Abstract
Nociceptive processing in trigeminal system is likely to undergo to significant changes in chronic pain disorders such as in primary headache.
The electrophysiological testing of brainstem reflexes seems to offer a valuable adjunct to the clinical evaluation, providing direct evidence for central dysmodulation of sensory and pain neurotransmission, i.e. dishabituation pattern and reduced pain threshold.
We review the contribution of trigeminofacial reflex studies in the light of those recent data supporting the view of a dysfunctional sensory processing in primary headache.
Introduction
The chapter of the primary headaches embraces some quite heterogeneous disorders. Activation of the trigeminovascular system, in the presence or absence of ipsilateral autonomic features, characterizes some of the most prominent headache syndromes such as migraine and cluster headache, as well as the so-called trigeminal autonomic cephalgias (TACs) (1). The highly prevalent tension-type headache, on the other hand, remains the least well-defined headache category, in which an imbalance between peripheral myofascial factors and central mechanisms could play a pathogenetic role.
In all the above headaches, the neuronal structures involved in the pain mechanisms range from peripheral structures (both the trigeminal afferent fibres innervating the cerebral vessels and dura mater, and those innervating muscles, tendons, ligaments and joints) to central structures, including brainstem nuclei and the cerebral cortex. A number of neurochemicals (calcitonin gene-related peptide, substance P, neurokinin A, prostaglandins) (2) are involved in the activation of the peripheral endings, and their sensitization is thought to be an important factor contributing, in terms of functional neuroplastic changes in pain sensitivity, to headache disorders (3).
Brainstem structures play a crucial role not only in conveying nociceptive impulses but, even more importantly, in modulating them. Descending modulation of sensory transmission occurs largely in the brainstem. Here, influences derived from higher centres, including the somatosensory cortex, hypothalamus, anterior cingulate cortex and other limbic structures, converge with those from structures within the brainstem itself (e.g. periaqueductal grey matter, locus coeruleus, nucleus raphe magnus and reticular formation) resulting in the integration of pain, emotion and autonomic responses (4).
The role of brainstem mechanisms in neurovascular headaches was also underlined in a positron emission tomography study of migraine subjects during attacks that reported areas of increased blood flow in the regions of the dorsal raphe nucleus and of the locus coeruleus (5). Since then, brainstem areas have been considered migraine generators or, at least, migraine modulators.
It is necessary for the development of new treatment strategies and for therapeutic purposes to be able to recognize clinically, and, possibly, to characterize, through full electrophysiological testing, the central and peripheral neuronal state in primary headache in order to assess objectively whether changes in the nociception process are primary or develop in the long term.
In the light of recent research that has identified critical neuronal elements and pathways in the brainstem as well as some of the mechanisms involved in the modulation and plasticity of craniofacial nociceptive transmission, our aim, in this paper, is to review the contribution of objective electrophysiological testing of the brainstem structures in primary headaches. In particular, we look at studies of trigeminofacial reflexes, mainly blink and corneal reflexes, which have been extensively applied in headache patients.
Trigeminofacial reflexes: anatomy and physiology
Anatomical pathways of trigeminofacial reflexes
The trigeminal nerve is the largest and most complex of the 12 cranial nerves. It has the greatest peripheral sensory distribution and the highest central brainstem representation. These essential anatomical features explain both the feasibility of electrophysiological investigation of the trigeminal nerve and the many studies that have explored the trigeminofacial circuits in several craniofacial pain disorders.
The trigeminal nerve conveys sensory information from most parts of the head and partly from the neck, allowing perception of the intensity, quality, location and duration of noxious stimuli. Nociception derives essentially from meningeal structures subserved by all the three divisions of the trigeminal nerve, but mostly from the ophthalmic division (6).
The blink reflex (BR) consists of bilateral eyelid closure in response to a stimulus applied in the area innervated by the trigeminal nerve or directly on a branch of the trigeminal nerve. Other types of stimulation (e.g. light, acoustic stimulus) can also induce a BR response (7).
The anatomical pathways subserving brainstem reflexes are largely known, and are shared by the trigeminofacial reflexes most commonly explored in the laboratory setting. They are exteroceptive reflexes with a sensory afferent limb made up, in the electrically evoked BR, of cutaneous trigeminal fibres, both exteroceptive and nociceptive A beta, and A delta and C fibres. The efferent limb consists of the motor fibres arising from the nucleus of the facial nerve. The afferents innervating cutaneous, intraoral, deep (i.e. joints, muscles, tendons) and cerebrovascular tissues project to the trigeminal brainstem complex. This can be subdivided into the main or principal sensory nucleus and the spinal tract nucleus, which comprises three subnuclei: oralis, interpolaris and caudalis. On the basis of anatomical, clinical and electrophysiological observations, the subnucleus caudalis is usually viewed as the principal brainstem relay site of trigeminal nociceptive information. The nociceptive inputs are conveyed predominantly at laminae I, II and V and VI.
Electrophysiological studies have revealed that many neurones respond to cutaneous nociceptive inputs. These nociceptive neurones have been categorized as nociceptive-specific (NS) neurones or wide dynamic range (WDR) neurones.
The NS neurones respond only to noxious stimuli (i.e. pinch, heat) applied to a localized craniofacial receptive field (RF) and receive small-diameter afferent inputs from A delta and/or C fibres. WDR neurones are excited both by non-noxious (i.e. tactile) and noxious stimuli, and may receive large- and small-diameter A-fibre inputs as well as C-fibre inputs. Many NS and WDR neurones in the subnucleus caudalis can be excited only by natural stimulation of cutaneous or mucosal tissues, and have properties consistent with a role in the detection, localization and discrimination of superficial noxious stimuli (8).
The extensive convergent afferent input patterns that are characteristic of temporomandibular joint (TMJ) or myofascial-activated NS and WDR neurones in the subnucleus caudalis may explain the poor localization of deep pain, as well as contribute to the spread and referral of pain, which are typical of deep pain conditions involving the TMJ and associated musculature. Analogous anatomical and physiological features characterize the spinal dorsal horn, which is also a laminated structure and which represents the integral component of spinal nociceptive mechanisms. Consequently, the subnucleus caudalis is now often termed the ‘medullary dorsal horn’. Although a crucial role in craniofacial nociceptive mechanisms has been documented for the subnucleus caudalis, recent studies have also implicated the rostral components of the trigeminal brainstem complex subnucleus oralis. However, the relative importance and role, in both deep and superficial pain mechanisms, of the rostral and caudal components of the trigeminal brainstem complex is still unclear (8).
Neurones at all levels of the trigeminal brainstem complex project to brainstem regions including the reticular formation and nerve motor nuclei; their connectivity to these particular regions provides the central substrate underlying autonomic and muscle reflex responses to craniofacial stimuli. The ventro-basal nociceptive neurones have properties and connections with the overlying somatosensory cerebral cortex, indicative of a role for most of them in the sensory–discriminative dimension of pain. In contrast, nociceptive neurones in the more medial nuclei (e.g. intralaminar nuclei; parafascicular nucleus) are generally considered to have properties and connections (e.g. with the anterior cingulate cortex) more indicative of a role in the affective or motivational dimensions of pain.
Components of the blink reflex
The BR consists of three components: an ipsilateral early component (R1) that
follows an oligosynaptic pontine pathway passing through the principal
trigeminal nucleus; a bilateral late component (R2) with a multisynaptic bulbar
pathway descending to the spinal trigeminal nucleus and then to the lateral
reticular formation as far as the facial nuclei; a bilateral ultralate component
(R3) that probably follows the same central pathways as R2. The R2 and, notably,
R3 are considered nociceptive components of the BR (7). However, very recently,
Ellrich et al. (9) showed that cutaneous A beta and A delta, but not C fibres constitute
the generators of the electrically evoked R3 component. As R2 and R3 components
can also be evoked by acoustic stimuli (10), some authors consider the
R3 component, which shows a marked tendency to habituate and tends to be
systematically suppressed after an alerting prestimulus
Corneal reflex
In the clinical setting the corneal reflex (CR) is commonly evoked by means of a light mechanical stimulus applied to the cornea using the folded corner of a piece of cotton. The ultimate function of bilateral eyelid closure is to protect the eyes against potential noxious stimuli.
In fact, the sensory projection from the cornea is mainly nociceptive, represented by free nerve endings in the stroma and epithelium of the cornea, which act as nociceptors (12). Some axons gain myelin just after this point, forming small myelinated fibres that, together with nonmyelinated fibres, project through the long ciliary nerves, the terminal branches of the ophthalmic division of the trigeminal nerve, and then enter the pons. From the pons they descend to the lateral region of the medulla oblongata as far as the outer laminae of the ipsilateral spinal trigeminal nucleus. The efferent arc of the reflex pathways goes back up to the pons to the homo- and contra-lateral motor facial nuclei, following a polysnaptic pathway in the lateral reticular formation on the medial side of the trigeminal trait and nucleus complex (13).
The CR is a typically nociceptive response. By means of electrical stimulation of the cornea, we induced the reflex activation of the orbicularis oculi muscle (14, 15). This alternative method of evoking a BR allows a precise quantitative analysis of the pure nociceptive function of the trigeminofacial circuits.
The CR consists of a polyphasic response of 40–60 ms latency made up of two late bilateral symmetrical components, probably equivalent to the R2 component (16).
Even though blink and corneal reflexes share most parts of the central trigeminofacial circuits, they differ in several respects. Firstly, the BR is made up of three distinct components, the early ipsilateral component (R1) and two bilateral components, the late and ultralate R2 and R3 (10), while the corneal reflex has only a bilateral component that is homologous, in terms of latency (about 10 ms later) and morphology, with the R2 component. Secondly, they differ at receptor level, as the BR is mainly a cutaneous reflex elicited by capsulated mechanoceptors with inputs travelling through A beta fibres. The CR, on the other hand, is nociceptive in nature, arising from free nerve endings embedded in the corneal ephithelium with nociceptive A delta afferent fibres. Moreover, the central connections of the CR are only partly shared with the R2 component, as the nociceptive afferents end on the second-order neurones of laminae I, II and V, VI of the subnucleus caudalis, while the primary A beta afferents end both on lamina III and lamina VI of the subnucleus caudalis, whereas tactile and nociceptive inputs converge on WDR neurones even at more rostral levels of the trigeminal complex (8). Finally, there is evidence suggesting the involvement of fewer fibres and interneurones in the CR, and thus that the BR offers a greater security factor than the CR (17, 18).
Neurotransmitters and trigeminal nociception
The activity of V nerve nociceptive brainstem neurones can be modulated or suppressed by endogenous neurochemicals such as opioids, serotonin (5-hydroxytryptamine; 5-HT), or gamma-amino butyrric acid (GABA). These modulatory influences are of clinical significance as they have been implicated as intrinsic mechanisms contributing to the analgesic effects of several therapeutic approaches, e.g. deep brain stimulation, acupuncture and opioid-related drugs.
A large number of experimental studies have investigated not only the anatomical pathways but also the neurotransmitters involved in the modulation of the BR.
In humans, an inhibitory effect on the trigeminofacial reflex after the administration of opiates, benzodiazepine (19), analgesics (20) and zolmitriptan (21) has been documented. TENS is also able to inhibit the BR (22).
Habituation and sensitization phenomena
Repetitive stimulation often results in a progressive decrease in the elicited response. This pattern, termed habituation, is a prominent feature of the BR. The response reduction that results from exposure to repeated and constant stimuli is a ubiquitously expressed feature at the level of the CNS neural network. The rate and degree of the decrement are proportional to the number of stimulus presentations, to the stimulation frequency and to the stimulus intensity (23).
Sensitization is defined as a response increment resulting from novel, strong or noxious stimulation. In contrast to habituation, stimulus intensity is a more important factor than stimulus repetition in determining sensitization (24).
Habituation is considered an elementary form of learning and a protective mechanism as, from a teleological standpoint, it serves to decrease the response to a stimulus whose informational value has decreased as a result of its repetition, thereby preventing overstimulation and/or a waste of energy. In contrast, sensitization serves to increase rapidly the response to a stimulus whose informational value is, on the basis of its initial novelty or strength, deemed high.
These concepts are nowadays well-established, but historically their first description by Peckham and Peckham in 1887 challenged the idea that reflexes are invariable, a notion that was generally accepted at the beginning of the twentieth century, and continued to prevail until Thomson and Spencer, in 1966, presented a list of nine criteria defining habituation (23).
It is generally assumed that peripheral mechanisms of sensory adaptation and effector fatigue are not part of the concept of habituation, the latter being considered an essentially central phenomenon. Cellular and neurochemical mechanisms underlying habituation are still not fully understood. Investigation of neuronal circuits of varying degrees of complexity has revealed not only that mechanisms of synaptic depression in sensory neurones are involved in the habituating network, but that interneurones may also contribute to it, mediating the influences descending from higher centres in diencephalic and telencephalic regions (25).
The main factors considered to be important at presynaptic level are the reduction in readily releasable transmitter and the inactivation of the voltage-gated calcium channel in the synaptic terminal (25).
Multiple neurotransmitter systems seem to play a role in habituation. Indeed, the pathways arising from brainstem and diencephalic nuclei projecting to thalamocortical areas are responsible for a modulatory effect on cortical information processing, sustained partly by the serotonergic drive, which regulates the cortical preactivation level (26).
Trigeminofacial reflex in primary headache
Blink reflex studies
An early study of BR in migraine patients (27) reported an increased R2 latency pointing to a brainstem dysfunction, findings that several studies since have failed to replicate (28–30) (Table 1).
Blink reflex and corneal reflex in primary headache
CCH, chronic cluster headache; CPH, chronic paroxysmal hemicrania; CTTH, chronic tension-type headache; ECH, episodic cluster headache; ETTH, episodic tension-type headache; HC, hemicrania continua; MA, migraine with aura; MO, migraine without aura.
Only during the headache phase were lower amplitude and area of the R2 responses reported bilaterally (29). By contrast, applying a new method developed in order to obtain a selective activation of the A delta fibres, Kaube et al. (31) showed a marked increase, more prominent on the pain side, of the R2 component bilaterally compared with the headache-free interval, partially reversed by acute treatment with zolmitriptan or lysine acetylsalycilate. These findings have been interpreted as suggesting a temporary sensitization of central trigeminal neurones during migraine attacks.
Similar findings of an increased area of the ultralate R3 component of the BR have been reported during a migraine attack, reversed by treatment with zolmitriptan but not with sumatriptan or placebo (21). In another study, the same authors also reported reduced R3 thresholds in both cluster headache and migraine patients interictally, findings that suggest a permanent derangement of the descending pain control system and/or a sensitization of the trigeminal network by the concomitant neurogenic inflammation.
A previous study exploring the recovery curve of the R2 component of the BR performed in patients during cluster period lends further support to the hypothesis of a hyperexcitability of the central loop of the BR (32). The authors assessed the effect of conditioning stimuli on the R2, both at trigeminal and peripheral level, using paired stimuli at increasing interstimulus intervals. They showed a markedly enhanced recovery curve with both peripheral and segmental conditioning stimuli compared with controls. With the segmental conditioning the R2 recovered even more rapidly on the symptomatic side, suggesting unilateral sensitization of the trigeminal pathway.
Another author reported only R2 amplitude asymmetry in cluster headache patients (33), but did not investigate any healthy controls. Previous data relating to the BR recovery cycle in migraine patients at interstimulus intervals of 100, 200, 300 and 500 ms, failed to show any difference compared with controls (30).
A very recent study applied a new method (selective activation of cutaneous nociceptive fibres by means of a novel concentric electrode) of eliciting the BR (31). The examinations were performed using both the novel and the standard procedures, during and 2 h after treatment of a migraine attack, and the results compared with values obtained in the headache-free interval. They revealed, with the novel but not the standard procedure, a shortening of R2 latencies and an increased reflex integral (area under the curve; AUC) during the acute migraine episode, more prominent on the pain side. The symptomatic treatment reversed these changes with normalization persisting in the pain-free interval. The authors considered their results indicative of a temporary sensitization of central trigeminal neurones that occurs during acute migraine attacks, as facilitated nociception-specific BR responses were found.
Corneal reflex in primary headaches
A reduced threshold of the electrically induced CR has been found in cluster headache patients during active period, with lower values on the pain side (14) compared with both healthy controls and with the same subjects in remission phase. These data have been interpreted as suggesting an enhanced trigeminal nociceptive neurone excitability, possibly concomitant with a derangement of integrative, nociceptive and autonomic functions of the pain control system during bouts.
More recently, this electrophysiological method was applied to a migraine sample including a subgroup of migraine patients with strictly unilateral pain attacks (15). The results showed a reduction of pain and reflex thresholds in migraine patients between attacks compared with controls. The lowest values were found on the symptomatic side in patients suffering from side-locked pain attacks. Because of the bilateral location of the abnormalities, even in the unilateral headache subgroup, the occurrence in migraine of a centrally located dysfunction was hypothesized.
The CR has also been explored in patients with less common forms of primary headache, which are currently being proposed for inclusion in the spectrum of so-called TACs because they show the same prominent cranial parasympathetic autonomic activation: hemicrania continua (HC) and chronic paroxysmal hemicrania (CPH). In both HC and CPH patients (34) abnormally decreased CR thresholds have been recorded bilaterally. These data are consistent with the results reported in cluster headache patients.
Habituation of the blink reflex in cluster headache
In migraine subjects, an abnormal pattern of habituation during stimulus repetition has been observed with different sensory modalities (35). It is hypothesized that the inability to adapt to environmental stimuli shown by migraineurs is due to an unstable sensory processing that may partly have a genetic basis.
The putative role even in the more common forms of migraine of P/Q calcium-type channel dysfunction (36, 37), known to control intracellular calcium levels and neuronal excitability and neurotransmitter release, may account for the subtle subclinical abnormalities encountered in migraine, such as the lack of habituation in cortical evoked responses.
Moreover, the evidence of a significant rise of lactate in human cortex during photic stimulation demonstrated in a 1H NMR spectroscopy study (38) and the decrease of cortical lactate levels as habituation takes place, clearly shows the protective nature of habituation against metabolic overload. Previous studies have, indeed, demonstrated a deranged energy metabolism in migraine brain and muscles, with mitochondrial abnormalities causing a reduction of the energy reserve (39, 40).
The convergence of abnormal functional changes as an interictal migraine trait (dishabituation in sensory processing) and the disruption of the cerebral state under circumstances of environmental triggers and/or biochemical overload may lead to cortical spreading depression-like phenomena and to the activation of the trigeminovascular system in a migraine attack (35).
A possible link between the lack of habituation and another electrophysiological hallmark of migraine subjects, the strong intensity dependence of cortical auditory evoked potentials (IDAP), a cortical response thought to be under serotonergic control (41), has been recently proven by means of a neural network model, showing that the same neuronal mechanisms, regardless of a genetically based gain or loss of Ca2+ channel function, underpin these two distinct electrophysiological abnormalities in migraine (42).
Furthermore, sensitization may also occur in central pain pathways, as demonstrated in the aforementioned study of reduced corneal reflex threshold in migraine patients (15). In fact, a sensitization process of second- and third-order neurones has been shown to take place during migraine attacks (43). Sensitization is possibly sustained by multiple neurochemicals released by neurogenic inflammation and by long-lasting nitrous oxide (NO) production. The long-term potentiation of nucleus caudalis neurones accounts for clinical allodynia and hyperalgesia that may last for many hours (44).
In spite of the fact that pathogenetic mechanisms partly shared by migraine and cluster headache have been hypothesized, investigations into habituation phenomena have been carried out only in migraine. Therefore, in the light of the trigeminal sensitization hypothesis, we very recently studied in our laboratory habituation of the BR in 19 patients suffering from cluster headache (45).
We explored the habituation pattern of the R2 component of the BR in 19 drug-free episodic CH patients in an active period by means of increasing frequency of stimulation (0.2, 0.3, 0.5, 0.7 and 1 Hz), and compared the results with those of 20 healthy volunteers. The habituation was computed as percentage decrements of the II and III blocks of five responses compared with the first block. The statistical analysis showed a significant difference in the adaptation of the R2 response at high frequency of stimulation, i.e. R2 habituation, with a reduced percentage decrement at 0.7 and 1 Hz on both (symptomatic and nonsymptomatic) sides in CH patients compared with controls (Fig. 1).
Changes in R2 area of the blink reflex (expressed as percentage changes of the I block of responses) in cluster headache patients at increasing frequency of stimulation: 0.7 Hz upper panel; 1 Hz lower panel.
These findings are not comparable with those of cortical evoked responses, observed in migraine, as here we are dealing with an exteroceptive brainstem reflex. These data do seem to point more to a sensitization of trigeminal pathways than to dishabituation in sensory processing. They suggest that during active periods of CH a central derangement in the pain control system may favour a lack of habituation of the BR as an expression of an enhanced neuronal excitability at brainstem level, with an amplification of trigeminal nociception. Indeed, the observed extensive release of neuroactive molecules (substance P, NO) during cluster pain attacks may affect the modulation and sensitivity to pain via N-metil-D-aspartate (NMDA)-mediated mechanisms (46).
Conclusions
Through trigeminofacial reflexes it is possible to explore the excitability of the pathways, including the nucleus caudalis, involved in the transmission of nociceptive messages at trigeminal level. Descending modulatory influences on these nuclei can be explored by these methods, as can the pharmacological effects induced by centrally acting drugs. The data obtained in several investigations carried out in migraine and cluster headache patients seem to confirm an involvement of trigeminal pathways in the pathogenesis of these diseases. Several studies support the recent theories on sensitization phenomena in primary headaches.
Recent studies by Burstein et al. (43, 44) have shown how functional sensitization processes may take place in the early stage of headache attacks and may progress and spread from peripheral nociceptors to central level up to the second- and third-order neurones. At the beginning, this process is reversible, and could involve both trigeminovascular and myofascial peripheral receptors, but after this it may persist interictally and lead to permanent central modification, especially at the level of the trigeminal nucleus caudalis, due to the high convergence of afferent fibres on it. Indeed, the incoming nociceptive impulses converging on the trigeminal nucleus caudalis result in temporal and spatial summation, a decrease in activation threshold and a parallel decrease in headache attack threshold and, finally, in a process of chronicization of the headache. The reduced central serotonergic activity and the role of several neurochemicals involved during acute pain, especially those inducing the NMDA-mediated augmentation of the nociceptive response, i.e. substance P (47), may underlie the derangement of the antinociceptive system opening up the way for central sensitization in the spinal trigeminal nucleus.
Psychopathological factors (anxiety, cognitive focalization, hypervigilance and anticipation of pain) acting through the limbic system, together with the common misuse of abortive treatments, may both contribute to a reduced endogenous control of pain.
These central modifications have been hypothesized in the long-term history of different types of headache but still remain difficult to demonstrate objectively. The electrophysiological approach seems to be the most suitable for this purpose, even though the data here reviewed are not, on the whole, conclusive, possibly because most of the study protocols applied earlier were not developed from this perspective. For instance, the high intensities of stimulation used to evoked reproducible responses may lead to an overload or a saturation of the afferent pathways and, consequently, to a loss of sensitivity and to more subtle dysfunction of central sensory and nociceptive neurotransmission. The states of the neuronal networks involved in headache differ greatly both from subject to subject and in successive attacks in the same individual over time.
The abnormal habituation of evoked potentials in migraineurs represents an electrophysiological marker, consistently found in studies exploring different sensory modalities. Our recent findings of reduced habituation of the BR in cluster headache is suggestive of a dysmodulation in trigeminal pain transmission. These two electrophysiological patterns pointing to an abnormal sensory and nociceptive information processing might share partly common pathomechanisms, i.e. possibly involving cortical disexcitability. Dishabituation and central sensitization are phenomena that have not been fully explained and there has, to date, been no exploration of possible correlations between them.
Clarification of these issues could help to further understanding of the heterogeneous phenotypic expression of primary headaches and could have important implications for preventive treatment strategies.