Abstract
Despite several reports on symptomatic cluster-like headache, there is no clear explanation of how different lesions thought to be causative are related to cluster-like headache. On the basis of two additional cases of symptomatic cluster headache, we discuss the possibility that an acute imbalance of the autonomic nervous system, namely a net overactivity of the parasympathetic system, may be able to trigger these headache attacks in patients who probably have an additional individual predisposition to react with a cluster-like headache. Such an imbalance can be due to an increase in parasympathetic tone (e.g. stimulation of parasympathetic fibres) or to a reduction of the sympathetic tone (e.g. a lesion of the sympathetic fibres).
Keywords
Introduction
The number of reports on symptomatic cluster-like headache has recently increased. The suspected causative lesions were often, but not exclusively located in the proximity of the internal carotid artery (1–8), where aneurysms, arteriovenous malformations, inflammation (e.g. aspergilloma, multiple sclerosis), internal carotid artery dissection and cavernous sinus metastasis were found. Other lesions involved the middle cerebral fossa, the cerebellum or the trigeminal nucleus (1, 5, 7, 9). It is still not clear how these lesions are related to the symptoms of cluster-like headache; however, several authors have proposed that the autonomic nervous system is involved. Indeed, in the superior pericarotid cavernous sinus plexus, fibres from the first trigeminal division as well as sympathetic fibres from the superior cervical ganglion and parasympathetic fibres from the greater petrosal nerve to the sphenopalatine ganglion are located next to each other. It has also been shown that idiopathic cluster headache is related to central or peripheral autonomic dysfunction (10–12), and in some patients there is a persisting sympathetic deficit after the cluster attacks (10, 13). Imaging studies have demonstrated changes in the posterior hypothalamus, an area also important for autonomic regulation (14), and evidence for the involvement of parasympathetic fibres has been reported (10). It is still not known whether activation of autonomic fibres starts the attack or if the parasympathetic activation observed during the attack is a result of primary activation of the trigeminovascular system and secondary activation of trigemino-parasympathetic reflexes.
Two patients presented to our clinic with cluster-like recurrent paroxysmal headache attacks. In one case the attacks were caused by a meningitis carcinomatosa and, in the other, by a spontaneous dissection of the internal carotid artery. In relation to these patients, we discuss how a primary autonomic dysfunction might trigger a cluster headache attack.
Case histories
A 60-year-old man reported having paroxysmal headache attacks lasting 30–90 min and occurring several times a day (two to five times), mostly in the evening and early at night. They had been accompanied by a very severe periorbital pain on the left side during the previous 5 weeks. The attacks always started with intense lacrimation, followed by sweating of the ipsilateral forehead and, some minutes later, a sharp pain in the orbita. If a miosis was present at this moment, it was not noticed by the patient or his spouse. The attacks responded well to intranasal zolmitriptan and subcutaneous sumatriptan. In between the attacks the patient complained of only a light, dull, persisting frontal headache (visual analogue scale 2/10). Clinically, paresis of the lateral rectus eye muscle on the left side had been present for 4 months due to an intraorbital metastasis of a prostatic carcinoma. The prostatic carcinoma had been present for 3 years; lymphatic as well as bone metastasis had been detected 2 years previously. Over the last 2 years the patient had received various chemotherapies for the carcinoma. At the time of admission to our clinic, he had an isolated paresis of the left lateral rectus eye muscle and a hypaesthesia/hypalgesia of the second and third trigeminal nerves on the left side, as well as weakness of the left masseter muscle, indicating involvement of the sphenopalatine fossa. There were no other neurological signs present at this time. Magnetic resonance imaging (MRI) as well as computed tomography scans revealed an orbital metastasis involving the left lateral rectus muscle, contrast enhancement and thickened meninges left frontally and left frontal basally with an oedema of the left frontal white matter. Furthermore, there were signs of a frontal sinusitis (Fig. 1). In the cerebrospinal fluid the protein level was elevated 2.5-fold and prostate-specific antigen was highly increased, providing additional support of a meningeosis due to the prostatic cancer. There were signs of inflammation with elevated C-reactive protein in the blood and other signs of the tumour disease. The patient was initially treated with antibiotics and 100 mg prednisolone daily, which immediately stopped the pain. When the cortisone dosage was reduced to 20 mg per day, the attacks started again. Despite radiation of the head and continuation of systemic chemotherapy, the patient died 6 weeks later. No autopsy was performed.
(a) T2-weighted transversal magnetic resonance images (1.5-T scanner). (b) T1-weighted contrast enhancement images of patient 1. The upper images show the left intraorbital metastasis (arrow) of the prostatic cancer and the lower images the frontal basal meningiosis as well as the sinusitis frontalis (arrows).
A 35-year-old, otherwise healthy man reported that he had developed a syncope with a fall and a short amnesia while attending his wife during the birth of his second child. In the days following he had awoken in the night due to an intense periorbital pain attack on the left side that lasted about 2 h. Normal analgesics did not lessen the pain. The patient also noticed touch allodynia on the ipsilateral forehead, an increase of the ipsilateral temperature by about 1°C in the outer meatus acusticus compared with the contralateral meatus (measured by a common electrical thermometer) and nasal congestion during the attacks. When further attacks occurred irregularly in the evening and night over the next days, the patient went to the emergency ward. There he was seen by an otolaryngologist who diagnosed sinusitis. Although treatment with antibiotics was initiated, the symptoms did not improve. The patient was also seen by a neurologist, who diagnosed cluster headache, and the attacks were treated with zolmitriptan tablets, which only slightly reduced the pain during the attack. No other triptan was used; other analgesics such as non-steroidal anti-rheumatics were not effective.
The patient was seen again 3 weeks after the first attack, at which time he complained of one to two attacks every second night. No mention was made any triggering of the attacks by alcohol or smoking. The neurological examination was unremarkable except for Horner's syndrome with miosis and slight ptosis on the left side. Since Horner's syndrome was present at the beginning of the first cluster headache, MRI was performed and revealed a cervical dissection of the internal carotid artery up to the siphon on the symptomatic side (Fig. 2). The patient was treated with anticoagulants (heparin intravenously) and since that time has suffered no further attacks. The only vascular risk factor present was nicotine abuse of two packs of cigarettes daily for more than 7 years.
(a) T2-weighted transversal images (1.5-T scanner) showing the typical semicircular hyperdensity (arrows) in the carotid wall indicating a haematoma in the vessel wall. (b) Reconstruction of the time of flight magnetic resonance imaging angiography at the same time demonstrates narrowing of the internal carotid artery (arrow).
Discussion
Both patients initially had attacks of acute recurrent periorbital pain with autonomic signs. This picture meets International Headache Society diagnostic criteria for cluster headache attacks. The history and the further clinical symptoms of patient 1 suggested a diagnosis of a symptomatic case. The second case was not as clear, although the persisting Horner syndrome at first manifestation of a primary cluster headache is uncommon. Although patient 1 was clearly a symptomatic case, he responded well to the ‘classic’ cluster headache therapy with triptans and cortisone. Such a positive response of symptomatic cluster headache patients to a classic treatment has been reported previously (15). Patients with symptomatic cluster headache due to dissection of the internal carotid artery (2, 4) or due to metastasis of the cavernous sinus have also responded well to this therapy (6).
The medical histories of our patients suggest something about the possible pathophysiology of symptomatic cluster-like headache. The first patient clearly stated that before the pain started there was always intense lacrimation and sweating on the symptomatic side; this indicates there was first a parasympathetic activation or local irritation of sympathetic fibres in the greater superficial petrosal nerve. Parasympathetic overactivity of the greater superficial petrosal nerve has also been discussed in primary cluster headache (13). It is not clear if, at the same time, there was a miosis as a sign of sympathetic dysfunction. The second patient noticed that the temperature of his ear on the symptomatic side increased during the attack compared with the opposite side; this is a sign of vasodilation in this area, probably due to loss of sympathetic vasoconstriction of the vessels in the area of the outer meatus acusticus.
Thus, in the first patient there was some indication of primary parasympathetic activation. Although it is not clear what triggered this parasympathetic activation, we suspect that in the first patient the reported infiltration of the sphenopalatine fossa due to the metastasis of the prostatic cancer might have induced this over-activation. In a recently published, somewhat similar case of a metastasis of cavernous sinus (6), a lesion of the sympathetic fibres travelling along the carotid artery was more likely to be the cause. Patient 1 did not show any sign of a permanent sympathetic deficit of the eye. A sympathetic deficit, however, was present in the second patient due to the carotid artery dissection, since here clearly the sympathetic fibres located in the wall of the artery were lesioned by the haematoma in the vessel wall.
In both situations there was relative over-activation of the parasympathetic system of the face, due either to over-activation of the parasympathetic system or to loss of sympathetic efferences to the face. The two patients reported having attacks mostly in the evening and at night, a time when the central parasympathetic system is normally more active than the sympathetic system. On the basis of the observation of a persisting deficit of the cervical sympathetic pathway to the face in a subgroup of cluster headache patients (10), Drummond has proposed that an autonomic disturbance could contribute to the rapid escalation of pain in cluster headache attacks, but surmised that the attack was the consequence of a primary trigemino-parasympathetic discharge (10). In a symptomatic cluster-like headache, as in our patients, it seems possible that the autonomic drive could be the primary trigger for an attack with subsequent activation of the trigeminovascular system. The effectiveness of treatment with sphenopalatine endoscopic ganglion blockage for primary cluster headache may support this interpretation (16), since the parasympathetic efferences to the face can be blocked by such treatment, suggesting that the trigeminal activation alone is not sufficient to induce the typical pain. Otherwise, Matharu and Goadsby have described a patient who experienced typical cluster headache attacks including autonomic signs despite complete ipsilateral sensory trigeminal root section (17). Thus, peripheral trigeminal activation is not necessary for the development of typical cluster symptoms. However, this does not exclude the possibility that peripheral activation of nociceptive afferents reaches central trigeminal structures by other pathways (e.g. the nervus intermedius). The recently described case reports with typical autonomic signs of cluster headache (18, 19), but without typical pain, also demonstrate that autonomic activation alone is not enough to induce the typical pain in cluster headache and that there is additional activation of the trigeminal nuclei, due to peripheral activation of trigeminal afferences or due to central pathways within the brainstem.
As regards our patients, we suggest that it is not the absolute activation level of the parasympathetic system alone, but rather the imbalance between the activation levels of the parasympathetic and sympathetic systems of the face, as well as the individual threshold for triggering the parasympathetic trigeminal reflexes, which determines whether a cluster-like headache occurs. This individual threshold for the activation of the trigeminal nerve may explain why some patients with carotid dissection do not develop a cluster-like headache. The therapy with anticoagulants (e.g. heparin in our patient) might have affected the sympathetic lesion in patient 2 with the carotid dissection by helping to dissipate the haematoma in the wall of the carotid artery. It was no longer visible in a repeat MRI 6 months later (not shown here). The treatment might have helped establish a new balance within the autonomic system.
Acknowledgements
The thank Mrs J. Benson for copyediting the manuscript, and the anonymous reviewers for their helpful comments. A.S. was supported by a grant of the BMFB in Germany (chronic pain).