Abstract
Purpose of review
Neurostimulation has emerged as a viable treatment for intractable chronic cluster headache. Several therapeutic strategies are being investigated including stimulation of the hypothalamus, occipital nerves and sphenopalatine ganglion. The aim of this review is to provide an overview of the rationale, methods and progress for each of these.
Latest findings
Results from a randomized, controlled trial investigating sphenopalatine ganglion stimulation have just been published. Reportedly the surgery is relatively simple and it is apparently the only therapy that provides relief acutely.
Summary
The rationale behind these therapies is based on growing evidence from clinical, hormonal and neuroimaging studies. The overall results are encouraging, but unfortunately not all patients have benefited. All the mentioned therapies require weeks to months of stimulation for a prophylactic effect to occur, suggesting brain plasticity as a possible mechanism, and only stimulation of the sphenopalatine ganglion has demonstrated an acute, abortive effect. Predictors of effect for all modes of neurostimulation still need to be identified and in the future, the least invasive and most effective strategy must be preferred as first-line therapy for intractable chronic cluster headache.
Keywords
Introduction
Cluster headache (CH) is a severe, disabling pain condition and a primary headache as defined by The International Headache Classification Committee, where it belongs to the trigeminal autonomic cephalalgias (TAC) (1). These are characterized by short-lasting, severe, strictly unilateral pain along the distribution of the first branch of the trigeminal nerve, accompanied by prominent cranial, parasympathetic, autonomic features. CH is a clinical diagnosis relying on diagnostic criteria and previous headache history. It has a prevalence of around 1% of the population (2,3) and a recent study found a male predominance with a ratio of 4.3:1 (3).
CH exists in two clinical forms, episodic cluster headache (ECH) and chronic cluster headache (CCH) with a ratio of about 6:1 (2,3). ECH involves periods of attacks (clusters) with periods of remission. Characteristically, CH attacks occur with circadian and circannual regularity often with a clock-like daily recurrence of attacks during cluster periods (4). The disorder has a substantial impact on the quality of life, economy and social functions of those affected (5).
Management of CH is divided into acute and prophylactic treatment, most often providing substantial, but not completely adequate, relief to a majority of patients. Prophylactics are required only in cluster periods, but in CCH a constant preventive therapy may be necessary. Unfortunately, a small fraction of CCH patients, presumably around 10%, seem to be resistant (drCCH) to all pharmacological treatments, fulfilling the proposed criteria for intractable headache (6). To provide relief to drCCH patients, various primarily destructive surgical procedures have been tried, aiming at trigeminal and parasympathetic pathways thought to be responsible for the pain and autonomic symptoms. These have included partial or complete surgical sectioning of the trigeminal root, radiofrequency trigeminal gangliorhizolysis, and microvascular decompression of the trigeminal nerve as well as comparable interventions aimed at the sphenopalatine ganglion (SPG). The effect is often sparse and side effects can be devastating (7,8).
In contrast, neurostimulation seems to offer a novel, superior alternative to these irreversible procedures. Hypothalamic deep brain stimulation (hDBS), occipital nerve stimulation (ONS), sphenopalatine ganglion stimulation (SPGS) and some others all seek to provide a nondestructive, reversible and adaptive way of helping drCCH patients. However, only two randomized, controlled studies have been conducted. Treatment with neurostimulation remains novel, but it seems that this may also, like medical treatment, be divided into prophylactic and acute effects, as will be shown below. Thus, a prophylactic effect would reduce the intensity and frequency of attacks whereas an acute effect would serve to abort an attack in progress.
This review aims to investigate the current progress concerning hDBS, ONS and SPGS in the treatment of drCCH and to compare these therapies.
Search strategy
A PubMed search with the English keywords “cluster headache,” “neurostimulation,” “occipital nerve stimulation,” “deep brain stimulation” and “sphenopalatine ganglion stimulation” was conducted. There was no time limit and only papers in English with CCH patients were included in the evaluation of neurostimulation. Searches were performed between summer 2010 and 2012. A single reference, Schoenen et al. (9), was changed in January 2013, as full results from a SPGS trial became available.
Pathophysiology of CH
The pathophysiological mechanisms behind CH are far from understood, but based on current knowledge, CH is now widely considered a chronobiological, neurovascular disorder, implying a crucial role of central brain mechanisms, specifically in the hypothalamus, in its pathophysiology (10,11).
Neuroimaging has suggested an abnormal or dysfunctional posterior hypothalamus as being central (12 –15). Clinical observations regarding the circadian and circannual symptomatology also point to pathology in the hypothalamus, especially the suprachiasmatic nucleus (SCN) (16,17). The SCN, situated in the anterior hypothalamus, works as an endogenous pacemaker and is responsible for controlling the circadian rhythmicity of hormone release and sleep-wakefulness cycles. It has extensive projections to other hypothalamic nuclei and a polysynaptic pathway to the pineal gland, responsible for melatonin production (18,19). Melatonin is thought to be the main biomarker of circadian rhythmicity (20), and its circadian secretory rhythm is driven by the SCN (21). A change in the circadian rhythm of plasma melatonin may indicate a desynchronization of circadian rhythms (21). It has been shown that the circadian pattern of melatonin and cortisol, as well as other hormones, are altered in CH patients (11,22).
A growing number of studies have shown specific hypothalamic involvement in CH. A landmark nitroglycerin-provocation study used positron emission tomography (PET) scans to visualize regional cerebral blood flow (rCBF) via an H2 15O tracer in 17 CH patients both in- and outside of cluster (12). The study showed significant activation of the ipsilateral, posterior, hypothalamic gray matter, but only in CH patients inside a cluster period. This specific hypothalamic activation has not been observed in experimentally induced trigeminal pain in healthy individuals nor in migraine patients, suggesting a specific role for the hypothalamus in CH. Additionally, a voxel-based, morphometric, magnetic resonance study found a significantly increased density and volume of the gray matter region thought to correspond to the inferior posterior hypothalamus in CH patients compared to healthy controls (13). Furthermore, a proton MR spectroscopy (1H-MRS) study measured N-acetylaspartate/creatine (NAA/Cr) ratios in three subgroups containing CCH patients and ECH patients in- and outside of cluster periods, respectively (14). NAA/Cr is generally considered to be a marker of neuronal health and is reduced in conditions of neuronal loss or dysfunction. All CH subgroups had lower NAA/Cr in the hypothalamus compared to healthy individuals and migraine patients. No differences between the three CH subgroups were seen (14). Recently, a functional magnetic resonance imaging (fMRI) study involving four ECH patients also showed significant hypothalamic activation ipsilaterally, attributable to cluster attacks (15).
Hypothalamic deep brain stimulation (DBS) in CCH
DBS is a method used since the 1990s especially in the management of movement disorders (23). The above-mentioned evidence of hypothalamic involvement in CH pathophysiology led to the use of hypothalamic DBS in drCCH patients. So far this treatment has been offered to drCCH patients in only a very few highly specialized centers. In short, a lead containing stimulation electrodes is placed targeting the posterior, inferior, ipsilateral hypothalamic gray matter, as identified in the PET and voxel-based-morphometry-magnetic resonance (VBM-MR) studies (24).
The mechanism of action in DBS is not completely clear. It is thought to work by decreasing somatic neural activity near the electrode by synaptic inhibition and at the same time increasing output from the stimulated neurons by directly activating the axons of local projection neurons and those passing nearby. Thereby, the intrinsic or spontaneous activity of stimulated neurons will be suppressed and replaced with an activity time-locked to the stimulus frequency, suppressing a possibly pathologic bursting and/or oscillatory activity (25).
Adjustable parameters in DBS include amplitude, pulse width, frequency and choice of active contacts. Typically in neurostimulation, the so-called charge injection is determined by amplitude and pulse width. By increasing the charge, injection fibers farther away from the active electrode can be excited. The frequency (Hz) probably plays a key role since observations in movement disorders DBS suggest that stimulation under 10 Hz has no effect while higher frequencies improve efficacy until a plateau is reached at around 200 Hz, at which point a higher frequency does not increase the efficacy of the stimulation (26).
Available studies regarding hypothalamic DBS
Available studies on deep brain stimulation for drug-resistant chronic cluster headache (CH).
As calculated by the authors of the present review.
Leone et al. and Broggi et al. (28,29) have reported long-term outcome from the first and largest study involving 16 drCCH patients. Thirteen of these are considered successes, of whom 10 are completely pain free and three patients now experience only sporadic attacks, indicating a shift from CCH to ECH (30). The report from Leone et al. has seen the best results of the currently available studies, with a success rate of 81%.
Bartsch et al., Sillay et al. and Schoenen et al. (31 –33) performed similar hDBS implantations in drCCH patients as the above mentioned. However, efficacy is reported lower, with a positive outcome around 50%–60%.
Brittain et al. and Owen et al. (34,35) account a total of three cases in their reports, all responders.
Hidding and May (36) report of a single patient in whom hDBS produced a constant frontal bilateral headache without alleviating the cluster headache. However, an existing polydipsia was relieved as long as the stimulation continued. The constant headache disappeared when the stimulator was turned off.
Fontaine et al. (37) performed the first sham-trial involving hDBS in drCCH. This was a randomized, prospective, crossover study involving 11 patients. The crossover period, in which active and sham stimulation was compared, was only one month, and no significant changes in headache attacks were found, thus efficacy of hDBS could not be confirmed. However, in a subsequent ten-month open-phase trial, six out of 11 were considered responders, suggesting that the experimental phase was too short. This supports the observations seen in the other studies, in which a positive response of hDBS typically requires more than one month of stimulation. It may, however, also be that some of those considered responders in the open phase went into spontaneous remission, as is known to happen in CCH (33).
Seijo at al. (38) report of five patients receiving hDBS to a modified target. This target was conceived to avoid the lateral ventricle wall and to decrease morbidity of potential hemorrhagic complications. Two patients became pain free, two had 90% or more reduction in frequency and one had a 50% reduction in frequency. There were no serious adverse events but permanent myosis and euphoria were seen in three patients.
Complications during electrode implantation
Macrostimulation in the posterior hypothalamus (pHyp) during implantation commonly produces oculomotor side effects like visual disturbances, mainly diplopia, accompanied by vertigo. This is the main side effect and a limiting element in increasing the amplitude in case of insufficient effect (30). Mood changes during the hDBS implantation procedure have been reported, and two patients suffered panic attacks, which in one led to abortion of the hDBS implantation (31,33). A micturition syncope was seen in one patient, one had a non-symptomatic transient hemorrhage visualized on computed tomography (CT) postoperatively (28), and yet another had transient hemiparesia and loss of consciousness during test stimulation with full remission. An intra-operative transient ischemic attack resulting in hemiparesis on the ipsilateral side of the electrode was seen in one patient; however, complete remission occurred after five minutes (39). The electrode implant procedure may also be fatal. One patient had an implantation-induced intracerebral hemorrhage shortly after surgery and died, confirming the risks of intracranial procedures (33).
Tolerability and adverse events following long-term hDBS
Long-term hDBS has generally proven to be without lasting side effects (30). Apart from relief of the above-mentioned polydipsia, no changes have been measured in body temperature, arousal, appetite, thirst or electrolyte balance. In a study of three hDBS patients, improved sleep structure and quality was found (40). An altered modulation in the mechanisms of orthostatic adaptation without affecting other cardiac systems is reported (41). Hormone levels like melatonin, cortisol, testosterone, prolactin and thyroid are not affected by long-term hDBS (30). No mood changes have been reported during long-term stimulation either, apart from one study in which euphoria/well-being was seen in three patients (38). One patient had an infection leading to removal of the stimulator and a subsequent full recovery (29). In conclusion, long-term hDBS seems to be safe and efficacious. Technical problems include electrode migration or dysfunction, and pulse generator battery problems after which pain attacks recurred (30,42).
Mechanisms and perspectives of hDBS
Hypothalamic DBS has so far been relatively successful in drCCH patients, with a positive response rate of approximately 60%. However, this leaves around 40% as nonresponders, a rather large proportion considering the invasive and potentially lethal implant procedure.
It has been questioned whether the stimulated area, according to the published stereotactic coordinates, indeed is the pHyp or is the adjacent mesencephalic gray. Recently Fontaine et al. used the Schaltenbrandt atlas and a stereotactic three-dimensional (3D) MRI atlas of the diencephalon-mesencephalon junction to precisely pinpoint the electrode in ten hDBS patients who had the implant procedure according to the published coordinates (43). They observed that the hDBS electrodes (both in responders and nonresponders) were located posterior to the mamillary body and mamillo-thalamic fasciculus, usually defined as the caudal border of the hypothalamus. Thus, placement was in the diencephalic-mesencephalic junction, posterior to the pHyp and not in the pHyp per se. The structures close to the active electrode in responders were the mesencephalic gray and the red nucleus but the location of the electrodes in responders as well as nonresponders did not differ significantly. Based on this, failure of hDBS may not necessarily be explained by a misplaced electrode. However, in some patients adjustments of stimulation parameters may be necessary, suggesting a non-optimal location of the active electrode.
Three mechanisms of hDBS have been proposed with regard to the known pathophysiology of CH: (1) A blockade of a cluster generator located in the hypothalamus or the nearby mesencephalic gray. (2) A nonspecific anti-nociceptive effect by activation of the periaqueductal gray (PAG) and/or rostral ventromedial medulla (RVM) pain modulatory system. (3) A long-term modulation of neuronal pain-processing pathways (44,45).
Ten drCCH patients treated with hDBS (of whom eight became pain free and two converted to having only sporadic attacks) were PET-scanned outside of an acute cluster attack (46). Scans compared the stimulator on vs. off. Stimulation induced significant activation in the ipsilateral posterior hypothalamic gray (site of stimulator), the ipsilateral trigeminal nucleus and ganglion, as well as activation and deactivation in other parts of the pain matrix. Of special interest, the PAG and RVM were not activated. This is important because it is evidence against a direct anti-nociceptive mechanism of hDBS mediated via those structures (44). Also, a direct inhibition of cluster generator activity alone seems unlikely with regard to the rich activation of the pain transmission systems that were seen during hDBS stimulation (44). Similarly, the latency of the effect as well as the absence of effect in acute stimulation argues against a direct inhibition of cluster activity. It seems a long-term modulation of pain processing or autonomic balance is most likely. Measurements in hDBS patients have shown decreased pain perception in peripheral limbs, but no lasting effect on pain threshold or nociceptive reflexes in the trigeminal territory (33). One study did find an increased threshold for cold pain along the ophthalmic nerve ipsilateral to the stimulation side, which suggests that some modulation along the ipsilateral trigeminal nucleus does happen (30).
The proposed cluster generator may be located in the pHyp, the SCN, the mesencephalic gray or even elsewhere. So far microrecording in the target area of hDBS has not shown a specific rhythmic pattern during the implant procedure. The most consistent finding is a 20-Hz discharge rate (30). However, Brittain et al. recently reported a significant 20-Hz increase in the pHyp occurring during a spontaneous CH attack during hDBS electrode implantation (34). It is possible that the constant high-frequency hDBS suppresses a pathologic, abnormal frequency or oscillating activity from possible dysfunctional cell bodies located in the pHyp, the SCN or the nearby mesencephalic gray, suspected to act as the cluster generator (44,47,48). In this manner hypothalamic cluster activity output is replaced with an activity time-locked to the stimulus frequency of hDBS, leading to stabilization of hypothalamic modulation of the trigeminal nucleus caudalis (TNC), in turn preventing abnormal activation of the trigemino-autonomic reflex (44,48). hDBS stimulation is typically required for weeks or months before any effect is seen, indicating that neural plasticity and changes in pain-processing pathways are involved and necessary for a positive outcome (30). This is confirmed by the fact that when stimulation is switched off, there can be a delay of up to three months before return to the original attack frequency.
In the past couple of years there has been an apparent decrease in publications on the topic of DBS in CH. The reason for this may be decreased interest because of the fatal outcome of one of the surgeries or publication bias against negative or inconclusive results of follow-up studies. Maybe DBS is not a viable long-term treatment for drCCH because of complications, diminishing effect over time or altogether other factors. The studies on the topic have, however, advanced the understanding of the disease significantly.
ONS in CCH
Peripheral nerve stimulation in the occipital region for intractable headache patients was first tried in 1999 in patients diagnosed with medically refractory occipital neuralgia (49). Most of these patients are later thought to have had chronic migraine, but the positive outcome led to ONS trials involving drug-resistant headache patients, including drCCH (50).
The main targets of stimulation are the distal branches of the C2–C3 roots corresponding to the lesser and especially the greater occipital nerve (GON). Leads containing electrodes are surgically inserted subcutaneously at the C1 level of the spinal cord (51). With the patients awake, the electrodes can be tested in order to ensure that relevant paresthesia is achieved, the main indicator of correct electrode placement (52). Adjustable parameters include amplitude, pulse width and frequency. Implantation may be uni- or bilateral (51).
Rationale for ONS
The pain in headache is thought to be caused by activation of afferent nociceptive fibers corresponding to the respective dermatomes of the pain (53). Nociceptive trigeminal afferents supplying the face, cornea, meninges and cranial vessels converge on second-order neurons in the medullary dorsal horn of the TNC in the brainstem. From there they project to relay in the thalamus on the way to the sensory cortex. Nociceptive cervical afferents from the greater and lesser occipital nerves branches of the C2 and C3 spinal nerve innervating cervical skin and muscles activate second-order neurons in the cervical dorsal horn along the C1, C2 and C3 level. Clinically, this is demonstrated when pain originating in the cervical region spreads to regions innervated by the trigeminal nerve and vice versa, confirming a functional connection between these areas. This is known as the convergence projection theory (53 –55).
Animal studies have shown that stimulation of the superior sagittal sinus, innervated by the ophthalmic branch of the trigeminal nerve, activates both second-order neurons in the TNC and second-order neurons in the dorsal horn of the C1 and C2 (56,57). Stimulation of nociceptive afferents from the GON, a branch of the C2 spinal root (50), leads to activation along the second-order neurons in the dorsal horn at the level of C1 and C2, but more interestingly, significant activation is also seen in the TNC (57,58). Also, in rats, a population of neurons in the C2 dorsal horn have been identified that receive convergent input from both trigeminal and cervical afferents (59,60).
In humans the convergence mechanism has been shown experimentally by injecting sterile water along the GON on one side (61). Immediately there was strong ipsilateral pain along the GON territory as expected. Furthermore, referred pain along the ipsilateral areas innervated by the ophthalmic branch of the trigeminal nerve was seen, in one subject, accompanied by cranial autonomic symptoms. This can be considered evidence supporting the existence of convergent cervical nociceptive projections to the TNC in humans (61). This may explain why nociceptive input from upper cervical segments can result in frontal head and facial pain.
Clinical studies also support the convergence theory. In a randomized, controlled trial (RCT) (62) as well as two other studies (63,64) GON blockades in CH patients showed effect in limiting duration, frequency and intensity of cluster attacks. This, and the subsequent modulation of the nociceptive blink reflex in healthy subjects as well as CCH patients (65,66), is thought to provide evidence for a functional influence of occipital inputs on trigeminal nociception (53). It supports the convergence mechanism in the trigeminocervical complex (TCC) and provides a rationale for ONS.
Available studies regarding ONS
Available studies on occipital nerve stimulation for drug-resistant chronic cluster headache (CH).
As calculated by the authors of the present review.
The largest series by Burns et al. includes 14 drCCH patients (52,67). All patients received bilateral ONS implantation. The first patient had a unilateral implant with good effect, but side shift in pain led to bilateral implantation in the patient and all subsequent patients. Ten of the 14 patients reported improvements; however, only five patients showed improvement of 50% or better. None have become completely pain free.
A study by Magis et al. involved 15 drCCH patients receiving unilateral ONS (68,69). Eighty percent of patients saw a 90% improvement, with 60% becoming pain free for prolonged periods. Two patients were nonresponders or described only mild improvement. There was no reported change in intensity of residual attacks. One patient had an infection of the device, which was subsequently explanted.
Schwedt et al. reported a study with 15 patients, among others, involving three drCCH patients (70,71). Two drCCH patients received unilateral and one bilateral ONS. A mean outcome after 20 months saw two of the three responders. A case report by the same authors found major improvements after unilateral implantation in one drCCH patient (72). Remarkably, he still had occasional autonomic attacks, but without the pain.
A German case series by Müller et al. (73) investigated seven drCCH patients receiving bilateral ONS. During follow-up, six out of seven patients had a positive effect. Intensity and frequency of attacks decreased to around 50% of baseline. Those six responders were able to reduce their use of acute attack medication by 77%. None, however, were completely pain free.
de Quintana-Schmidt et al. (74) report of four patients who saw a reduction of frequency, intensity and duration of their attacks with bilateral ONS.
Fontaine et al. (75) report of 13 patients of whom ten were responders with an improvement > 50%. One patient had an infection that led to explant of the stimulator.
Strand et al. (76) used an implantable rechargeable microstimulator and report of three patients, two of whom were responders.
Wolter et al. (77) report of a series of patients who were implanted with high cervical epidural electrodes activated by a subcutaneous impulse generator. A total of seven patients were implanted and improvement started immediately after electrode implantation. All patients showed significant improvement.
Overall results
Compared to hDBS, ONS seems safer and complications are fewer and milder (42,50,52,68). All patients experience ONS-induced paresthesias along the GON region thought to be required for effect. Most patients tolerate this minor side effect, but one patient without effect of the ONS felt the paresthesias were unbearable and had the device explanted (68). Local neck stiffness and discomfort are common (50). In ONS lead migration is the biggest problem. This often requires surgical revision after one to three years. Reports on the frequency of this varies between 0 (69,75) and 30% (67). Battery depletion is an expected event that requires replacement of the device (51). This may occur in as many as 64% (69) of patients. Also battery and electrode malfunction are reported (52). The positive effect of ONS requires weeks, but more often months of therapy, indicating that brain plasticity might be responsible for the effect. However, when the battery is depleted or malfunctioning the pain often returns to baseline almost immediately, suggesting that constant stimulation is necessary (52,67,68).
Mechanism and perspectives of ONS
The mechanism of action behind peripheral neurostimulation is not fully understood. ONS has shown effect in drCCH, chronic migraine and hemicrania continua (HC) (50,78), suggesting a nonspecific pain-relief mechanism (79). In drCCH and HC compared with chronic migraine, ONS is needed for months before any effect is seen (50,52,67,68,78). A direct anti-nociceptive mechanism could be responsible for the pain relief. According to the well-known gate control theory, the relative activity of large sensory afferents compared with small-fiber nociceptive afferents mediates pain perception (71). A direct effect on the nerve itself by slowing conduction velocity, increasing the threshold and decreasing response probability, is possible (80). Also, by stimulating large afferents, ONS may suppress small nociceptive fibers by pre-synaptic inhibition, thereby providing pain relief. However, unchanging cephalic and extra-cephalic pain thresholds in drCCH patients tested one month after ONS implantation argue against a general nonspecific analgesic effect (42,68). Also, the long interval before effect is seen contradicts this hypothesis. Some evidence also seems to indicate that supraspinal pain-modulating mechanisms may play a part (81).
Remarkably, it has been observed that although paresthesia is restricted to the dermatomes of the occipital nerves, ONS efficacy is not restricted to pain in this location (52). This suggests the involvement of convergence of cervical and trigeminal input in the TCC and supports the convergence projection theory as part of the explanation (52,82). It is not clear if ONS induces changes along the TCC or in higher centers relevant for CH pathogenesis or implicated in endogenous pain control (42). That some modulation in the TCC is thought to happen is supported by an increased latency in the nociceptive blink reflex in CH patients after one month of ONS. This may reflect modulation in the brainstem along the TCC but could also be a random finding (83).
In another study, ONS-treated chronic migraine patients with good effect were PET scanned with the stimulator on, off and partly activated (84). The results showed that pain during migraine attacks was inversely correlated with paresthesia in the GON region. Also, increasing paresthesia and thereby pain relief was associated with increased activation in the left pulvinar nucleus of the thalamus, together with the anterior cingulate cortex, a structure involved in the affective dimension of pain. The authors suggested that neuromodulatory changes may be responsible for the pain-relieving effect in ONS. Although these were migraine patients, the same mechanism may be true for drCCH patients (84).
In conclusion, the precise mechanism of ONS is yet to be determined. The response may reflect a reduction in trigeminal activation and mobilization of central pain modulatory centers, resulting in a possible inhibition along the TCC (70,85). It may be that over time ONS induces neuromodulatory changes in central pain-processing structures (42). However, the effect is reversible as pain may return to normal shortly after abruption of the stimulator (42,52,67). This last observation suggests that the effect is symptomatic only (86).
SPG
The SPG (also called the pterygopalatine ganglion) is a neural structure situated extracranially behind the maxillary sinus in the pterygopalatine fossa. It contains sensory, motor and autonomic fibers, of which only the parasympathetic nerves synapse in the ganglion. Its implication in the pathophysiology of CH is well recognized (11,87,88).
Rationale for SPGS
Efferents from the SPG are the final common pathway for parasympathetic activation in CH (87) and as a mediator of such responses a number of therapeutic methods have been aimed at the structure in the attempt to treat CH: intranasal lidocaine (89), transnasal injection of lidocaine and other substances (90,91), nerve blocks (92), radiofrequency ablation (93 –95), surgical resection (96), gamma knife surgery (97,98) and cryosurgery (99). Most of these interventions have shown only a transient effect, needing repeated intervention at varying intervals, or have proven to be associated with moderate to severe and irreversible side effects. They have, however, proven that it is possible through such manipulation to treat attacks, even in drCCH.
A view of the anatomy and physiology of the SPG provides evidence why it plays a pivotal role as the final common pathway in the pathophysiology of cluster headache. It has direct and indirect connections to the hypothalamus, superior salivatory nucleus (SSN), trigeminal-vascular system, meninges and various somatic and autonomic nerves innervating structures in the head, face and eye. As mentioned above, autonomic symptoms from these structures are a main feature of CH attacks. Precisely how this parasympathetic outflow is connected with the actual headache attacks is not fully understood at this time. The possibility exists that it could merely be an epiphenomenon.
Parasympathetic preganglionic neurons projecting to the SPG have their somas in the SSN, located in the pons. The SSN receives strong afferents from the limbic system and olfactory areas via the hypothalamus, relayed through the dorsal longitudinal fasciculus. Such pathways are likely responsible for the lacrimal response in weeping whereas sensory input to the trigeminal nucleus may stimulate the SSN via other afferents to result in tearing, as when the eye is irritated.
Efferents from the SSN travel in the intermediate nerve, divide in the geniculate ganglion and become the greater petrosal nerve and chorda tympani nerve. First-order parasympathetic neurons in the greater petrosal nerve are joined by postganglionic sympathetic fibers from the deep petrosal nerve, forming the vidian nerve. The preganglionic parasympathetic neurons then synapse with second-order parasympathetic neuronal cell bodies in the SPG. The parasympathetic fibers from the SPG link with sensory fibers of the maxillary nerve (V2) to innervate the gingiva, mucous membranes of the hard and soft palate and the auditory tube, the mucous glands of the palate, nose, pharyngeal and lacrimal gland. The SPG also allows passage of postganglionic sympathetic fibers to innervate blood vessels in the area.
Available studies regarding SPGS
There are currently reports from two studies available concerning SPGS of drCCH patients.
Ansarinia et al. 2010 (100) studied six patients with refractory chronic CH. They were treated with electrical stimulation of the SPG during an acute spontaneous or provoked attack. The stimulation was made possible by a needle placed in the pterygopalatine fossa through which a temporary single-contact stimulation electrode was advanced toward the SPG. Stimulation was delivered at varying frequencies, pulse widths and intensities. In this manner paresthesias could be induced and the location of the stimulator could be verified. Eighteen attacks were studied, and SPGS resulted in complete resolution of the headache in 11 of these. The response was fast, within one to three minutes of stimulation. The stimulation parameters that most often resulted in resolution of the headache was 50 Hz, pulse width of 300 µs and an amplitude below 2 V. The results were verified by placebo, to which one patient responded. The investigators also found that autonomic symptoms resolved along with the pain. There were no serious complications during the study.
Schoenen et al. (9) report of a total of 32 enrolled patients who were implanted with an on-demand SPG neurostimulator. During the randomized, experimental period the patients would receive either sham, subperception or full therapy. Twenty-eight patients completed this part of the study and in 67.1% of attacks treated with full stimulation the result was pain relief. This was the case for only 7.4% and 7.3% for the sham- and subperception-treated attacks, respectively; 68% of the patients experienced a clinically significant improvement. It is noteworthy that in addition to the above acute effect, 10 of the patients also experienced a ≥50% reduction in attack frequency. With regards to side effects and complications, 81% experienced transient mild to moderate loss of sensation in distinct maxillary nerve regions. However, 65% of these events resolved within three months. Serious side effects were reported in five patients and included misplacement of the lead during implantation and lead migration.
Mechanisms and perspectives of SPGS
The mechanism by which SPGS provides relief is not completely clear and is likely different from both ONS and hDBS. It is clear that if nervous outflow from the SPG is inhibited, CH attacks can be aborted. SPGS appears to have prophylactic properties, as seen in the study reported by Schoenen et al., but also shows efficacy acutely, during the attack, which is not seen in the other therapies.
A possible mechanism for the neuromodulatory effects necessary for prophylaxis could be via activation of sensory fibers from the V2 division of the fifth cranial nerve. These axons pass through the SPG and converge on second-order neurons in the TNC together with afferents from the V1 division. The acute effect of high frequency (80–120 Hz) stimulation of the SPG on the CH attack is likely by a completely different mechanism, possibly by inhibiting parasympathetic outflow, or in an antidromic fashion blocking connections from the SSN and the hypothalamus (100). The importance of the SSN in the pathophysiology of CH has been established relatively recently (87). The dramatic effect seen within minutes may be attributable to a disruption of abnormal electrical or neurochemical activity causing CH. The observed effect on the autonomic symptoms suggests an effect on vascular tone as both periorbital swelling and nasal congestion has been seen to resolve quickly during stimulation.
It would seem that SPGS in the treatment of CH is promising. Based on these new findings it seems to be both effective and safe, with a minimum of complications. Further studies need to be conducted to establish predictors for efficacy.
Efficacy of neurostimulation in CCH
hDBS, ONS and SPGS have so far seen positive outcomes in drCCH. Common for hDBS and ONS, but especially the latter is the fact that a positive outcome may take weeks to months of stimulation, pointing at long-term brain alterations in pain processing as a possible mechanism (30,42,52,67,68,101). Also noticeable is the fast relapse in both hDBS and ONS in which the pain and autonomic symptoms return to baseline levels when the stimulator is turned off, regardless of whether it is intentional or as a result of a device malfunctioning (30,52,68). No information regarding this is available on SPGS. Both these clinical observations and the fact that these patients have had drCCH for years suggest more than a simple placebo effect. However, it is noteworthy that out of all the studies mentioned in this review, only two are randomized and controlled (9,37).
It is unknown why hDBS and ONS work in only some drCCH patients and why time to effect differs. So far no predictors regarding treatment response have been found (30,52,67,68). Occipital tenderness was a predictor for GON block efficacy (63) but GON block does not predict ONS response (71). One study reported that those drCCH patients with the best outcome after ONS had a relatively fast response (weeks) while those having a more moderate effect took a longer time to respond (months) (67). One can speculate whether the role of central mechanisms and peripheral nerve structures differs individually in CH patients, suggesting subtypes of CH. Clinically CH patients respond differently to medication, GON block, surgery and now neurostimulation (10,102,103). Also, atypical cases of CH have been reported in which CH patients do not experience the clock-like regularity of headache attacks as well as the restlessness and agitation typically associated with the hypothalamus (104,105). CH pain and autonomic features, however, can also be caused by an entirely central mechanism independent of peripheral signals, proven by a patient having received complete surgical sectioning of the trigeminal root, but still experiencing CH attacks on the same side (106).
Perhaps subtypes of CH exist. Just like ECH sometimes is seen to evolve into CCH, perhaps occasionally CH progresses from involving a certain amount of peripheral mechanisms to exclusively involving central mechanisms, the latter perhaps the case in some patients with drCCH (10). These differences in central and peripheral mechanisms may explain why ONS and hDBS work in only some patients. Ultimately understanding these mechanisms will help decide which treatment option will be best suited for the individual drCCH patient.
Conclusion
Conservatively, there is room for optimism in helping individuals tormented by drCCH through various methods of neurostimulation. Other invasive options exist; however, the complications and side effects of destructive surgery may be severe and irreversible.
The question when hDBS, ONS or SPGS should be offered remains. Possible side effects and complications like the fatal hemorrhage in hDBS must be considered. Proposals for patient selection exist, suggesting that neurostimulation is an option in select CCH patients only who have not had satisfactory effect of current, recognized, pharmacological prophylactics. Deciding if either hDBS, ONS or SPGS should be offered is difficult since no predictors for outcome have yet been identified. It seems that SPGS has a higher efficacy than hDBS and ONS; there are, however, problems regarding post-surgical sensory disturbances and lead placement that may improve as the implant procedure is refined. Long-term studies and comprehensive data are still lacking.
Lastly, side-shifts in pain have been reported after hDBS and ONS, thus forcing physicians to consider whether implantation should be uni- or bilateral. The exact rate of side-shifting for the individual therapies is unknown at present. That side-shift may occur spontaneously in CH further complicates the picture (107). Thus far only two RCTs with neurostimulation in drCCH have been conducted and placebo is known to play a role in both acute and prophylactic treatment (108). True RCT studies can be difficult to perform but should always be considered.
In order to fully be able to help drCCH patients, the pathophysiology of CH and mechanisms of neurostimulation need to be better understood. Given that a certain proportion of drCCH patients may be nonresponders to these therapies, a critical issue for future studies will be identifying predictors of outcome that can prospectively distinguish responders from nonresponders. Mechanistic studies may be helpful in this regard.
Clinical implications
Neurostimulation has emerged as a viable treatment option for drug refractory chronic cluster headache and stimulation of the posterior hypothalamus, the occipital nerve and the sphenopalatine ganglion have been investigated. The exact methods of action of these therapies are not yet completely understood. Deep brain and occipital nerve stimulation offer prophylactic benefit whereas sphenopalatine ganglion stimulation seems to offer both prophylactic and acute relief. There are still no certain predictors of effect and this needs to be investigated in future studies.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflicts of interest
Jeppe Lyngholm Pedersen and Mads Barloese have nothing to declare. Rigmor Jensen is a member of the advisory board of Autonomic Technologies, Neurocore, Medotech and Linde Gas; is a director of LTB and EHMTIC and president of EHF; and has given lectures for Pfizer, Berlin Chemie, Allergan and Merck.