Hypothalamic activation in cluster headache attacks

Arne May, Anish Bahra, Christian Büchel, Richard S J Frackowiak, Peter J Goadsby

University Department of Clinical Neurology (A May MD, A Bahra MRCP, P J Goadsby MD), and Wellcome Department of Cognitive Neurology (C Büchel MD, R S J Frackowiak FRCP), Institute of Neurology, National Hospital for Neurology and Neurosurgery, Queen Square, London, UK

Correspondence to: Dr Arne May, Institute of Neurology, National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK (e-mail: amay@ion.ucl.ac.uk)

Summary
Introduction
Methods and patients
Results
Discussion
References

Background Cluster headache, one of the most severe pain syndromes in human beings, is usually described as a vascular headache. However, the striking circadian rhythmicity of this strictly half-sided pain syndrome cannot be readily explained by the vascular hypothesis. We aimed to assess changes in regional cerebral blood flow (rCBF) in patients with cluster headache. Methods We used positron emission tomography (PET) to assess the changes in rCBF, as an index of synaptic activity, during nitroglycerin-induced cluster headache attacks in nine patients who had chronic cluster headache. Eight patients who had cluster headache but were not in the bout acted as a control group. Findings In the acute pain state, activation was seen in the ipsilateral inferior hypothalamic grey matter, the contralateral ventroposterior thalamus, the anterior cingulate cortex, and bilaterally in the insulae. Activation in the hypothalamus was seen solely in the pain state and was not seen in patients who have cluster headache but were out of the bout. Interpretation Our findings establish central nervous system dysfunction in the region of the hypothalamus as the primum movens in the pathophysiology of cluster headache. We suggest that a radical reappraisal of this type of headache is needed and that it should in general terms, be regarded as a neurovascular headache, to give equal weight to the pathological and physiological mechanisms that are at work.

The pain of cluster headache is perhaps the most severe known to human beings. Women who have such headaches describe each attack as being worse than childbirth. The syndrome is clinically well defined¹ and despite its recognition in published work for more than two centuries² its pathophysiology is poorly understood. The excruciatingly severe one-sided pain is likely to be mediated by activation of the first (ophthalmic) division of the trigeminal nerve, whereas the autonomic symptoms are a result of activation of the cranial parasympathetic outflow from the VIIth cranial nerve.³ The relapsing-remitting course,⁴ its seasonal variation,⁴ and the clockwise regularity⁵ are characteristic but unexplained features of the disorder. The striking circadian rhythmicity of cluster headache has led to the suggestion of a central origin for its initiation.^6,7 Substantially lowered concentrations of plasma testosterone during the cluster headache period in men provided the first evidence of hypothalamic involvement in cluster headache.⁸ This finding was further supported by a reduced response to thyrotropin-releasing hormone⁹ and a range of other circadian irregularities that have been reported in patients who have cluster headaches.¹⁰ Melatonin is a marker of the circadian system and a blunted nocturnal peak melatonin concentration and complete loss of circadian rhythm have been reported in patients who have cluster headache.¹⁰ The endogenous circadian rhythm is run by an oscillator in the suprachiasmatic nuclei in the ventral hypothalamus and reacts to temporal environmental cues of light conditions via a retino-hypothalamic pathway. The hypothalamus, or a closely related structure, is a candidate site for triggering the acute attack of cluster headache. Positron emission tomography (PET) is probably the best technique for visualising in-vivo changes in regional cerebral blood flow (rCBF) in human beings. Modern high-resolution PET allows the detection of subtle changes in rCBF during defined behavioural tasks and provides an index of synaptic activity relating networks of regions to tested brain functions.¹¹ Cluster headache attacks can be elicited with nitroglycerin during the active period without significant side effects.⁵ Nitroglycerin-provoked and spontaneous cluster attacks are comparable^3,12 and nitroglycerin does not substantially alter rCBF.¹³ The headache can be rapidly and effectively aborted with sumatriptan. This approach was therefore used to detect brain regions with increased blood flow during nitroglycerin-induced cluster attacks, focusing our interest on the hypothalamic region.

Nine right-handed men (age 2562 years, mean 43 years) with active chronic cluster headache, according to the Headache Classification Committee of the International Headache Society,¹ were studied during an induced acute headache attack (study group). Acute cluster headache was provoked by inhalation of nitroglycerin (1·01·2 mg).⁵ All patients studied were not treated prophylactically for cluster headache and were otherwise healthy. Eight patients with a history of cluster headache but who were not having a headache bout (age 3661 years, mean age 49 years) had a PET scan by the same study design (control group). None of the control patients had a cluster headache attack after nitroglycerin was applied. Informed consent was obtained from all patients and the study was approved by the Ethics Committee of the National Hospital for Neurology and Neurosurgery, London.

During the active headache period each of the nine study patients had 12 or 13 consecutive scans at four times: baseline; after application of nitroglycerin; after onset of headache; and when headache-free, after treatment with subcutaneous sumatriptan (6 mg). Each of the eight controls had 12 consecutive scans by the same design. Since none of the patients in this group had a cluster headache attack after taking the nitroglycerin, we defined scans for the second and third condition according to the mean number of scans for these conditions in the study group. For each scan, patients rated their headache intensity with a visual analogue scale (0=no pain, 10=the most severe pain). Participants had their eyes closed during all scans.

PET scans were done with an ECAT EXACT HR+ scanning system (CTI Siemens, Knoxsville, TN, USA) in three-dimensional mode with septa retraced. An antecubital vein cannula was used to administer the tracer, about 350 mBq of H₂¹⁵O. The activity was flushed into patients over 20 s at a rate of 10 mL/min. The data were acquired in one 90 s frame beginning 5 s before the peak of the head curve. The interval between scans was 815 min. Attenuation correction was done with a transmission scan done at the beginning of each study. Images were reconstructed by filtered-back projection into 63 images planes (separation 2·4 mm) and into a 128 by 128 pixel image matrix (pixel size 2·1×2·1 mm²). Statistical Parametric Mapping 97 (SPM'97; http://www.fil.ion.ucl.ac.uk/spm) was used for data analysis. Images were realigned with the first image as the reference and then coregistered with the patient's structural magnetic resonance imaging (MRI) image and finally spatially normalised into the space defined by the atlas of Talairach and Tournoux.¹⁴ The normalised images were smoothed with a Gaussian filter of 10 mm full width at half maximum. Statistical parametric maps were derived with pre-specified contrasts,¹⁵ to compare rCBF during headache versus rCBF during the non-headache phase after nitroglycerin application. We also addressed the question of significant rCBF differences in the study group relative to the control group with a group-by-condition interaction analysis. Because headache is a strictly lateralised syndrome¹ we mirrored PET and MRI scans in patients with right-sided headache in the sagittal plane to be able to analyse all patients in the same analysis. The uncorrected threshold of p<0·001 was chosen because of a strong regional a-priori hypothesis based on the clinical and experimental data cited in the text.

Activated brain region*	Brodman	Talairach co-ordinates (mm)			Z score of

	area	x	y	z	peak

					activation

Left hypothalamus		-2	-18	-8	3·68

Right thalamus		-6	-12	-6	5·01

Right cingulate cortex	24	-2	-22	-24	4·9

Right frontal lobe	10	-26	-54	-6	4·06

Left primary motor area	6/44	-40	-2	-32	3·29

(face)

Right insula	13	-32	-10	-2	4·28

Left insula	13	-40	-12	-8	4·37

Left basal ganglia		-20	-12	-2	4·22

Cerebellum (vermis)		-2	-38	-10	3·09

Each location is the peak within a cluster (defined as the voxel with the highest

Z-score). *p<0·001 for all regions.

Increases in blood flow during an induced attack of acute cluster headache compared with the pain-free state

Of the nine patients in the bout, five experienced a cluster headache attack on the left side and four patients on the right side after nitroglycerin spray. Typical concurrent autonomic symptoms such as ipsilateral miosis, lacrimation, and rhinorrhoea confirmed the presence of a classic cluster headache attack. All patients described the provoked attack as being similar to spontaneous attacks. In the nine patients who had attacks of acute cluster headache, significant activations in the acute attack compared with the headache-free state were found in the ipsilateral hypothalamic grey area, bilaterally in the anterior cingulate cortex, in the contralateral posterior thalamus, the ipsilateral basal ganglia, bilaterally in the insulae (figures 1 and 2), and in the cerebellar hemispheres (table). Figure 1 shows that significant activation was detected next to the third ventricle slightly lateralised to the left and rostral to the aqueduct. The activation is ipsilateral to the pain side, lies in the diencephalon, and coincides, in the Talairach atlas,¹⁴ with the hypothalamic grey matter. Figure 2 shows that significant activation was detected in the right frontal lobe (Brodman's area 10), bilaterally in the insula, in the cerebellum/vermis and in the hypothalamic grey matter. The activation in the hypothalamic grey area was seen only in the patients with a cluster headache attack but not in the control patients (p<0·001).

Figure 1: Comparison of nitroglycerin-induced acute cluster-headache attack and rest (no pain) in nine patients with active chronic cluster headache

Activations during the attack are shown as statistical parametric maps that show the areas of signfiicant rCBF increases (p<0·001) in colour superimposed on an anatomical reference derived from a T1-weighted MRI. The left side of the picture is the left side of the brain.

Figure 2: Comparison of nitroglycerin-induced acute cluster headache attack and rest (no pain) condition in nine patients with active chronic cluster headache

The activations during the attack are shown as statistical parametric maps which show the areas of significant rCBF increases (p<0·001) in colour superimposed on an anatomical reference derived from a T1-weighted MRI. The anterior part of the brain corresponds to the right side of the picture, the posterior parts to the left side, the left side of the brain corresponds to the top of the image, and the right side to the bottom.

We further confirmed this result with a group (study group vs control group) by condition (headache-no headache) interaction (p<0·001). The difference in rCBF for the hypothalamic grey area comparing headache and headache-free conditions was significantly greater for the study than for the control group (figure 3).

Figure 3: Condition by group interaction

*p<0·001 for rCBF increase in the hypothalamic area comparing headache and headache-free conditions.
rCBF values are based on average grand mean set to 50 mL
100 g¹ min¹.

We observed areas of activation in acute cluster headache that fall into two broad groups: areas known to be involved in pain processing or response to pain, such as cingulate and insula cortex and thalamus; and areas activated specifically in cluster headache but not in other causes of head pain, notably the hypothalamic grey areas. These data suggest that primary headache syndromes share some processing pathways but equally can be distinguished on a functional neuroanatomical basis by areas of activation specific to the clinical presentation. Studies with PET have repeatedly given results that show activation of the anterior cingulate cortex on the sensation of somatic or visceral pain that are attributed to the emotional response to pain.^13,16,17 Activations in the insula have been shown after application of heat,^16,18 subacutaneous injections of ethanol,¹⁹ somatosensory stimulation,²⁰ and during cluster headache.¹³ Given its anatomical connections, the insula has been suggested as a relay of sensory information into the limbic system and is known to play an important part in the regulation of autonomic responses.²¹ Painful stimuli are significantly effective in activating the anterior insula, a region closely associated with both somatosensory and limbic systems. Such connections may provide one route through which nociceptive input is integrated with memory to allow full appreciation of the meaning and dangers of painful stimuli. In the acute pain state the thalamus is a site where activations would most be expected. Activation of the contralateral thalamus as a result of pain is known from studies on animals²² and functional imaging studies in human beings.^16,17 The acute pain in cluster headache, induced activation bilaterally in the cerebellar hemispheres and in the vermis. There seems to be no direct nociceptive input to the cerebellum,²³ and there is no clinical evidence that cerebellar lesions or stimulation affect pain sensation in human beings.¹⁶ However, there are some PET studies that report an activation in this area during experimental pain.^16,24 In contrast to migraine,²⁵ no brain stem activation was found during the acute attack compared with the resting state. This finding is remarkable because migraine and cluster headache are often discussed as associated disorders and similar compounds, such as ergotamine and sumatriptan, are used in the acute treatment of both types of headache. These data suggest that while primary headaches, such as migraine and cluster headache, may share a common pain pathway (the trigeminovascular innervation), the underlying pathogenesis differs substantially as might be inferred from the different patterns of presentation and responses to preventive agents.²⁶ Substantial activations ascribable to cluster headache were observed in the ipsilateral hypothalamic grey area when compared with the headache-free state. Just as it is striking that no brain-stem activation occurs, which is in contrast to acute migraine,²⁵ we have seen no hypothalamic activation in experimental pain induced by capsaicin injection into the forehead.²⁷ Injection into the forehead would activate first division (ophthalmic) afferents which traverse the trigeminal division responsible for pain activation in cluster headache. Thus two other types of first division trigeminal nerve pain, while sharing neuroanatomical pathways with cluster headache, do not give rise to hypothalamic activation. Moreover, in the eight control patients who did not experience a headache after taking nitroglycerin, rCBF in the region of the hypothalamic grey area was not increased. This finding implies that the activation we have observed is involved in the pain process in a permissive or triggering manner rather than simply as a response to first division nociception per se. Hypothalamic activation in traumatic nociception has been observed in the hypothalamus proper and is a different more rostral area than we report.¹⁹ Moreover, Hsieh and colleagues¹⁹ report changes contralateral to the pain, whereas we report changes that are ipsilateral and in the hypothalamic grey area in the region of the circadian pacemaker neurons which is, therefore, an anatomically distinct area on the opposite side of the brain. Given that this area is involved in circadian rhythm and sleep-wake cycling, our data establish an involvement of this area of the hypothalamus as a primum movens in the acute cluster attack. Cluster headache has been attributed to an inflammatory process in the cavernous sinus and tributary veins.²⁸ Inflammation has been thought to obliterate venous outflow from the cavernous sinus on one side, thus injuring the traversing sympathetic fibres of the intracranial internal carotid artery and its branches. According to this theory, the active period ends when the inflammation is suppressed and the sympathetic fibres partially or fully recover. This theory is based on abnormal findings with orbital phlebography in patients with cluster headache,²⁹ and the fact that nitroglycerin and other vasodilators can induce a cluster attack.⁵ However, given the circadian rhythmicity and unilaterality of the symptoms, a purely vasogenic cause cannot explain the entire picture of cluster headache.³⁰ Moreover, the frequency and pattern of pathological findings at orbital phlebography in cervicogenic headache, migraine, and tension-type headache is similar to that in cluster headache.³¹ Given that we have found an increased signal in the region of the cavernous sinus in the patients with acute cluster headache in this study and after capsaicin injection to the forehead in another PET study,²⁷ it seems likely that the vascular changes are an epiphenomenon of activation of the trigeminovascular system.³² A radical reappraisal of the pathophysiology of cluster headache is needed. Our data establish that cluster headache, far from being a primarily vascular disorder, is a condition the genesis of which is to be found in the central nervous system in pacemaker or circadian regions of the hypothalamic grey matter. Further, we suggest that both cluster headache and migraine might usefully be regarded are neurovascular headaches to include the neural contribution to these important clinical syndromes.

All five investigators contributed to the design of the study and to the writing of the paper. Arne May was involved in planning, study coordination, and analysis. Anish Bahra was involved in recruiting the patients and scanning. Christian Büchel was involved in the statistical analysis. Richard Frackowiak and Peter Goadsby were involved in planning, study coordination, and review of the data.

The authors wish to thank the radiographers of the Functional Imaging Laboratory, Queen Square, for technical support. This work was supported by the Wellcome Trust and the Migraine Trust. AM is the International Headache Society Cluster Headache Research Fellow (Doppelfeld Stiftung); AB is a Zeneca Clinical Research Fellow; CB is a Wellcome Research Fellow; RSJF is a Wellcome Principal Research Fellow; and PJG is a Wellcome Senior Research Fellow.

1 Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia 1988; 8: 196. [PubMed] 2 Isler H. Episodic cluster headache from a textbook of 1745: Van Swieten's classic description. Cephalalgia 1993; 13: 17274. [PubMed] 3 Goadsby PJ, Edvinsson L. Human in-vitro evidence for trigeminovascular activation in cluster headache. Neuropeptide changes and effects of acute attacks therapies. Brain 1994; 117: 42734. [PubMed] 4 Kunkle EC, Pfieffer J,Wilhoit WM, Hamrick J. Recurrent brief headache in cluster pattern. Trans Am Neurol Assoc 1952; 27: 24043. [PubMed] 5 Ekbom K. Nitroglycerin as a provocative agent in cluster headache. Arch Neurol 1968; 19: 48793. [PubMed] 6 Ekbom K. Patterns of cluster headache with a note on the relations to angina pectoris and peptide ulcer. Acta Neurol Scand 1970; 46: 22537. [PubMed] 7 Kudrow L. The cylic relationship of natural illumination to cluster period frequency. Cephalalgia 1987; 7: 7678. [PubMed] 8 Nelson RF. Testosterone levels in cluster and non-cluster migrainous patients. Headache 1978; 18: 26567. [PubMed] 9 Leone M, Partuno G, Vescovi A, Bussone G. Neuroendocrine dysfunction in cluster headache. Cephalalgia 1990; 10: 23539. [PubMed] 10 Chazot G, Claustrat B, Brun J, Jordan D, Sassolas G, Schott B. A chronobiological study of melatonin, cortisol, growth hormone and prolactin secretion in cluster headache. Cephalalgia 1984; 4: 21320. [PubMed] 11 Frackowiak RS, Friston KJ. Functional neuroanatomy of the human brain: positron emission tomography--a new neuroanatomical technique. J Anat 1994; 184: 21125. [PubMed] 12 Fanciullacci M, Alessandri M, Figini M, Geppetti P, Michelaci S. Increases in plasma calcitonin gene-related peptide from extracerebral circulation during nitroglycerin-induced cluster headache attacks. Pain 1995; 60: 11923. [PubMed] 13 Hsieh JC, Hannerz J, Ingvar M. Right-lateralised central processing for pain of nitroglycerin-induced cluster headache. Pain 1996; 67: 5968. [PubMed] 14 Talairach J, Tournaux P. Coplanar stereotaxic atlas of the human brain. New York: Thieme, 1988. 15 Friston KJ, Holmes AP, Worsley KP, Proline JB, Frith CD, Frackowiak RS. Statistical parametric maps in functional imaging: a general linear approach. J Hum Br Map 1995; 2: 189210. [PubMed] 16 Casey KL, Minoshima S, Berger KL, Koeppe RA, Morow TJ, Frey KA. Positron emission tomography analysis of cerebral structures activated specifically by repetitive noxious heat stimuli. J Neurophysiol 1994; 71: 80207. [PubMed] 17 Rosen SD, Paulesu E, Frith CD, et al. Central nervous pathways mediating angina pectoris. Lancet 1994; 344: 14750. [PubMed] 18 Minoshima S, Morrow TJ, Koeppe RA, Casey KL. Involvement of the insular cortex in central autonomic regulation during painful thermal stimulation. J Cereb Blood Flow Metab 1995; 15: 859. 19 Hsieh JC, Stahle Backdahl M, Hagermark O, Stone Elander S, Rosenquist G, Ingvar M. Traumatic nociceptive pain activates the hypothalamus and the periaqueductal gray: a positron emission tomography study. Pain 1996; 64: 30314. [PubMed] 20 Burton H, Videen TO, Raichle ME. Tactile vibration activated foci in insular parietal opercular cortex studied with positron emission tomography. Somatosens Mot Res 1993; 3: 297308. [PubMed] 21 Mesulam MM, Mufson EF. The insula of Reil in man and monkey. Architectonics, connectivity and function. New York: Plenum, 1985. 22 Goadsby PJ, Zagami AS, Lambert GA. Neural processing of craniovascular pain: a synthesis of the central structures involved in migraine. Headache 1991; 31: 36571. [PubMed] 23 Ekerot CF, Garwicz M, Schouenborg J. Topography and nociceptive receptive fields of climbing fibres projecting to the cerebellar anterior lobe in the cat. J Physiol 1991; 441: 25774. [PubMed] 24 Hsieh JC, Belfrage M, Stone Elander S, Hansson P, Ingvar M. Central representation of chronic ongoing neuropathic pain studied by positron emission tomography. Pain 1995; 63: 22536. [PubMed] 25 Weiller C, May A, Limmroth V, et al. Brain stem activation in spontaneous human migraine attacks. Nat Med 1995; 1: 65860. [PubMed] 26 Goadsby PJ, Silberstein SD. Headache. New York: Butterworth-Heinemann, 1997. 27 May A, Kaube H, Buechel C, et al. Experimental cranial pain elicited by Capsaicin: a PET-study. Pain 1998; 74: 616. [PubMed] 28 Hardebo JE. How cluster headache is explained as an intracavernous inflammatory process lesioning sympathetic fibres. Headache 1994; 34: 12531. [PubMed] 29 Hannerz J, Ericson K, Bergstrand G. Orbital phlebography in patients with cluster headache. Cephalalgia 1987; 7: 20711. [PubMed] 30 Goadsby PJ, Lance JW. Brainstem effects on intra- and extracerebral circulations. Relation to migraine and cluster headache. In: Olesen J, Edrinsson L, eds. Basic mechanisms of headache. Amsterdam: Elsevier Science Publishers, 1988: 41327. 31 Bovin G, Jenssen G, Ericson K. Orbital phlebography: a comparison between cluster headache and other headaches. Headache 1992; 32: 40812. [PubMed] 32 Goadsby PJ, Duckworth JW. Effect of stimulation of trigeminal ganglion on regional cerebral blood flow in cats. Am J Physiol 1987; 253: 27074. [PubMed]