Due to its central role in rhythm regulation and integration of the autonomic nervous function, the hypothalamus has been hypothesized to be involved in CH pathogenesis (36). The temporal pattern of CH together with altered neuroendocrine, vascular, and pain control indicates that circadian and circannual rhythm regulation is affected. The cluster periods tend to occur at regular intervals, in a seasonal pattern, and they are sometimes interrupted by changing work schedules and sleep patterns. The “clockwise regularity” of CH attacks also appears to be governed by the dark-light cycle, sleep, and activity. Accordingly, there is an afternoon peak of CH attacks in Italy (28), which is not observed in Scandinavia (44), where there is no siesta during the work day. Neuroendocrine data show altered 24-hour secretory patterns with phase shifts/advances for a number of hormones during active periods but also during remission, as well as altered responses to various neuroendocrine tests (51). Twenty-four-hour data for certain hormones, blood pressure, body temperature, or pain sensitivity show that more patients than controls lack a significant circadian rhythm when the data are analyzed by cosinor rhythmometry (16). Some of the neuroendocrine changes may in part be related to pain, stress, or interrupted sleep; others, such as blunted nocturnal prolactin secretion during remission and permanently reduced melatonin urinary concentrations around the year, are interpreted to reflect central pathology and hypothalamic involvement in CH.

Functional neuroimaging with positron emission tomography (PET) and anatomic imaging with voxel-based magnetic resonance imaging (MRI) morphometry have made it possible to show directly that the posterior hypothalamic gray matter indeed is a key area in CH. A PET study using H₂¹⁵O as a marker showed increased activity in the ipsilateral hypothalamus during nitroglycerin-induced attacks of CH (32), but not during the cluster period between attacks or in a control group in remission after a negative nitroglycerin provocation. Recently, it was reported that hypothalamic activation also occurred in a patient during a spontaneous attack of CH (48). A similar activation pattern has been seen with functional MRI in a series of patients with short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT) (33) but not in migraine (53) or other spontaneous or experimentally induced cranial pain (35). Activation of this particular area in the hypothalamus appears to be unique to the trigeminal autonomic cephalalgias.

With voxel-based morphometric analysis of T1-weighted MRI scans it was possible to demonstrate an increase in gray matter volume in a small region coinciding with the inferior hypothalamus in right-handed men with CH compared with healthy controls (31). The gray matter change was bilateral, but when mirror images were used to normalize for pain side, the structural change seemed slightly lateralized to the pain side. Although conventional imaging was normal in all patients, the voxel-based morphometry was capable of showing subtle changes in the gray matter by averaging across subjects. The nature of the
structural changes is unknown. The activation of the inferior posterior hypothalamus and corresponding structural changes ask fundamental questions about what is a “primary” headache disorder and has inspired a completely new approach to treatment of intractable CH (15).

Deep brain stimulation of posterior hypothalamus ipsilateral to the side of attacks has been shown to improve otherwise drug-resistant chronic CH (26,52). Five patients with intractable chronic CH were treated with long-term, high-frequency electrical stimulation of the posterior hypothalamus ipsilateral to pain. In one patient with bilateral pain an additional contralateral implant was required to achieve pain relief. Electrodes were implanted 3 mm behind and 5 mm below the midcommissural point and 2mm lateral to the midline. After 2 to 22 months of follow-up, two of five patients had remained pain free without any medication, whereas three patients required low doses of methysergide or verapamil (26). Pain disappearance was never immediate but occurred from a few hours up to 4 weeks after starting stimulation, and the insertion of the electrode as such appeared not to affect pain attacks. In this series of patients there were no major adverse events recorded. Another study reported preliminary good results in four pilot patients but, unfortunately, a fifth patient died from an intracerebral hemorrhage (52). Since there was no immediate cessation of attacks by deep brain stimulation, a more complex mechanism involving several brain structures rather than simple inhibition or stimulation of hypothalamic nuclei was suggested. This would be consistent with the results of recent PET studies showing that hypothalamic stimulation activate certain brain areas and deactivate others (34).

A very recent publication reports a polymorphism of the hypocretin receptor 2 (HERTR2) gene to be associated with CH (43). Interestingly, this gene is primarily expressed in the hypothalamus. The study suggests that the HCRTR2 gene or a linked locus significantly modulates the risk for CH.

One of the main diagnostic criteria for CH attacks is the strict unilaterality, suggesting a locus, either constitutional or acquired, where pain is generated or to which it is referred. Rarely does a CH shift sides for a complete cluster period. The risk of having a cluster period on the previously asymptomatic side is about 200 times higher (46) than the overall incidence. This indicates an increased vulnerability in subjects already suffering from CH (18,21,39). Infections, trauma, or toxic effects have been put forward as possible initiators of CH. Studies of the disorder during the very first episode of CH may help to clarify the initiating etiologic event.

The maximum intensity of pain is generally localized behind the eye, radiating toward the temple or to the upper cheek. It is described as excruciating, as almost intolerable, and as if the eye is pushed out of the orbit or a knife is being turned around. During pain most patients appear restless or agitated.

The ophthalmic, anterior cerebral, and middle cerebral arteries dilate during attacks of CH (12,20,50). Vascular dilation together with lowering of the pain threshold by sensitization of pain receptors may contribute to pain. Vascular pain, however, would be expected to be throbbing, which is rather uncommon in CH (11). Alternatively, pain may be caused by dilated and edematous vessels pressing against surrounding tissues in narrow passages such as the bony carotid canal and the pterygopalatine fossa, or by obstructed venous outflow from the cavernous sinus (21). CH has been proposed to be caused by a remitting venous phlebitis in the cavernous sinus based on pathologic orbital phlebograms (18,19), and inflammatory signs in blood (18) and cerebrospinal fluid (22) in some patients during the cluster period. However, there was no consistent correspondence between phlebopathic signs and the symptomatic side, and similar phlebographic findings have been reported in healthy controls (1).