Cluster headache is a clinically well-defined disorder in which patients suffer extremely painful headaches with clock-like regularity one to three times or more a day for perhaps several months, usually followed by a period of remission (Headache Classification Committee of the International Headache Society, 1988). Cluster headache has been recognized for at least 250 years (Isler, 1993), yet until recently its mechanism has been poorly understood. Closely related conditions include paroxysmal hemicrania and the SUNCT syndrome, which also feature relatively short-lasting headaches with evidence of activation of the cranial parasympathetic outflow (Goadsby and Lipton, 1997).
The three major aspects of the pathophysiology of cluster headache are as follows: the trigeminal distribution of the pain, the autonomic features, and the episodic pattern of the attacks (which are in many ways the defining clinical signature of the disorder compared to migraine) (Lance and Goadsby, 1998).
These features raise the classic issues of the location of the lesion and the generic terminology that should be applied to these and related headaches, such as migraine. The classic observation during angiography of a patient suffering an acute cluster headache demonstrating changes in the internal carotid artery (Ekbom and Greitz, 1970) suggested a pathological focus in the region of the cavernous sinus.
The arguments for this locus for the disease have been set out elsewhere (Hardebo, 1994; Moskowitz, 1988). By outlining what is known about cluster headache, I will make an argument for it to be considered as a central nervous system disorder involving the posterior hypothalamic circadian cycling mechanisms.
Figure 5 – 3 Activation of the rostral brain stem in patients with spontaneous migraine has been demonstrated using positron emission tomography. This region (red area) was active during the acute attack, and activation persisted after successful treatment but was not present between attacks (Weiller et al., 1995). These findings suggest that there are brain stem regions that play a pivotal role in either initiation or termination of the acute attack of migraine. Indeed, migraine may well be a defect of normal control mechanisms for suppressing input because similar regions in the experimental animal gate trigeminal nociceptive information (Knight and Goadsby, 1999). Key to abbreviations: DRN, dorsal raphe nucleus; LC, locus coeruleus; NRM, nucleus raphe magnus; SSN, superior salivatory nucleus. (Reproduced with permission from Neurology Ambassador Programme, American Headache Society.)
Perhaps the key feature of cluster headache is the cycling of short-lasting attacks of pain or cycling between bouts of pain and periods of complete freedom from attacks. Kudrow (1976, 1977) was the first to point out that testosterone levels were altered in cluster headache patients during bouts; this implicated hypothalamic dysfunction. Nelson (1978) could not confirm this result.
Leone and colleagues (1990) identified reduced responses to stimulation by thyrotropin-releasing releasing hormone, and there are interesting observations of disordered circadian rhythm for cortisol, growth hormone, lutenizing hormone, and prolactin (Chazot et al., 1984; Leone and Bussone, 1993; Leone et al., 1995). One area involved in human clock systems is the suprachiasmatic nucleus in the hypothalamic gray, which sits at the base of the third ventricle (Moore-Ede, 1983).
Melatonin is produced by the pineal gland and has a strong circadian rhythm that is regulated by the suprachiasmatic nucleus (Moore, 1997). Connections between the retina and the hypothalamus are thought to provide light cues for the circadian rhythm (Hofman et al., 1996). The characteristic nocturnal peak of melatonin secretion is blunted during the active phase of cluster headache (Waldenlind et al., 1987). Neuroendocrine changes in cluster headache have, therefore, implicated this region of the brain in the pathophysiology of the disorder.
Cluster headache attacks can be triggered with nitroglycerin spray (Ekbom, 1968), and recent PET scan studies in patients with cluster headache have taken advantage of this fact, although the pattern of activation in the brain is no different from that of spontaneous attacks (May et al., 2000). The areas that were observed to be activated fell into three categories: areas generally associated with pain, an area that seems specific to cluster headache, and vascular structures (May et al., 1998a).
The anterior cingulate was significantly activated during acute cluster headache, as would be expected since in most human PET studies with pain, activation of the anterior cingulate is observed, perhaps as a part of the affective response (Derbyshire et al., 1997). Activation was also observed in the frontal cortex and insulae and the thalamus contra-lateral to the side of the pain. In addition, there was activation in the ipsilateral basal ganglia. This is not the first observation of basal ganglia changes associated with pain (Chudler and Dong, 1995; Derbyshire et al., 1997), and it may simply relate to movement or the wish to move, which is common in cluster patients, or even some deliberate inhibition of movement.
The only activated area that is particular to cluster headache is the ipsilateral hypothalamic gray. This is in contrast to the results of Hsieh et al. (1996), who did PET scans on four cluster headache patients, two with right-sided and two with left-sided headaches. They observed cingulate cortex and frontal activation but not hypothalamic activation or thalamic activation. The data were not mirrored in that study. May and colleagues (1998a) observed ipsilateral activation at the base of the third ventricle in the hypothalamic gray. This area is of obvious interest because of its role in the control of the circadian rhythm of neurons, as outlined above. Moreover, voxel-based morphometry, a method of comparing sets of brains for subtle differences, demonstrated a small enlargement of the grey matter region of the posterior hypothalamus in patients with cluster headache (May et al., 1999a), which is totally consistent with the functional imaging data.
In the PET study of acute cluster headache, pooling of tracer in the region of the cavernous sinus/internal carotid artery was noted. This suggests vasodilation and, in view of classical ideas concerning cluster headache pathophysiology, is dealt with below, specifically in the vascular section. Suffice to say that pooling is in no way specific to cluster headache and suggests that the disorder is neurovascular, not primarily vascular, in nature.
Department of Neurology, Centre Hospitalier Universitaire Vaudois, Lausanne , Switzerland;
Department of Neurology, Medical University of Warsaw, Warsaw , Poland
Cerebrovasc Dis 2006;22:91–100