Abstract
Chronic daily headache is a major worldwide health problem that affects 3–5% of the population and results in substantial disability. Advances in the management of headache disorders have meant that a substantial proportion of patients can be effectively treated with medical treatments. However, a significant minority of these patients are intractable to conventional medical treatments. Occipital nerve stimulation (ONS) is emerging as a promising treatment for patients with medically intractable, highly disabling chronic headache disorders, including migraine, cluster headache and other less common headache syndromes. Open-label studies have suggested that this treatment modality is effective and recent controlled trial data are also encouraging. The procedure is performed using several technical variations that have been reviewed along with the complications, which are usually minor and tolerable. The mechanism of action is poorly understood, though recent data suggest that ONS could restore the balance within the impaired central pain system through slow neuromodulatory processes in the pain neuromatrix. While the available data are very encouraging, the ultimate confirmation of the utility of a new therapeutic modality should come from controlled trials before widespread use can be advocated; more controlled data are still needed to properly assess the role of ONS in the management of medically intractable headache disorders. Future studies also need to address the variables that are predictors of response, including clinical phenotypes, surgical techniques and stimulation parameters.
Keywords
Introduction
Chronic daily headache (CDH) is an umbrella term for headache disorders that occur very frequently. The term refers to headaches disorders that occur on 15 or more days per month for more than 3 months. CDH can be further subclassified as secondary headaches, which have an identifiable underlying pathology that can be established with appropriate investigations, or primary headaches, which do not have an identifiable underlying pathological process other than the functional derangement of the pain neuromatrix. CDH is a major worldwide health problem that affects 3–5% of the population [Scher et al. 1998] and results in substantial disability. According to a report by the World Health Organization a patient suffering from severe migraine is as disabled as a patient with dementia, active psychosis or quadriplegia [Menken et al. 2000]. Advances in the management of headache disorders have meant that a substantial proportion of patients can be effectively treated with medical treatments. However, a significant minority of these patients are intractable to conventional medical treatments. Thus, there is a clear need for novel approaches for the management of this patient group. Neurostimulation therapies that entail peripheral or central nervous system targets are emerging as very promising approaches. The peripheral targets that have been used recently include the occipital, supraorbital and vagus nerves and the sphenopalatine ganglion. The use of occipital nerve stimulation (ONS) to treat primary headaches is reviewed in this article.
Anatomy of the occipital nerves
The anatomy of the nerves of the occipital region has been well described [Natsis et al. 2006]. There are three nerves that innervate the occipital region, namely the greater, lesser and least occipital nerves. The greater occipital nerve is a branch of the C2 spinal root. It proceeds between the inferior oblique and the semispinalis capitis muscle in a superomedial fashion. The nerve then crosses above the rectus capitis posterior major muscle and arises medial to the semispinalis capitis muscle, which it occasionally pierces. It then penetrates through the trapezius muscle to join the occipital artery [Natsis et al. 2006]. It provides innervations to an occipito-parietal area 6–8 cm wide and ascending paramedially from the subocciput to the vertex [Poletti, 1991]. The lesser occipital nerve is composed of branches of the C2 and C3 spinal roots. It runs lateral to the greater occipital nerve, crossing over the sternocleidomastoid muscle, and courses superolaterally towards the region behind and above the ear. The medial branch of the posterior division of the C3 root gives off a branch called the least occipital nerve, which pierces the trapezius and ends in the skin of the lower part of the back of the head. It lies medial to the greater occipital nerve and communicates with it.
There is an anatomical and functional overlap of trigeminal and cervical afferents throughout the trigeminocervical complex from the level of the caudal trigeminal nucleus to at least the C2 segment [Kerr and Olafson, 1961]. This convergence explains how nociceptive activation at either end of this structure can result in both trigeminal and cervical distribution pain. Similarly, ONS could modulate pain not only in the territories innervated by the occipital nerves but also areas innervated by the trigeminal nerve.
Operative techniques
ONS for the treatment of medically intractable headaches was introduced by Weiner and Reed [Weiner and Reed, 1999]. ONS is typically performed with the equipment normally used for spinal cord stimulation (SCS), which includes electrodes and their leads, anchors to fasten the leads to connective tissue, and the implantable pulse generator (IPG). Cylindrical-style and paddle-style electrodes can be used. Cylindrical electrodes are thin and can be inserted through a needle, while paddle electrodes (flat and broad) are associated with a lower incidence of lead migration but require more extensive surgical dissection for placement. Electrodes can be programmed to function as either cathodes or anodes to direct the flow of current adjacent to the stimulated structure.
A stimulation trial is performed before the permanent implantation in some centres, with the view to improving selection of the candidates for a permanent stimulation. The procedure involves inserting percutaneous leads into the epidural space via a Tuohy needle and externally powering them for 5–7 days. If the trial is successful in terms of significant pain improvement, the patient is offered a permanent implantation. However, in primary headache syndromes, unlike in neuropathic pain, there can be a considerable delay of several weeks to months before the response emerges and therefore the utility of a stimulation trial in selecting patients for permanent implantation remains questionable for now.
The permanent implantation procedure to site the electrodes can be performed via both midline and retromastoid approaches. Electrodes are placed subcutaneously superficial to the cervical muscle fascia, transverse to the affected occipital nerve trunk at the level of C1, usually using fluoroscopic guidance. The standard procedure is typically performed in two stages. The first stage, carried out under local anaesthesia with sedation, is used to test the stimulation and determine optimal placement of electrodes. The second part, which involves insertion of the rest of the ONS system, is carried out under general anaesthesia. However, a recent report of a small case series described successful placement of ONS systems entirely under general anaesthesia while still achieving the desired occipital region stimulation [Trentman et al. 2010].
The IPG can be located at various regions, including the buttock, mid-axillary thoracic region, or low abdomen. The battery can be nonrechargeable (with a lifespan of 2–5 years) or rechargeable (with a lifespan close to 10 years). The patients control the ONS with a handheld remote control by which they can turn the device on or off besides adjusting the stimulator parameters. The stimulation parameters, including frequency, pulse width and voltage, are adjusted such that patients experience comfortable paraesthesia in the stimulated area. Stimulation can be continuous, turned on and off as needed, or continuous with alterations in parameters as needed. There is a wide variation in the stimulation settings used, with the amplitude ranging from 0.1 to 10 V, the frequency ranging from 3 to 130 Hz and pulse width ranging from 90 to 450 ms [Trentman and Zimmerman, 2008]. Optimal stimulator settings need to be better defined since there are no data on impact of specific parameters on outcome.
Miniaturized devices are a potential new option for delivering peripheral nerve stimulation. One such device that has been used for ONS is the bion. The bion is a rechargeable, self-contained, battery-powered, telemetrically programmable, current-controlled mini-neurostimulator with an integrated electrode and battery that are encased in a small device which can be implanted over the occipital nerves [Schwedt et al. 2006]. Recently Trentman and colleagues described the implantation technique and the stimulation parameters of the bion microstimulator in nine patients with medically intractable primary headache disorders [Trentman et al. 2009]. Their results showed that the bion may provide effective occipital stimulation without requiring anchoring or tunnelling of extensions to remote power sources. It may also minimize common adverse events such as lead migration. However, the microstimulator needs frequent recharging, has limited choices in terms of electrode combinations and can become encapsulated, hence increasing the energy required to stimulate the occipital nerve. The role of these miniaturized devices in the treatment of headache syndromes remains unclear because their efficacy and safety need to be studied further.
Evidence for efficacy of occipital nerve stimulation in primary headache syndromes
ONS has been used in various primary headache syndromes, including migraine, cluster headache, hemicrania continua (HC) and short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT) (see Table 1). ONS has also been used in secondary headaches, including cervicogenic headache, post-traumatic headache and occipital neuralgia.
Occipital nerve stimulation studies in primary headache syndromes.
IPG, implantable pulse generator; ONS, occipital nerve stimulation; SUNCT, short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing; SUNA, short-lasting unilateral neuralgiform headache attacks with cranial autonomic symptoms.
Migraine
Popeney and Alo [Popeney and Alo, 2003] used C1 through C3 peripheral nerve stimulation to treat 25 patients meeting International Headache Society (IHS) criteria for episodic migraine and suggested criteria for transformed migraine [Saper, 1990]. Migraine had been present for a mean duration of 10 years (range 1–30 years). All patients gave a history of progressive pain into the posterior occipital, vertex, or retro-orbital regions. The patients reported that migraine had failed to respond to an average of seven pharmacological treatments. Nineteen patients (76%) reported overuse of abortive treatments. All patients completed a successful 5–7-day trial of outpatient stimulation with an externalized quadripolar electrode system before they underwent a permanent implant. The mean stimulation parameters were a frequency of 55 Hz, a voltage of 3.2 V, and a pulse width of 400 µs. Sixty percent of patients used the stimulator intermittently while the remaining 40% used it continuously. The patients were followed up for an average of 18 months (range 9–36 months). Eighty-eight percent of patients reported at least 50% reduction in headache frequency or severity after the ONS was implanted. The average 3-month headache frequency reduced from 76 days pre ONS to 38 days post ONS, while the average severity decreased from visual rating score (VRS, 0–10) 9.32 to 5.72. Moreover the average migraine disability assessment score dropped from grade IV (severe disability) to grade I (little disability). After stimulation, most patients used less than 15 symptomatic medication doses per month for residual symptoms. Complications included traumatic lead migration in six patients, spontaneous migration in three patients, and infection in one patient.
Oh and colleagues used ONS in 10 patients with transformed migraine [Oh et al. 2004]. Migraine had been present for a mean duration of 12 years (range 2–25 years). All patients had failed to respond to preventive treatments and physical therapy. Nine patients had complete but transient pain relief with greater occipital nerve blocks while one patient reported partial benefit (80%). Seven of these patients had previous success with dual percutaneous cylindrical electrodes that migrated and all 10 patients were then implanted with dual paddle style electrodes. Nine of 10 patients reported more than 90% pain relief, and one reported 75–90% pain relief at 1 month. At 6-month follow up, seven patients reported more than 90% pain relief, two patients reported 75–90% relief, and one was lost to follow up. All patients stated they would have the operation again. Adverse events included infections in two patients, which led to explantation of the stimulator in one patient and a replacement 2 months later.
Weiner and Reed reported a series of cases of intractable occipital neuralgia responding to ONS [Weiner and Reed, 1999], though detailed phenotyping of eight patients who took part in a functional imaging study by Matharu and colleagues reported that these patients had chronic migraine [Matharu et al. 2004]. The patients could vary their stimulation parameters; the range of stimulation settings used were pulse width of 90–180 μs, amplitude of 1.5–10.5 V, and frequency of 60–130 Hz. These eight patients were followed up for an average of 1.5 years. ONS resulted in excellent response in four patients (complete suppression or rare breakthrough headaches), very good response in two (complete suppression most of the time with breakthrough headaches on approximately 10 days per month), and good response in two (continued constant headaches but with severity reduced by 50–75%). All patients reported that they had managed to either completely stop or considerably reduce the headache medications they were taking. However, it is important to appreciate that this cohort was selected for the functional imaging study, the inclusion criteria for which was a good response to ONS and therefore the rapid response reported is not representative for this patient group. Three patients had lead migration requiring surgical revision and one patient developed an abdominal haematoma due to implantation of the IPG.
Schwedt and colleagues [Schwedt et al. 2007a] used ONS in eight patients with IHS-defined chronic migraine that was intractable to medical treatments. Each patient suffered from head pain that involved the C2 distribution with or without pain in other regions of the head. All patients underwent a 5–7-day percutaneous stimulator trial prior to permanent placement. Three patients had unilateral and five patients bilateral stimulation with percutaneous cylindrical leads. Patients were able to vary the stimulator parameters and to make changes in their headache medications during the follow-up period. After a mean follow up of 18 months, 3-month headache frequency reduced from 90 to 60 days while the severity decreased from VRS 6.8 to 4.5. The patients reported a significant reduction in disability and depression after the ONS operation. Three patients required revision for lead migration and one required IPG revision. One patient reported neck stiffness and one reported incision site pain.
These open-label studies of ONS in medically intractable chronic migraine, often in association with medication overuse, were very encouraging with 43 of 51 patients (84%) reporting at least 50% improvement. This led to the assessment of ONS for the Treatment of Intractable chronic Migraine (ONSTIM) in a multicentre, randomized, single-blind, controlled study [Saper et al. 2011]. The study recruited 66 patients who met the revised International Classification of Headache Disorders (ICHD-II) criteria and had responded to occipital nerve blocks. The patients experienced migraine for an average of 22 years (range 1–51 years) and chronic migraine for an average of 10 years prior to the study enrolment (range <1–30 years). Patients were randomized in a ratio of 2:1:1 to adjustable stimulation, preset stimulation and medical management. The responder rate was defined as 50% reduction in headache days/month or at least three-point drop (on VRS 0–10) in pain intensity at 3 months. The responder rate was 39% in the adjustable stimulation group compared with 6% in the preset stimulation group and none in the medical management group. Lead migration occurred in 12 of 51 patients (24%). These data suggest that occipital nerve blocks are not predictive of response to ONS. This feasibility study had various limitations. Blinding is difficult in ONS trials and it is possible that paraesthesia secondary to stimulation and the ability to control the stimulation parameters resulted in a significant placebo effect in the adjustable-stimulation group. Nonetheless, these results suggest that ONS might represent an important therapeutic option for at least a subset of patients with chronic migraine, though further studies are required to properly define the therapeutic response and adverse effect profile.
Cluster headache
Magis and colleagues prospectively studied eight patients with medically intractable chronic cluster headache (CCH) treated with unilateral ONS [Magis et al. 2007]. The mean duration of cluster headaches was 13.6 years, with a mean duration of CCH of 5.1 years. The mean stimulation parameters used were amplitude of 6.36 V (range 2.4–10 V), frequency of 66 Hz (range 40–100 Hz) and pulse width of 364 μs (range 270–450 μs). All patients had continuous stimulation. Headache diaries were completed prospectively. After a mean follow up of 15 months (range 3–22 months), two patients were pain free, three patients had a 90% reduction in attack frequency, two patients had improvement of around 40% and one had no benefit. Cessation of ONS was followed within days by recurrence and increase of attacks. Weekly headache frequency was reduced from 13.4 pre to 2.8 post ONS. Attack intensity was reduced from 2.62 to 1.47 post ONS [scale from 1 (mildest) to 4 (worst pain)]. All the patients who responded to ONS were able to substantially reduce their preventive drug treatment, but only one was able to stop them completely. Treatment effects from ONS were often delayed for 2 months or more after implantation. Two patients with relief of ipsilateral cluster attacks began to have the attacks on the contralateral side, which disappeared after suboccipital injections of steroids, and one patient experienced isolated painless autonomic attacks. When stimulators were turned off or batteries became depleted during the study period, the attacks worsened within days. Battery depletion occurred in half of the patients during the follow-up period. Lead displacement and electrode migration were reported in one patient each. No serious adverse effects were reported.
Burns and colleagues treated 14 patients with medically intractable CCH using bilateral ONS [Burns et al. 2009, 2007]. Patients had CCH for a median duration of 6 years (range 2–17 years). Patients were able to vary the stimulation parameters and a wide range of settings were used: the range for amplitude was 0–10.5 V, pulse width 60–450 μs, and frequency 3–130 Hz. Twelve patients used continuous stimulation while two used intermittent stimulation. The median follow-up period was 17.5 months (range 4–35 months). Ten of 14 patients reported an improvement, although 11 patients recommended the treatment to other patients with CCH. Three patients noticed a marked improvement of 90% or better, three reported a moderate improvement of 40–60%, and four reported a mild improvement of 20–30%. Four patients did not notice any benefit. Improvement occurred within days to weeks for those who responded most and patients consistently reported that their attacks returned within hours to days when the device was off. One patient found that ONS helped to abort acute attacks. The mean battery life was 15 months, therefore almost half of the patients required battery replacement. Lead migration was reported in four patients (29%). Other complications included muscle recruitment, neck stiffness, skin discomfort, superficial infection and painful paraesthesias.
Schwedt and colleagues used ONS in three patients with CCH [Schwedt et al. 2007a]. Two had unilateral and one had bilateral ONS. Mean follow up was 20 months (range 10–39 months). The efficacy of ONS was inferior compared with the other studies, albeit that this is a very small series. The headache frequency remained unchanged in two patients and increased in the third, while severity decreased in two patients (VRS 8 to 5 and 8 to 3.5) and remained unchanged in one. Lead migration was also confirmed as the main adverse event in these patients.
Though the published open-label studies report encouraging results of ONS in CCH, they have also highlighted some issues regarding the design of the trial, the proper patient selection and the assessment of the outcome. In the prospective pilot study of patients with CCH by Magis and colleagues [Magis et al. 2007], the average attack per day was quite low and the follow up after the operation was very short for some of the patients. In the retrospective study of eight patients with CCH [Burns et al. 2007] there was a discrepancy between patient estimation of benefits and objective improvement. This led Leone and colleagues to make the following recommendations about the criteria for the use of neurostimulation in primary headaches [Leone et al. 2007]: patients should have daily or almost daily attacks over 2 years; all reasonable drugs must be tried at sufficient doses for sufficient time periods unless contraindicated; an adequate length of post-implant follow up is required (at least 1 year); a psychological assessment is required before surgery; patients should keep a prospective headache diary (frequency, severity and duration of headache attacks, and analgesic consumption); and, quality of life measurements and the self-assessment of pain are required.
Hemicrania continua
Schwedt and colleagues described a patient with a 12-year history of HC who had to discontinue indometacin because of intolerance [Schwedt et al. 2006]. She had a unilateral bion device implanted that provided continuous stimulation with a pulse width of 300 ms, frequency of 45 Hz and amplitude of 3–7 V. The patient had significant improvement in pain, becoming pain free at baseline with superimposed severe headaches occurring five times in 3 months. The patient reported episodes of autonomic activation without headache. Schwedt and colleagues also treated two patients with HC with ONS [Schwedt et al. 2006]. Both patients had responded to indometacin but had to discontinue it because of intolerance. One patient had unilateral stimulation and the other had bilateral. At follow up of 13 and 21 months, the 3-month headache frequency had dropped from 90 days in both patients to 10 and 12 days while the pain severity had reduced from VRS 7.5 down to VRS 3–7. Both patients had lead migration and one had infection.
Burns and colleagues described six patients with HC with a suboccipital bion device implanted ipsilateral to their headache [Burns et al. 2008]. The patients had HC for a median duration of 14 years (range 6–36 years). The patients had continuous stimulation at a frequency of 60 Hz and pulse width of 250 μs. A crossover study design was used: the bion was on for 3 months, off for the fourth month and on again during long-term follow up. At a median follow-up of 13.5 months (range 6–21 months) four patients reported substantial improvement (80–95%), one reported 30% improvement and one reported that the pain had worsened by 20%. The onset of the benefit was delayed by days to weeks, and the headaches did not recur for a similar period when the device was switched off. Adverse events were mild and associated with transient overstimulation. While the authors claim that all these patients satisfied the revised ICHD-II criteria for HC, a review of the medical notes revealed that one of the patients had an incomplete response to oral, rectal and parenteral indometacin tests. This patient has chronic migraine rather than HC and is the patient who worsened with the bion device. Hence, all the patients with HC reported thus far have improved with ONS albeit that the improvement was relatively mild in one patient.
SUNCT and SUNA
Matharu and colleagues reported the outcome of seven patients with medically intractable SUNCT and one patient with short-lasting neuralgiform headache attacks with autonomic symptoms (SUNA) treated with bilateral ONS [Matharu et al. 2010]. These patients failed to respond to at least five of the following preventive treatments: lamotrigine, topiramate, gabapentin, pregabalin, carbamazepine, oxcarbazepine, mexiletine and melatonin. At a median follow up of 24 months (range 4–29), four patients reported a substantial improvement (95–100%), one reported moderate benefit (50%), one reported a temporary marked benefit (50%) for 6 months followed by recurrence of headache at the pre-ONS baseline, and one patient’s condition failed to respond. The onset of the benefit was rapid (within 2 weeks) with attacks recurring rapidly when the stimulator was switched off or malfunctioned. One patient developed HC 1 month after implantation and was successfully treated with indometacin. No major adverse events were reported.
Safety
ONS is a relatively safe procedure with no reports of any serious adverse events, except for the development of HC in one patient, though this was readily treatable with indometacin [Shanahan et al. 2009]. Other adverse events reported include lead migration, lead site pain, myofascial incision site pain, neck stiffness, discharged battery, battery site pain and contact dermatitis. Jasper and Hayek reviewed the safety of ONS [Jasper and Hayek, 2008]. They reported that lead migration occurred frequently with percutaneous cylindrical leads, being described in 26% of patients, but occurs in only 6% of patients implanted with paddle-type leads. Recent development of anchoring techniques will likely reduce cylindrical lead migration. Battery failure due to depletion is an expected event requiring IPG replacement and therefore should not be considered a complication. However, rapid battery depletion (less than 1 year) is reported and highlights the need for less expensive and longer-lasting power sources, especially for patients who require high voltage and frequency to control the pain [Trentman and Zimmerman, 2008]. The advent of rechargeable devices will obviate this in clinical practice.
Mechanism of action
The mechanisms by which peripheral neurostimulation mediates the antinociceptive effect are poorly understood. Several sites of action within the peripheral and central nervous system have been proposed, including the peripheral nerve, spinal segmental level and supraspinal levels. It is likely that peripheral neurostimulation exerts its effect by multiple mechanisms and that these mechanisms may differ in the various headache and pain syndromes.
Direct effects of neurostimulation on peripheral nerve fibre excitability have been described, including transient slowing of conduction velocity, increase in electrical threshold and decrease in response probability [Ignelzi and Nyquist, 1979]. However, ONS did not significantly modify pain thresholds in CCH [Magis et al. 2007], which argues against a diffuse analgesic effect.
A widely accepted theory for the antinociceptive effect of neurostimulation is the gate-control theory of pain which proposes that the activation of large diameter afferent nerve fibres in the spinal dorsal horn inhibit onward transmission in small diameter primary afferent nociceptive fibres, thereby preventing the nociceptive signals from reaching the higher neural centres and being interpreted as pain [Wall, 1978; Melzack and Wall, 1965]. Indeed, a number of physiological studies have confirmed that afferent activity set up by peripheral neurostimulation blocks nociceptive transmission in the spinal cord [Garrison and Foreman, 1996; Chung et al. 1984b; Woolf and Wall, 1982]. The explanation for the antinociceptive effect of SCS according to the gate-control theory is that nociceptive input from the periphery could be inhibited at the first dorsal horn relay by stimulation-induced antidromic activation of collaterals of large dorsal column fibres projecting onto the same spinal segment [Dubuisson, 1989; Foreman et al. 1975; Hillman and Wall, 1969]. However, the gate theory does not adequately explain some of the animal and human experimental data and therefore several additional mechanisms of action for neurostimulation have been postulated. Some of the theories proposed include: activation of supraspinal mechanisms; alteration of putative neurotransmitter levels; and blockade of sympathetic mechanisms [Linderoth and Meyerson, 2002; Hansson and Lundeberg, 1999; Krames, 1999].
The involvement of supraspinal sensory pathways is a requisite for the orthodromic transmission of the activation resulting from neurostimulation. The key issues are whether the ascending and descending pain pathways are involved in mediating an antinociceptive effect and, if so, then which supraspinal structures are involved. Antinociception in animal models produced by sensory afferent stimulation is reduced by spinal transaction, thus implicating the involvement of supraspinal mechanisms [Chung et al. 1984a; Woolf et al. 1980]. Similarly, with SCS it has been argued that the inhibitory effects on nociceptive transmission in the spinal dorsal horn cannot be entirely attributed to antidromic activation of the dorsal columns because they persist after dorsal column transection caudal to the stimulating electrode [Saade et al. 1985]. Furthermore, on the basis of animal studies, various supraspinal structures have been proposed as candidates for mediating the antinociceptive effect including the periaqueductal grey (PAG) [Stiller et al. 1995] and thalamus [Nyquist and Greenhoot, 1973].
A positron emission tomography (PET) study investigated the effect of SCS in patients with refractory angina pectoris [Hautvast et al. 1997]. The study was performed when the patients were not in pain and therefore the regional cerebral blood flow changes reflect the effects of SCS solely. During stimulation, activation was noted in the PAG, dorsomedial and the pulvinar nuclei of the thalamus, prefrontal cortex, medial temporal gyrus, posterior caudate nuclei and posterior cingulate cortex, while a relative decrease in regional cerebral blood flow was observed in the insulae and anterior cingulate cortex. Furthermore, a functional magnetic resonance imaging study in three patients with chronic pain syndromes showed primary and secondary somatosensory cortex and anterior cingulate cortex activation with SCS [Kiriakopoulos et al. 1997].
A PET study investigated the brain structures modulated by ONS in chronic migraine [Matharu et al. 2004]. Eight patients with a marked beneficial response to bilateral ONS were studied in three states: during stimulation when the patient was pain free; during pain with the ONS switched off; and during partial stimulation and varying levels of pain and paraesthesia. Stimulation suppressed the headache within 30 min and pain recurred within 20 min of switching off the device. Stimulation evoked local paraesthesia, the presence of which was a criterion of pain relief. There were significant changes in regional cerebral blood flow in the dorsal rostral pons, anterior cingulate cortex and cuneus correlated to pain scores, and in the anterior cingulate cortex and left pulvinar correlated to stimulation-induced paraesthesia scores.
The activation pattern in the dorsal rostral pons in this study is highly suggestive of a role for this structure in the pathophysiology of chronic migraine. However, this brainstem region may also be a locus for neuromodulation by ONS. The PAG has long been proposed as a candidate for mediating the antinociceptive effects of neurostimulation. Stiller and colleagues performed microdialysis studies on transmitter release in the PAG of rats receiving SCS [Stiller et al. 1995]. They observed that SCS affected a decrease in gamma-aminobutyric acid (GABA) levels but not of serotonin or substance P. As GABA neurons in the PAG exert a tonic depressive effect on the activity in descending pain inhibitory pathways, the authors proposed that a decreased GABA level in this region following repeated SCS might indicate increased pain inhibition.
Magis and colleagues studied 10 patients with medically intractable CCH treated with ONS and 39 drug-free healthy volunteers using 18-fluorodeoxyglucose (FDG) PET [Magis et al. 2011]. The 10 patients with CCH underwent an 18 FDG PET scan after ONS, at delays varying between 0 and 30 months. All were scanned with ongoing ONS and with the stimulator switched off. After 6–30 months of ONS, three patients were pain free and four had a reduction of attack frequency of at least 90% (patients who responded). In all patients compared with controls, several areas of the pain matrix showed hypermetabolism, including the ipsilateral hypothalamus, midbrain and ipsilateral lower pons. All normalized after ONS, except for the hypothalamus. Switching the stimulator on or off had little influence on brain glucose metabolism. The perigenual anterior cingulate cortex was hyperactive in patients who responded to ONS compared with those who did not respond.
These results may support the hypothesis that ONS exerts its beneficial effects via slow neuromodulatory processes in the central pain matrix. The finding of a possible selective perigenual anterior cingulate cortex (PACC) in patients who responded raises the possibility that ONS activates descending pain control systems in a top-down manner and restores an equilibrium in antinociceptive opioidergic pathways. The study also reported persistent hypermetabolism of the ipsilateral posterior hypothalamus outside of an attack, which might be a hallmark of cluster headaches and also explains why attacks rapidly recur after interruption of ONS.
Conclusions
ONS is emerging as a promising treatment modality for medically intractable primary headache syndromes, including migraine, cluster headache, HC and SUNCT/SUNA. It has already provided a better life for some patients and gives hope to many more. However, the ultimate confirmation of the utility of a new therapeutic modality should come from randomized, double-blind, placebo-controlled trials. This poses a special problem in designing blinded studies of treatment with stimulation since there is no placebo equivalent for the paraesthesia that accompanies stimulation. Any credible sham device would, therefore, be required to produce a discernible stimulus, which could then be criticized for providing neurostimulation. Nonetheless, it is possible to work around this challenging problem with particular trial designs such as that utilized in the ONSTIM trial. Further large-scale controlled trials are required to differentiate the neuromodulatory effects from the nonspecific effects, such as placebo response, regression to the mean and spontaneous improvement.
Ideally, results from controlled trials should be available before widespread open-label use of ONS can be recommended, though it might be several years before these studies are available. Given the highly challenging nature of the management of medically intractable headache syndromes, most of which are very disabling, and taken together with the promising efficacy data for ONS, it seems reasonable to cautiously proceed with the open-label use of this procedure, preferably in tertiary referral settings. Patients who are considered for ONS should satisfy the following criteria: clearly established headache diagnosis using International Headache Society (IHS) classification criteria, with a careful workup including brain imaging, lumbar puncture and indometacin tests where appropriate; medically intractable as defined by the consensus statements [Goadsby et al. 2006], which suggests failure of a minimum of four classes of preventive treatments in migraine and cluster headache; and the patient is severely disabled by the headache syndrome.
With regard to future prospects, it would be helpful to have predictors of success with ONS. Occipital nerve blocks do not appear to be correlated with response [Schwedt et al. 2007b]. It is likely that large-scale studies will be required to tease out these predictors. The role of the surgical technique used, including the anchoring method, electrode type and optimal stimulation parameters are all variables that require further exploration. Finally, the mode of action of ONS is poorly understood and further studies are required to elucidate the underlying mechanisms by which the antinociceptive effect is exerted.