventricular or sinoatrial block) can occur (21). There may be no autonomic symptoms in up to 3% of cluster headache sufferers (18).
Physical characteristics
The typical heavy facial features of many of those subject to cluster headache were noted by Graham (22). Deep nasolabial furrows, ‘peau d’orange' skin, and telangiectasia led to the description of ‘leonine facies’. Women with cluster headache often have a masculine appearance, according to Kudrow (23). Many of the characteristic facial features believed typical of those subject to cluster headache are probably due to heavy tobacco and alcohol use, which are also characteristic of this population. Kudrow (24) reported that two-thirds of the patients in his large series had hazel-coloured eyes. He also noted that many of those subject to cluster headache are tall—about three inches above average. Given that exceptions are common, these physical characteristics are of no diagnostic specificity or therapeutic utility.
High gastric acid production and an increased incidence of peptic ulceration are also typical of cluster headache patients, but could be due to alcohol overuse.
Trigger factors
Once a cluster period begins, individual headache attacks can, in many patients, be triggered or precipitated by ingestion of alcohol and by other vasodilators, notably nitroglycerin (25) and histamine. Alcohol rarely precipitates an attack during a remission period; most sufferers will avoid alcohol as soon as a drink triggers an attack and will remain abstinent until the cluster period is over. The mechanism whereby alcohol induces an attack is not understood. Nitroglycerin is a prodrug for nitric oxide that can activate the trigeminal vascular system.
Allergies, food sensitivities, hormonal changes, and stress do not play a large part in the pathogenesis of cluster headache. Head trauma has been recognized as a cause of cluster headache (26, 27) but it is hard to prove a cause-and-effect relationship.
Cluster patients are typically hard drinkers and chronic cigarette smokers. These social vices may be the needed trigger to initiate a first-ever cluster attack. A smoking history has been found in up to 85% of cluster patients, while alcohol consumption is higher in cluster headache patients than in controls.
Differential diagnosis
Cluster headache, in its typical form, is unmistakable; when it is atypical, other disorders should be considered in the differential diagnosis. Migraine may present with recurrent unilateral headache with ipsilateral autonomic symptoms, particularly during severe attacks. However, the frequency and duration of cluster headache attacks differ from migraine. Cluster headache attacks are short-lived (45–90 min) compared with migraine, which can last from 4 h to 72 h. Cluster attacks are almost always unilateral, frequently nocturnal, can occur several times a day, and are associated with less nausea, vomiting, and aura than migraine. Alcohol typically triggers an attack of cluster headache in 20–30 min while headache is often delayed in migraineurs.
Temporal arteritis pain is usually continuous but may wax and wane, and is often associated with systemic symptoms, such as fever, polymyalgia, and weight loss. Trigeminal neuralgia is characterized by paroxysmal electric shock-like jabs of unilateral pain, most commonly limited to the distribution of the second or third divisions of the trigeminal nerve. In an acute attack, the cluster headache sufferer is typically agitated and moves about, in contrast to the migraineur, who is typically quiescent. The pain can be triggered by stimulation of limited areas of facial skin or oral mucosa. Sinusitis, glaucoma, intracranial aneurysms, tumours, arteriovenous malformations, dissection of the cervicocephalic cerebral blood vessels (carotid or vertebral), and even cervical cord lesions (meningioma) or infarctions can mimic cluster headache. In many of these instances, however, the history and examination disclose features that suggest a secondary cause of headache, and the history lacks the typical stereotyped periodicity of attack and remission phases.
A number of primary headache syndromes, such as chronic and episodic paroxysmal hemicrania, SUNCT syndrome, and even hemicrania continua, may resemble cluster headache (Table 2). Collectively, these disorders have been referred to as ‘trigeminal-autonomic cephalalgias’ because of the trigeminal distribution of pain and the associated autonomic signs (28). They are characterized by discrete, short-lasting, episodic attacks of intense, unilateral, orbital-temporal headaches associated with robust ipsilateral autonomic signs. These syndromes may be associated with nocturnal attacks and may be precipitated by alcohol. Short-lasting unilateral neuralgiform pain with conjunctival and tearing syndrome is the only other primary headache disorder that is more common in men. Compared with cluster headache, these disorders have a higher attack frequency and shorter attack duration; as attack frequency increases, attack duration tends to decrease (Table 2). Cluster headache differs from the other paroxysmal hemicranias in its response to therapy. The paroxysmal hemicranias and hemicrania continua often respond dramatically to indomethacin; patients with SUNCT syndrome derive no benefit from indomethacin or drugs typically used to treat cluster headache.
Cluster headache: comparison with other trigeminal-autonomic cephalalgias
CPH, Chronic paroxysmal hemicrania; EPH, episodic paroxysmal hemicrania; SUNCT, short-lasting unilateral neuralgiform pain with conjunctival injection and tearing; ppt, precipitated.
Preventive treatment for cluster headache
Surgical procedures for cluster headache
Genetics and family history
In contrast to migraine, cluster headache has not previously been considered to be an inherited condition. However, between 1947 and 1985, 12 studies demonstrated a family history in 47 of 1182 (4%) of patients (29). Newer family studies have found a positive family history in approximately 7% of cluster sufferers, which represents a 14-fold increased risk of cluster headache in first-degree relatives of probands with cluster headache (29–31) and a twofold increased risk in second-degree relatives. Twin studies have shown 100% concordance in five pairs of monozygotic twins (32–35). Complex segregation analysis suggests that an autosomal-dominant gene has a role in the inheritance of cluster headache in some families (29).
Epidemiology
Cluster headache is rare and the exact prevalence is uncertain because of the paucity of epidemiologic surveys. A large study of 9803 Swedish army recruits, all 18-year-old men, revealed a prevalence of 92 per 100 000 (confidence interval (CI) 42–174) (36), while the prevalence in a large population-based survey from the republic of San Marino was 69 per 100 000 (CI 39–114) (37). Since cluster headache predominates in men, the difference in prevalence between these two studies may be explained by the gender difference in the study populations. The Swedish data were extrapolated by Kudrow to the distribution of age of onset from headache clinic data to yield a prevalence rate of approximately 0.4% (8). A third study from Rochester, Minnesota, based on a retrospective chart review, revealed a much higher prevalence of 401 per 100 000 (38). In this study, the age-adjusted incidence of cluster headache was 15.6 per 100 000 person-years for men and 4.0 per 100 000 person-years for women. The diagnosis of cluster headache was based on a review of charts from patients who had been seen and diagnosed by a wide variety of clinicians. Diagnostic inaccuracy may have led to the high figures obtained in this study.
Demographics
Cluster headache has long been recognized to predominate in men. The gender ratio has varied between 5.0:1 and 6.7:1 in the larger case series that have included at least 200 patients (8, 39–43). The largest case series (Horton's in 1956 of 1176 patients and Kudrow's in 1980 of 425 patients) have reported male:female ratios of 6.7:1 and 5:1, respectively. However, recent evidence suggests a progressively decreasing male preponderance or an increasing incidence of cluster headache in women. In a study of 482 patients from Parma, Italy, Manzoni (44) reported a male:female ratio of 3.5:1. In analysing the male to female ratio based on the year of onset, the ratio decreased from 6.2:1 in the 1960s to 2.1:1 in the 1990s. It is unclear whether this represents a true rise in the incidence of cluster headache in women or whether it is just becoming better recognized by physicians and thus diagnosed in women. Specific male behavioural and societal traits have been blamed as possible cluster triggers. Until recently men have had more stressors of daily living compared with women and were heavier smokers and drinkers compared with their female counterparts. Perhaps this was why men developed cluster more often than women. The changing role of women in society and in the workplace, especially since the 1970s, may be in part responsible for the rise of cluster headache in women.
Cluster headache in women
The clinical characteristics of cluster headache in women are not well recognized. The small number of studies that have looked at cluster headache in women have shown that women have their first-ever cluster headache earlier in life than men (7, 45, 46). Women also have two age peaks of cluster onset (second and sixth decade), while men only have one (third decade). Most women with cluster are Caucasian, but more African–American women develop cluster than African–American men (8, 46). Episodic cluster is the dominant cluster subtype in women, as it is in men. Chronic cluster, which the old literature stated was almost non-existent in women, does occur in women and may do so to the same extent as in men (8, 43, 46). Although women have the same number of cluster attacks a day as men, the attack duration appears to be shorter in women (43, 46). The autonomic symptoms that are associated with a cluster attack appear to be less prominent in women than in men. Rozen et al. (46) found that women with cluster experienced less miosis and ptosis during attacks than men, but had an equal frequency of lacrimation, nasal congestion, and rhinorrhoea. This suggests that women cluster patients have less sympathetic dysfunction but parasympathetic activation is equal to that in men. ‘Migrainous symptoms’ in cluster are very prominent in women, most of whom do not have a concomitant history of migraine. Nausea and vomiting are much more common in women cluster patients (43, 46), while photophobia and phonophobia are prominent symptoms in both sexes, approaching the prevalence of these symptoms in migraine. Women cluster patients may respond differently to cluster therapy than men. Rozen et al. (47) found that women responded less frequently to oxygen therapy than men (59% vs. 87%), and women achieved a headache-free state, on any combination of cluster preventive therapy, less frequently than men (41% vs. 58%).
Pathogenesis
A unifying pathophysiologic explanation of cluster headache is not yet available
Based on the clinical features of the disorder, three conclusions can be drawn. First, because the pain of cluster headache is invariably centred around the eye and forehead, it is probable that the ipsilateral trigeminal nociceptive pathways are integrally involved. Second, the ipsilateral autonomic features suggest activation of the cranial parasympathetic system (lacrimation and rhinorrhoea) and dysfunction of the ipsilateral sympathetic nerves (ptosis and miosis). The cavernous carotid artery was suggested as a likely site of involvement, since it is here that the trigeminal, parasympathetic, and sympathetic fibres converge. Finally, the remarkable and often clockwork consistency and seasonal predilection of attacks strongly suggest that a central pacemaker (the suprachiasmatic nucleus) must be integrally involved in the genesis of this disorder.
Pain and autonomic features
Blood flow/cavernous sinus hypothesis
Cluster headache has traditionally been thought of as a ‘vascular’ headache disorder. A number of observations have indicated that there is vasodilation of the ipsilateral ophthalmic artery during a cluster headache attack. These include increased corneal indentation pulse amplitude, intraocular pressure, and skin temperature around the eye, as well as decreased blood flow velocities on ultrasonography (48–52). Magnetic resonance angiography performed during spontaneous attacks of cluster headache revealed marked dilation of the ophthalmic artery ipsilateral to the pain (53). This confirmed Ekbom's original observation of marked dilation of the ipsilateral ophthalmic artery in a patient undergoing catheter angiography during a spontaneous cluster headache (54). Positron emission tomography (PET) studies during precipitated attacks of cluster headache have demonstrated bilateral activation (more marked on the side of the headache) in the region of the cavernous sinus (55), thought to be indicative of increased flow in the cavernous portion of the internal carotid arteries (56). Involvement of the cavernous carotid artery would be consistent with the location of the pain and the third order neurone postganglionic Horner's syndrome often seen during attacks.
Hardebo (57) suggested that the mechanism of cluster headache is an inflammatory process in the cavernous sinus and tributary veins. This theory was based principally on the observation of abnormal orbital phlebography in patients during cluster headache attacks (52, 58), the fact that nitroglycerin and other vasodilators can induce an attack of cluster headache (25), and the finding that cluster headache patients have a narrower anterior/middle cranial fossa and possibly a narrower cavernous sinus loggia, which might favour disturbances in local venous drainage from the cavernous sinus region (59). An inflammatory ‘venous vasculitis’ was thought to obliterate the venous outflow from the cavernous sinus and lead to vascular congestion. The vascular congestion within the arterial and venous circulation in the cavernous sinus may lead not only to pain, but also to injury of the traversing sympathetic fibres that richly invest the carotid artery. The active period ends when the inflammation is suppressed and the sympathetic fibres partially or fully recover.
However, a magnetic resonance imaging (MRI) study of cluster patients found no definite pathologic changes in the area of the cavernous sinus (60). Furthermore, parasellar hyperactivity using single positron emission computed tomography was seen in 50–80% of cluster headache patients and in more than 70% of migraine patients, indicating that activation in this region is not specific to cluster headache (61). In addition, identical orbital phlebography findings have been demonstrated in patients with Tolosa Hunt syndrome (62), hemicrania continua (63), SUNCT syndrome (64, 65), chronic paroxysmal hemicrania (63), cervicogenic headache, migraine, and tension-type headache (66). PET has shown the same pattern of activation in migraine as well as in healthy controls with experimentally induced first division trigeminal pain elicited with capsaicin (67). This strongly suggests that the flow changes seen during cluster headache attacks are an epiphenomenon of trigeminal activation, are not specific, and do not play a large role in the genesis of cluster headache.
Nitroglycerin induces attacks of cluster headache that are indistinguishable from spontaneous attacks after a latency of approximately 30–40 min, well after the vasodilatory effect of the drug has resolved. The provocative action of nitroglycerin may be due, in part, to activation of the trigeminovascular system (68).
Trigeminovascular and cranial-parasympathetic pathways
Intracranial pain-producing structures are predominantly served by the first (ophthalmic) division of the trigeminal nerve with cell bodies in the trigeminal ganglion (69, 70). The cerebral blood vessels and dura mater are innervated by the trigeminal neurones containing the neuropeptides, substance P, calcitonin gene related peptide, and neurokinin A (71, 72). In addition, both parasympathetic and sympathetic nerves innervate these vessels, distinguishable by specific neuropeptide populations.
The cranial parasympathetic innervation of the intracranial vessels arises from primary-order neurones located in the superior salivatory nucleus. The efferent fibres exit the brainstem via the seventh cranial nerve (73), traverse the geniculate ganglion, and synapse in the sphenopalatine, otic, and carotid miniganglia (74). Parasympathetic vasomotor efferents then travel via the ethmoidal nerve to innervate the cerebral blood vessels. Secretomotor efferents innervate both the lacrimal and nasal mucosal glands, thus providing the anatomic basis for the cranial autonomic symptoms (lacrimation, nasal congestion, rhinorrhoea) seen in patients with cluster headache and other trigeminal-autonomic cephalalgias (28). Cranial parasympathetic fibres contain vasoactive intestinal polypeptide and nitric oxide synthase, both of which co-localize within sphenopalatine ganglion neurones and cerebrovascular parasympathetic nerve fibres (75).
Stimulation of the trigeminal ganglion causes a cerebral vasodilator response, with an increase in brain blood flow. This effect is mediated by antidromic activation of trigeminal afferents with release of calcitonin gene related peptide, as well as through stimulation of parasympathetic outflow, which is accomplished through a functional reflex between the trigeminovascular (nucleus caudalis) and cranial parasympathetic (superior salivatory nucleus) systems (76, 77). Stimulation of either the superior saggital sinus, a C-fibre innervated pain sensitive structure (78), or the trigeminal ganglion (79) results in the release of both calcitonin gene related peptide and vasoactive intestinal polypeptide into the cranial circulation.
In humans, activation of the trigeminovascular system, which is marked by an increase in the level of calcitonin gene related peptide in the cranial venous circulation, occurs during attacks of migraine (80), cluster headache (79), and chronic paroxysmal hemicrania (75). Parasympathetic activation, with dramatically elevated levels of vasoactive intestinal polypeptide, occurs during attacks of cluster headache (79) and chronic paroxysmal hemicrania (75) and is associated with robust ipsilateral autonomic features. In patients with migraine, vasoactive intestinal polypeptide levels are elevated only if there are accompanying signs of autonomic activation, such as tearing or nasal congestion (28).
These findings support the involvement of both the trigeminovascular and cranial parasympathetic systems in cluster headache (79). Activation of these pathways provides the anatomical basis for the expression of the first division trigeminal pain and the ipsilateral autonomic symptoms that occur during cluster attacks.
Periodicity
The signature feature of cluster headache is the unmistakable circadian and circannual periodicity of the disorder. This clock-like rhythmicity is difficult to reconcile or explain on the basis of haemodynamic mechanisms and leads to the belief that cluster headache initiation must be central in origin (7, 10, 81).
Cluster headache attacks occur one-to-eight times a day, often with clock-like regularity. Cluster periods are frequently observed to occur cyclically, often at the same time each year. The frequency of cluster period onset has been found to be related to photoperiod duration, increased in July and January, shortly after the longest and shortest days of the year, respectively. Conversely, cluster period onset decreases following the 1-h resetting of clocks for daylight-saving and standard times in April and October, respectively (82).
This distinctive periodicity suggests involvement of the biological clock or pacemaker, which, in humans, is located in the hypothalamic grey in an area known as the suprachiasmatic nucleus. Hypothalamic regulation of the endocrine system involves rhythmic and phasic modulation of the hypophyseal hormones and melatonin to maintain homeostasis. Lowered concentrations of plasma testosterone during the cluster headache period in men provided the first evidence of hypothalamic involvement in cluster headache (83). Alterations in secretory circadian rhythms of leutinising hormone (LH), cortisol, and prolactin, as well as altered responses in the production of cortisol, LH, follicle stimulating hormone (FSH), prolactin, growth hormone (GH), and thyroid stimulating hormone (TSH) to diverse challenges also occur in cluster headache (84).
Melatonin (a sensitive surrogate marker of circadian function in humans) and its rhythmic secretion are under the control of the suprachiasmatic nucleus (85). The principal environmental stimulus for the entrainment and rhythmic secretion of melatonin is light intensity. Photic information reaches the suprachiasmatic nucleus from a direct retinal-hypophyseal pathway. The circadian rhythm for the release of melatonin from the pineal gland is closely synchronized with the habitual hours of sleep. Melatonin levels are normally low during the day and increase during the hours of darkness and sleep. In patients with cluster headache, the 24-h production of melatonin is reduced; the nocturnal peak in melatonin concentration is blunted during cluster periods, and the acrophase (the time from midnight to the moment of peak hormone level) is moved forward (86). Pain-induced stress cannot explain this decrease, because stress causes a release in endogenous norepinephrine, which is known to increase melatonin production.
Low melatonin may be due to reduced availability of serotonin, which is needed for its synthesis. Impaired function of the serotonergic system in cluster headache may occur; an increase in plasma serotonergic metabolism with low platelet 5-HT and increased 5-HIAA has been reported (87). Leone et al. (88), using m-CPP challenge, found an altered cortisol secretion pattern during the active cluster headache, suggesting impaired central serotonergic function.
The most direct and convincing evidence for the role of the hypothalamus in cluster headache has come from functional and morphometric neuroimaging. Using positive emission tomography (PET) to detect areas of functional activation, May et al. (55) demonstrated marked activation in the ipsilateral ventral hypothalamic grey matter during attacks of acute cluster headache induced by nitroglycerin. This finding is specific for cluster headache and other related trigeminal autonomic cephalalgias; it also occurs with SUNCT syndrome (56). This pattern of activation does not occur with migraine or experimentally induced ophthalmic (first division) pain induced by capsaicin injection into the forehead of control subjects (55).
Voxel-based morphometric analysis of the structural T1-weighted MRI scans of 25 right-handed patients with cluster headache revealed a significant difference in hypothalamic grey-matter density between these patients and 29 right-handed healthy male volunteers (56). Cluster headache patients had an increase in hypothalamic volume. This structural difference was located in the inferior posterior hypothalamus, almost identical to the area of activation seen on PET during an acute cluster headache attack. The co-localization of morphologic and functional changes within a discrete hypothalamic region has identified the anatomical location for the central nervous lesion of cluster headache. It might explain the circadian rhythmicity of this syndrome. The nature of the hypothalamic disturbance is as yet unclear.
Mitochondrial dysfunction
Phosphorus magnetic resonance spectroscopy has shown defective brain and muscle energy metabolism in patients with migraine with and without aura during and between attacks (89–91). A defect of brain and muscle mitochondrial respiration has also been demonstrated in cluster headache patients both during and after the cluster periods (92). Two cases of cluster headache occurring in conjunction with large scale mitochondrial deletions and a point mutation in platelet mitochondrial tRNALeu(UUR) have been reported (93, 94). One patient had a therapeutic response to ubidecarenone (94). Abnormal energy metabolism of excitable cells (brain and muscle) may represent a characteristic of migraine and cluster headache and suggests a shared pathogenesis. Recent reports on the presence of premonitory symptoms (16), photophobia and phonophobia (95), nausea, and visual aura (17) in patients who have cluster headache appear to underscore the potential for a shared mechanism in these two disorders.
Management
Successful cluster headache treatment requires patient education, prophylactic medication, and symptomatic treatment.
Patient education
Patients should be told that most cluster attacks may be prevented by prophylactic measures and that breakthrough attacks can be quickly terminated with acute therapies. However, the cluster period itself can be neither prevented nor shortened.
Patients should be instructed to avoid afternoon naps and alcoholic beverages, including wine and beer, since alcohol induces acute attacks during active cluster periods. They should be cautioned about prolonged exposure to volatile substances, such as solvents, gasoline, or oil-based paints, during cluster periods. Dietary influences, with the exception of alcohol, appear to have little importance in cluster headache.
Altitude hypoxaemia at levels above 5000 feet may induce attacks during cluster periods. Cluster attacks due to higher altitudes may be prevented by oral administration of acetazolamide, 250 mg twice a day for 4 days, starting 2 days before altitude is reached.
Finally, patients may be advised that the onset of cluster periods often follows a long period of sleep alteration, such as changes resulting from vacation trips, work-shift changes, or new occupations. Although many variables are associated with an altered life-style, the different sleep–wake patterns that often accompany these changes may be the most important.
Acute (symptomatic) therapy
Because of the rapid onset and short time to peak intensity of cluster headache pain, fast-acting symptomatic therapy is needed. Oxygen, subcutaneous sumatriptan, and intramuscular dihydroergotamine provide the most rapid, effective, and reliable relief for cluster headache attacks.
Oxygen
Oxygen inhalation has been the standard of care for the symptomatic relief of cluster since it was introduced as an effective therapy by Horton (39). If delivered at the onset of an attack via a non-rebreathing facial mask at a flow rate of 7–10 l/min for 15 min, approximately 70% of patients will obtain relief within 15 min (96, 97). In some patients, oxygen is effective if taken when the pain is at maximal intensity, while in others the attack is delayed for minutes to hours rather than completely aborted. A small, placebo-controlled study of hyperbaric oxygen (2 a.t.m.) delivered over 30 min demonstrated efficacy in six of seven patients within 5–13 min, with these patients reporting complete or partial interruption of the cluster period (98). These therapies have obvious practical limitations, in that treatment is not always readily available, and although small portable cylinders are available for use at work or when out of the house, some patients find this to be cumbersome and inconvenient.
Sumatriptan
Subcutaneous sumatriptan is the most effective self-administered medication for the symptomatic relief of cluster headache. In a placebo-controlled study, 6 mg of sumatriptan delivered subcutaneously was significantly more effective than placebo, with 74% of patients having complete relief by 15 min compared with 26% of placebo-treated patients (99). In long-term, open-label studies, sumatriptan is effective in 76–100% of all attacks within 15 min, with no evidence of tachyphylaxis or rebound, even after repetitive daily use for several months (100, 101) However, sumatriptan is not effective when used before an expected attack in an attempt to prevent an oncoming attack, nor is it useful as a prophylactic agent (102). Sumatriptan has been reported to be approximately 8% less effective in patients with chronic cluster headache than in patients with episodic cluster headache (101). Although generally well tolerated, sumatriptan is contraindicated in patients with ischaemic heart disease or uncontrolled hypertension. Caution must be exercised since cluster headache predominates in middle-aged males who often have risk factors for cardiovascular disease, particularly tobacco abuse, which is present in up to 88% of cluster headache sufferers (103).
Sumatriptan nasal spray 20 mg is less effective than subcutaneous injection. In an open, randomized study comparing the effectiveness and satisfaction of subcutaneous sumatriptan 6 mg with intranasal sumatriptan 20 mg, 49 of 52 treatments with injection resulted in complete relief of pain within 15 min, with a mean time to pain relief of 9.6 min (57). The remaining three attacks were reduced by a mean of 87% at 15 min. By comparison, only seven of 52 treatments with nasal spray in the nostril ipsilateral to the pain resulted in complete relief within 15 min, with a mean of 13 min. There was no pain relief at 15 min in 27 attacks.
Dihydroergotamine
Dihydroergotamine is available in the USA in injectable and intranasal formulations. Dihydroergotamine administered intravenously provides prompt and effective relief of cluster headache within 15 min. The intramuscular and subcutaneous routes of administration provide slower relief because of the time to maximal concentration. Because of the rapid peak intensity and relatively short duration of each cluster headache attack, intravenous dihydroergotamine is a less attractive and feasible option than sumatriptan, because the patient would have to get to an emergency department or physician's office to have an i.v. line placed. It is not a feasible long-term solution, as the attacks are often daily and multiple during a cluster period, which may last months, particularly if preventive therapy is not optimal or effective. However, i.v. or s.c. dihydroergotamine can be self-administered and is an option.
A double-blind crossover trial compared intranasal dihydroergotamine (1 mg) with placebo and found no effect on headache frequency or duration, but the pain intensity was significantly reduced with dihydroergotamine compared with placebo (104). The effect, however, was not dramatic. It has been suggested that the dosage used (1 mg) was lower than the recommended dosage for migraine (2 mg) and less than the currently available commercial preparations of dihydroergotamine nasal spray (4 mg). Therefore, dihydroergotamine at a dose of 2 mg may be more effective than 1 mg, but this has not been studied in a controlled fashion.
Zolmitriptan
Zolmitriptan is an effective oral agent for the acute treatment of migraine. Recently, a double-blind controlled trial compared the efficacy of 5 and 10 mg of oral zolmitriptan and placebo for the treatment of acute cluster headache attacks (105). With headache response defined as a two-point reduction on a five-point pain intensity scale at 30 min, response rates following placebo and 5 and 10 mg of zolmitriptan were 29%, 40%, and 47%, respectively. The difference reached statistical significance for 10 mg of zolmitriptan compared with placebo. Significantly more patients reported mild or no pain 30 min after treatment with 5 and 10 mg of zolmitriptan (57% and 60%, respectively) than after placebo (42%). Although these efficacy rates do not approach those of oxygen or s.c. sumatriptan, zolmitriptan is the first orally administered triptan to demonstrate efficacy in the treatment of cluster headache and remains a therapeutic option for patients who desire an oral medication or cannot tolerate oxygen, s.c. sumatriptan, or s.c. dihydroergotamine.
Lidocaine
Because cocainization of the pterygopalatine ganglion has been helpful in aborting cluster headache attacks (106), intranasal lidocaine has been utilized as an adjunctive therapy since Kitelle's (107) study. However, whether applied via a spray bottle or by dropping 4% viscous lidocaine in the nostril ipsilateral to the pain, it produces only a moderate reduction in pain in less than one-third of patients (108). While it may be useful as adjunctive therapy, it is not a front-line ‘stand-alone’ therapy for relief of acute cluster attacks.
Preventive pharmacotherapy
The importance of an effective preventive regimen during cluster periods cannot be overstated. During cluster periods, individual cluster attacks often occur daily for several weeks to months. Since many patients have more than one attack a day (up to eight), and the attacks are severe, short-lived, and peak rapidly, repeated attempts at abortive therapy become an exhaustive exercise. In addition, abortive therapies may be contraindicated, ineffective, or not tolerated, or they may simply delay the attack. Treating frequent daily attacks may result in overmedication or toxicity, and finally, repeated attacks of severe pain may unnecessarily prolong suffering.
The primary goals of preventive therapy are to produce a rapid suppression of attacks and to maintain that remission over the expected duration of the cluster period. Secondary objectives are to reduce the headache frequency and the attack severity and duration. To achieve these primary goals, preventive therapy can best be thought of in terms of transitional and maintenance prophylaxis.
Transitional prophylaxis
Ergotamine derivatives
Both ergotamine tartrate (2 mg) and dihydroergotamine (1 mg) are effective in achieving rapid suppression of attacks when administered daily for a short period of time. Patients often tolerate these medications for a period of 2–3 weeks without the risk of rebound. Ergotamine tartrate is more convenient because of its oral route of administration and may be particularly useful when given 1–2 h prior to bedtime for attacks that occur predominantly or exclusively during sleep (3, 23, 109). Both agents may also be administered in divided daily dosages (not to exceed 4 mg of ergotamine tartrate or 3 mg of dihydroergotamine) when attacks are multiple or occur throughout the day. Both are contraindicated during pregnancy and in patients with peripheral vascular disease, coronary artery disease, or uncontrolled hypertension. They should not be used for the duration of the cluster period and are not intended for long-term preventive use. They may limit preventive and acute options since they are contraindicated within 24 h of using sumatriptan, and some do use them concomitantly with methysergide.
Corticosteroids
Corticosteroids (prednisone and dexamethasone) are the fastest-acting preventive agents. They are a very effective initial prophylactic option, rapidly suppressing attacks during the time required for the longer-acting maintenance preventive agents to take effect. Standard preventive therapy may not be effective until 2 weeks after treatment is initiated. The largest open-label study reported marked relief of cluster headache in 77% of 77 patients with episodic cluster headache and partial relief in another 12% of patients treated with prednisone (8). Prednisone provided marked relief in 40% of patients with chronic cluster headache and was more effective than methysergide. Treatment is usually initiated with 60–80 mg of prednisone per day for 2–3 days followed by 10 mg decrements every 2–3 days.
Dexamethasone at a dose of 4 mg twice a day for 2 weeks followed by 4 mg a day for 1 week has also been shown to be effective (110). However, when dexamethasone or prednisone is tapered, the cluster attacks frequently recur. Therefore, corticosteroids are primarily useful for inducing a rapid remission in patients with episodic cluster headache, although they may provide a brief respite for patients with chronic cluster headache. Long-term use of corticosteroids in these patients must be resisted.
Maintenance prevention
Maintenance prevention refers to the use of preventive medications throughout the anticipated duration of the cluster period. The preventive medications are started at the onset of the cluster period in conjunction with either corticosteroids or ergotamine derivatives, but are continued after these initial suppressive agents are discontinued.
Verapamil
Verapamil is often used as the first preventive therapy for both episodic and chronic cluster headache. It is generally well tolerated and can be used safely in conjunction with sumatriptan, ergotamine, corticosteroids, and other preventive agents. In an open-label trial involving 48 patients, 69% of patients improved by more than 75% during treatment with verapamil (111). A recent double-blind, placebo-controlled trial evaluated the efficacy of verapamil (360 mg in three divided doses) over a 14-day period. A statistically significant reduction in headache frequency and analgesic consumption was seen in the verapamil-treated patients, with a greater reduction in the second week of treatment (112).
The initial starting daily dosage is 80 mg three times a day or 240 mg of sustained release a day. Dosages employed range from 240 g to 720 g a day in divided dosages. Both the regular and extended released preparations have been shown to be useful, but no direct comparative trials are available. Delayed release verapamil at dosages up to 720 mg may be effective in cases of refractory cluster headache (113). Because of this apparent dose–response relationship, a total daily dosage of between 480 mg and 720 mg is recommended before the medication is deemed a failure. Some push the dose as high as 1 gm if tolerated. Constipation is the most common side-effect, but dizziness, oedema, nausea, fatigue, hypotension, and bradycardia may also occur.
Lithium carbonate
Lithium carbonate therapy is effective for cluster headache based mainly on open clinical trials. Collectively, in over 28 clinical trials involving 468 patients, good-to-excellent results were found in 78% or 304 patients with chronic cluster headache (6). The lithium efficacy in patients with chronic cluster persists up to 4 years after treatment (12). Upon interruption or cessation of therapy with lithium, a transition from chronic to episodic cluster headache occurs in this group (40).
Although with somewhat less robust response than in patients with chronic cluster headache, lithium has been shown to induce a remission in 63% of 164 patients with episodic cluster (40). A double-blind crossover study compared verapamil (360 mg daily) with lithium (900 mg daily) in 30 patients and found equal efficacy (114). A single double-blind, placebo-controlled trial failed to show superiority of lithium (800 mg sustained release) over placebo. However, this study was stopped 1 week after treatment began, and there was an unexpectedly high placebo response rate of 31% (115). The treatment period was therefore too short to be conclusive.
The initial starting daily dosage is either 300 mg three times a day or 450 mg sustained release. Trials comparing the two formulations are not available, but the longer half-life affords the option of a once-daily dosage regimen, which enhances compliance.
Lithium is often effective at serum concentrations (0.4–0.8 mEq/l) lower than those usually required for the treatment of bipolar disorder. Most patients will benefit from dosages between 600 g and 900 mg a day.
Lithium has many side-effects and a narrow therapeutic window. The serum concentration should be measured 12 h after the last dose and should not exceed 1.0 mEq/l. Renal and thyroid function must be measured prior to and during treatment; side-effects such as tremor, diarrhoea, and polyuria must be monitored; and caution must be exercised when other drugs, such as diuretics and non-steroidal anti-inflammatory drugs, are prescribed.
Methysergide
Methysergide is an effective preventive drug for the treatment of cluster headache, but because of the potential for fibrotic complications, it is not commonly employed for long periods of time (> 3 months) in patients with chronic cluster headache. In patients with episodic cluster headache, good-to-excellent results occur in 70% of patients (8, 116), but the drug appears to lose its effectiveness with repeated use in up to 20% of patients (8).
Methysergide is a prodrug of methylergometrine, and should be used with caution when patients are receiving other ergotamine derivatives or vasoconstrictive agents. The short-term side-effects include nausea, muscle cramps, abdominal pain, and pedal oedema. Long-term side-effects include fibrosis of the retroperitoneum or the pleural and pericardial lining. The daily dose is usually 2 mg in three divided dosages, but up to 12 mg may be used if tolerated.
Valproic acid
Valproate is an anticonvulsant that is also effective for mania and migraine. Valproic acid (600–2000 mg) was effective in an open-label study of 15 patients with cluster headache, with a 73% favourable response rate (117). Nine of 15 patients had complete suppression of their attacks; the time to pain relief was short, ranging from 1 to 4 days. Treatment was well tolerated; only nausea was reported, but weight gain, hair loss, tremor, and lethargy are potential side-effects. Patients whose cluster headaches are accompanied by migrainous features, such as nausea, vomiting, photophobia and phonophobia, may preferentially respond to valproic acid (118).
Valproate, in the form of divalproex sodium, is usually started in divided dosages of 250 mg twice a day, and a 250-mg dose increment is recommended to find the lowest effective dose so as to minimize side-effects. Pancreatitis, platelet dysfunction, thrombocytopenia, and hepatic dysfunction have been rarely described with this medication, necessitating baseline complete blood counts and liver function testing. Follow-up studies are recommended only if they are clinically warranted.
Topiramate
Topiramate, in a recent open-label study, was associated with rapid improvement in 10 cluster headache patients (119). Cluster period remission occurred in 1–3 weeks in nine patients, two of whom had chronic cluster headache. All patients responded to relatively small dosages (50–125 mg a day given twice a day), which were well tolerated. Starting at low dosages and making small increments can minimize both the total daily dosage and the potential for side-effects. Somnolence, dizziness, ataxia, and cognitive symptoms are the most commonly reported side-effects. Topiramate is a weak carbonic anhydrase inhibitor, and renal calculi and paresthesias have been reported. This favourable preliminary report must be followed up by further corroborative study.
Melatonin
Serum melatonin levels are reduced in patients with cluster headache, particularly during a cluster period (53, 86). Based upon these observations, the striking circadian rhythmicity of cluster headache, and the importance of the hypothalamus in the pathogenesis of this disorder, the efficacy of 10 mg of melatonin orally was evaluated in a double-blind, placebo-controlled trial (88). Cluster headache remission within 3–5 days occurred in five of 10 patients who received melatonin compared with zero of 10 patients who received placebo.
Capsaicin
Capsaicin was superior to placebo in reducing attack frequency and severity in a double-blind study when delivered at a dose of 0.025% twice a day via a cotton-tipped applicator in the ipsilateral nostril for 7 days (120). Since there are more easily administered and effective agents available, and because of the intense local discomfort it causes, it has not enjoyed widespread use in the treatment of cluster headache.
Indomethacin
Despite the fact that other trigeminal-autonomic cephalalgias, such as chronic and episodic paroxysmal hemicrania, respond in an absolute way to indomethacin, it has not been systematically evaluated for cluster headache prevention. Anecdotal evidence suggests that some cluster headache patients respond to indomethacin, but the response rate appears to be less than that seen with these other disorders.
Others
Small open-label studies or case reports have suggested the efficacy of methylphenidate, anti-spasticity drugs (tizanidine and baclofen), clonidine, diltiazem, flunarizine, histamine, somatostatin, and pizotifen for patients with cluster headache. Further evidence is needed before recommendations can be made to support their routine use in cluster headache. However, all medical options should be considered in patients with treatment-resistant cluster headache before an ablative surgical procedure is attempted.
Refractory patients
Medical therapy
Approximately 10% of patients develop chronic cluster headache that does not respond to monotherapy. In addition, patients who have episodic cluster headache with frequent cluster periods may develop resistance, intolerance, or contraindications to preventive and/or acute medications and may require a surgical procedure for pain control.
Before considering surgery, combination therapy should be tried. Lithium, methysergide, divalproex, or topiramate may be combined with verapamil. Melatonin may be a useful adjunct, as it has few side-effects and its long-term use appears to be associated with few adverse events. Three medications may be used in combination, such as ergotamine, verapamil, and lithium. This is usually not a feasible long-term therapeutic option, as the toxicity of these medications may become cumulative and the side-effects intolerable. In addition, ergotamine is not recommended for long-term use and the use of lithium, ergotamine, or methysergide may restrict the usage of sumatriptan as an abortive agent.
Repetitive intravenous dihydroergotamine administered in an inpatient setting over a period of 3 days may be very useful in some patients who have both episodic and chronic cluster headache. In one study of 54 intractable cluster headache patients (31 of whom had chronic cluster headache) with a median hospital stay of 6 days, all patients were headache-free after repetitive i.v. dihydroergotamine, and at 12-month follow-up, 83% and 39% of episodic and chronic cluster headache patients, respectively, remained free of headache (121).
Histamine ‘desensitization’ has been used to treat patients with intractable cluster headache, with mixed results. This therapy usually entails a prolonged hospital stay of at least 1 week with repetitive administration of i.v. histamine (122). This treatment modality does not enjoy widespread use at this time.
Surgery
For the most intractable patients who have failed outpatient and inpatient therapy, or for whom contraindications or intolerance limits the use of effective medications, surgery may be a feasible option. Patients must be carefully selected. Only patients whose headaches are exclusively unilateral should be considered for surgery, as patients whose attacks have alternated sides are at risk for a contralateral recurrence after surgery. In addition, only patients with a stable personality and psychological profile with low addiction proneness should be considered.
Many surgical procedures have been used; those directed toward the sensory trigeminal nerve have been the most successful. The procedure of choice is radiofrequency thermocoagulation of the trigeminal ganglion. It is generally preferred over glycerol gangliorhizolysis, as the extent and precision of the lesion can be better controlled with the former and it entails a lower risk of aseptic meningitis or subarachnoid haemorrhage.
The overall results of radiofrequency rhizotomies are encouraging; approximately 75% of patients have good-to-excellent results (123, 124). The durability of the procedure is also quite favourable, with a long-term recurrence rate of only 20% and some patients remaining pain-free even after 20 years (125). The best results may require complete analgesia or dense hypalgesia. If the pain is primarily orbital in location, V1 and V2 lesions appear to be adequate, but if the pain also involves the temporal or auricular region, a V3 lesion may also be necessary for optimal results. Patients whose pain is located primarily around the ear, temple, or cheek may not have as good a result (125). Transient complications include diplopia, hyperacusis, ice-pick pain, and jaw deviation; longer-term complications include corneal anaesthesia and, in less than 4% of cases, anaesthesia dolorosa. Aggressive long-term ophthalmic follow-up and eye care are needed.
Gamma knife radiosurgery was reported to be effective in six medically recalcitrant cluster headache patients (126). The time to effective relief was either immediate or within 1 week. Four patients were pain-free at follow up after more than 8 months. This study has not yet been duplicated, and the overall efficacy, safety, and durability of the procedure are as yet unknown. However, it is a non-invasive procedure with fewer side-effects than ablative surgery and may represent a viable alternative in some patients prior to a destructive procedure.
Microvascular decompression of the trigeminal nerve with or without microvascular decompression or section of the nervus intermedius was recently reported to be effective in chronic cluster headache by Lovely et al. (127). In their series, 28 patients, two of whom had bilateral cluster headache, underwent 39 operations for microvascular decompression of the trigeminal nerve, alone or combined with microvascular decompression or section of the nervus intermedius. Twenty-two of the 30 first-time procedures resulted in 50% relief or better, but long-term follow-up (average 5.3 years) showed a fall in good-to-excellent relief to 46%. Repeat procedures were ineffective. Three patients who responded with pain reduction of > 50% with microvascular decompression of the trigeminal nerve improved to better than 90% after a microvascular decompression or section of the nervus intermedius. These procedures require the skill of a very experienced surgical team and require a craniotomy. Further experience with this combined approach is needed.
Finally, a number of authors have reported that patients with refractory chronic cluster headache have benefited from section of the sensory trigeminal nerve at the root exit zone (123, 128–130). Trigeminal sensory rhizotomy via a posterior fossa approach was employed by Onofrio & Campbell (123) in 10 patients, with pain relief in six and failure in four. In the largest series, Kirkpatrick et al. (130) reported complete or near complete pain relief in 12 of 14 patients who underwent a sensory trigeminal rhizotomy using a similar approach. The mean duration of follow-up was 5.6 years. One patient developed a contralateral recurrence of attacks. Seven patients who had a partial nerve root section required a second procedure for complete resection. A complete section was more likely to provide relief than a partial section, but total loss of sensation of all three divisions of the trigeminal nerve did not guarantee relief from the attacks.
Natural history
Information regarding the natural history of cluster headache is relatively scarce. Manzoni (45) evaluated 189 consecutive patients who had cluster for over 10 years. Approximately 13% of patients with episodic cluster headache converted to a chronic pattern, while another 6% evolved into an intermediate pattern. The remaining 80% continued to have episodic cluster headache. Chronic cluster headache evolved into an episodic or intermediate pattern in 33% and 14%, respectively. The prognosis in patients with chronic cluster headache appears to be favourable, with over one-third reverting to an episodic pattern of attacks. It is unclear whether pharmacotherapy contributes to this conversion, but some authors feel that lithium carbonate in particular may play a role in this transformation.
Kudrow (131) found that approximately one-third of patients who had cluster headache for 20 years or longer experienced a complete remission. In another third, the attacks lessened in severity so as to obviate the need for medication, while in the final third, the pattern of attacks remained unchanged. A Japanese study of 68 cluster headache patients with a mean cluster headache history of 18.9 years, followed over an 18-year period, found no change in headache attack severity, frequency, or duration or associated symptoms (132). There was no change in cluster period onset or duration. However, in all but five patients, the remission periods lengthened from a mean of 1.1 years to a mean of 3.3 years. This change appeared to occur most commonly around the age of 42 years.
Cluster headache is a lifelong disorder in the majority of patients. In a substantial proportion of patients, a complete remission or a lengthening of remission phases may occur, while in one-third of patients, chronic cluster may evolve into an episodic pattern.