Abstract
Glossopharyngeal neuralgia (GN) triggered by non-noxious stimuli at multiple cephalic and extracephalic sites with positron emission tomography (PET) evidence for involvement of the upper brainstem has never been reported. We present such a patient, a 73-year-old man who since the age of 50 had suffered from GN with a high recurrence rate and very severe unilateral, non-familial GN episodes with very easy trigger zones widely extending beyond the n IX territory. Extensive neuroimaging and neurophysiological tests detected no precise underlying cause. PET scan revealed activation in the upper brainstem on extracephalic triggers. Single-fibre electromyography data will be discussed. We hypothesize that deficient inhibition as seen in trigeminal nociceptive reflexes on the level of brainstem interneurons, a functional lesion in the primary somatosensory cortex-sensory thalamic nuclei circuit and the dorsal column-thalamic pathway both activated by light touch may in part be involved in the extra- cephalic triggering.
Introduction
Glossopharyngeal neuralgia (GN) [International Classification of Headache Disorders, 2nd edn (ICHD-II) code 13.2](1) is a rare disease with an estimated incidence of 0.8/100 000, i.e. 0.2–1.3% of that found for trigeminal neuralgia (TN) (2, 3). The majority of GN cases are thought to be idiopathic, but in some of them neurovascular compression can be demonstrated on magnetic resonance imaging (MRI). The most frequent associated conditions are TN and hypertension (2, 3). Familial occurrence has not been reported. GN mostly presents in (ir)regular periods, and two-thirds of patients have only one episode in their life. Data on long-term follow-up are sparse and mainly concern the results obtained with neurosurgical interventions (4, 5). It has nonetheless been estimated that, after a first period of pain, the cumulative 5-year incidence of a second episode is 18% (2). The pain in GN is sharp, shooting and stabbing, and in the territory of one glossopharyngeal nerve, i.e. deep throat and ear. In 25% of patients GN might be bilateral (3). As in TN, the pain attacks are brief, lasting seconds or tens of seconds, and they occur spontaneously or are triggered by non-noxious mechanical stimuli in the sensory innervation territory of the nerve, so-called trigger zones. We present a patient who suffered from recurring periods of unilateral, very severe GN with remarkable clinical and functional imaging features.
Case report
This male patient, aged 73 years, was admitted to hospital in 2001 because of recurrence of very intense and disabling GN with gradual increase in pain intensity over the previous 6 weeks.
He was diagnosed to have had left-sided GN for 23 years and had suffered from several painful periods per year lasting about 2–3 months. His past medical history revealed arterial hypertension, myocardial infarction and a coronary artery bypass graft, for which he was treated with simvastatin, amlodipine, hydrochlorthiacide, isosorbide dinitrate and acenocumarol. He did not smoke or abuse alcohol. His family history was unremarkable and, in particular, devoid of subjects suffering from neuralgias or primary headache disorders.
Routine computed tomography (CT) scan and MRI of the head, routine blood analyses and ENT examinations had been found normal in the past, except for hyperintense deep white matter lesions detected on T2-weighted MRI, which were attributed to microangiopathy.
Over time, the neuralgia had been treated successfully with carbamazepine up to 850 mg/day, which was used for each pain period at variable duration. The present GN episode occurred despite treatment with carbamazepine 800 mg/day.
During the last 6 weeks, the sharp, lancinating pain attacks in the left deep throat region had progressively increased in intensity and lasted 15–30 s, or occasionally longer. The pain, scored by the patient at the maximum of 10 on a visual analogue scale, was in part synchronous with the cardiac rhythm and sometimes associated with paraesthesias of the left tongue.
His usual triggers were food–throat contact, especially with cold drinks, talking and sometimes head movements. At the time of hospitalization he had in addition multiple spontaneous GN attacks that occurred at any time of the day–night cycle. For the first time since suffering from the disorder the patient reported that stimulation of a number of zones outside the oropharyngeal area were now able to trigger pain in the left throat, e.g. non-painful cutaneous stimuli of the whole head, particularly the left jaw or (peri)auricular region, anterior neck and left shoulder region, but also mild and intense touching of his feet, distal legs and knees. Moreover, elevating the arms and bowing the trunk could provoke the typical pain. Palpation of the right radial artery at the wrist evoked a pulsating, left-sided neuralgic pain. He was able to sleep only for short periods in a sitting position. He reported abnormal sound perception in his left ear, but there was no hearing loss or any other abnormality on renewed ENT examination. Food intake had not been possible for weeks and needed a naso-gastric tube.
Neurological examination was normal. Blood pressure was 140/70 and heart rate 75/min. Autonomous function analysis revealed mild orthostatic hypotension. Extensive blood and cerebrospinal fluid examinations were normal, excluding in particular inflammatory or infectious diseases. Newly performed CT scans and MRI of the head and upper cervical region revealed the known deep white matter lesions without infarction.
CT scans centred on the base of the skull and C1–C2 vertebral levels were normal. Selective vertebrobasilar angiography was normal.
Recordings of blink reflex and upper and lower limb somatosensory evoked potentials were normal. Single-fibre electromyography (SFEMG) was out of the normal range because blocking was found in 5% of fibres and the mean of mean consecutive differences in latencies was 51 ms.
Positron emission tomography (PET) scanning of regional cerebral blood flow (the integrated brain radioactivity counts of the H2 15O scans were uses as a measure) with a high-resolution PET scanner (Siemens ECAT Exact HR+, 3D mode) was performed in two conditions. Each scan started around 30 s after the injection of 300 MBq H2 15O and lasted 1 min. The first scan was acquired after the induction of the GN pain by touching the left knee and left elbow region. After 15 min (the period for decay of the tracer) the baseline scan was acquired. The patient did not suffer from pain at that moment. Data were reconstructed according to the standard protocol at that time. Images were transformed to the Montreal Neurological Institute template space and subtracted (6). Voxels showing a difference of z > 2 were identified. Clusters of > 200 voxels in size are reported. Hyperperfused regions were found in the left orbitofrontal gyrus, the pons, the calcarine region, the right uncus, amygdale and the border of the right inferior and middle frontal gyrus. Hypoperfused regions were found in the bilateral cerebellum, right postcentral gyrus, cingulated gyrus and bilateral medial frontal gyrus (Tables 1 and 2 and Fig. 1).
Changes in regional cerebral blood flow between the two scans: hyperperfusion (during headache) and hypoperfusion (without headache).
Regions and voxels with hypoperfusion during the headache condition
Threshold: z > 2.00; cluster size > 50.
Regions and voxels with hyperperfusion in the headache condition
Threshold: z < −2.00; cluster size > 50.
A condition with spontaneous GN pain could not be induced. Anti-neuralgic medication when the PET study was performed comprised gabapentin 3200 mg/day and baclofen 15 mg/day. Subsequently, carbamazepine was replaced by varying combinations of gabapentin 3200 mg/day, buprenorfine 0.8 mg/day, amitriptyline 75 mg/day, acetazolamide 1000 mg/day, sodium valproate 2000 mg/day and baclofen 30 mg/day—all without pain reduction. Because of lack of efficacy of drug treatments a radiofrequency lesion of the proximal extracranial part of the glossopharyngeal nerve was performed and resulted in 9 days' pain freedom. The patient was discharged using the nasal tube. After recurrence of the pain, carbamazepine 400 mg/day was restarted and resulted in disappearance within some days of the GN.
At the present time, 8 years later, there has been no recurrence of pain. [Correction added after online publication 9 April 2009: The last sentence of the Case history has been changed.]
Discussion
The patient reported here suffers from a severe type of GN because of the high recurrence rate of pain episodes since the age of 50, the high daily frequency of spontaneous and triggered attacks and their severe intensity, which necessitated artificial feeding through a nasogastric tube. Despite extensive and repeated investigation, no precise underlying cause for the GN could be detected, suggesting that the diagnosis is ‘classical’ GN (ICHD-II code 13.2.1).
Nevertheless, this patient is unique in his non-classical and/or hitherto non-reported clinical and imaging features.
The easy triggering of the GN pains by non-noxious heterotopic cutaneous stimuli applied almost anywhere over the body is remarkable since, to the best of our knowledge, it has been reported only twice before in patients with trigeminal neuralgia (7, 8). Generalized triggering did not occur in all GN episodes and was a characteristic feature only during the last pain episode that led to the patient's hospitalization. The precise mechanisms of TN and GN are still incompletely understood.
Hyperexcitability of trigeminal nociceptors is a common denominator, but the question whether the culprit is bursting of ganglion neurons, the so-called ‘trigeminal ganglion ignition hypothesis’(9), or ephaptic transmission in lesioned afferents close to the dorsal root entry zone (10) or abnormal firing of second-order nociceptors in trigeminal nucleus caudalis (11) remains controversial. A common feature of TN is that pain can be triggered by light touch in the affected territory, i.e. by activation of large myelinated Aβ fibres. Again this could be due to the induction of autorhythmic firing in sensory ganglion cells or in more proximal sites involving, for example, polymodal nociceptors in the trigeminal nucleus caudalis. Allodynia, thought to be due to central sensitization, is not found in affected trigeminal territories, suggesting that the pathophysiology underlying the neuralgias is different. The possible location of trigger zones outside the affected branch is not exceptional in TN and GN (12), contrary to the triggering of pains in extracephalic sites.
The widespread triggering of paroxysmal pains by light touch and limb movement suggests that low-threshold tactile and proprioceptive afferents were able to trigger the pathophysiological process. Large cutaneous and deep myelinated Aβ fibres are known to inhibit pain at the segmental level. To the best of our knowledge, there is no direct anatomical connection between peripheral limb afferents and second- or third-order trigeminal nociceptors. It is known, however, that non-noxious heterotopic stimuli are able to inhibit trigeminal nociceptive reflexes involving brain stem interneurons (13, 14) and that this inhibition is deficient in migraine and tension-type headache (15). We have suggested that the underlying pathway involves descending connections to trigeminal interneurons from upper brainstem centres including the periaqueductal grey matter (PAG) (13). On the other hand, corticofugal output from the primary somatosensory cortex activates third nociceptors in sensory thalamic nuclei of normal rats (16), and the dorsal column–thalamic pathway contributes to thalamic hyperexcitability in rats with an experimental mononeuropathy (17). Both dorsal columns and somatosensory cortex are activated by light touch in peripheral limbs, and the above-mentioned experimental data might be relevant for the findings in our patient. However, in both studies only homotopic areas were explored, and it remains to be demonstrated that heterotopic non-noxious stimuli are equally able to enhance activity in trigeminal nociceptive processing.
An alternative explanation for our clinical findings might involve the low-threshold unmyelinated tactile C fibre system (18), which is known to mediate affective and autonomic aspects of pain (19). This fibre system projects indeed to the insula (20), which in turn is known to project to the PAG (21). Assuming a dysfunction of central pain control systems in our patient, this pathway could potentially lead to increased discharges of nociceptors in trigeminal nucleus caudalis. Among the latter, those innervated by polymodal afferents are indeed activated by descending input from the PAG, in contrast to nociceptors with exclusive C fibre input, which are inhibited (22).
Whatever the mechanisms underlying the triggering of pain attacks by extracephalic tactile stimuli, the PAG seems to occupy a central position in the suspected neuronal circuitries. It is thus of major interest that GN attacks triggered by extracephalic brush stimuli were associated on PET scanning with hyperactivity in the rostral part of the brainstem that includes PAG. Activation in this area is known to occur in spontaneous (23) and experimental pain (24), during allodynia (25) and migraine attacks (26). The pathophysiological mechanisms might vary between these different conditions, and none of them is phenotypically identical to the clinical presentation of our patient. Nonetheless, the above discussed pathways involving large afferent fibres or tactile C fibres may in theory both trigger GN pains via activation of PAG.
Unfortunately, we cannot exclude that drug treatments at the time of recordings were totally or in part responsible for the abnormal SFEMG findings in our patient, and for this reason we will not emphasize them here. Assuming that the SFEMG abnormalities were genuinely linked to the underlying disorder, it suffices to mention that a decrease of the safety factor at the neuromuscular junction as indexed by SFEMG has been reported in migraine with aura and cluster headache (27). The molecular underpinnings of these subclinical abnormalities are not known, but the fact that they are corrected by acetazolamide, which also has clinical benefit in certain neurological channelopathies, suggests that ion channels and/or pumps may be involved (28). The latter are known to be dysfunctioning in monogenic forms of migraine (29) and neuropathic pain (30). It can therefore not be excluded that the abnormal SFEMG results, if they are not drug related, might reflect a hitherto unrecognized genetic dysfunction of ion channels.
In summary, we have described a patient with GN in whom, at the most advanced and treatment-resistant stage of the disorder, trigger zones to light touch were present both at cephalic and at widespread extracephalic sites. Triggering of attacks by extracephalic stimuli was associated with PET activation in the upper brainstem. We hypothesize that the extracephalic triggering involves large myelinated cutaneous afferents or tactile cutaneous C fibres, which would via upper brainstem structures, including PAG, induce firing of second- or third-order trigeminal nociceptors.