7 The ascending pain pathways

Pain is an essential survival mechanism, protecting the individual from tissue damage or from further injury to already damaged tissue. However, the relationship of pain to injury is complex and variable, both under normal circumstances and in pathological pain states where anatomical, molecular, and biochemical changes inevitably occur (Wall, 1979).

Noxious stimulation provokes a series of physiological changes in the organism accompanied by a range of affective responses and motivational changes designed to minimize the risk of injury. At the behavioural level, there is usually a rapid withdrawal from the site of stimulation on top of which motor behaviour such as attack or flight and autonomic activity characterized by changes in respiration, heart rate and blood pressure, and muscle tone may occur. This wide range of modulations also prepares the body for emergency and escape. The nature, location, and intensity of the pain will also be registered and the environmental context in which the pain occurred will be stored for future reference.

If noxious stimulation results in injury, then pain sensitivity around the wound usually increases as a result of both local changes in the small diameter nociceptive sensory neurons and alterations in sensory processing within the spinal cord and brain. The increase in pain sensitivity around damaged tissue during healing is also accompanied by motivational changes that encourage rest during recovery. Pain processing at spinal levels is at all times under the control of descending pathways from the brain which registers both the environmental and bodily status (Craig, 2002; Bester et al., 2001, 2000; Hunt and Mantyh, 2001). This contributes to a balance between pain sensitivity and survival after injury. Chronic pain states, particularly when the pain remains after recovery from injury or results from direct injury to the nervous system, have in largely unknown ways, subverted the normal controls that modulate pain sensitivity.

This complex response to noxious stimulation over time implies that nociceptive information is processed at different levels of the brain and spinal cord to deliver an integrated response. Much of the literature on the ascending pain pathways concentrates on different aspects of this processing but it remains unclear as to whether particular pathways can be ascribed a discrete functional role (Treede, 2002; Willis et al., 2002; Lahuerta et al., 1994).

Nociceptive information leaves the spinal cord by a number of distinct pathways and terminates within other areas of the spinal cord and discrete areas of the brainstem including the thalamus. Historically, it had been appreciated that cutting the anterolateral tract or hemisection of the spinal cord in humans led to selective loss of pain, temperature, and itch contralateral and caudal to the lesion (Craig, 2002; Aminoff, 1996; Lahuerta et al., 1994; Denny–Brown, 1979). This approach was exploited neurosurgically to control pain but the (albeit variable) return of pain within a variable period after cord surgery limited the general usefulness of the approach (Bowsher, 1999, 1996). Return of pain implied considerable plasticity within ascending pathways and also suggested that information related to painful sensation could reach the brain through pathways normally not accessed but uncovered in time by spinal lesions. There was also the implication that changes had occurred in the brain that left the potential for re-emergence of a pain state independent of peripheral stimulation (Lenz et al., 1995; Katz and Melzack, 1990). Further experimental work and clinical observation also indicated that the loss, and possibly the return of pain sensation after cordotomy was not simply tied to loss of an ascending pathway but also involved descending pathways from the brainstem. Indeed, pathways known to be essential for the control of spinal excitability were inevitably destroyed by the procedure (Urban and Gebhart, 1999; Aminoff, 1996; Denny–Brown, 1979).

Pain has been described as having affective and discriminative components that, at least in man, have been shown to be dissociable by hypnosis or following cortical lesions (Rainville, 2002; Hofbauer et al., 2001; Price, 2000; Rainville et al., 1999). The discriminative dimension defines the location and describes the feeling of the pain-burning, stabbing, and so on, while the affective dimension attributes an ‘unpleasantness’ value to the pain. The primary sensory cortex is thought to perform at least the first levels of sensory discriminative analysis — that is, the ‘what’ and ‘where’ of the stimulus. The limbic system, including the limbic cortex and subcortical structures such as the amygdala, attribute an affective value to the incoming information based on the past history of the animal and the current behavioural context. Affective behaviours are motor responses that include an integrated autonomic response through activation of brainstem and spinal centres that control sympathetic and parasympathetic outflow to the body.

Many chronic pain states also carry with them a memory of past peripheral injury, for example, with painful events that occurred before amputation of a limb (Katz and Melzack, 1990). Pain memories can also be recovered in the absence of on-going pain by stimulation of particular areas of the thalamus, and retain both a discriminative and affective dimension (Lenz et al., 1995). Pain memories thus may be triggered by peripheral stimulation but do not seem to rely purely on activity within ascending pathways from the spinal cord and may require cortical involvement.

Nociceptive information must gain access to the areas of the spinal cord and brain that are concerned with patterning the fully elaborated and integrated pain response, but how this is achieved remains controversial.

This chapter summarizes and synthesizes recent findings in the field and is intended to complement a number of recent and older reviews (Gauriau and Bernard, 2002; Villanueva and Bernard, 1999; Willis and Coggeshall, 1991; Willis, 1985).

Somatosensory information reaches the spinal cord and areas of the brainstem through primary afferent sensory fibres (Hunt and Mantyh, 2001; Snider and McMahon, 1998; Hunt and Rossi, 1985; Nagy and Hunt, 1983). Sensory fibres terminate within the ten designated laminae (I–X) of the spinal cord in a highly reproducible and characteristic fashion depending upon their diameter, biochemical composition, and receptive field properties (Fig. 7.1). Nociceptive information is relayed to spinal cord neurons throughout the dorsal horn, particularly laminae I–II and V–VII (Fig. 7.2.) but largely avoids populations of neurons within laminae III and IV which receive, almost exclusively, non-nociceptive projections from Aβ sensory afferents. Neurons within laminae I and II and V and VI receive nociceptive information through unmyelinated C fibres and finely myelinated Aδ sensory afferents. C fibres form the major part of the sensory input to the cord and can be further divided up into almost equal numbers of peptide-containing and non-peptide-expressing neurons that stain for the lectin IB4. Nociceptive information from the skin is distributed between laminae I, II, and V, while visceral input is almost exclusively peptidergic and terminates largely within laminae I and V, avoiding lamina II (Lawson, 2002; Cervero and Laird, 1999; Cervero, 1994; Sharkey et al., 1987; Hunt and Rossi, 1985; Nagy and Hunt, 1983). However, while projection neurons in lamina I receive almost exclusively peptidergic C-fibre input from the body, it should be noted that both lamina I neurons and lamina V neurons — the origins of two of the major ascending pathways — potentially receive all types of nociceptive input.

Neurons that relay nociceptive information to the brain are located primarily in laminae I, III, V–VII, and X (Gauriau and Bernard, 2002; Villanueva and Bernard, 1999; Willis and Coggeshall, 1991). Considerable attention has recently been directed towards the pathway arising largely from lamina I neurons such that many of the functions once ascribed to the deeper pathway are now being reallocated to this pathway arising from the most superficial layer of the dorsal horn. There are a number of reasons for this. Retrograde and anterograde pathway tracing techniques have greatly improved in sensitivity and forced a reappraisal of the connections of spinal projection neurons. Species differences are also important, as well as the level of the spinal cord at which the analysis is made, cervical cord often being very different to the lumbar cord but having traditionally received more attention. However, there has also been a shift in the way that function has been allocated to particular pathways particularly in terms of the affective, discriminative, and homeostatic dimensions of pain (Craig, 2002). Finally, chemical lesioning techniques have been introduced which specifically destroy subpopulations of lamina I neurons and effectively dissociate the lamina I projection from the deeper lying nociceptive pathways that project upon the brainstem withoutdisrupting descending pathways (Nichols et al., 1999; Mantyh et al., 1997). This dissociation has permitted the behavioural assessment of the role particular pathways play in nociceptive behaviour and led to a rethinking of the contribution that the different spinal projections make to acute and chronic nociceptive behaviours.

The lamina I pathway has been repeatedly re-described in numerous studies (see below) and in a variety of animals including humans, where it is thought to have reached the highest degree of differentiation. Several sites of termination of the pathway have been described which hint at the function of this pathway in both homeostatic regulation and in supplying information to areas of the brain concerned with discrimination, affect and autonomic regulation (Fig. 7.3).

In a number of recent reports, Craig and his colleagues (Craig, 2002) have made a convincing argument in favour of a role for the lamina I pathway in homeostasis and of being, in one sense, an ‘afferent sympathetic’ pathway directing both interoceptive and extereoceptive information to areas of the CNS concerned with autonomic regulation and affective expression. In rodents, only a small percentage (∼10%) of lamina I neurons have been shown to be projection neurons (Todd et al., 2000). These neurons receive nociceptive information segmentally from all areas of the body and have axons that terminate within the sympathetic preganglionic motor column of the thoracic cord, in the medulla, and throughout other regions of the brainstem as far rostral as the thalamus. These areas of termination can be broken down, somewhat artificially, into areas concerned with sympathetic regulation and affect, motor control, cognition, and sensory discrimination. Obviously, sudden injury is followed by a closely orchestrated sequence of responses in the animal including arousal accompanied by cardiovascular changes, fight or flight responses, and suppression of ongoing pain to aid escape (Keay and Bandler, 2002; Janig and Habler, 2000  a,b). The success of these survival responses is linked to their close coordination, and this is reflected by substantial interconnections between subsystems within the nervous system.

Recent evidence also supports the view that the lamina I pathway is crucial for the regulation of spinal cord excitability, and therefore pain behaviour, through the activation of descending inhibitory and excitatory pathways from the brainstem (Suzuki et al., 2002; Bester et al., 2001).

Injections of retrograde tracers into brainstem areas known to receive spinal nociceptive input almost tend to label neurons in more than one lamina of the spinal cord. This implies that pathways originating in deep and superficial laminae overlap to some extent at their sites of termination within the brain. This is however not always the case. Indeed, if one considers projections to the parabrachial area, it is now clearly established that lamina I and lamina V neurones send projections to distinct subregions (Gauriau and Bernard, 2002; Bourgeais et al., 2001b; Villanueva and Bernard, 1999; Bernard et al., 1996, 1995; Feil and Herbert, 1995; Bernard and Besson, 1990). Lamina I neurons project mainly (80%) to a contralateral parabrachial region centred around the external lateral nucleus, whereas lamina V neurones project bilaterally to the internal nucleus (∼500 μm medial and dorsal to the ipsilateral external lateral nucleus).

The nociceptive projection pathway originating from the superficial dorsal horn (the ‘lamina I pathway’) arises from neurons mainly in lamina I (see Bester et al., 2000, 1995) but with a small contribution from laminae III and IV neurons (Todd, 2002; Todd et al., 2000; Naim et al., 1997; Ding et al., 1995). Strikingly, projection neurons which are only made of 10% of the lamina I neuron population have been shown to be largely nociceptive-specific, receiving input from C and Aδ nociceptors responding to noxious thermal and mechanical stimulation (Lawson, 2002; Bester et al., 2000; Snider and McMahon, 1998). Temperature-specific neurons have been described in primates. Receptive fields are small and the majority (up to 90%) of these projection neurons express the NK1 (‘substance P’) receptor (Todd et al., 2000). C-Fos histochemistry combined with noxious stimulation at a variety of deep (joints, muscles) and cutaneous sites in the body have indicated that in the rat, lamina I NK1-positive neurons are primarily concerned with the intensity of noxious stimulation rather than the location or tissue of origin of the pain (Doyle and Hunt, 1999  a,b). Destruction of these neurons with intrathecally delivered saporin–substance P conjugate does not result in obvious changes in behaviourally assessed nociceptive thresholds but does reduce the increased behavioural sensitivity associated with inflammation or neuropathic nociception (Nichols et al., 1999). Axons from superficial projection neurons either cross the spinal cord immediately to ascend in the contralateral ventrolateral and dorsolateral fasciculi or in the dorsolateral fasciculus of the same side.

Within the spinal cord, ascending fibres terminate around the sympathetic preganglionic motor neurons of the intermediolateral column in thoracic cord segments (Craig, 2002, 2000, 1996) and continue on to terminate in association with discrete areas of the caudal brainstem (see Fig. 7.3). Heavy termination is found in areas concerned with cardiovascular and visceral regulation, particularly the nucleus tractus solitarius (NTS)(Gamboa–Esteves et al., 2001) and ventro lateral medulla (Lima et al., 2002; Bourgeais et al., 2001a). Painful stimulation results in responses in the cardiovascular and respiratory systems and afferents from both of these systems terminate within the NTS (Janig and Habler, 2000  a).

Other areas of termination include the dorsal reticular nucleus, also known as the subnucleus reticularis dorsalis (SRD), and possibly the lateral reticular nucleus and adjacent reticular formation which has motor functions related to the cerebellum. The SRD receives bilateral input from deep and superficial layers of the spinal cord which is largely nociceptive and forms part of a ‘pronociceptive’ pathway that both projects to the medial thalamus and projects back upon the spinal cord modulating nociceptive transmission (Lima and Almeida, 2002;Monconduit et al., 2002). This is one of many brain areas that both receive nociceptive input from the spinal cord and project back upon the dorsal horn to regulate the flow of pain-related information. Such ‘neural loops’ connecting the brain and spinal cord appear to be crucial to the ways in which pain sensitivity is regulated in the behaving animal. SRD neurons have extremely large receptive fields, in some cases covering the whole body. The dorsal part of the nucleus receives largely ipsilateral nociceptive input from laminae I and X, and destruction of the nucleus depresses nociceptive responses to acute and inflammatory nociception.

Further rostrally, axons from the lamina I pathway terminate heavily within the contralateral parabrachial area (PB) (Bernard et al., 1995; Feil and Herbert, 1995) and periaqueductal gray (PAG) (Villanueva and Bernard, 1999). The PB area probably receives the densest set of terminal projections originating in spinal lamina I. Considerable emphasis has recently been placed on the extensive termination of lamina I neurons in the external medial lateral PB. PB is an area that integrates information from the NTS and visceral and somatic information from the body through the spinal cord. Neurons respond almost exclusively to nociceptive stimulation and their receptive fields can be extremely large, including the whole of the body area (see Fig. 7.3) (Bester et al., 1995; Matsumoto et al., 1996; Menendez et al., 1996; Bernard and Besson, 1990) as well as responding to visceral stimulation, such as following colorectal stimulation (Bernard et al., 1992). PB neurons are therefore exquisitely sensitive to intensity of the noxious stimulus rather than location or nature of the stimulus.

PB projects heavily upon areas of the limbic system such as the ventromedial hypothalamus (Bester et al., 1997a) and the central nucleus of the amygdala (Bernard et al., 1993), providing the nociceptive input to areas of the brain classically associated with affect (Davis and Shi, 1999; Shi and Davis, 1999; Bester et al., 1995; Bernard and Besson, 1990). The ventromedial hypothalamus has been associated with rage and aggression and projects upon the PAG (Keay and Bandler, 2002). The amygdala regulates autonomic function and is necessary for fear conditioning (Roeling et al., 1994; Adamec and Stark–Adamec, 1983).

Together these areas may be important in regulating pain-related emotional responses, as well as in modulating longer-term functions such as motivational and metabolic states that follow injury. Importantly, these structures give rise to powerful descending projections that modulate nociceptive processing at the level of the spinal cord.

The primary brainstem route for modulating and coordinating pain behaviours is through the PAG of the midbrain (Urban and Gebhart, 1999; Basbaum and Fields, 1984). As would be expected, apart from receiving direct nociceptive inputs from lamina I neurons and neurons of the lateral spinal nucleus, the PAG also receives substantial inputs from PB and the ventromedial hypothalamus (see Fig. 11).

The organization of nociceptive inputs to the thalamus and, therefore, to the cerebral cortex remains controversial (Gariaut and Bernard 2004a,b; Craig and Blomquist, 2002; Jones, 2002; Treede, 2002; Willis et al., 2002). Classically, it had been maintained that there were distinct areas in the medial and lateral thalamus that received nociceptive input. The medial pathway terminated within nuclei of the intralaminar and medial thalamic nuclei which projected upon areas of limbic cortex linked to affect and motivation, whereas the lateral thalamic nuclei projected to primary somatosensory cortex concerned with discriminative aspects of pain processing (Melzack and Casey, 1968). Indeed, in the human brain activation of the somatosensory cortex, insula and cingulate corteces have been seen by positron emission tomography following various types of noxious stimulation of the body (see Fig. 13).

However, recent data has blurred this simple mediolateral distinction. Recent anterograde tracing studies following small injections of tracer into lamina I of the spinal cord (which by their nature only label a very small population of neurons in the superficial spinal cord) indicated:

That is, there is claimed to be a projection from lamina I to both medial and lateral thalamic territories with access to both limbic and somatosensory cortex (Gauriau and Bernard, 2004  a,b, 2002; Craig, 2002; Willis et al., 2001).

PoT has attracted considerable attention. In the rat, this area of the thalamus also receives both superficial and input from SRD, a small input from lamina V spinal neurons, and projects upon SII and insular cortex. PoT and insula together are essential for the acquisition of fear conditioning generated by foot shock (Shi and Davis, 1999). Both nocispecific (NS) and neurons that respond to both noxious and non-noxious stimulation (NNS) have been identified. NS neurons project upon the second somatosensory cortex, while NNS neurons terminate in the insular cortex (Gauriau and Bernard, 2004  a), presumably reflecting input from both deep and superficial dorsal horn projection pathways. In primates, a posterior part of the ventromedial nucleus, Vmpo (part of the ventrocaudal nucleus, VMpc, in man) has been identified, which is claimed to receive input only from lamina I nociceptive and temperature-sensitive neurons and to project to insula cortex (Craig, 2002; Craig and Blomquist, 2002; Craig et al., 2000, 1994). The homologue in rodents, if it exists, is unclear.

In a recent series of articles it has been claimed that the projection area of Vmpo within the insula cortex represents a primary sensory representation for pain, temperature, itch, and other feelings from the body and that in humans, stimulation of Vmpo evokes pain and temperature sensation and that ‘lesions of Vmpo or its cortical field reduce these sensations specifically’ (Craig, 2002).While this idea has been debated and remains controversial (Willis et al., 2002), it re-introduces the possibility of ‘hard-wired’ pathways (Perl, 1998) uniquely dedicated to the processing of particular types of nociceptive and non-nociceptive sensory information.

Deeper-lying neurons, particularly lamina V and the adjacent laminae IV and VI–VII neurons of the spinal cord, also receive substantial nociceptive input and give rise to a major ascending pathway to the brain (Fig. 7.4). Compared to lamina I neurons however, they generally have a wide dynamic range of response and are rarely nociceptive specific, have larger receptive field sizes, and respond to a variety of noxious and nonnoxious stimuli, as well as showing viscero-somatic convergence (Treede, 2002; Willis and Coggeshall, 1991; Maixner et al., 1989; Besson and Chaouch, 1987; Le Bars et al., 1986; Hoffman et al., 1981). This has led to the suggestion that they are well fitted to providing information about the intensity but not the location of the stimulus. Yet they are the most closely studied of spinal nociceptive neurons and give rise to projections that overlap some of the areas of termination of lamina I neurons. Lamina V projections had not been studied using anterograde tracing until very recently, and these studies only partially agree with previous retrograde tracing studies (Gauriau and Bernard, 2004  b).

Nevertheless, within the brainstem, retrograde labelling has identified deep spinal projections to the lateral reticular nucleus and SRD, as well as to areas of the lateral reticular formation and pontine and deep mesencephalic reticular nuclei (Gauriau and Bernard, 2004  b; Lima and Almeida, 2002; Bernard et al., 1990). Many of these areas project onto classical motor structures such as the cerebellum and cranial motor nuclei. Terminations are also found in the PB (bilaterally with substantial ipsilateral projections) and adjacent areas. It is clear that many of these areas also receive from lamina I neurons, although the degree of overlap may be restricted. Moreover, certain PB neurons tend to be overwhelmingly nociceptive (Bester et al., 1997b, 1995; Matsumoto et al., 1996; Menendez et al., 1996; Bernard and Besson, 1990), suggesting that at the single-cell level there may well be dedicated nociceptive input from lamina I neurons. This consideration emphasizes the prominent role of the PB area in integrating noxious input from the whole body.

The functional importance of the vast majority of nociceptive information pouring into the reticular formation is unclear. Many reticular nuclei project upon the midline and intralaminar thalamic nuclei, and these nuclei project upon the cortex and also the striatum and globus pallidus, again emphasizing that lamina V neurons seem to terminate largely within areas of the brain considered to have motor and arousal functions. Neurons within the reticular formation and the SRD also project upon the medial and intralaminar nuclei of the thalamus, providing a substantial nociceptive input. Neurons of the intralaminar nuclei, which project to the cingulate cortex, have very large receptive fields, perhaps reflecting input from SRD and PB. The SRD, as described above, receives projections from deep, but also superficial, spinal pathways and projects extensively upon the thalamus, including PoT and areas that are probably homologous to the VMpc of primates. These areas of the reticular formation could therefore be considered nociceptive relays, transferring nociceptive information from the deep spinal cord to the thalamus (Villanueva and Bernard, 1999).

Other midbrain projections of deep neurons include the anterior pretectal nucleus and peripeduncular tegmental areas (Rees and Roberts, 1989). The pretectal area has been implicated in antinociception, and the pedunculo-pontine area is known to be involved in motivational behaviours (Bechara and van der Kooy, 1992), reinforcing the notion that the processing of nociceptive information has to involve many different levels of analysis.

The thalamic connections of deep dorsal horn neurons have recently undergone something of a revision. For many years lamina V neurons were considered to be the major source of the spinothalamic tract, but recent work has begun to suggest that the primary input from deep dorsal horn nociceptive neurons is to the medullary and mesencephalic brainstem, and that nociceptive information is relayed from these sites to the thalamus — for example, from SRD to the intralaminar thalamus and VM (Gauriau and Bernard, 2004  b).However, retrograde labelling studies following injections of tracer into the thalamus have reported labelling of deep dorsal horn neurons as well as lamina I–III neurons. Neurons located deep within the dorsal horn have been found in most studies (Kobayashi, 1998; Willis et al., 1979) particularly in the cervical spinal cord. In primates, a strong projection was found from lamina V neurons to the ventrobasal thalamus (VB) which receives non-nociceptive lemniscal projections and projects upon the somatosensory cortex (Willis et al., 2001). However, a recent anterograde tracing study in rodents (and this may reflect a species difference) of lamina V projections by Gauriau and Bernard (2004b) has failed to confirm this observation. These authors maintain that the major input to VB comes from lamina I neurons and that the areas of non-nociceptive lemniscal and nociceptive lamina I input are separated within the ventrobasal complex. Thus in the rat, anterograde labelling of deep dorsal horn neuron projections showed very few thalamic sites of termination apart from the centrolateral nucleus (CL), which also receives projections from the lateral spinal nucleus. There were, however, projections to the globus pallidus, the principle output from the striatum. CL projects upon the motor and premotor cortex and to the dorsal striatum underlining the importance of the extrapyramidal motor system in the response to injury and attack.

Recently, lamina X neurons, which lie around the central canal of the spinal cord, have been shown to give rise to an ascending pathway running within the dorsal columns and synapsing within the dorsal column nuclei (Willis and Westlund, 2001;Willis et al., 1999; Al–Chaer et al., 1998). Lamina X neurons respond to noxious stimulation of the viscera, as do neurons within the cuneate nucleus. These project through the medial lemniscus to the ventrobasal complex. Because these ascending fibres run within the most medial aspects of the dorsal columns, it was suggested that the success of midline lesions and commissural myelotomies in relieving pelvic visceral cancer pain might be due to section of these ascending fibres. Functional MRI revealed that changes in blood flow in the thalamus and cortex, following colorectal distention, could be eliminated by midline lesions of the dorsal columns. However, similar visceral pain relief in humans has also been reported following anterolateral cordotomy, although feasibly through a different mechanism such as interruption of descending pathways.

Other ascending nociceptive pathways have been reported which project to the hypothalamus (Burstein et al., 1990) and which are derived from both deep and superficial neurons within the rostral cervical cord or spinal trigeminal nucleus. Anterograde tracing studies have suggested that these projections are modest though, when compared to the density of hypothalamic nociceptive projections originating in the PB area. Overall, these projections are probably involved in direct modulation of hypothalamic mechanisms such as the release of hormones, the regulation of circadian rhythms, and perhaps feeding, drinking, and grooming activity, as well as the generation of aggressive behaviour, and possibly antinociception.

Much has been made of the proposed differences between the laminae I–III projection and the pathway that arises from deeper lying laminae V–VII dorsal horn neurons. Essentially, it has been proposed that the superficial pathway carries information more specifically linked to discriminative aspects of pain processing than the deeper lying neurons. Superficial neurons are nocispecific (Christensen and Perl, 1970) with small receptive fields, and subpopulations of neurons also respond to temperature and a range of intense stimuli. Laminae I–III neurons project as far rostrally as the thalamus and terminate in a number of thalamic sites that relay to somatosensory cortex where location and identification of the stimulus would occur. However, there are major inputs from these neurons to the parabrachial nucleus (Bester et al., 2000; Bernard et al., 1995; Feil and Herbert, 1995) and to the SRD (Lima and Almeida, 2002; Villanueva et al., 1996). The neurons in these areas have extremely large receptive fields that can cover the whole body and, in the case of PB, project upon the hypothalamus and amygdala (Bester et al., 1997b, 1995; Bernard et al., 1993; Bernard and Besson, 1990) areas of the brain classically associated with affect and motivation rather than discrimination. The receptive field sizes and the fact that most of these neurons respond to all types of noxious stimulation and are exquisitely sensitive to the intensity of the stimulus would also tend to rule out a role in discrimination but reinforce a role in affective and motivational behaviours. Feasibly, it could be argued that there are distinct subpopulations of lamina I neurons that project to different targets and that the superficial nociceptive pathway can concurrently provide information to both somatosensory cortex and limbic cortex through distinct thalamic relay nuclei.

However, this argument effectively relegates the deeper lying lamina V pathway to a non-discriminative role with largely motor and arousal functions. Lamina V neurons respond to both noxious and non-noxious stimuli and possess larger receptive fields. Moreover, the receptive field properties of deep neurons are to an extent dependent on lamina I neurons (Suzuki et al., 2002). Uniquely, there is an extensive zone of termination of lamina V neurons within the brainstem reticular formation where electrical stimulation can elicit escape behaviours (Roberts, 1992; Casey, 1971) and from where there are direct projections to areas of the thalamus and forebrain (PoT, CL, GP) concerned with affect and motor coordination and learning.

Recent data indicates, however, that selective lesioning of laminae I–III neurons did not disturb sensory discrimination (Nichols et al., 1999). In rodents, and probably primates, the vast majority of laminae I–III projection neurons express the NK1 (substance P) receptor. Using saporin-substance P conjugate, it was possible to selectively lesion the laminae I–III ascending pathway (Fig. 7.5). The behaviours of the rats several weeks later were not obviously different. Response to hot plate and noxious mechanical stimulation were not altered, effectively implying that this pathway was not essential for pain perception. However, when inflammatory or neuropathic pain states were established, the degree of increased sensitivity (hyperalgesia and allodynia) were substantially reduced by destruction of the ascending laminae I–III pathway. To date, no selective lesion of lamina V projection neurons has been made, but it is clear from these experiments that they confer some discriminative capacity, perhaps by virtue of secondary projections to the somatosensory thalamus from sites of termination within the caudal brain stem.

How might these results be explained? It has often been noted that ascending projections from laminae I–III are to areas of the brain that send a reciprocal projection preferentially back to the spinal cord and are closely involved in the control of spinal sensitivity (Hunt and Mantyh, 2001). PAG receives inputs from the hypothalamus, PB, and amygdala and projects upon the rostroventral medulla an area which is in part serotonergic, and projects to spinal cord, both inhibiting and exciting spinal neurons (Suzuki et al., 2002; Basbaum and Fields, 1984). Other areas of the brainstem, such as the SRD, are also reciprocally connected to the dorsal horn, and lateral areas of the reticular formation send descending noradrenergic projections to the spinal cord (Proudfit and Clark, 1991; Sagen and Proudfit, 1984).

These descending controls can be activated following input from spinal neurons as well as from areas of the brain concerned with monitoring the behavioural state of the animal such as the amygdala and hypothalamus. Disconnecting the laminae I–III pathway from the brain would therefore effectively render the brain incapable of modulating spinal sensitivity because there was no longer sufficient information coming from the spinal cord concerning ongoing nociceptive activity (Hunt, 2000). Those parts of the brain that are notably receiving less nociceptive input would be limbic areas associated with affect and motivation. These are parts of the brain that closely monitor the behavioural context of the animal and have considerable influence on spinal processing of nociceptive information alongside autonomic control and modulation.

The detailed anatomy of the ascending pain pathways is still not fully understood, but we can identify projections that terminate in a large number of different areas of the brain associated with autonomic, motor, discriminative, affective, cognitive, and motivational aspects of pain behaviour.

Deep and superficial pathways from the dorsal horn terminate in discrete areas of the brainstem. The superficial laminae I–III pathway, which reaches its highest level of differentiation in primates, seems to have a particular role in delivering nociceptive information to areas of the brain concerned with discrimination, affect, cognition, and motivation and from which extensive descending pathways emerge to modulate activity in many other areas of the brain and spinal cord. Deeper lying dorsal horn neurons can support nociceptive behaviour but the resetting of nociceptive sensitivity following skin or peripheral nerve damage critically requires the laminae I–III pathway and the engagement of descending pathways that regulate spinal sensitivity. How pain is perceived and located is still largely unknown but is assumed to require cortical processing of nociceptive information that could be provided through deep and superficial spinal pathways and from polysynaptic pathways within the brainstem.