9.2 Pathophysiology of pain in cancer and other terminal illnesses

Acute pain involves a series of excitatory events at peripheral and central levels that faithfully transmit signals about noxious events, leading to clinical reports of pain that usually have an overt relationship to identifiable peripheral damage, lesion, or abnormality. In contrast, clinically important chronic pains have altered neurophysiological and pharmacological substrates at many levels, from the periphery to the central nervous system (CNS), and the relationship between noxious phenomena and the level of pain experienced can become altered. Clinically, pain may be present, or unexpectedly severe, in the absence of an explanatory peripheral process. Plasticity, the ability of the nervous system to alter in response to injury-evoked dysfunction, leads to changes that can be observed throughout the pathways involved in the perception of pain.

Exploitation of continually developing pharmacological, anatomical, molecular, and genomic techniques is providing a basis for understanding the molecular and cellular mechanisms that contribute to the pain associated with varied pathophysiologies. Some chronic pains arise from persistent damage to tissue, such as that from arthritis, whereas others arise in the absence of these changes. Plasticity may alter pain responses in all these cases, but may be particularly notable with chronic neuropathic pains, which can arise from injury, a lesion, or a disease to either the peripheral or central nervous somatosensory systems.

Patients with cancer may develop injury to somatic or visceral tissues, or damage to neural structures. The aetiology may be related to the tumour itself, to various antineoplastic therapies, or to comorbid disorders. Cancer pain, therefore, often has a mix of mechanisms. Some mechanisms are broadly characterized as nociceptive, and some of these are inflammatory. Others are neuropathic. This complexity, which is shared by other types of advanced illnesses, such as HIV/AIDS, can complicate treatment but also offer more treatment options—a double-edged sword.

Recent animal studies are shedding light on some of the specific mechanisms underlying cancer pain. Importantly, whereas nociceptive pain results from activation of sensory afferents, neuropathic pain originates from damage to nerves and may prominently feature disturbances in ion channels. Thus, treatments aimed at the peripheral mechanisms are different for these two types of pain.

Plasticity in the nervous system in response to injury may lead to characteristic symptoms that contribute to the pain experience. These include expanded receptive fields, increased amplitude of response to a given stimulus (hyperalgesia), pain elicited by normally innocuous stimuli (allodynia), and spontaneous pain in the absence of external stimuli. Sensory deficits can also exist in neuropathic pain. In addition, as pain persists, affective and emotional responses must be considered along with the sensory aspects of the stimulus. It is clear that, although the sensory and psychological aspects of pain are separable, the neural pathways that contribute to these aspects of pain are interlinked. Furthermore, at both peripheral and central sites, there are mechanisms that can amplify and prolong the painful stimulus so that the pain becomes greater—this can result in severe pain in the presence of relatively minor peripheral pathology. This chapter considers these signalling systems and changes therein in the context of pain in cancer.

The dorsal horn receives sensory information from somatosensory receptors in the periphery via primary afferents. The area of the dorsal horn in which these primary afferents terminate is determined by the type of primary afferent and, therefore, the nature of the information that they carry. Different sensory inputs are carried by fibres of different thickness, from thick myelinated to thin and unmyelinated. Due to the differing degrees of myelination, these different groups exhibit differing conduction velocities at which they transmit a stimulus. The largest of the afferent sensory fibres, with thick myelin sheaths, are the Aβ fibres that carry information from muscle and tendons; these are the fastest conducting. A subset of thickly myelinated fibres carries mostly information from cutaneous mechanoreceptors; these usually do not transmit nociceptive signals. Neurons that carry nociceptive information include the thinly myelinated Aδ nociceptors and the thin unmyelinated C fibres. The latter two types of fibres are therefore pivotal in detection of potentially harmful stimuli in the external environment.

Aδ and C fibres terminate primarily in the superficial laminae of the dorsal horn, namely lamina I, which is an area intrinsically important in pain processing due to its large output to supraspinal areas. The large majority of neurons found within lamina I are nociceptive-specific; these neurons have small receptive fields and respond to only noxious pinch and/or heat stimulation. This region also contains a smaller population of polymodal neurons that are also cold responsive (Andrew and Craig, 2002; Craig and Andrew, 2002) and also a small population of so-called wide dynamic range (WDR) neurons that code throughout innocuous and noxious stimulus intensities (Seagrove et al., 2004). Finally, other neurons that respond purely to itch-inducing stimuli or to non-noxious heat have been noted (Light et al., 1993; Andrew and Craig, 2001).

Lamina I neurons have been shown to project to areas in the brain, such as the periaqueductal grey (PAG), lateral parabrachial nucleus, thalamus, nucleus tractus solitarius, and the medullary reticular formation (Todd, 2002). A large number of projection neurons from lamina I express the receptor for substance P, which is also known as neurokinin 1 (Todd et al., 2000). This group of neurons is the origin of a spinobulbospinal loop that ascends from the cord to the brain and then drives descending controls back to the cord; in this way the circuit can control dorsal horn excitability from higher centres (Bannister et al., 2009).

Deep dorsal horn neurons are mostly WDR neurons and consequently have larger receptive fields than the neuronal populations of the superficial dorsal horn (SDH). Projections from the deep dorsal horn neurons have been shown to be mainly to the reticular nuclei (Raboisson et al., 1996) and to the thalamus in the spinothalamic pathways. These nuclei of the brain have good connections with areas concerned with primary somatosensory cortex and therefore discriminatory perception of pain.

Peripheral receptors and channels involved in transduction of nociceptive stimuli in the periphery (see Fig. 9.2.1). The diagram depicts a C fibre and a polymodal nociceptor comprising numerous receptors and channels activated by voltage changes (voltage-gated ion channels) or chemical mediators of pain. The latter mediators include adenosine (acting at P2Y), bradykinin (B2), prostaglandin (endoprostinoid receptor (EP)), noradrenaline (β2), protons (acid sensing ion channel (ASIC)/TRPV1), heat/capsaicin (TRPV1), adenosine triphosphate (ATP) (P2X), and nerve growth factor (NGF) (trkA/p75). The precise molecular identity of a mechanoreceptor is still unclear.

In order to sense the external environment, it is necessary to convert peripheral stimuli into signals that can be carried by nerves to the CNS, a process known as transduction. There are a number of transduction molecules on the peripheral neuron that allow detection of a wide range of both exogenous and endogenous stimuli. While the full pharmacology and physiology of each of these peripheral sensory transducers falls outside the scope of this chapter, Table 9.2.1 illustrates a number of mechanisms through which a peripheral neuron can sense the peripheral environment. Needless to say, the actions of these transducers have a large part to play in pathological states where tissue and nerves are damaged, including cancer pain.

Transduction molecules seem to be highly preserved throughout evolution, with homology found throughout non-mammalian and mammalian species. This suggests the huge importance of an animal’s ability to sense its surroundings (Caterina and Julius, 2001). Peripheral tissue damage and subsequent local inflammation can cause the release of a wide variety of chemical factors that are able to sensitize primary afferent fibres. Pro-inflammatory compounds also can be released by nerve endings themselves in a process known as neurogenic inflammation.

Neurogenic inflammation is one of the mechanisms of peripheral sensitization and can further amplify the peripheral response of nociceptors. Peripheral terminations of nociceptors may arborize over a large area. Activation of peripheral afferents may cause neuromodulator release from nearby peripheral branches into peripheral tissues. These include factors such as substance P, neuropeptide Y, calcitonin gene-related peptide (CGRP), ATP, and glutamate. These compounds may act on peripheral blood vessels, mast cells, and sympathetic nerve fibres, leading to an increase in vasodilation, vascular permeability, and therefore plasma extravasation, causing erythema and oedema. Serotonin, bradykinin, glutamate, NGF, and other cytokines in the inflammatory infiltrate can cause further activation of primary afferent fibres and help propagate nociception when tissue is damaged and the production of these molecules is promoted.

Peripheral sensitization also involves change in ion channels. TRPV1 is a ligand-gated ion channel that is responsive to noxious heat and capsaicin, the pungent component of chilli peppers (Caterina and Julius, 2001). This channel is able to drive a neural response when exposed to noxious heat in the normal physiological setting. When inflammation occurs, however, a decrease in local tissue acidity can potentiate the channel’s response so that it is active at temperatures nearer body temperature (Caterina and Julius, 2001). This is a good example of how inflammation may cause a lowering of the nociceptive threshold in peripheral fibres and how hyperalgesia can result. A hugely intriguing point of note is that a receptor for noxious mechanical transduction, a much more obvious clinical issue, has yet to be fully elucidated. As mechanical allodynia presents such a large problem in the clinic, a cognate receptor for mechanical noxious stimuli may be of great therapeutic benefit.

There is very strong evidence pointing to the key role of ion channels in the production of electrical activity within sensory nerves and altered function of these channels after nerve damage (Suzuki et al., 2002). This evidence ranges from preclinical studies, the actions of drugs used in patients, and the discovery of familial pain disorders. Ion channels are important in the occurrence of altered transduction and disordered neural activity in damaged and intact fibres when neuropathy occurs.

The opening of sodium channels depolarizes neurons and generates the action potential. Injury to peripheral nerve can alter the normal arrangement of these channels along the length of a nerve. This is particularly notable in the development of neuromas after axons have been injured. Neuromas are associated with ‘ectopic’ electrical activity, which results from the accumulation of sodium channels at this site of injury. There also are many reports of altered distribution and levels of these channels in adjacent nerves, not just the damaged ones. Some inherited pain disorders arise from genetic mutations that either increase or decrease the functioning of a specific sodium channel subtype known as Nav 1.7 (Yang et al., 2004; Cox et al., 2006). Less dramatic polymorphisms in this channel impact on the level of pain experienced in several groups of pain patients. The latter observation suggests that inherited variations in channel function might be behind some of the variability in pain within patient groups (Reimann et al., 2010). Blockers of the channels Nav1.7 and Nav1.8, the major pain-related channels, have been described but have yet to reach the clinic. Thus, treatments presently use non-selective blockers, such as lidocaine and some anticonvulsants (e.g. carbamazepine, which is approved for the treatment of trigeminal neuralgia (Suzuki et al., 2002).

While peripheral mechanisms play a large role in development of pain states, a large amount of interest has been generated in the CNS’s abilities to amplify the inputs it receives from the peripheral nervous system and therefore cause an increased perception of pain. The mechanisms are multiple and complex, and again, involve changes in ion channel functioning. At spinal levels, upregulation and enhancement of transmitter release occurs via calcium channels. The drugs gabapentin and pregabalin modulate the function of these channels and, in this way, reduce transmitter release and excessive hyperexcitability.

One mechanism at the spinal level that has relevance to pain perception is called ‘wind up’. ‘Winding up’ of neuronal responses is made possible by the physiological properties of the N-methyl-D-aspartate (NMDA) receptor (Seagrove et al., 2004). The NMDA receptor is a ligand-gated ion channel whose central pore is, under normal neuronal activity, blocked by a magnesium ion. Due to this, the NMDA receptor plays little part in normal neuronal activity. However, after prolonged peripheral C-fibre nociceptive drive, increased presynaptic release of neurotransmitters, such as glutamate and substance P, causes depolarization of the postsynaptic neurons via their actions on the α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) and neurokinin 1(NK1) receptors, respectively (Bannister et al., 2009). This membrane depolarization allows the release of the magnesium ion blocking the pore of the NMDA receptor, and calcium to flow through the pore, further increasing postsynaptic excitability. The influx of calcium through the NMDA receptor allows short-term changes, such as phosphorylation of AMPA and NMDA receptors in the postsynaptic membrane. All these events lead to the potentiation of postsynaptic response. NMDA receptor blockers, such as ketamine and MK-801, have been shown to be effective in the reduction of neuronal actions and pain behaviours in animal models of acute and chronic pain, as well as in human acute and chronic pain states. However, due to the widespread nature and important role that the NMDA receptor plays in many other physiological systems, NMDA blockers may cause unacceptable neurological side effects, resulting in a limited utility in the clinic.

Pain is prevalent in populations with active cancer or other serious or life-threatening illnesses. As noted, cancer pain may be due directly to the tumour (tissue or nerve destruction) or may occur as a result of cancer therapy (e.g. chemotherapy-induced neuropathic pain). Bone is a common site of metastatic disease, exceeded only by lung and liver (Tubiana-Hulin, 1991), and bone pain is the most common cause of pain among patients with active cancer. Metastatic disease in bone occurs in 64–80% of those with solid tumours (Mercadante et al., 1997), and cancer-induced bone pain affects 28% of hospice inpatients, 34% of those patients in cancer pain clinics (Banning et al., 1991), and 45% of advanced cancer patients followed at home (Mercadante et al., 1997). This epidemiology highlights the need to find better drug therapies to combat pain in the clinical setting and recent progress in understanding bone pain illustrates the potential of translational research that links discoveries in the laboratory with potentially useful clinical treatments.

Until recent years, advances in the treatment of malignant bone pain were hindered by the lack of knowledge of the basic mechanisms of disease (Fig. 9.2.2). Original attempts at modelling this pathology involved administering a systemic bolus injection of metastatic tumour cells. This, however, led to systemically unwell animals, from which it was hard to draw conclusions about underlying mechanisms specific to pain rather than those related to systemic cancer (Kostenuik et al., 1993; Sasaki et al., 1998). As a consequence of these recognized deficiencies in earlier models, a number of novel approaches were developed to elucidate the mechanisms of cancer-induced bone pain. These new models rely on injecting a bolus of a variety of different tumour cells into either the long bones or the calcaneum of rodents. In general, this leads to the progressive and reliable development of pain-like behaviours to either mechanical or thermal stimuli in the postoperative period. This model has been used to explore a number of pharmacological, genetic and anatomical manipulations. As a result, some basic mechanisms have now been uncovered.

Originally, evidence supporting innervation of tumours was limited, and therefore the precise peripheral mechanisms underlying bone pain were of great debate (see also Chapter 13.2). While there were suspect players implicated in the generation of this particular pain state, such as primary afferents, interactions in the bone/cancer microenvironment, tumour-associated macrophages and others, none had substantial support. The information was insufficient to determine whether bone pain was related to neural mechanisms, to inflammation, or to other processes. Indeed, to this day the question remains: ‘Is bone cancer pain one of neuropathic or inflammatory origin?’ Answering this question will involve dissecting apart the various mechanisms implicated thus far.

Bone is not simply a framework to support and protect the body’s internal organs. It is an active tissue and plays key roles as a reservoir for calcium and phosphate and as a source of blood cells from the bone marrow. Its mechanical characteristics permit movements via actions of the muscles. The multitude of functions carried out by bone is reflected in its complex physiology and innervation.

There are two main types of bone which differ in structure and density: cortical bone and trabecular bone. Cortical bone is the dense outer layer of all bones, representing nearly 80% of all skeletal mass; it has a high resistance to torsion and bending forces. The periosteum forms the fibrous sheath surrounding the outer surface of cortical bone. Trabecular bone is found in the epiphyseal regions of long bones and constitutes a large proportion of the bone tissue of the ribs, spine, and skull. Paradoxically, this tissue type represents 20% of the skeletal mass, yet 80% of its surface area. Trabecular bone has a much less dense, woven appearance, created by interspersed trabeculae (plates) and bars of bone adjacent to red marrow cavities. For these reasons, it has more of an elastic characteristic compared with cortical bone. The cavities are connected through canaliculi, through which they receive their blood supply. Trabecular bone undergoes a greater amount of constitutive remodelling compared with the dense cortical bone and therefore bone pathology is often largely evident in bone of this type.

The remodelling of bone is reliant on an equilibrium of two main cell groups (Blair, 1998; Mackie, 2003): osteoclasts and osteoblasts. Osteoblasts are derived from primitive mesenchymal cells and are responsible for bone formation through the secretion of an array of extracellular matrix proteins (type I collagen, proteoglycans). Once osteoblasts have finished their function, they either apoptose or terminally differentiate into osteocytes, which remain viable surrounded by the bone matrix. Osteoblasts also have the interesting role of interacting with osteoclast progenitors and therefore regulate osteoclast activity.

Osteoclasts are derived from the monocyte-macrophage lineage and are the primary bone resorption cells. They are of great interest in bone pathologies such as osteoporosis and cancer-related bone pain. While a specialized cell for bone degradation may seem counterintuitive, it permits regulation of extracellular calcium and periodic bone repair, as well as remodelling in response to mechanical loads (Blair, 1998). An acidic extracellular microenvironment is highly important for an osteoclast to function properly as it is involved in the predominant mechanism through which osteoclasts degrade the base mineral hydroxyapatite. High expression of the vacuolar-(v) type electrogenic ATP-H+ channel is found along the ruffled border of the resorptive surface of an osteoclast, permitting the required development of an acidic environment of around pH 4.0–4.5 (Blair, 1998).

Even though it has been shown since the 1500s that nerve fibres are present in mineralized bone and the marrow cavity, tracing along the paths of blood vessels, the consensus of thought had been that pain arising from bone was principally the result of dense periosteal innervations (Mundy, 2002; Foley, 2004). While the periosteum is the most densely innervated structure, the bone marrow space receives the highest number of sensory and sympathetic fibres (Mantyh et al., 2002). Mineralized bone also receives a high volume of sympathetic and sensory fibres, more so than that of the densely innervated periosteum. All of the bone marrow, mineralized bone, and the periosteum receive both myelinated and unmyelinated sensory afferent fibres, as well as sympathetic fibres. Interestingly, of these small diameter unmyelinated fibres (presumably C-fibre population), only the CGRP trkA expressing peptidergic neurons are found to innervate bone and not the non-peptidergic IB4 labelled populations.

The theory that pain arising from bone metastases is only the result of structural weakness leading to mechanical distortion of the periosteum by innocuous stressors did not explain pain arising from bone with little or no radiographic evidence of periosteum involvement. Models of cancer-induced bone pain have now highlighted a number of other mechanisms that may be important both in peripheral and central sites.

Changes in the periphery have been shown to cause peripheral sensitization of the primary nociceptive afferents, and this peripheral sensitization can, in turn, drive central changes and hyperexcitability. When considering a tumour seeded within a bone, it is important to recognize that this includes not only cancer cells but also an inflammatory infiltrate, including macrophages, neutrophils, and T-lymphocytes (Mantyh et al., 2002). The immune-mediated response to the tumour leads to the release of a plethora of factors, such as cytokines, interleukins (ILs), chemokines, prostanoids, growth factors, and endothelins (Suzuki and Yamada, 1994; Safieh-Garabedian et al., 1995).

Peripheral nociceptors have an array of receptors that respond to these algogenic agents in the periphery. These factors are therefore able to sensitize and/or directly excite nociceptive fibres by acting on these peripheral receptors and lowering their threshold for activation. Pharmacological manipulation of these factors in a murine model of cancer-induced bone pain has shown promise in reducing measures of pain behaviour. Antagonism of endothelins, tumour necrosis factor alpha (TNFα), and bradykinin, all reduced pain behaviours (Baamonde et al., 2004; Wacnik et al., 2005), and endothelin antagonism also reduced central neurochemical markers that have been associated with the development of cancer-induced bone pain (Sorkin et al., 1997). As previously stated, the peptidergic CGRP-expressing neurons are the exclusive group of unmyelinated neurons that innervate the bone (Davar, 2001). This group of neurons express the receptor for NGF, namely trkA/p75. Macrophages, tumour cells, and other immune cells associated with the tumour mass have been previously shown to express NGF (Vega et al., 2003). The use of a NGF sequestering antibody attenuated both early and late phases of pain in the murine model of cancer-induced bone pain (Halvorson et al., 2005; Sevcik et al., 2005b) and reduced central markers of this pain state. Due to the exclusivity of trkA expressing peptidergic neurons in the bone, antagonism of this fibre type means that there can be no compensatory mechanisms in a differing fibre type.

Most importantly, as a tumour grows, there is a progressive increase in the innervation of the tissue, driven by NGF that is secreted by tumour-associated stromal cells. This pathological sprouting leads to a hyper-innervation of bone. Since almost all of the fibres express the receptor for NGF, sequestration of NGF has been shown to block both the reorganization of the sensory fibres and the associated nociceptive responses seen in models of cancer pain (Jimenez-Andrade et al., 2010).

Recently, emphasis has turned to ATP, which is present within all cells and is a purine that is algogenic and will be released into local tissues by damage. ATP has long been recognized as a local pain mediator and it is now clear that the P2X2/3 receptors are activated at peripheral and central levels in a model of cancer-induced bone pain. The data suggest that ATP may increase the activation of the central terminals of sensory afferents and so enhance spinal hyperexcitability (Kaan et al., 2010). Another receptor for ATP is the P2X7 receptor and this has been implicated in a number of animal pain models and also as a contributor to pain levels in postoperative and osteoarthritis patients. However, in a bone cancer model, deletion of the receptor had no effects on the pain behaviour, again reiterating the point that cancer pain is different from other pain conditions (Hansen et al., 2011).

With an uncertain relationship between bone destruction and pain, a number of other mechanisms for the generation of this pain state have been studied, one of which involves the activation of osteoclasts. Osteolytic tumours have been widely studied in the animal literature and have been shown to involve the recruitment of osteoclasts within the bone, leading to bone resorption. Primary tumours in bone (i.e. osteosarcoma) and secondary tumours in bone (i.e. metastatic spread from primary lung, breast, or prostate tumours) each have a profile of effects on the remodelling of bone. While some have a mainly osteolytic profile (i.e. osteosarcoma) others have a predominately osteoblastic profile (i.e. prostate carcinoma). However, in both osteolytic and osteoblastic tumours abnormal osteoclast regulation has been proposed as both a mechanism through which tumours destroy bone and a process that generates pain in cancer patients.

Cancer-induced bone destruction has been shown to be osteoclast-mediated, and, in a proportion of cases, dependent on the receptor activator of nuclear factor kappa B (RANK)/RANK ligand (RANKL) regulatory axis. In the non-pathological situation, osteoblast and osteoclast activity are in equilibrium so that normal bone remodelling can occur (Wacnik et al., 2005). In the presence of the growth factor colony stimulating factor-1 (CSF-1), osteoblasts expressing RANKL bind to RANK on local osteoclasts and osteoclast progenitor cells to stimulate bone resorption. This in turn stimulates nearby osteoblast activity and local bone formation (Boyle et al., 2003). It has been shown that metastatic cancer cells release a number of factors that may disrupt this axis, among the most important of which is parathyroid-hormone-related peptide (PTHrP). Metastatic breast cancer in bone has a higher expression of PTHrP than metastases in soft tissue (Powell et al., 1991). In light of this, it is apparent that PTHrP causes up-regulation of RANKL on osteoblast cells, which causes terminal differentiation of osteoclast progenitor cells (Guise, 2000). Activated T lymphocytes in the immune infiltrate of the tumour mass may also express RANKL and cause further osteoclast activation (Kong et al., 1999).

Given the importance of the RANK/RANKL regulatory axis, it is unsurprising that osteoprotegerin, the soluble ligand of RANKL, has shown efficacy in preventing cancer-induced bone destruction in animal models of osteolytic skeletal destruction (Clohisy et al., 1995). Additionally, osteoprotegerin attenuates development of pain behaviours in a murine model of osteolytic sarcoma bone pain (Honore et al., 2000). As mentioned previously, osteoclasts rely on an acidic extracellular microenvironment at the osteoclast/bone interface to facilitate resorption (Delaisse and Vaes, 1992). Moreover, increased osteoclast activation leads to a decreased extracellular pH. CGRP fibres that innervate the marrow or mineralized bone express ASIC (Safieh-Garabedian et al., 1995), as well as TRPV1 (Tominaga et al., 1998). Both of these channels are either sensitized or excited by protons and therefore likely to cause nociceptive transmission due to a decrease in extracellular pH. The increase in osteoclast actions may not be solely responsible for the decrease in extracellular pH, as tumours themselves lower the extracellular pH in order to assist invasion into surrounding tissues (Stubbs et al., 2000).

Both TRPV1 antagonism and TRPV1 knock-out in murine models of bone-cancer pain show attenuation of pain behaviours (Ghilardi et al., 2005). However, in these models, both osteoclast inhibition and TRPV1 antagonism does not completely attenuate all facets of the pain behaviours seen, even in light of the fact that osteoprotegerin almost completely prevented bone destruction and osteoclast activation (Guise, 2000; Kaan et al., 2010). This suggests that while osteoclast-induced acidosis and structural weakening may play an important role in the development of malignant bone pain, it is not the sole mechanism through which this pain is generated.

Increased osteoclast activity in this setting may also cause structural weakness. Mechanical stress upon the periosteum and its distension due to tumour burden may well result in peripheral fibre activation and the sensation of pain (Mach et al., 2002). It would be expected that the tumour growing within bone would damage the distal processes of nerves within the bone marrow, mineralized bone, and the periosteum. Studies in animal models suggest that this is indeed the case. A marker for neuronal cell injury ATF-3, which is up-regulated in the dorsal root ganglion in peripheral neuropathic pain models, is also found to be up-regulated in models of malignant bone pain. Of interest is that gabapentin, a drug that has been shown to be efficacious in models of neuropathic pain, has also been shown to be of benefit in models of cancer-induced bone pain (Sevcik, 2004; Donovan-Rodriguez et al., 2005). This suggests that nerve injury may play a role in the development of bone cancer pain. Mechanisms underlying neuropathic pain will be discussed in greater depth later in this chapter.

It is also clear that the CNS undergoes changes that aid the maintenance of this pain state. The early murine models of cancer-induced bone pain involving confinement of tumour (NCTC 2472 sarcoma cells) within the femur established the neurochemical ‘fingerprint’ of cancer-induced bone pain. Confinement of tumour to within the bone not only leads to development of postoperative behavioural signs of pain, but increased osteoclastic bone destruction in the periphery (Schwei et al., 1999). However, of greater interest in this murine model of malignant pain were the immunohistochemical studies showing a number of central cellular and neurochemical changes in the segments of the spinal cord relating to the peripheral input. The spinal cord segments that receive afferent input from tumour-laden femur showed a massive astrocyte hypertrophy and elevation of the pro-hyperalgesic peptide dynorphin (Schwei et al., 1999). These changes were seen exclusively in the side of the spinal cord ipsilateral to the affected limb and not on the contralateral side. Glia, a family of which astrocytes are a member, are in the normal situation quiescent. Upon becoming activated, glia release a myriad of pro-inflammatory cytokines, including IL-1, TNF, IL-6, reactive oxygen species, nitric oxide, prostaglandins, excitatory amino acids, and ATP (Watkins and Maier, 2003). This in turn can cause enhanced second-order neuron excitability within the dorsal horn and further exaggerate primary afferent neurotransmitter release. It has also been demonstrated in murine models that normally non-noxious palpation of the affected femur not only produced nocifensive behaviour but an increase in substance P receptor internalization and an increase in c-Fos expression in lamina I neurons of the dorsal horn (Schwei et al., 1999).

These findings provide a weight of evidence showing that primary afferent fibres are sensitized following tumour growth in the periphery. This astrocyctosis, increased dynorphin expression, increased substance P internalization, and increased c-Fos expression has been shown in models of inflammatory and/or neuropathic pain. While substance P levels in primary afferent neurons have been shown to increase in inflammatory models and decrease in neuropathic models, levels remain unchanged in cancer-induced bone pain states. However, the coexistence of all these features in cancer-induced bone pain provides evidence that this is a unique pain state that may have mechanisms similar to inflammation and neuropathy (Schwei et al., 1999). Furthermore, this may well be the basis of reasoning behind why conventional treatments have failed thus far in the battle with malignant bone pain and further highlights the need for unique pharmacotherapy (Fig. 9.2.3).

By recording second-order neurons in the dorsal horn of the spinal cord using in vivo electrophysiology, we can gain an idea of the supra-threshold response to peripheral stimuli that cannot be ascertained using behavioural techniques (Urch et al., 2003; Donovan-Rodriguez et al., 2005). Neurons can be characterized, based on their responses to mechanical, thermal, and electrically evoked stimuli. The SDH is predominantly populated with nociceptive specific (NS) neurons, which respond to nociceptive stimuli. These cells are distinguished from WDR neurons, which respond to a wide range of both noxious and innocuous stimuli (Seagrove, 2004). Establishment of cancer-induced bone pain changes the ratio of WDR:NS neurons in the SDH from the 26% WDR:74% NS in a sham animal to 47% WDR to 53% in the pathological setting (Stubbs et al., 2000). The phenotype shift seen in the superficial dorsal horn was also paralleled by the development of superficial and deep dorsal horn neuronal hyperexcitability to mechanical, thermal, and electrical stimuli, further suggesting ongoing central sensitization. Furthermore, these lamina I neurons that become hyperexcitable after cancer-induced bone pain now show a de novo or increased responsivity in the innocuous range (Stubbs et al., 2000). Thus pain selective neurons can now be activated by innocuous stimuli. This may allow the limbic areas of the brain, concerned with affective/emotional aspects of pain, to have an influence via the parabrachial pathways, and also increase the effects of low-threshold stimuli. Plausibly, this may result in affective areas of the brain now being dominated by painful messages and so relate to the distress and co-morbidities such as fear, anxiety, mood changes and sleep disorders caused by pain.

This hyperexcitability in lamina I also plays a role in further maintaining neuronal dorsal horn hyperexcitability through descending facilitations. Spinal events are not only controlled by afferent input but also by descending controls from higher centres (Bannister et al., 2009). Higher-order cognitive and emotional processes such as anxiety, mood, and attention can influence perceived pain. Such phenomena are enabled by the convergence of somatic and limbic systems into such descending modulatory systems. Areas in the midbrain and brainstem, such as the PAG and the rostroventral medial medulla (RVM), are key structures in the descending modulatory repertoire (Bannister et al., 2009). Such a system is important as it provides neural networks by which cognitive and emotional states can influence pain processing at the level of the spinal cord (Suzuki et al., 2004; Bannister et al., 2009). In short, these circuits allow the brain to exert some control over spinal pain events. Recent animal studies suggest that in addition to inhibitory systems, there are important descending facilitations that can be engaged by external and internal processes, and act to enhance intrinsic spinal mechanisms of pain. Lamina I neurons expressing the substance P receptor NK1 form the origin of a spinobulbospinal loop, which relays through the RVM (Todd, 2002; Bannister et al., 2009). The RVM has been highlighted as a key area involved with descending facilitations which are thought to be mediated through the 5-hydroxytryptamine type 3 (5-HT₃) receptor. Blockade of spinal 5-HT₃ receptors with intrathecal ondansetron reduces the mechanical and thermal evoked responses of superficial and deep dorsal horn neurons (Bannister et al., 2009). This suggests that descending facilitations are indeed important in amplifying nociceptive transmission from the dorsal horn to higher centres (Suzuki et al., 2004; Bannister et al., 2009).

Current therapies give researchers a great insight into the possible mechanisms underlying cancer pain. Opioids are the mainstay of treatment of severe malignant pain in the clinic (Mercadante et al 1997). However, while the benefits to be gained from opioid therapy are obvious, opioid treatment is associated with a large number of side effects, such as nausea, vomiting, constipation, sedation, and delirium. At the high doses sometimes required to relieve persistent cancer pain, these side effects become even more problematic to control and, in some cases, may be the reason for inadequate analgesia. In line with clinical evidence, opioids are effective in reducing pain-like behaviours in a number of animal models of cancer-induced bone pain.

Early data from murine models suggest that higher doses of morphine were required to attenuate cancer-induced bone pain than those needed in inflammatory pain, and this point was used to further highlight the mechanistic differences between these two pain states (Luger et al., 2002). The relatively low efficacy of acute morphine has since been further validated (Mercadante et al., 1997; Vermeirsch et al., 2004); however, a situation that more closely mimicked the clinical paradigm was sought. A bi-daily injection schedule over 5 days after the establishment of the pain state has been shown to be highly effective in reducing behavioural signs of cancer-induced bone pain (Urch et al., 2005). This regimen also was found to be more efficacious than a single acute dose (Sevcik et al., 2004). However, even in light of this, behavioural measurements taken pre and post the final morphine administration, at peak disease progression, indicated a significantly reduced analgesia between consecutive morphine administrations. This is not to say that the analgesia provided by morphine completely wore off between doses. Behavioural signs of pain were still, 12 hours after the last dose, significantly reduced from that of the vehicle-treated animals and no different from that of the acutely treated group. However this may help to explain the requirement for escalated doses of opioids to treat severe cancer-induced bone pain.

Chronic, but not acute, gabapentin administration also has been shown to be effective in these models (Donovan-Rodriguez et al., 2005; Peters et al., 2005). Gabapentin acts by binding to the calcium channel accessory subunit α2δ. This implies a role of α2δ in the pathology underlying this pain state. Yet looking at the response of neurons in chronic morphine and chronic gabapentin regimens highlights two differing mechanisms involved in cancer-related bone pain. While chronic morphine and gabapentin administration both lead to behavioural attenuation of pain, chronic morphine treatment was unable to completely reset dorsal horn excitability and the associated phenotype shift towards the WDR neuronal population in superficial dorsal horn (Urch et al., 2005). The bias of superficial neurons toward the WDR phenotype, even in the presence of morphine analgesia, suggests that low-threshold inputs to the spinal cord may still access areas of the brain concerned with both pain affect and perception. This is a possible physiological mechanism through which breakthrough pain may remain refractory to morphine analgesia in the clinic. In contrast, gabapentin completely attenuated dorsal horn excitability and reversed the phenotype shift of WDR:NS ratio back towards that seen in a normal animal (Donovan-Rodriguez et al., 2005). This suggests that while morphine may cause behavioural attenuation of pain, it is not acting on the mechanisms that are intricately involved in producing those behavioural signs. On the other hand, gabapentin is able to inhibit the mechanisms that lead to cancer-induced bone pain rather than merely blocking sensory inputs as seen with morphine. Upon the termination of gabapentin treatment, pain behaviours and dorsal horn excitability returned to their hyperexcitable ‘pathological’ state, implicating neuronal physiological mechanisms rather than anatomical changes to be key in development of cancer-induced bone pain.

Peripheral opioids may also have a role to play in the management of malignant cancer pain. In support of this potential is evidence showing that loperamide, the peripheral mu opioid receptor agonist, is able to attenuate thermal hyperalgesia in a murine model of cancer-induced bone pain (Menendez et al., 2003).

Non-steroidal anti-inflammatory drugs (NSAIDs) are the first step on the World Health Organization’s (WHO) ladder for the use of analgesics for the relief of cancer pain. Although NSAIDs are still used as an analgesic additive even after increased pain severity (Mercadante et al., 1997), the clinical data supporting their efficacy is limited (Urch, 2004). NSAIDs inhibit the cyclooxygenase (COX) enzyme and therefore attenuate the synthesis of prostaglandins from arachidonic acid. This ultimately prevents sensitization or activation of primary afferents by prostaglandins produced locally by tumour cells and/or the immune response. Inhibition of the COX-2 enzyme, the local inducible form of COX, using three different COX-2 selective inhibitors in a rat and murine models of cancer-induced bone pain has been shown to be effective in ameliorating behavioural pain signs. In both these studies, chronic COX-2 inhibition has been shown to be better at attenuating pain-like signs than a single acute dose. Chronic administration reduced tumour burden and bone destruction (Sabino et al., 2002; Fox et al., 2004), suggesting that COX-2 inhibition may be acting at multiple sites rather than at a single point in a pathway.

Bisphosphonates are used widely in the clinical setting and prevent bone resorption by osteoclast inhibition. Bisphosphonates were able to reduce pain scores, as well as bone and sensory nerve destruction, in both rat and mouse models of cancer-induced bone pain (Sevcik et al., 2004; Walker et al., 2002). However, tumour burden was found to not be affected, with both tumour burden and tumour necrosis increasing (Sevcik et al., 2004). Bisphosphonates, therefore, may have a role to play in reducing bone pain in malignancy, especially if combined with other analgesics (Mercadante et al., 1997).

Radiotherapy of tumours within bone causes analgesia but the mechanism is still unclear. Radiotherapy in the murine model of bone cancer pain reduced tumour burden by more than 75% (Goblirsch et al., 2005). However, it did not affect the osteoclast density, suggesting that radiotherapy causes analgesia via direct mechanisms on tumour cells themselves.

While pain arising from bone is a common clinical problem, pain arising from soft tissue can also be painful. The pathophysiology in this pain state, however, is still not clear. A novel model of pancreatic cancer pain has been developed to elucidate the relationship between disease progression and pain development (Lindsay et al., 2005). This model uses a transgenic mouse that develops pancreatic cancer and shows changes similar to that seen in the human condition. These include tumour growth, increased innervation, macrophage infiltration, weight loss, and pain. More significantly, pain was only evident at a point at which cancer progression was highly advanced. Sensory innervation has now been quantified and models such as these will hopefully provide mechanistic insights in to the development of pancreatic cancer pain in the future.

The International Association for the Study of Pain (IASP) defines neuropathic pain as ‘pain initiated or caused by a primary lesion or dysfunction in the nervous system’. This may not be a complete definition of neuropathic pain, and the definition is changing to be ‘a lesion or disease’ affecting the ‘sensory nervous system’ (Treede et al., 2008). These definitions hint at broad mechanisms. After nerve injury occurs, there are large-scale changes that occur within the peripheral and central nervous systems. This nerve plasticity has been heavily scrutinized and some of these plastic changes have been proposed as possible important mechanisms for the generation of pain in neuropathic states.

Neuropathic pain may arise in the palliative care setting for a multitude of reasons. Nerve damage is known to occur as a result of tumour compressing a nerve, surgical resection, radiotherapy, and chemotherapy. Numerous pre-clinical animal models have been developed in an effort to elucidate the basic mechanisms involved in the generation and maintenance of neuropathic pain. These include models of peripheral nerve injury (common in traumatic neuropathic pain) and chemotherapy-induced neuropathic pain. Peripheral lesions to nerves are obvious, and therefore many models of traumatic peripheral nerve injury exist. Consequently, a multitude of mechanisms have been implicated in painful peripheral nerve lesions. It would be wrong to believe that mechanisms of peripheral nerve injury are the same whether caused by chemical injury, constriction, or by the surgeon’s scalpel. It is for these reasons that the pharmacotherapy of different types of neuropathic pain should differ. One only has to look at the different responsiveness of trigeminal neuralgia and other neuropathic pain syndromes to drugs such as carbamazepine to see that mechanisms underlying these neuropathies must differ. Iatrogenic neuropathic pain caused by various drug treatments is well recognized. Theoretically, it should be of no surprise that chemotherapeutic agents and antiretrovirals, compounds which broadly act through inhibiting cellular processes, may have neuropathic side effects.

The chemotherapy agents vincristine and paclitaxel and the nucleoside reverse transcriptase inhibitors (NRTIs) such as dideoxycytidine have a well-established history of causing painful neuropathy (Berger et al., 1993; Forsyth et al., 1997), with the incidence of painful neuropathies in paclitaxel-treated patients suggested to be around 22% (Forsyth et al., 1997). Paclitaxel binds to microtubules and promotes hyperpolymerization. This interferes with the mitotic spindle formation and promotes cellular arrest in the metaphase-anaphase transition, and consequently apoptosis (Jordan et al., 1996; Yvon et al., 1999). As the precise mechanism of paclitaxel-induced neuropathy is still not clear, it has been assumed that the binding of paclitaxel to microtubules of neurons, which consequently prevents anterograde and reterograde axonal transport, could lead to a neuropathic state. However, these proposed mechanisms of both tumour apoptosis and neuropathy have been questioned, and the precise mechanisms are still not clear (Komiya and Tashiro, 1988; Fan, 1999). Another chemotherapeutic agent, vincristine, is said to act via a similar mechanism, inhibiting normal polymerization of β-tubulin, leading to abnormal spindle function. Early studies showed that epineural injection of paclitaxel caused oedema and axonal degeneration (Roytta and Raine, 1986) along with interference of axonal transport and microtubule anomalies. However, a number of sources have doubted the theory that axonal degeneration and microtubule anomalies occur, citing the unusually high concentrations of paclitaxel in the epineural injections responsible for the local axonal reactions (Polomano and Bennett, 2001). These local epineural injections allowed paclitaxel to bypass the liver, so that any metabolites usually involved in the development of pathology would not be present.

Mitochondria have now been shown to play an important role in the establishment of some pain states, especially chemotherapy and NRTI-induced neuropathic pain. Change in mitochondrial function has been shown to be involved in the pathogenesis of neurodegenerative diseases; however, this has not previously been widely studied in pain. In patients, neuropathies solely due to disarray in mitochondrial function have been shown to have an increased incidence of developing pain (Finsterer, 2004).

These mitochondrial effects may be relevant to the mechanisms described recently in models of chemotherapy-induced neuropathic pain that are more akin to the clinical scenario. Low-dose systemic injections of paclitaxel produced mechanical hypersensitivity, yet even at the stage of peak pain-like behaviours, there were no signs of axonal degeneration, markedly altered microtubules, or impairment of axonal transport (Flatters and Bennett, 2006). However, abnormal mitochondria in C fibres and myelinated axons were noted and this was suggested to be the cause of paclitaxel’s actions. In this model, it was proposed that paclitaxel’s binding to mitochondria caused opening of the mitochondria permeability transition pore, causing calcium ion influx into the cytosol. This mitochondrial calcium efflux has been suggested to be both the cause of mitochondrial swelling as well as of primary afferent excitability and pain behaviour (Flatters and Bennett, 2006). The T-type calcium channel blocker ethosuximide, α2δ calcium channel subunit ligand, gabapentin, and calcium chelators, all block neuropathic pain behaviours caused by chemotherapeutic agents (Flatters and Bennett, 2004; Xiao et al., 2006).

Mitochondrial damage is also the proposed mechanism for the neuropathy seen after treatment with NRTIs for HIV/AIDs (Joseph et al., 2004; Joseph and Levine, 2006). The mitochondrial electron transport chain and its end product, ATP, have also been a suggested pathological player in the development of neuropathic pain due to both chemotherapy and antiretroviral treatment; however, the downstream targets in the pathological setting are still to be elucidated (Joseph and Levine, 2006). It has also been noted that both paclitaxel- and vincristine-evoked painful peripheral neuropathies show a loss of innervation of the epidermal sensory fibres as well as the Langerhans cells, the skin’s resident immune cells. Whether this is a causal or merely consequential finding is yet to be shown (Siau et al., 2006).

Iatrogenic neuropathic pain due to treatment with NRTIs is not the only cause of neuropathic pain in HIV/AIDS patients. Around 10–15% of all HIV-1 infected patients have a symptomatic distal polyneuropathy (Verma, 2001), a proportion of which experience pain as a result. It has been shown that direct infection of neurons by HIV has a negligible incidence rate and it is for this reason that the attention of researchers turned to the neuroimmune system. An envelope protein of HIV-1, gp120 has been widely shown to produce hindpaw hypersensitivity to thermal and mechanical stimuli when injected into the intrathecal space (Milligan et al., 2001), as well as directly to the sciatic nerve of the rat (Herzberg, and Sagen, 2001). Astrocytes and microglia have been shown to bind to the virus epitope gp120, which in turn causes their activation. Milligan et al. showed that attenuating glial activation ameliorates the development of behavioural hypersensitivity of animals once exposed to gp120 (Milligan et al., 2001). In this study, the response of microglia and astrocytes to gp120 was blocked. As mentioned previously, activation of glia in the spinal cord by exogenous factors, such as gp120, causes the transcription and release of factors such as prostaglandins, excitatory amino acids, IL-1, and IL-6 as well as inducing nitric oxide synthetase expression in glial cells (Kreutzberg, 1996; Kong et al., 1996). The release of these factors from glia in the dorsal horn may well be able to directly excite local sensory neurons and propagate the sensation of pain, even without the presence of a peripheral pathology. The development of spinal sensitization has been put forward as an explanation for why pains in HIV-1 infected patients commonly present without obvious peripheral pathologies and are vague and diffuse in nature (Breitbart et al., 1996; Hewitt et al., 1997).

Many changes in the periphery have been implicated in the development of neuropathic pain, but none have been more heavily studied than the role of ion channel dysregulation on peripheral nerves. Normally, tactile stimulation of sensory nerve terminals in the skin, viscera, or bone leads to the propagation of an action potential and sensory signalling. However, after nerve injury, many peripheral nerves display ectopic discharge, which may lead to an increased barrage of nociceptive signalling onto dorsal horn transmission neurons without a peripheral stimulus. Electrophysiologically, neuronal responses of second-order deep dorsal horn neurons are heightened after nerve injury in rodent models, with an increase in receptive field size and an increased response to natural stimuli applied to the hindpaw. This goes hand in hand with increased spontaneous activity and hyperexcitability of these neurons.

Changes in the sodium channel populations on peripheral nerves and their subsequent aberrant activity have been of great interest due to their key role in setting neuronal excitability and therefore the development of pain states (Suzuki et al., 2002; Cox et al., 2006; Cummins et al., 2007). Expression of sets of sodium channels on the peripheral neurons shows plasticity after nerve injury. The mRNA of NaV1.3, a usually embryonic TTX-sensitive current, has been found to increase after axotomy and this may play a role in the generation of ectopic activity in peripheral nerve fibres. The sensory neuron specific sodium channel, NaV1.8 is also thought to play a key role in the generation of abnormal sensory signalling following nerve injury. NaV1.8 protein is markedly decreased in the dorsal root ganglion of predominantly small fibres after nerve injury. This is paralleled by an increase in immunoreactivity of the channel in the distal axons and nerve terminals, representing a redistribution of the channel to the distal sites of the neuron, where it may take part in the development of hyperexcitability and increased nociceptive transmission. Both in humans and animals, sodium channel accumulation has been shown to occur around the neuroma formed at the site of the nerve lesion. This has long been the pharmacological basis for the use of drugs such as carbamazepine, lamotrigine, and local anaesthetics in patients with neuropathic pain. After nerve injury, demyelination and abnormal trafficking of sodium channels occurs along the membrane of injured nerves and maybe in the uninjured neighbours. This may lower the threshold for activation and induce ectopic activity in the peripheral nerve. This contributes to the development of central sensitization and amplification of peripheral events, possibly leading to the allodynia and hyperalgesia seen in patients (see Suzuki et al., 2002; Cox et al., 2006).

Voltage-gated calcium channels (VGCCs) may also play a large hand in the increased peripheral nociceptive drive in neuropathic pain. VGCCs play a key role in permitting neurotransmitter release from the presynaptic terminal and, therefore, the postsynaptic propagation of the sensory signal. As with sodium channels, there are a large number of calcium channels that play a role in neuronal excitability. Activation of calcium channels by peripheral electrical events causes the inward flow of calcium and allows neurotransmitter vesicle exocytosis and thus postsynaptic depolarization. However, VGCCs not only have presynaptic actions but also act at the postsynaptic site, allowing activation of postsynaptic second messenger cascades. This may lead to altered gene expression, protein synthesis, and therefore long-term plastic changes. Long-term potentiation of dorsal horn neurons, caused by repetitive afferent stimulation, may be of importance in maintaining exaggerated neuronal responses for long periods after increased peripheral drive has subsided (Rygh et al., 2006).

Specific targeting of these VGCCs and their accessory subunits seems to be beneficial in the treatment of neuropathic pain, with novel drugs acting at these sites now available. Gabapentin and its newer analogue pregabalin have shown benefit to patients with neuropathic pain. Their target is thought to be the α2δ accessory subunit of calcium channels. Gabapentin has been shown to be effective in reducing neuronal responses in a model of neuropathic pain (Suzuki et al., 2005). However, gabapentin’s inability to reduce neuronal responses in normal animals highlights a clear state dependency of gabapentin’s action and implicates a role for the α2δ subunit in neuropathic pain pathology. The α2δ subunit has been shown to up-regulate after nerve injury and this correlates not only with the development of behavioural allodynia in these animals but also with gabapentin’s behavioural antiallodynic efficacy (Luo et al., 2001). Gabapentin’s actions have now been characterized in many differing animal models of neuropathic pain including a model of chemotheraputic paclitaxel-induced neuropathic pain (Xiao et al., 2006).

Another target for the treatment of neuropathic pain is the N-type calcium channel. It was shown that blockade of the N-type calcium channel with ω-conotoxin-GIVA reduced dorsal horn neuronal responses in animals with neuropathic pain and that the release of substance P and CGRP from primary afferents was N-type dependent (Matthews and Dickenson, 2001). An analogue of ω-conotoxin-GIVA, ziconotide, has now been developed and is now licensed for the treatment of neuropathic pain.

In addition to spinal changes, higher brain centres are able to facilitate dorsal horn neuronal activity. An ascending descending facilitatory pathway seems to play a critical role in chronic pain states, such as neuropathic pain. Superficial dorsal horn NK1 neurons project to higher brainstem nuclei, such as the parabrachial nucleus, which receives connections from amygdala and hypothalamus and may explain the ability of emotion to affect pain processing (Bannister et al., 2009). Along with the affective implications, these brainstem nuclei form a part of a spinal-bulbo-spinal loop, which through the RVM, can facilitate dorsal horn neuronal responses. Ablation of these NK1-expressing projection neurons, using a saporin toxin conjugated to substance P, is able to attenuate dorsal horn neuronal responses to stimuli evoked in the periphery (Suzuki et al., 2005). Using an antagonist of the excitatory 5-HT₃ receptor, it is clear that these descending facilitations are in the large part serotonergic acting at the 5-HT₃ receptor in the spinal cord. These descending facilitations contribute to maintaining central sensitization in pathological pain states and may well aid the development of tactile allodynia seen in patients with chronic pain. The analgesia produced by tricyclic antidepressants and serotonin-norepinephrine reuptake inhibitors are no doubt due to interactions with these descending controls, which are altered in many pain states, including cancer pain (Dickenson and Ghandehari, 2007; Bannister et al., 2009).

Astonishingly, gabapentin’s actions in reducing neuronal responses in nerve-ligated animals were blocked by both the ablation of these superficial projection neurons and by the application of ondansetron (Suzuki et al., 2005). However, the activation of the 5-HT₃ receptor allowed gabapentin to reduce neuronal responses in normal animals, suggesting that this supraspinal loop needs to be in place for gabapentin’s action to be unmasked.

Any advances in the understanding and treatment of pain in cancer and other terminal illnesses will need to be based on a better understanding of pain mechanisms, so that existing therapies can be used with greater efficacy. Further knowledge of these mechanisms is a basis for the development of future therapeutic approaches based on controlling the pathophysiology, as well as the pain itself.

This chapter illustrates how important advances in understanding the pathophysiology of pain have been made in recent years and provide a basis for improvements in the treatment of the pain, distress, and co-morbidities. Tumour growing in the periphery elicits a series of changes that run from peripheral tissue and nerve, causes profound changes in spinal cord function, and in the final experience of pain, involves complex circuits in the brain that link pain with affective function.

The authors acknowledge funding from the Wellcome Trust London Pain Consortium.

Complete references for this chapter are available online at <http://www.oxfordmedicine.com>.