Article Text


Trigeminal autonomic cephalalgias: fancy term or constructive change to the IHS classification?
  1. P J Goadsby
  1. Correspondence to:
 Prof P J Goadsby
 Institute of Neurology, Queen Square, London WC1N 3BG, UK;

Statistics from

A classification based on pathophysiology is a useful aid to differential diagnosis and effective treatment planning

For the neurologist faced with the day to day grind of clinical work a change to terminology may seem like the academics “at it again”. I will try to set out this change and illustrate a physiology that may be attractive to understand, and hopefully one that enhances, clinical practice. Appreciating the physiology of the trigeminal-autonomic reflex can make patients presenting with varying degrees of cranial autonomic activation, such as lacrimation, conjunctival injection, nasal congestion or rhinorrhoea and the like, comprehensible at the bedside.1

The trigeminal autonomic cephalalgias (TACs) is a grouping of headache syndromes recognised in the second edition of the International Headache Society (IHS) classification.2 The term was coined to reflect a part of the pathophysiology of these conditions that is a common thread—that is, excessive cranial parasympathetic autonomic reflex activation to nociceptive input in the ophthalmic division of the trigeminal nerve.1 The TACs are classified in section III of the second edition of the classification,2 and include cluster headache,3 paroxysmal hemicrania, and short lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT).4 In an early draft, hemicrania continua was included5 but this was finally classified in section IV. I will briefly review the underlying physiology of the trigeminal-autonomic reflex that underpins these conditions and set out their classification and differential diagnosis. I will point out some limitations and some directions for future research. Their therapy is beyond the scope of the present paper, but it has been recently reviewed.4


Any pathophysiological construct for TACs must account for the two major shared clinical features characteristic of the various conditions that comprise this group: trigeminal distribution pain and ipsilateral cranial autonomic features.1 The pain producing innervation of the cranium projects through branches of the trigeminal and upper cervical nerves6,7 to the trigeminocervical complex8 from whence nociceptive pathways project to higher centres.9 A reflex activation of the cranial parasympathetic outflow provides the efferent loop.

Experimental studies

Stimulation of the trigeminal ganglion in the cat produces cranial vasodilation and neuropeptide release, notably calcitonin gene related peptide (CGRP) and substance P.10 The dilation is mediated by antidromic activation of the trigeminal nerve (20% of the effect) and orthodromic activation through the cranial parasympathetic outflow via the facial (VIIth) cranial nerve, for the other 80%.11 The afferent arm of the trigeminal-parasympathetic reflex traverses the trigeminal root,11 synapses in the trigeminal nucleus and then projects to neurones of the superior salivatory nucleus in the pons.12 There is a glutamatergic excitatory receptor in the pontine synapse13 and projection via the facial nerve14 without synapse in the geniculate ganglion. The greater superficial petrosal nerve supplies classic autonomic preganglionic fibres to the sphenopalatine (pterygopalatine in humans) and otic ganglia.15 The sphenopalatine synapse involves a nicotinic ganglion that is hexamethonium sensitive.15 VIIth cranial nerve activation is associated with release of vasoactive intestinal polypeptide (VIP)16 and blocked by VIP antibodies.17 Changes in the flow of blood in the brain depend on the frequency of stimulation18,19 and are independent of cerebral metabolism.20 There is VIP in the sphenopalatine ganglion,21 as well as nitric oxide synthase, which is also involved in the vasodilator mechanism.22

Human studies

The basic science work outlined above implies an integral role for the ipsilateral trigeminal nociceptive pathways in TACs and predicts in some patients cranial parasympathetic autonomic activation. The ipsilateral autonomic features seen clinically are consistent with cranial parasympathetic activation (lacrimation, rhinorrhoea, nasal congestion, and eyelid oedema) and sympathetic hypofunction (ptosis and miosis). The latter is likely to be a neurapraxic effect of carotid wall swelling23,24 with cranial parasympathetic activation. Some degree of cranial autonomic symptomatology is, therefore, a normal physiological response to cranial nociceptive input.25–27 Indeed other primary headaches, notably migraine,28 or patients with facial pain, such as trigeminal neuralgia,29 would be expected to have cranial autonomic activation, and they do. The distinction between the TACs and other headache syndromes is the degree of cranial autonomic activation, not its presence alone.30 This is why some patients with migraine have minor cranial autonomic activation that leads to the term cluster-migraine, when most such patients have migraine with cranial autonomic activation.

Permitting trigeminal-parasympathetic activation

What is the basis for the cranial autonomic symptoms being so prominent in the TACs? Is it due to a central disinhibition of the trigeminal-autonomic reflex?30 Functional imaging studies—positron emission tomography studies in cluster headache31–33 and a functional magnetic resonance imaging (MRI) study in SUNCT syndrome34—has demonstrated ipsilateral posterior hypothalamic activation. Posterior hypothalamic activation seems specific to these syndromes and is not seen in episodic35–37 or chronic38 migraine, or in experimental ophthalmic trigeminal distribution head pain.39 There are direct hypothalamic-trigeminal connections40 and the hypothalamus is known to have a modulatory role on the nociceptive and autonomic pathways, specifically trigeminovascular nociceptive pathways.41 Hence, cluster headache and SUNCT syndrome are probably due to an abnormality in the region of the hypothalamus (fig 1) with subsequent trigeminovascular and cranial autonomic activation. Imaging data with paroxysmal hemicrania are keenly awaited. Cranial autonomic features are not invariably linked with trigeminal pain and may persist after lesions of the trigeminal nerve.

Figure 1

 Brain imaging of two trigeminal autonomic cephalalgias. Changes in the posterior hypothalamic grey are revealed with (A) positron emission tomography in patients with chronic cluster headache31 and (B) with blood oxygen level dependent (BOLD)-functional MRI in a patient with short lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT) in whom multiple attacks were captured.34


The TACs need to be differentiated from secondary TAC producing lesions, from other primary headaches, and from each other. The differentiation from secondary causes is not a problem if one images patients but can be extremely difficult if one does not. An MRI of the brain with attention to the pituitary fossa and cavernous sinus will detect most secondary causes. It is easy to make an argument given the rarity of paroxysmal hemicrania and SUNCT that MRI would be a reasonable part of the initial work-up of such patients. It is more complex for cluster headache. There are no clear studies, and our impression from a cohort that now exceeds 400 (the National Hospital for Neurology and Neurosurgery, London) is that MRI would detect no more than 1 in 100 cases of lesions in episodic cluster headache, so we cannot recommend its routine use. For chronic cluster headache, an MRI seems reasonable given that very difficult nature of the long term management and developments in neuromodulation as a treatment.42

For other primary headaches, migraine is the single biggest problem in the differential diagnosis of cluster headache. Migraine can cluster and despite the best intentions of the IHS classification committee short attacks do occur. Cranial autonomic symptoms are well reported,28 and the neuropeptide changes are the same43 as in cluster headache.44 The occurrence of attacks together does not seem to have the seasonal preponderance that is so typical of cluster headache,45,46 and this can be a useful differential diagnostic feature. I regard the term cluster-migraine as unhelpful and I am yet to see a convincing case of a distinct biological entity usefully described by this name. The criterion for the effect of movement was added to cluster headache to sharpen the difference with migraine. The committee hoped this would draw attention to the fact that most cluster headache patients feel restless or agitated,47 whereas most migraine patients are quiescent, as IHS-I recognised.48

In clinical practice, this symptom, and the periodicity, are extremely helpful in differential diagnosis. The other feature of cluster headache, and this is a feature of TACs when compared with migraine, is that patients with TACs often complain of unilateral, homolateral photophobia, whereas patients with migraine more often complain of bilateral photophobia. Bilateral photophobia in patients with TAC could be speculated to occur in about 25% purely by the chance of them having some migrainous biology.

The TACs themselves (table 1) can often be differentiated by their attack length. This is certainly true when comparing cluster headache with SUNCT/short lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms (SUNA). The IHS criteria for TACs does betray an uncomfortable biological naivety with regard to the timing. The A, C, D, E/F criteria are rather similar for each TAC (tables 2–4). It seems neat in some way to have SUNCT be up to four minutes long, paroxysmal hemicrania from two to 30 minutes and cluster headache from 15 minutes onwards. The overlap seems minimal. It almost goes without saying that this must be wrong in absolute terms, biology rarely provides such neat rules, but it does provide a useful way to identify cases of sufficiently similarity to make biologically meaningful studies.

Table 1

 Clinical features of the trigeminal autonomic cephalalgias (TACs)

Table 2

 Cluster headache

Table 3

 Paroxysmal hemicrania

Table 4

 Short lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT)


The classification and biology of the TACs have come a long way in a short time. The syndromes are well established, and although rare compared with migraine they are sufficiently common, with cluster headache affecting about 0.2% of the population,49 to demand a neurological and headache specialist’s attention. There are some particular issues of classification that are not currently clear.

Cluster headache

A patient with a first attack of cluster headache is now simply classified as cluster headache (3.1). This takes the top-down view—that is, diagnose what you can and fill in the detail as available. Such cases are unsuitable for almost any study except natural history studies where they are ideally the starting point. A similar problem is how to refer to patients who have one type of TAC, typically an episodic form, and then evolve to the chronic form. The old classification differentiated primary from secondary chronic cluster headache depending on whether there was a period of episodic headache first. This argument would apply equally to chronic paroxysmal hemicrania. There seems little evidence that the clinical characteristics or therapeutic behaviour of primary or secondary chronic cluster headache are different, and the terminology secondary in headache parlance generally implies an underlying pathology. Moreover, the main clinical imperative when the timing alters would be review, perhaps with investigation, but this is a generic principle in headache management. For the moment the distinction has been dropped.

Paroxysmal hemicrania (PH)

The diagnosis of PH by the IHS criteria requires a response to indometacin. This is very difficult. It is not clear what the basis for the indometacin effect is, although it is perfectly clear that the effect is clinically very meaningful (table 5). Patients with PH who are treated with indometacin have an almost unbelievably spectacular resolution. This response seems so distinct that reserving the diagnosis of PH for these patients seems reasonable. Given varying sensitivity to indometacin, we have seen a requirement for a single dose given first thing in the morning of 300 mg indometacin to produce a complete response—perhaps there are unrecognised dosing requirements. There is certainly a timing requirement and again we have seen patients turn off, but only after 10 days at the dose of 275 mg daily.

Table 5

 Effects of treatment on trigeminal autonomic cephalalgias


For SUNCT the most immediate challenge must be to define the phenotype properly. We have seen patients who fulfil criteria for SUNA (table 6) but not SUNCT (see table 4). Typically the eye is not red, but we have also seen, for example external auditory canal swelling and periaural flushing as the sole cranial autonomic symptom, as has been reported for PH.50 It seems possible, given the relative proportion of patients with cluster headache who have lacrimation and conjunctival injection as compared with other cranial autonomic symptoms,47 that these symptoms are for some reason biologically more likely. This is supported by the same relative changes being seen in experimentally induced head pain.51 Thus research criteria for a more encompassing syndrome are proposed (see table 6).

Table 6

 Short lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms (SUNA)


The TACs represent a great success story in headache. From a classification point of view, the syndromes share much biology so their agglomeration in section III draws attention to them and to the trigemino-parasympathetic reflex. It is highly desirable that headache classification moves to a more biological and pathophysiological basis and the TACs are a step in that direction. The TACs also represent excellent clinical opportunities to take a careful history and offer effective therapy to otherwise highly disabled, suffering patients. Lastly, further investigations of the TACs are bound to illuminate physiological processes whose understanding will be useful to the range of primary headache syndromes.

A classification based on pathophysiology is a useful aid to differential diagnosis and effective treatment planning


View Abstract


  • Competing interests: PJG is a Wellcome Trust Senior Research Fellow.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.