4.1.1 Differentiation of Cluster Headache from Migraine

For those with an interest in, and knowledge of, headache diagnosis, it often is difficult to understand how the diagnosis of cluster headache can be missed entirely, or be delayed for so long, in so many patients, given that the phenotype of CH is so characteristic and, once learnt, should never be forgotten. However, there are several potential reasons for this. Firstly, most doctors receive very little training in headache medicine and, therefore, many may never have learnt about cluster headache, in the first place. Also, migraine is very common, especially in women, while CH is rare, but more common in men with a male/female ratio of 3.5:1 [7]. Despite the fact that the phenotype of CH is essentially the same for men and women, women with CH are more frequently misdiagnosed than men (61.1% vs. 45.5%, p < 0.01) [6]. As noted above, CH is most often misdiagnosed as migraine as opposed to any other headache condition, and while there are overlapping features that they share, there nevertheless remain distinct differences, even in these shared features, that clearly separate them.

Cluster headaches (and the other TACs) are unilateral, side-locked headaches in 69–92% of cases [8], whereas this only occurs in 20.8% of migraine sufferers [9]. A shift in the side of the headache within an attack occurs in only between 1 and 8% in CH [7, 10] but is more common in migraine. Cluster headache sufferers can also experience the same accompanying symptoms as in migraine, such as nausea, photophobia, and phonophobia, although less commonly. However, in CH the sensitivity to light and sound, and in particular to light, is ipsilateral to the headache much more commonly than in migraine. In one study [11], whereas only 2/54 (4%) episodic migraine patients had unilateral photophobia or phonophobia, or both, this occurred in 10/21 (48%) of chronic, and 4/5 (80%) episodic, cluster headache patients, respectively.

While the presence of ipsilateral autonomic features is one of the diagnostic criteria for the diagnosis of CH (and the other TACs) [1], autonomic features can also occur in migraine. While the frequency of autonomic features in CH ranges from 72% for rhinorrhoea to 91% for lacrimation [10], one or more autonomic features were seen in 226/841 (26.9%) in a population-based sample of migraineurs [12]. Importantly, in trying to distinguish migraine with autonomic features from CH, it needs to be emphasized that the autonomic features are much more often lateralized, and side-locked, with the headache in CH, whereas in migraine they are more often bilateral and less prominent [13]. The presence, or not, of aura is not particularly helpful since it occurs in 14–21% of CH patients [7, 10]. Perhaps, the most telling differences in the clinical picture of a CH patient, compared to a migraine patient, are their demeanour and behaviour during an attack. In up to 97% of migraine without aura patients, movement makes the headaches worse [14] and therefore most migraine patients prefer to rest and be still. In dramatic contrast, in CH up to 93% of patients will be restless or agitated during an attack or have no worsening of the headache with movement [10] and therefore are much more likely to pace about and not be still. In another study, only 0.8% of CH patients did not have any sense of agitation during their attacks [7]. Having described the diagnostic differences between migraine and CH which help the clinician to confidently differentiate between them, the next major task when dealing with a TAC, such as CH, or a TAC-like syndrome, is whether the headache is primary or secondary. In the following sections, we will briefly summarize the recent literature on secondary TACs and try to identify if there are any “red flags” that might suggest that an otherwise typical TAC might have an underlying lesion and thus whether all, or only some, patients should have neuroimaging.

4.2.1 Summary of Previous Reviews of Symptomatic, or Secondary, Trigeminal Autonomic Cephalalgias

Since 2004 there have been five comprehensive reviews of “secondary” or “symptomatic” TACs [3, 4, 15–17]. Here we summarize the findings of these publications. Trucco et al. in 2004 [15] reviewed cases from 1980 to 2001 and identified 22 CPH, 9 HC and 7 SUNCT patients in whom coexisting pathology was identified. They did not look for symptomatic CH cases but rather referred to a review of such cases by Giraud et al. [18]. For the diagnosis of CPH, the criteria of the International Classification of Headache Disorders: second edition [19] were used, while for HC and SUNCT those suggested by Goadsby and Lipton [20] were used. Of the 22 CPH patients, 6 fulfilled all (definite CPH) and 10 fulfilled all, bar 1 (probable CPH), criteria. In the other six, there was lack of sufficient information, or else they clearly did not fulfil the required criteria. The lesions identified in the definite CPH group were a Pancoast syndrome [21]; left sella turcica gangliocytoma [22]; right cavernous sinus meningioma [23]; maxillary cyst [24], cerebral metastasis of parotid epidermoid carcinoma [25]; and Meckel’s cave non-Hodgkin’s lymphoma [15]. In the probable group, the lesions were right internal carotid artery aneurysm [15], vasculitis [26], intracranial hypertension [27], AV fistula [15], tuber cinereum hamartoma [28], ipsilateral occipital infarction [29], pituitary microadenoma [24], possible cerebral vasculitis [25], ipsilateral intraorbital and cavernous sinus granulomata [25], and head injury [30].

Of the nine HC cases, six cases fulfilled all, and two all but one, of the Goadsby and Lipton criteria [20]. The lesions implicated in the definite group were right mesenchymal chondrosarcoma [31] and sphenoid sinusitis [32], while in the probable group they were C7 nerve root compression [23], HIV infection [33] and head trauma in four patients [34]. Of the seven SUNCT patients, five had definite and two possible SUNCT [20]. The lesions identified in the definite group were ipsilateral AVMs in two patients [35, 36], ipsilateral para-pontine cavernous angioma [37], HIV infection [38] and craniosynostosis [39], while in the possible cases, they were ipsilateral dorsolateral brainstem infarction [40] and basilar invagination due to osteogenesis imperfecta [41].

From their review of these cases, the authors concluded that it was generally not possible to confidently establish a causal link between the pathology identified and the presenting TAC, or TAC-like, headache. They commented that posterior fossa lesions seemed somewhat common in SUNCT patients, while lesions in the region of the cavernous sinus seemed more common in CPH-like headache patients, although the numbers were small. Finally, they recommended neuroimaging in patients with “atypical” TACs, in terms of the phenotype, or with an uncharacteristic response to indomethacin in CPH or HC patients.

Favier et al. in 2007 [16] reported on 31 patients with secondary TAC, or TAC-like, headaches in whom a structural lesion was identified which, when treated successfully, resulted in a significant improvement, or even complete resolution, of the headache. They reviewed the literature from 2001 to 2005 and collated 27 cases, as well as 4 new patients from the records of their own institutions. Of the 31 patients, there were 19 patients with typical cluster, and 4 with cluster-like, headache. They also identified one PH-like, one typical CPH and one CPH-like patient and four SUNCT patients. They did not investigate patients with symptomatic, HC or HC-like, headache.

The underlying lesions in the four typical CH patients were an ipsilateral upper cervical meningioma [42], a recurrent nasopharyngeal carcinoma involving the ipsilateral internal carotid artery [43], an ipsilateral temporal lobe AVM [44] and an aspergilloma of the sphenoid sinuses [45]. The single CH-like patient had a mycotic aneurysm of the intracavernous portion of the ipsilateral internal carotid artery [46].

In the four typical ECH cases, the lesions were an infected foreign body in the ipsilateral maxillary sinus [47], a thrombosed aneurysm of the ipsilateral posterior communicating artery [48], an AVM of the ipsilateral frontal lobe [44] and a pituitary adenoma (prolactinoma) extending into the ipsilateral cavernous sinus [49], while in the three ECH-like patients, the lesions were an ipsilateral occipital lobe AVM [50] and two ipsilateral pituitary adenomas: one prolactinoma [16] and one growth hormone-secreting [51].

In the six typical chronic CH patients, the following abnormalities were identified: pituitary adenomas (prolactinomas) in three patients [16, 52, 53], a parasellar meningioma extending ipsilaterally [54], an ipsilateral meningioma of the tentorium cerebelli [55] and a benign tumour of the ipsilateral posterior fossa [56], while in the three chronic CH-like patients, there were an aneurysm of the ipsilateral vertebral artery [57], an ipsilateral subclavian steal syndrome [58], and a cavernous haemangioma of the ipsilateral orbit [16]. Both the typical cluster-tic syndrome patient [16] and the chronic cluster-tic-like patient [59] had pituitary adenomas (both prolactinomas).

There was a single patient with PH-like headaches with bilateral attacks with an AVM of the parietal lobe [60], a typical CPH patient with an aneurysm of the contralateral carotid artery and a dilated ipsilateral carotid artery [53], a typical CPH patient previously reported [22] and a patient with CPH-like headache with a mucocoele of the ipsilateral maxillary sinus [61]. Three of the four SUNCT patients had ipsilateral pituitary adenomas (two prolactinomas and one non-functioning) [62–64] and the other, a brainstem pilocytic astrocytoma extending to the ipsilateral cerebro-pontine cistern [65].

The authors concluded from their study that typical TACs, even those that respond to the usual pharmacologic TAC treatments, such as indomethacin, can be due to structural lesions. Moreover, only 10 of the 31 patients had atypical features. The other striking finding they noted was that 11 of the 31 patients had pituitary tumours, 10 of which were secretory and 9 of these were prolactinomas. On the basis of these observations, they recommended neuroimaging in all patients presenting with TACs or TAC-like headaches.

In 2009 Cittadini and Matharu published a review of 37 cases of symptomatic TACs [3]. These included 24 CH patients, 3 PH patients and 10 SUNCT patients. They selected cases where they believed the associated, underlying lesion was likely to be causal to the TAC, as evidenced by the fact that the lesion was ipsilateral to the TAC in all cases, although in 3 CH patients (1 cerebral vein thrombosis (CVST), 1 idiopathic granulomatous hypophysitis and 1 sphenoidal aspergilloma) and 1 SUNCT patient (metastatic carcinoid), the lesion involved the other side as well and that, in all cases, treatment of the underlying lesion resulted in significant improvement in the headache. In fact, in all of the CH and PH patients and in 8 of the 10 SUNCT patients, there was complete resolution of the headache, although reported follow-up times varied considerably. One SUNCT patient with a prolactinoma [63] had a marked, but not complete, response to treatment (partial resection and radiotherapy), while another SUNCT patient with bilateral intraorbital metastatic bronchial carcinoid had only transient improvement for 1 month after radiotherapy [66].

Twenty-four of the 37 cases had already been described in 1, or other, of the 2 previous reviews [15, 16]. Thus, there were 13 new cases identified: 6 CH (4 subtypes unclear and 2 CCH), 1 PH and 6 SUNCT patients. Interestingly, three of the four unclassifiable CH (duration of headache less than a year) patients had ipsilateral internal carotid artery dissections [67, 68], while the fourth had CVST [69]. In the single new CPH patient [70] and in four of the six SUNCT patients [71–74], the underlying lesion was a pituitary tumour. In the CPH patient, this was a macroadenoma (prolactinoma) and in the SUNCT patients there were two microadenomas (both prolactinomas), one growth hormone-secreting adenoma and a non-functioning macroadenoma. The lesions in the two other SUNCT patients were an orbital cystic lesion [75] and a pilocytic astrocytoma [76], both ipsilateral.

The authors again highlighted the predilection for pituitary lesions in all three of the TACs studied, with 8/24 (33%) of CH patients, 2/3 (66%) PH and 7/10 (70%) of SUNCT patients having these. They also emphasized that in more than half of CH and SUNCT patients, and in all cases of PH, an atypical headache phenotype and/or abnormal physical examination was noted. Finally, they noted that a poor response to, or the need for higher than usual doses of, appropriate medication should be a red flag for a secondary, or symptomatic, TAC. In terms of which patients should receive neuroimaging, they concluded that MR imaging should be done in all patients with “an atypical symptomatology, abnormal examination, and poor response to the appropriate treatments” [3].

The most recently published review of symptomatic TACs and TAC-like headaches is by de Coo et al. and was published in 2015 [17]. It was the first to use the ICHD-III beta criteria when evaluating reports of symptomatic TACs. They reviewed cases from 2009 to 2015 and separated them into probably secondary, possibly secondary and unknown. Probably secondary was used when there was a “dramatic improvement of the headache after treatment of the underlying lesion” and possibly secondary, when there was improvement but not complete resolution of the headache or when a causal effect was deemed possible by the authors. We will not discuss their last category, unknown.

In the 12 probably symptomatic CH patients, there were 7 patients with tumours: 2 had ipsilateral pituitary adenomas, 1 macroprolactinoma [77] and 1 non-functioning [78], 1 an intrasellar arachnoid cyst [79], 1 a hypothalamic cystic tumour related to sarcoid [80], 1 an ipsilateral glioblastoma multiforme [81], 1 an ipsilateral carotid paraganglioma [82] and an angiomyolipoma infiltrating the ipsilateral face [83]. There were two vascular lesions: one ischaemic stroke related to moyamoya [84]; and one due to neurovascular compression of the ipsilateral C3 nerve root and vertebral artery [85]; two inflammatory cases, both due to acute ipsilateral maxillary sinusitis [86, 87]; and one attributed to obstructive sleep apnoea [88]. In the seven possibly symptomatic CH patients, the underlying lesions were tumours in 2:1 ipsilateral macroprolactinoma [89] and one angiomyolipoma infiltrating the ipsilateral face [83]; two ocular causes; recurrent posterior scleritis and aseptic meningitis [90] and post intraocular lens implant [91]; two cases of multiple sclerosis [92, 93]; and one dissection of the ipsilateral distal internal carotid artery [94].

There were 14 probable cases of symptomatic SUNCT/SUNA and 12 possible cases. As is the case with other TACs, there were four cases of probable symptomatic SUNCT/SUNA due to tumours: three pituitary lesions – one ipsilateral prolactinoma [95], one ipsilateral macroprolactinoma [96] and one ipsilateral mixed gangliocytoma and pituitary adenoma [95]—and an epidermoid tumour in the ipsilateral cerebellopontine angle [97]; and likewise, two vascular lesions affecting the internal carotid artery, one aneurysm of the cavernous portion of the ipsilateral internal carotid artery [98] and an ipsilateral cavernous dural AV fistula [99]. The striking finding, however, was the high number of cases attributed to neurovascular conflict where there was compression of the ipsilateral trigeminal nerve by the superior cerebellar [7], or anterior inferior cerebellar [1], artery, eight in total [100–102]. In the possibly symptomatic group, there were five cases due to tumours: three ipsilateral pituitary adenomas, including two prolactinomas [95], one ipsilateral meningioma [103] and one lung adenocarcinoma [104]. Again, there were four cases attributed to neurovascular conflict [101, 102]. The other three underlying lesions were lesions due to multiple sclerosis [105], viral meningitis [106] and a case of mild hypothalamic/pituitary dysfunction related to ipsilateral optic nerve hypoplasia [107]. All patients in the probably symptomatic group became pain-free, bar 1 who improved considerably, whereas only five in the possibly symptomatic group became pain-free.

There were two cases of probably symptomatic, and three cases of possibly symptomatic, HC all of whom, by definition, were absolutely responsive to indomethacin. The underlying lesions in the probable group were a cerebral vein thrombosis [108] and cerebral metastases due to lung adenocarcinoma [109]. In both of these patients, after definitive treatment of the underlying condition, they were rendered pain-free and were able to cease indomethacin completely. In the three possibly symptomatic HC patients, the lesions were post-traumatic head injury, post-craniotomy for evacuation of traumatic subdural haematoma and postabdominal surgery done under spinal anaesthesia [110]. All three patients remain headache-free, but only on indomethacin.

De Coo et al. [17] concluded from their review of a total of 53 TACs that tumours, particularly of the pituitary, especially prolactinomas and other adenomas, were relatively common. They reported only a single case of cerebral artery dissection in their CH patients, in distinct contrast to other reviews. They highlighted that 8/14 patients with probable secondary SUNCT and 4/12 with possible SUNCT, that is, more than 45% of their SUNCT/SUNA cases, had evidence of ipsilateral neurovascular conflict. Moreover, they emphasized that of those patients who underwent microvascular decompression, the vast majority had excellent outcomes, often being rendered entirely pain-free. This dramatic response to microvascular decompression is similar to that seen in classical trigeminal neuralgia, with which SUNCT and SUNA share several other features as well. Thus, in some patients first division trigeminal neuralgia will enter into the differential of SUNCT/SUNA patients. While there are clear differences, such as the presence of a refractory period in TN, but not in SUNCT/SUNA, it has been suggested that these disorders may be variants of the same disorder [111].