17 Neuropathic pain

Pain signalled by a normal sensory system, nociceptive pain, serves a vital protective function. The peripheral and central nervous somatosensory systems permit rapid localization and identification of the nature of painful stimuli, prior to appropriate action to minimize or avoid potentially tissue damaging events. A reduction or absence of pain resulting from neurological disease emphasizes the importance of this normal protective function of pain. For example, tissue destruction occurs frequently in peripheral nerve diseases which cause severe sensory loss such as leprosy, and in central disorders such as syringomyelia. Neuropathic pain results from damage to somatosensory pathways and serves no protective function. This chapter provides an overview of neuropathic pain, considering its context, clinical features, pathophysiology, and treatment.

In the peripheral nervous system, neuropathic pain is caused by conditions affecting small nerve fibres, and in the central nervous system by lesions of the spinothalamic tract and thalamus, and rarely by subcortical and cortical lesions. The clinical feature common to virtually all conditions leading to the development of neuropathic pain is the perception of pain in an area of sensory impairment, an apparently paradoxical situation. The exception is trigeminal neuralgia (Sections 19.2.1; 20.1.4).

Neuropathic pain is heterogeneous clinically, aetiologically, and pathophysiologically. Within a given diagnostic category, whether defined clinically or aetiologically, there are wide variations in reports of pain by patients. This heterogeneity poses one of the greatest challenges in understanding the mechanisms of neuropathic pain. Knowledge of the pathophysiology is an obvious pre-requisite to the development of effective treatments. The goal of a pathophysiologically based understanding of the symptoms and signs of neuropathic pain is, of course, just such a rational and specific approach to treatment. While this is not yet achievable, clinical-pathophysiological correlations have led to some recent advances in treatment.

Pain is a common symptom in patients with neurological disease; neuropathic and nociceptive pains frequently coexist. Reliable epidemiological data on neuropathic pain are difficult to obtain, due to problems of case ascertainment, retrospective rather than prospective studies, inclusion variations and biases in different case series, small sample size, differing defining thresholds for considering pain a leading symptom, and historical disagreements about the definition of neuropathic pain. In many patients with neurological disease, pain is but one of several symptoms and not necessarily the leading symptom. Some conditions are characteristically painful, while others are not, and it is in the latter category that the incidence and prevalence of neuropathic pain have tended to be underestimated.

For example, while the prevalence of peripheral neuropathies is 2.4 per cent in the general population, rising to 8 per cent with age (Martyn and Hughes 1998), the prevalence of neuropathic pain within this group is uncertain. One study has calculated a point prevalence for peripheral neuropathic pain in the general population as high as 5 per cent (Daousi et al. 2004). Studies of the incidence of postherpetic neuralgia demonstrate the problem of an agreed definition of the condition, including the interval following the acute attack of shingles, and underline the need for prospective studies.

In relation to central nervous system disease, several studies have indicated a prevalence rate of neuropathic pain of around 71 per cent in spinal cord injury (Bonica 1991). Epidural cord compression by metastatic carcinoma is a common example of a pathology that may cause both neuropathic pain, be it myelopathic or radicular, and nociceptive somatic pain due to involvement of skeletal structures and soft tissues. In one study, pain was a first symptom in 96 per cent of such patients (Gilbert et al. 1978). Estimates of neuropathic pain as a symptom in patients with multiple sclerosis vary widely, from 14.5 to 82.1 per cent (Clifford and Trotter 1984; Vermote et al. 1986; Kassirer and Osterberg 1987; Moulin et al. 1988), emphasizing variations in symptom characterization, definition of pain type, and case selection bias. Finally, the incidence of central post-stroke pain, formerly known as thalamic pain, was found in a prospective study to be as high as 8.4 per cent in a group of 191 patients. In the sub-group with sensory deficits, 42 per cent of the stroke group, 18 per cent of patients had such pain (Andersen et al. 1995).

There is a continuing debate about the definition of neuropathic pain. Matters are not helped by the use of a multiplicity of terms, some with identical or overlapping meanings. This situation exists partly for historical reasons, partly because a relatively recent change in definition broadened the definition of neuropathic pain (IASP 1994) and partly because neuropathic pain is defined on a clinical basis rather than pathophysiologically. A mechanism-based understanding of the symptoms and signs of neuropathic pain will clarify the nosology of neuropathic pain and assist in the development of specific treatments (Woolf et al. 1998; Woolf and Mannion 1999; Scadding and Koltzenburg 2005).

Neurogenic pain. This refers to all neurological causes of pain, both peripheral and central. It is defined as ‘pain initiated or caused by a primary lesion, dysfunction, or transitory perturbation in the peripheral or central nervous system’ (IASP 1994). Although an acceptable term, its use has declined, possibly because of the inclusion of states of ‘transitory perturbation’ and ‘dysfunction’ of the nervous system, for neither of which is there a definition or agreed description. This broad definition of neurogenic pain, with ill-defined limits, has questionable clinical usefulness. For example, it has led some to consider conditions such as fibromyalgia, as neuropathic pains.

Neuropathic pain. This was the term originally restricted to refer to pain due to peripheral neuropathies and plexopathies. As part of an initiative to produce a comprehensive taxonomy of terms used to describe all painful conditions, the definition was broadened to include ‘pain initiated or caused by a primary lesion or dysfunction of the nervous system’ (IASP 1994). Under this definition, neurogenic and neuropathic are now virtually synonymous, but as with neurogenic pain, the 1994 definition includes a category of ‘dysfunction of the nervous system’. This would include complex regional pain syndrome type 1, known previously as reflex sympathetic dystrophy (Section 17.5), which shares many clinical features with pain that undoubtedly has a primary nervous system cause. These features include severe pain, allodynia, hyperalgesia, sometimes accompanied by vasomotor and sudomotor disturbances. Within the current broad definition, neuropathic pain is subdivided into peripheral and central, to denote the location of the causative lesion.

Neuralgia. This term describes ‘pain arising in the distribution of a nerve or nerves’ (IASP 1994). It has become restricted to describe neuropathic pain due to lesions of specific nerves, for example intercostal, sciatic, femoral, or trigeminal, and of roots in the case of postherpetic neuralgia. Neuralgia is thus a sub-category of neuropathic pain.

Central pain. Behan (1914) was the first to use the term central pain. Riddoch’s (1938) later refined description of pain arising from central nervous system lesions is applicable to this day. Central pain is subsumed within the broader definition of neuropathic pain.

Working definition of neuropathic pain. In the light of this confusing terminology and in order to avoid ambiguity, the working definition of neuropathic pain used in this chapter is ‘pain arising as a direct consequence of a lesion affecting the somatosensory system’. This reflects a growing consensus among pain clinicians and scientists. Importantly, this working definition excludes any reference to vague ‘dysfunction’ of the nervous system, an aspect of the current definition that has proved so contentious.

In the absence of a comprehensive mechanism-based nosology of neuropathic pain, classifications remain anatomical and aetiological. For peripheral neuropathic pains, various classification schemes have been proposed. The anatomical distribution pattern of the affected nerves provides valuable differential diagnostic clues as to possible underlying causes. Neuropathic pain comprises stimulus- independent, or ongoing, and stimulus-dependent pains. Mechanisms underlying these are now partly established, so that an anatomical and aetiological classification can be supplemented, but not yet replaced, by pathophysiological data (Section 17.4). These data are more complete and reliable in relation to peripheral than for central neuropathic pain (Woolf et al. 1988; Koltzenburg 1996; Boivie 2005; Scadding and Koltzenburg 2005; Siddall 2005).

Tables 17.1 and 17.2 list the many causes of neuropathic pain, classified anatomically. Descriptions of many of the conditions listed in the tables are to be found elsewhere in this book. Postherpetic neuralgia is discussed in Section 19.2.3.

Neuropathic pain is multi-dimensional, comprising various types of ongoing stimulus-independent pains, and the evoked stimulus-dependent pains of allodynia, hyperalgesia, and hyperpathia. The latter are frequently major components of the overall pain complaint. The presence and clinical differentiation of stimulus-independent and stimulus-dependent pains, and their relative contributions to a patient’s complaint of pain is often indicated by the history. Examination includes careful evaluation of the evoked sensory phenomena that frequently accompany neuropathic pain, in addition to the usual assessment of modality specific sensory impairment (Bennett 2001; Rasmussen et al. 2004). Neuropathic pain is frequently associated with comorbidities which also require

careful evaluation. The combination of neuropathic pain and its comorbidities, together with associated neurological deficits in many patients, often results in major impairment of quality of life.

Patients with neuropathic pain often find it difficult to characterize the qualities of their painful symptoms, because these fall outside their previous lifelong experience of nociceptive pain. The most commonly described ongoing symptoms are deep aching in the extremities and a superficial burning, stinging, or pricking pain. Verbal descriptors from the McGill Pain Questionnaire that are used significantly more frequently by patients with neuropathic pain than those with nociceptive pain include electric shocks, burning, tingling, itching, or pricking. Descriptors such as dull, heavy, and tiring are more commonly reported by patients with nociceptive pains (Boureau et al. 1990). Accompanying paroxysmal, shock-like, or lancinating pains, sometimes radiating through a whole limb are fairly common. Stimulus-independent pains may be continuous, intermittent, or paroxysmal.

Paroxysmal pains are characteristic of certain types of neuropathic pain, for example trigeminal neuralgia, in which pains are both spontaneous and evoked, the lightning pains of tabes dorsalis and the painful crises of the neuropathy in Fabry’s disease. They are also common in other neuropathic pains, for example stump and phantom limb pains in amputees and may occur as part of almost any type of neuropathic pain, of peripheral or central cause. Painful paraesthesiae, ongoing or evoked, often accompany neuropathic pain.

Stimulus-dependent pains include allodynia, hyperalgesia, and hyperpathia, evoked by mechanical, thermal, or chemical stimulation. Stimulus-dependent pain is a hallmark of both inflammatory states and of neuropathic pain (Kilo et al. 1994; Koltzenburg et al. 1994; Woolf and Mannion 1999). Several terms are used to describe stimulus-dependent pains:

Allodynia is pain resulting from a stimulus that does not normally provoke pain.

Hyperalgesia is an increased response to a stimulus that is normally painful; noxious stimuli are often associated with a lowering of the pain threshold, together with an exaggerated perception of pain. In clinical practice, the term hyperalgesia tends to be loosely used to describe abnormally painful responses to stimuli that are normally not painful, so that these really fall into the category of allodynia rather than hyperalgesia. Several subdivisions of hyperalgesia are recognized. In static hyperalgesia, gentle pressure on the skin causes pain. In punctate hyperalgesia, stimuli such as pinprick evoke pain. In dynamic hyperalgesia, light brushing of the skin evokes pain; strictly, this is a form of allodynia rather than hyperalgesia. In heat and cold hyperalgesia, warm and cool stimuli respectively evoke pain. In the physical examination to elicit these signs, hot and cold stimuli are used that are not normally painful, at 40 and 20˚C respectively. Painful reactions to these stimuli are again further examples of allodynia.

The basis for all these types of hyperalgesia is sensitization of nociceptors. Brush-evoked allodynia is mediated by an A beta fibre input, but depends on a state of central sensitization, albeit initially established by a nociceptor input (Table 17.3).

In hyperpathia there is a raised sensory threshold, delay in perception of a stimulus, an abnormally painful reaction with summation causing increasing pain to a repetitive stimulus, and a painful after-sensation, sometimes longlasting. Hyperpathia is often severe and frequently has an explosive character, due to rapid summation. The raised sensory threshold of hyperpathia results from partial loss of afferent input, while the summation and after-sensation are due predominantly to central sensitization. So-called wind-up pain can be induced by repetitive C fibre stimulation in normal human skin, or by normally innocuous stimulation in hyperalgesia due to inflammation, as well as in neuropathic pain (Mendell and Wall 1965; Gottrup et al. 1998). Wind-up and painful after-sensations are thought to be the result of abnormal activity in wide dynamic range neurons in the dorsal horn of the spinal cord, this effect being mediated by N-methyl D-aspartate, NMDA receptors (Dickenson and Sullivan 1987).

In clinical practice, pain scales are not routinely used, though visual analogue scales can provide information about spontaneous fluctuations of pain severity and treatment-related changes. The McGill Pain Questionnaire is widely used as a research tool (Melzack 1975); the short form McGill Pain Questionnaire is robust and well validated (Melzack and Katz 1999). A pain scale specific for neuropathic pain has been developed (Galer and Jensen 1997).

Neuropathic pain is often confined to the anatomical area corresponding to the causative neurological lesion, either peripheral or central, and this is helpful diagnostically. However, neuropathic pain may radiate beyond the causative neural territory, either spontaneously, or more commonly as an evoked sensation. There is a direct relationship between the severity of pain and the extent of radiation (Laursen et al. 1997). Recruitment of wide dynamic range neurons over several segments of the spinal cord may be the basis for this clinical phenomenon (Lamotte et al. 1991; Jensen and Gottrup 2003).

Neuropathic and nociceptive pains frequently coexist. It is necessary to characterize the different components comprising a patient’s pain, as investigation and treatment are guided by accurate diagnostic assessment. A common example is cervical and lumbar spine disease, in which pain is often of both musculoskeletal, nociceptive, and neuropathic types. In this situation, the description of the pain and its distribution may not discriminate between the two types, particularly when the pain is unilateral. The presence of a neurological deficit, together with the results of electrophysiological testing and imaging may be needed to elucidate such presentations. Even following these investigations, there can be continuing doubt about the relative contributions of different types of pain, and these situations are further complicated when previous cervical or lumbar surgery has been performed.

Other painful consequences of neurological disease include arthropathies, skeletal deformities, spasticity, contractures, and dystonia. For example, in patients with painful diabetic peripheral neuropathy, distal pain in the lower legs is neuropathic, but pain may also arise from vascular insufficiency, foot and ankle arthropathies, or diabetic skin ulceration. This emphasizes the need for clinical awareness and careful diagnostic assessment, supported by investigations, in order to identify the various pathologies contributing to a patient’s overall complaint of pain.

Sensory loss is usually found in a distribution that corresponds anatomically to the causative lesion, but as described above, neuropathic pain and sensory signs may radiate well beyond the expected anatomical area. Furthermore, sensory impairment can be mild and difficult to detect, particularly when overshadowed by allodynia, hyperalgesia, or hyperpathia. The degree of sensory loss and the severity of ongoing and evoked neuropathic pain are not closely related. Trigeminal neuralgia (Section 19.2.2) is the only neuropathic pain in which cutaneous sensation is characteristically normal on routine physical examination.

Impairment of pinprick and temperature, small fibre functions sometimes highly selectively, is found in peripheral neuropathies causing neuropathic pain (Scadding and Koltzenburg 2005), though involvement of small fibres may not be selective. In the case of central lesions in the spinal cord or brain, sensory loss is of spinothalamic type, again sometimes selectively (Jensen and Lenz 1995; Vestergaard et al. 1995).

The role of sympathetic efferent activity as a causal factor in the initiation and maintenance of persistent pain in man remains controversial. There is substantial evidence for a sympathetic influence in nerve injury in animal experiments (Devor 2005). Disturbances of vasomotor and sudomotor activity are frequently seen in causalgia, complex regional pain syndrome type 2, and in complex regional pain syndrome type 1, formerly known as reflex sympathetic dystrophy (Baron 2005). Signs of altered sympathetic activity in these conditions, and occasionally in neuropathic pain states other than causalgia, include swelling, smooth glossy skin, excessive sweating, and vasomotor instability ranging from warm extremities with vasodilatation and redness of the skin, to cool vasoconstricted extremities (Fig. 17.1). Sympathetic symptoms and signs, and the role of the sympathetic nervous system in the generation and maintenance of neuropathic pain and in the pathogenesis of complex regional pain syndrome type 1 are considered in Section 17.5.8.

The sensory examination in patients with neuropathic pain requires care, but can be completed using simple standard equipment, without the need for quantitative sensory testing in the great majority of patients. Thresholds and suprathreshold responses to light touch and pin prick punctuate stimulation are assessed, before dynamic and repetitive stimulation to assess the presence of allodynia and hyperpathia.

Cold and heat are best tested either with metal rollers kept at 20 and 40˚C respectively, or with water-filled tubes. These also test touch punctuate and dynamic mechanical stimulation respectively. To overcome this, radiant heat sources or lasers can be used for selective thermal testing in selected patients. Thermal threshold testing using a thermocouple with a Peltier element is now performed routinely in many neurophysiological laboratories, supplementing clinical examination.

Examination of vibration and joint position helps with anatomical localization, and together with cutaneous sensory modalities, provides information about the density and selectivity of sensory loss. It is important to identify the degree of deafferentation and the balance of peripheral and central mechanisms responsible for neuropathic pain, as this influences the approach to treatment.

The comorbidities of chronic pain contribute to loss of function and impair quality of life in many patients with neuropathic pain. Depression is very common, present in up to 100 per cent of patients in some series (Romano and Turner 1985). The coexistence of pain and depression in neurological disease is well established (Fishbain et al. 1997; Williams et al. 2004). For example, in a study of patients with peripheral neuropathies, depression, anxiety, altered sleep patterns, social isolation, and reduced employment status were important comorbidities (Meyer-Rosberg et al. 2001).

Pain can be a symptom of primarily psychiatric disease, for example atypical facial pain and somatoform disorders, and in patients with chronic pain there is an association with somatoform disorders (Fishbain 1995) (see Section 4.7.6). Anxiety is a frequent feature, related to pain severity, delayed diagnosis, poor response to treatment, fear of progression, and social and financial consequences of the illness. In patients presenting with various types of pain, anxiety is a significantly associated symptom (Gureje et al. 1998). Conversely, in panic disorder, pain is a presenting complaint in up to 81 per cent of patients (Katon 1984). Substance abuse is an important cause of comorbidity in patients with chronic pain (Fishbain et al. 1992). Adverse effects of prescribed medications also frequently produce comorbidity: sedation, fatigue, dysphoria, and depression are the common complaints.

A wide variety of scales measuring the comorbidities of chronic pain are available, used mainly as research tools. Comorbidity measurement is reviewed by Williams (1999). The Beck Depression Inventory (Beck et al. 1961) and the Hospital Anxiety and Depression Scale (Zigmond and Snaith 1983) provide quantitative information about mood and affect. Of the wider multidimensional scales, the most commonly used are the Short Form 36 of Medical Outcomes Study (Ware et al. 1993), the Sickness Impact Profile (Bergner et al. 1981) and the Multidimensional Pain Inventory (Kerns et al. 1985). The Oswestry Low Back Questionnaire (Fairbank et al. 1980) has been validated in patients with back pain.

Detailed discussion of the mechanisms contributing to the development of neuropathic pain lies beyond the scope of this chapter. As mentioned earlier, symptoms and signs in patients with neuropathic pain cannot yet be tightly linked with underlying cellular and neuropharmacological pathophysiologies. Animal experiments have yielded a great number of candidate mechanisms that may underlie neuropathic pain, and increasingly, investigative techniques such as microneurography and skin biopsy in peripheral neuropathies are contributing data of direct relevance to human disease.

Table 17.3 outlines the major pathophysiological changes likely to contribute to peripheral neuropathic pain. These include: abnormal impulse generation in damaged primary afferent axons and their cell bodies in the dorsal root ganglia, with the development of mechanical and noradrenergic sensitivities; fibre interactions in damaged nerves; and secondary central effects in the spinal cord. Table 17.4 summarizes the relationship between symptoms, primary afferent type, and pathophysiological properties. Full accounts of these changes can be found in McMahon and Koltzenburg (2005).

In relation to mechanisms of central neuropathic pain, knowledge is much less complete. Table 17.5 summarizes the main likely contributing properties. Data concerning the factors governing the development of neuropathic pain with cerebral lesions have reached the level of brain region excitation/inhibition anatomico-physiological correlations. And while positron emission tomography and functional MRI have opened up the area of pain perception, in both cognitive and emotional aspects and now permit psychophysical investigation in painful states in man, neuropathic pain has as yet been relatively little studied. The major changes responsible for the development of central neuropathic pain include: loss of normal inhibitory controls at cord and thalamic levels; inflammatory changes in spinal cord neurons leading to increased excitability; increased excitatory amino acid activation and abnormal expression of sodium and calcium channels; and imbalance of activity in central pathways. However, it should be appreciated that the links between pathophysiological changes and symptoms in central neuropathic pain are much less certain than for those established for peripheral neuropathic pain. McMahon and Koltzenburg (2005) provide a detailed account of the current understanding of the mechanisms of central neuropathic pain.

Complex regional pain syndrome, often referred to as CRPS, is the term introduced recently (Boas 1996) to refer to a group of conditions previously known as reflex sympathetic dystrophy, causalgia, algodystrophy, Sudeck’s atrophy, and a number of other diagnostic terms (Table 17.6). Clinical features common to all these conditions, in very variable proportion, include pain associated with allodynia and hyperalgesia, autonomic disturbances, trophic changes, oedema, and loss of function of the affected part, usually a limb. The term causalgia literally means burning pain and was used by Weir Mitchell to describe the severe burning pain, hyperaesthesia, glossy skin, and colour changes in the limbs of soldiers following injury to major nerves, sustained in the American Civil War (Mitchell et al. 1864; Richards 1967). Later, it became clear that limb injuries not involving nerves could produce a very similar clinical picture, becoming known as reflex sympathetic dystrophy (Evans 1946). The definition and nosology of these conditions remains almost as problematic as it was in the 1940s (Livingston 1943), underlining the fact that, as with neuropathic pain, it is not possible to tightly link symptoms and signs with specific pathophysiological properties (Woolf and Mannion, 1999). A diagnosis of complex regional pain syndrome describes a clinical state, without making unjustifiable pathophysiological assumptions, as was the case with the older term reflex sympathetic dystrophy.

Complex regional pain syndrome describes a variety of painful conditions that usually follow injury, occur regionally, have a distal predominance of abnormal findings, exceed both in magnitude and duration the expected clinical course of the inciting event, often result in significant impairment of motor function, and show variable progression over time (Boas 1996). Complex regional pain syndrome is divided into types 1 and 2:

These descriptive definitions, based on symptoms and signs, avoid unwarranted inclusion of pathophysiology, but lead to difficulty in recognizing the clinical limits of these conditions, particularly complex regional pain syndrome type 1. Confusion has also resulted from making complex regional pain syndrome type 2 and causalgia synonymous, and emphasizes the problems caused by applying a term originally intended to refer to a single symptom of burning pain, to a clinical syndrome. For the moment, however, this terminology prevails.

The causes of complex regional pain syndrome are listed in Table 17.7. The great majority of cases are secondary to peripheral tissue injury such as fractures and soft tissue injury. Complex regional pain syndrome type 2 represents a small minority of cases. Rarely, complex regional pain syndrome occurs with central nervous system lesions.

The large prospective study of Veldman et al. (1993) provides an indication of the relative frequency of the many symptoms and signs of complex regional pain syndrome, at two time intervals following the inciting event (Table 17.8), divided into four categories. Such prospective studies are few, but suggest an under-recognition of the syndrome, while also raising important questions about the limits of the diagnosis (Bickerstaff and Kanis 1994). Spontaneous and evoked pains are common to all patients at some time. Pain quality is often burning, aching, or throbbing. In complex regional pain syndrome type 2, additional paroxysmal pains are common. Allodynia, hyperalgesia, and hyperpathia are often very severe, leading to immobilization and avoidance of any skin contact or pressure on the affected limb. Immobilization itself may in turn result in secondary muscle wasting and joint stiffness, over and above the dystrophic changes that occur as part of the condition.

Autonomic signs are variable in complex regional pain syndrome, but occur at some time in the course of the illness. They may be subtle in some patients and gross in others. Abnormalities of colour, temperature, and sweating are frequently described. Oedema is common but not universal (Fig. 17.1) (Table 17.8).

Objective motor signs are difficult to elicit, because both passive and active movements provoke severe pain. Tendon reflex examination is often impossible. Wasting and weakness are common. Tremor, incoordination, muscle spasms, and dystonia affect some patients.

Dystrophic changes include skin thinning, sometimes with a shiny appearance, or thickened, flaky skin. Hair may either be lost or become coarse, and nails may become thickened. Osteoporosis is common, recognized in its extreme form as Sudeck’s atrophy.

Less common features of complex regional pain syndrome include a migratory or relapsing pattern, recurrent skin infections associated with chronic oedema, increased skin pigmentation, nodular fasciitis of the palmar or plantar skin, and nail clubbing (Bentley and Hameroff 1980; Veldman et al. 1993).

The disproportionate pain and loss of function of complex regional pain syndrome, together with lack of clarity about the pathogenesis of the condition have led to consideration of a psychological contribution to its development and perpetuation. Patients with conversion disorder or factitious illnesses may present with symptoms that can closely resemble complex regional pain syndrome. Not all

the clinical features listed in Table 17.8 need to be present in order to establish the diagnosis, and it is perhaps not surprising that some patients with primary psychiatric morbidity are erroneously diagnosed as suffering from complex regional pain syndrome.

The severe pain and loss of function in complex regional pain syndrome, together with frequent delay in diagnosis causes anxiety, fear, and depression in many patients. Issues of secondary gain may be raised. However, there is no evidence that the condition is primarily psychologically determined (Covington 1996). Some of the patients reported by Ochoa and Verdugo (1995) as having pseudoneuropathy, presenting with clinical features mimicking complex regional pain syndrome, were likely to have suffered from primarily psychiatric illnesses. This emphasizes the need for repeated careful evaluation of patients in whom there is continuing diagnostic doubt.

It has been suggested that three clinical stages of complex regional pain syndrome can be recognized (Blumberg and Janig 1994):

However, not all patients follow this course, the duration of these phases is very variable and not all patients progress through all three stages (Veldman et al. 1993).

As already stated, the diagnosis of complex regional pain syndrome is based on clinical features. Claims have been made for the usefulness of three-phase bone scans as a diagnostic test, but while this investigation is frequently abnormal (Goldsmith et al. 1989), a normal scan does not exclude the diagnosis.

The clinical limits of complex regional pain syndrome are difficult to define and so data concerning the incidence of the condition are unreliable. Careful prospective studies, for example the series of 829 patients reported by Veldman et al. (1993), and the study of 274 patients with a Colles fracture of Bickerstaff and Kanis (1994), indicate the prevalence of symptoms at various intervals following the inciting injury. At what point the symptoms and signs are judged to be disproportionate in severity and duration to the inciting event is a matter for debate in the different conditions that may lead to complex regional pain syndrome, and this once again emphasizes the imprecision of the current diagnostic criteria.

Involvement of the sympathetic nervous system: sympathetically maintained pain. Leriche (1916) described the relief of causalgia in a patient with a brachial plexus injury and thrombosis of the brachial artery, by periarterial surgical sympathectomy. Pain relief was accompanied by an improvement in discolouration and sweating changes following sympathectomy in a further series of patients (Leriche 1939), and pre-ganglionic sympathectomy became established as standard treatment for painful nerve injuries.

In relation to neuropathic pain, including complex regional pain syndrome type 2, three sites of sympathetic-sensory interaction after nerve injury have been identified in experimental animal studies: the region of nerve damage itself, undamaged fibres distal to the nerve lesion, and the dorsal root ganglion (Devor 2005). The findings are consistent with expression of alpha adrenoreceptors on regenerating and partly damaged nerve fibres. Denervation supersensitivity may contribute to the magnitude of the agonist effect of circulating and locally released catecholamines.

There is also evidence of a sympathetic influence on neuropathic pain in man. For example, intraoperative stimulation of the sympathetic chain exacerbates causalgia (Walker and Nulsen 1948). In patients with successfully treated causalgia, intracutaneous injection of noradrenaline rekindled their original pain (Wallin et al. 1976). Intracutaneous injection of adrenaline or noradrenaline in dermatomes affected by postherpetic neuralgia increases both ongoing pain and allodynia (Choi and Rowbotham 1997). Temporary relief of neuropathic pain by sympathetic blockade has often been reported, usually of peripheral origin but also some central neuropathic pains (Bonica 1990; Arner 1991). However, the existence of this influence has been questioned, on the basis of poor design of some studies, including the lack of proper controls and an underestimation of psychological factors (Ochoa et al. 1994).

By definition, complex regional pain syndrome type 1 develops in the absence of an initiating nerve injury. As described below, there is substantial evidence of inflammation in the pathogenesis of this condition. Some components of cutaneous inflammation and hyperalgesia are enhanced by alpha adrenoreceptor stimulation (Drummond 1995; Kinman et al. 1997). In addition, noradrenaline causes release of prostaglandins, which are major mediators of inflammatory responses. This effect may be dependent on the pre-existing state of nociceptor activity (Baron et al. 1999). Finally, there is an increased alpha adrenoreceptor density in skin biopsies form patients with complex regional pain syndrome type 1 (Drummond et al. 1996).

In studies on patients with complex regional pain syndrome type 1 whose pain is relieved by temporary sympathetic block the severity of ongoing pain and allodynia is increased by sympathetic stimulation produced by whole body cooling, indicating a peripheral noradrenergic interaction with primary sensory afferents (Baron et al. 2002). This and other evidence indicate a peripheral sympathetic influence in complex regional pain syndrome type 1 (Baron 2005). The term sympathetically maintained pain is now widely used to describe the sympathetic agonist effect in chronic pain states.

Central autonomic dysregulation. Several observations are consistent with a central disturbance of autonomic control in patients with complex regional pain syndrome type 1. Hyperhidrosis is found in many patients, both in resting states and in response to physiological stimulation (Birklein et al. 1997). Studies of centrally mediated sympathetic reflexes show that complex regional pain syndrome type 1 is associated with an abnormal unilateral inhibition of cutaneous sympathetic vasoconstrictor neurons, leading to the development of a warm limb, present in some patients. In patients with cold limbs, temperature and vascular perfusion are both reduced in the affected limb during the full range of sympathetic stimulation procedures (Baron 2005).

Inflammation and immune stimulation. Clinical signs of inflammation are obvious in many patients with complex regional pain syndrome type 1 (Table 17.8). Response of type 1 to systemic corticosteroids, in the early stages, at less than 13 weeks duration, has been reported (Christensen et al. 1982), though complex regional pain syndrome is refractory to steroids at later stages.

Neurogenic inflammation is an important component of complex regional pain syndrome, leading to oedema, vasodilatation, and increased sweating. For example, plasma extravasation of radiolabelled immunoglobulins has been shown (Oyen et al. 1993); there is evidence of neurogenic inflammation in joints (Weber et al. 2001); and in the fluid of artificially produced blisters, higher levels of interleukin-6 and TNF alpha have been reported in the affected limb, compared with the unaffected contralateral limb (Huygen et al. 2002).

The possibility that the inflammatory features of complex regional pain syndrome type 1 might represent a chronic post-infectious state is suggested by the presence of higher titres of serum antibodies to intestinal pathogens than in healthy controls (Goebel 2001). In addition, Borrelia infection has been reported in association with complex regional pain syndrome with marked bone dystrophy (Sudeck’s atrophy) (Bruckbauer et al. 1997). A preliminary report indicates response of the pain to treatment with intravenous human immunoglobulin (Goebel et al. 2002), though an analgesic effect was also found in patients with a variety of other types of pain, both nociceptive and neuropathic, indicating a rather non-specific effect of the non-blinded treatment in this study.

Antibodies to Campylobacter jejuni and increased tissue specific antibodies in sera from patients with complex regional pain syndrome, with disease duration of less than 1.5 years, are further evidence of immune activation resulting from antecedent infection (Goebel et al. 2005). Finally, it has been shown that serum from a patient with complex regional pain syndrome type 1, whose pain responded on three separate occasions to IVIG, produced pain behavioural changes when injected into mice (Goebel et al. 2005).

While these must be regarded as preliminary observations, they indicate the probable importance of inflammatory processes and possibly of immune-mediated changes in the pathogenesis of complex regional pain syndrome.

Central sensitization. As in peripheral neuropathic pain, prolonged noxious inputs from peripheral tissues in complex regional pain syndrome type 1 will lead to central sensitization (Tables 17.3 and 17.4). In addition, functional MRI studies have demonstrated adaptive changes in the thalamus (Fukomoto et al. 1999) and cortex (Maihofner et al. 2003), the latter resolving with successful treatment of the pain (Maihofner et al. 2004).

Motor abnormalities. The prospective investigation of Veldman et al. (1993) emphasized the frequency of motor abnormalities in complex regional pain syndrome: tremor, paresis, and dystonia. Kinematic analysis of certain motor functions indicates that the problem lies in central motor processing (Baron 2005).

The treatment of causalgia, complex regional pain syndrome type 2, is as for neuropathic pain (Section 17.6). The following comments relate to the treatment of complex regional pain syndrome type 1. It seems intuitive that early treatment improves outcomes. However, there are difficulties in identifying the onset of the condition, and there is no controlled study addressing this important issue. Clearly, the management of conditions known to have the potential of developing into complex regional pain syndrome should be optimized in the early stages, but there is no evidence that it is exclusively poorly managed patients who develop the syndrome. Immobilization and disuse probably contribute to the development of complex regional pain syndrome type 1, so it makes sense to minimize these factors with physiotherapy and aid restoration of function at as early stage after injury as possible.

Sympatholytic procedures. Temporary sympathetic block can be achieved either by injection of local anaesthetic around sympathetic ganglia or by intravenous regional block, using guanethidine or other noradrenaline-depleting drugs (Hannington-Kiff 1974). These techniques were used for many years, for neuropathic pains of various types and for complex regional pain syndrome type 1, before being subjected to controlled study. Most of the studies cited here have included patients with neuropathic pain, as well as patients with complex regional pain syndrome type 1.

There are seven randomized controlled trials of intravenous regional block (Jadad et al. 1995). In most of these, guanethidine was used, but some studies included reserpine, bretylium, droperidol, or ketanserin. In four of these, no analgesic effect was found. Recruiting patients who had responded to guanethidine in open trial, to a controlled trial of the same treatment did not reveal a therapeutic effect (Jadad et al. 1995). Sixty patients randomized to treatment with intravenous regional block using guanethidine and lignocaine, versus saline and lignocaine, showed no difference in pain relief between the two groups (Ramamurthy and Hoffman 1995). Interestingly, both groups showed improvements in oedema, sudomotor, vasomotor, and trophic changes. A tourniquet inflated to suprasystolic pressure alone relieves hyperalgesia, but temperature sensation is not altered, indicating that hyperalgesia is mediated by A beta mechanoreceptor fibres, these being most susceptible to pressure block by the cuff. In patients with complex regional pain syndrome type 1, four intravenous regional block with guanethidine and pilocarpine versus placebo were ineffective (Livingstone and Atkins 2002). A controlled trial of sympathetic ganglion local anaesthetic block in complex regional pain syndrome type 1 failed to demonstrate any immediate effect, though at 24 h, the lignocaine-treated patients were better than controls (Price et al. 1998).

Systematic reviews of sympatholysis in complex regional pain syndrome type 1 and neuropathic pain have concluded that the treatment is ineffective (Kingery 1997; Perez et al. 2001).

Surgical sympathectomy can no longer be recommended; no study has demonstrated lasting analgesia (Baron 2005). A return of pain, sometimes worse than the original pain for which the procedure was performed, at an interval following sympathectomy has been frequently observed. The initial warm limb produced by sympathectomy is replaced by a cold limb, despite evidence of no sympathetic reinnervation (Baron and Maier 1996). This change, and the recurrence of pain are probably due to the development of denervation supersensitivity to circulating catecholamines.

Systemic drugs. The evidence base for systemic drug therapy of complex regional pain syndrome type 1 is very limited. Drawing on the limited number of published trials available, not all of which are randomized controlled trials (Baron 2005), only tentative conclusions can be reached at present.

Mild standard analgesic drugs such as paracetamol have only marginal effect. Non-steroidal anti-inflammatory drugs are often used and found to be useful by some patients. Neither class of drug has been investigated in controlled trials. The same applies to opioids, but in routine clinical practice these drugs are found to be helpful by many patients and may be the only treatment that has an effect on the severe pain. A trial of tramadol, followed if ineffective by either slow release morphine or fentanyl patches is justified.

If the development of complex regional pain syndrome type 1 can be confidently identified at less than 3 months, a trial of prednisolone, initially at high dose, can be recommended (Christensen et al. 1982). Treatment at later times is ineffective. There is some evidence that calcium-regulating drugs have an effect. These include bisphosphonates ((Varenna et al. 2000; Manicourt et al. 2004), and intranasal calcitonin (Gobelet et al. 1992). Two trials of gabapentin have shown an analgesic effect in complex regional pain syndrome type 1 (Mellick and Mellick 1995; van de Vusse et al. 2004). There are reports that free radical scavengers are effective, either oral N-acetyl cysteine or topical dimethylsulfoxide (Perez et al. 2003). Uncontrolled studies have suggested therapeutic effects for phenoxybenzamine, tricyclic antidepressants, phenytoin, and nifedipine (Scadding 1999).

There are many clinical similarities in the nature of the ongoing and evoked pains of complex regional pain syndrome type 1 and of neuropathic pain. Thus despite the absence of firm evidence of effectiveness, it is justifiable to recommend that the drugs shown to be effective for neuropathic pain should also be tried in patients with complex regional pain syndrome, with a clear understanding on the part of both clinician and patient that the drugs in question are being used for an unlicensed indication. This recommendation reflects both the severity of the pain in complex regional pain syndrome and the current very restricted armamentarium of effective treatments.

Epidural and intrathecal treatment. Intrathecal morphine is effective in the treatment of severe pain due to complex regional pain syndrome, refractory to all other measures (Becker et al. 1995). Epidural clonidine, an alpha 2 receptor agonist, has been reported to be effective for pain in complex regional pain syndrome affecting either upper or lower limbs (Rauck et al. 1993). Intrathecal baclofen may help the dystonia sometimes associated with complex regional pain syndrome (van Hilten et al. 2000).

Electrical stimulation. Uncontrolled observations in clinical practice indicate that transcutaneous electronical nerve stimulation has a useful effect in some patients. The effect of spinal cord stimulation is unpredictable, but this can be effective (Kemler et al. 2000).

Physiotherapy. Efforts to achieve pain relief should always be accompanied by attempts to mobilize the affected part and restore function; the clinical features of complex regional pain syndrome type 1 are probably to an extent perpetuated by disuse and immobility. Physiotherapy and occupational therapy have both been shown to be helpful (Oerlemans et al. 2000).

Attention and mirror visual feedback. Distraction from pain by various means has been advocated and exploited as a therapeutic intervention for many years. The mechanisms and potential for treatment have recently come under scrutiny (McCabe et al. 2005). Mirror visual feedback, a method in which patients can utilize the movements of their normal limb to promote active movement of a limb affected by complex regional pain syndrome type 1, has shown promise in short-term treatment (Ramachandran 2005). Further studies are awaited.

Psychological measures. The comments concerning cognitive behavioural therapy (Section 17.6.7) in relation to the treatment of neuropathic pain apply equally to complex regional pain syndrome type 1. Combined physical therapy and coggitive behavioural therapy has been shown to be effective (Lee et al. 2002).

Amputation. Patients with very severe intractable pain in a limb rendered useless by complex regional pain syndrome type 1 sometimes raise the question of amputation. While amputation will rid them of a painful hyperaesthetic limb, they should be warned that there is a high risk of developing stump and phantom limb pain. Although it has not been proven, there is a strong clinical impression that those with painful limbs prior to amputation are more likely to develop these new problems, which for some patients may be just as incapacitating as their original pain.

Information concerning prognosis of complex regional pain syndrome is limited and it is difficult to compare directly results of different studies, due to case inclusion bias and initial severity. Many patients remain incapacitated by their symptoms for years and in a proportion, problems persist for decades or even lifelong. In one study, at follow up after 5.5 years, 62 per cent of patients remained markedly troubled by pain and impaired function (Geertzen et al. 1994). Factors adversely affecting outcome include severity at the time of diagnosis, female gender, and affection of the lower limb. It is in these patients that the most serious complications are likely to occur, including skin ulceration, infection, chronic oedema, and dystonia (van der Laan et al. 1998).

Complex regional pain syndrome type 1 is an uncommon condition in children and thus prone to delayed diagnosis. The lower limb is much more frequently affected than the upper limb in a ratio of about 5:1, and girls are more commonly affected than boys, in a ratio of approximately 4:1. Those affected are typically pubertal adolescent girls. Many children with complex regional pain syndrome have participated in competitive sports and other physical activities, placing them at increased risk of musculoskeletal injury. There has been interest in the possible psychological gain to a child of a persistent injury enabling escape from the stress of competition and parental expectations, as a factor influencing the development of the condition. The prognosis for complete recovery with physiotherapy, transcutaneous nerve stimulation and cognitive behavioural therapy is very much better in children than in adults (Wilder 1996; Sherry and Weisman 1988).

Neuropathic pain is much more difficult to treat than most nociceptive pains and the range of specific therapies is still limited. Once established, neuropathic pain is often a lifelong complaint. Regardless of the presence or absence of additional neurological deficits, neuropathic pain and its comorbidities impose a substantial burden for many patients.

When neuropathic pain arises from compressive lesions of peripheral nerves, spinal roots and sometimes of the spinal cord, surgical decompression can partly or completely relieve pain. However, this is certainly not always the case, particularly if compression has been prolonged. There is a poor correlation between the severity of the neurological deficit and the degree of pain, both pre- and post-surgery. The place of other surgical interventions for neuropathic pain is very limited. Ablative operations that produce new lesions in the somatosensory system, designed to relieve neuropathic pain, may do so temporarily, but are themselves potent causes of neuropathic pain. Such pain develops at variable intervals following surgery, ranging from days to years.

When neuropathic pain is anatomically limited in extent and associated with painful evoked symptoms—allodynia, hyperalgesia and hyperpathia—every effort should be made to employ local measures. Not only may these be partially effective, but they will also reduce or obviate the need to consider systemic drug treatment. All classes of drugs given for neuropathic pain may cause treatment-limiting adverse effects and while by no means confined to older patients, these represent a particular problem in this age group.

Many patients with chronic neuropathic pain require a multi-modality approach to their treatment. This reflects both the multidimensional nature of neuropathic pain and the partial effectiveness of individual treatments. Support and follow up over long periods are needed. Many patients are most appropriately managed in an integrated multidisciplinary setting that includes input from neurologists, anaesthetists, psychologists, physiotherapists, and occupational therapists.

Local anaesthetic injections. Local anaesthetic blocks of peripheral nerves, plexuses, or spinal roots may be helpful diagnostically, though do not always permit accurate localization of causative lesions. For example, an effective sensory root block may mask a more peripheral lesion. The effect of a local anaesthetic block is usually shortlived and thus of limited therapeutic value.

Topical local anaesthetic. In patients with limited areas of allodynia, local-anaesthetic-impregnated patches, or EMLA cream are indicated. A less elaborate but sometimes surprisingly effective alternative is simple 5 per cent lignocaine ointment. Topical local anaesthetic has been shown to be beneficial in painful polyneuropathy, postherpetic neuralgia, and a variety of other neuropathic pains (Galer et al. 1999 ; 2002).

Topical capsaicin. This pungent extract of chilli peppers, binds to vanilloid receptors and causes depolarization of afferent C fibres, with release of substance P. This agonist action leads to burning pain on first application, and this frequently limits its clinical use. However, following repeated topical application, there is prolonged depletion of substance P and this probably accounts for desensitization of afferent C fibres. Topical capsaicin 0.075 per cent can be helpful in patients with neuropathic pain associated with limited areas of allodynia. It has been shown to be effective in painful diabetic neuropathy (Capsaicin Study Group 1991), postherpetic neuralgia (Watson et al. 1993), and post-surgical pain (Ellison et al. 1997), but not in HIV neuropathy (Paice et al. 2000).

Until relatively recently, the literature abounded with therapeutic claims for a wide variety of systemic drugs in neuropathic pain, based on anecdotal, open label reports or poorly controlled studies using limited pain outcome measures. An appreciation of the need to consider the multidimensional aspects of neuropathic pain, and improved trial methodology has led to clinical trial results of greater relevance to patients with neuropathic pain (discussed in several recent systematic reviews: Kingery 1997; McQuay and Moore 1998; Sindrup and Jensen 1999, 2000; Finnerup et al. 2005). Nonetheless, comparison of data from different trials remains problematic.

A useful calculated measure of effectiveness of treatments, now widely used, is the number needed to treat, NNT, defined as the number of patients who have to be treated to produce pain relief in one patient (McQuay and Moore 1998). The degree of analgesia is usually defined as 50 or 30 per cent, the latter figure equating to a value patients describe as at least ‘moderate’ pain relief. Similarly, the number needed to harm, NNH, provides a useful measure of safety and acceptability of a drug. This is defined as the number of patients who need to be treated for one patient to drop out due to adverse effects. NNT and NNH can only be calculated from trials with dichotomous data.

Calculation of NNT permits data from different trials to be pooled, but does not necessarily give a consistent measure of relative effectiveness of different drugs. This is due to methodological factors including trial design, whether parallel or cross-over, different drug dosages, variation in the scales and instruments used for assessment of pain severity, baseline pain scores at the time of entry to the trial, variable inclusion of comorbidity data, heterogeneous patient inclusion, incomplete reporting of results including adverse effects, and variable magnitude of placebo effects in different trials. NNT data thus need to be interpreted with caution. Additional confounding factors include an absence of the most relevant head-to-head trials in certain instances, and under-reporting of trials produces negative results.

Multiple mechanisms contribute to the development of neuropathic pain (Tables 17.3–17.5), so it should not be surprising that a single drug rarely produces more than partial analgesia. In clinical practice this frequently leads to treatment with drug combinations, associated with a substantial incidence of adverse effects.

Table 17.9 summarizes NNT values for systemic drug treatment of peripheral, central, and mixed neuropathic pain, together with NNH values where these have been calculated.

Antidepressant drugs. The mechanism by which tricyclic antidepressant drugs exert an analgesic effect in neuropathic pain remains uncertain. A serotoninergic action, enhancing the inhibitory effect of the descending pathway from brainstem to the dorsal horn of the spinal cord (Fields and Basbaum 1978) is often invoked. However this is unlikely to be the sole mechanism, as the selective serotonin reuptake inhibitors, which should theoretically be more potent analgesics due to this action, have a weaker analgesic effect

than tricyclics in neuropathic pain. Tricyclics also block the uptake of noradrenaline, and may potentiate noradrenergic inhibitory mechanisms at the spinal level (Rang et al. 2001).

Chronic pain and depression frequently coexist, with a complex relationship between the two conditions (Monks and Merskey 1999). That the mechanisms of action of tricyclics in chronic pain and depression are different and distinct is indicated by an earlier onset of an analgesic than an antidepressant action (Langohr et al. 1982), an analgesic effect without relief of depression in some patients (Lascelles 1966), and an analgesic effect in patients who are not depressed (Lance and Curran 1964).

Amitriptyline, nortriptyline, imipramine, desipramine, clomipramine, and maprotiline have all been demonstrated to reduce neuropathic pain, with a combined NNT of 3.1 and a combined NNH of 14.7 (Finnerup et al. 2005). Adverse effects, particularly the anticholinergic actions, often limit the dose that is tolerated. There is little to choose between the different tricyclics for the treatment of neuropathic pain, based on current evidence.

Of the selective serotonin reuptake inhibitors, paroxetine and citalopram have been shown to be weakly effective in peripheral neuropathic pain, with a combined NNT of 6.8 (Finnerup et al. 2005), but in one trial fluoxetine was no more effective than placebo in patients with painful diabetic neuropathy (Max et al. 1992). The more recently available serotonin and noradrenaline reuptake inhibitors, venlafaxine and duloxetine, appear to be of similar efficacy to selective serotonin reuptake inhibitors in painful peripheral neuropathy, with a combined NNT of 5.5 (Finnerup et al. 2005).

Neuroleptic drugs. These were reported to have analgesic effects in both peripheral and central neuropathic pain in the 1950s (Sadove et al. 1955; Margolis and Gianascol 1956; Sigwald et al. 1959). There is limited evidence of efficacy of neuroleptics when used in combination with antidepressants, for both nociceptive and neuropathic pains (Monks and Merskey 1999), but overall, the evidence for a substantial independent analgesic effect of neuroleptics in neuropathic pain is slender. The serious long-term adverse effects of neuroleptics, together with the relative lack of evidence of efficacy, has led to a reluctance to use neuroleptics for the treatment of chronic neuropathic pain.

Anticonvulsant drugs. Carbamazepine and phenytoin have membrane-stabilizing actions mediated by non-specific sodium channel blockade, and theoretically, might have multiple sites of action in relation to the mechanisms underlying neuropathic pain.

The remarkable effect of carbamazepine in trigeminal neuralgia (Blom 1963) led to the hope that this and other antiepileptic drugs might be as effective generally in neuropathic pain. Sadly, this has not proved to be the case, either for peripheral or central neuropathic pain (McQuay et al. 1995; Finnerup et al. 2005). Although early trials suggested efficacy, these were methodologically flawed. Open trial in numerous patients over the last three decades has demonstrated the very limited usefulness of both drugs as analgesics in both peripheral and central neuropathic pains, with the notable exception of trigeminal neuralgia.

Gabapentin and pregabalin bind to the alpha 2 delta subunit of voltage-dependent calcium channels. Their mode of action in neuropathic pain is uncertain, but they may modulate neurotransmitter release from primary afferent terminals, via an action on interneurones in the dorsal horn of the spinal cord. Both drugs have been extensively studied in painful diabetic neuropathy and postherpetic neuralgia in large, well-controlled trials that have included multidimensional pain measures. A combined NNT for gabapentin is 4.7, with an NNH of 17.8 (Finnerup et al. 2005). With pregabalin, the NNT for postherpetic neuralgia is 5.5, and for painful diabetic neuropathy 7.7 (Dworkin et al. 2003; Rosenstock et al. 2004; Sabatowski et al. 2004; Richter et al. 2005).

Lamotrigine exerts an inhibitory effect on voltage-sensitive sodium channels, and also inhibits release of the excitatory amino acids glutamate and aspartate. By these actions it stabilizes neuronal membranes. In painful diabetic neuropathy, Eisenberg et al. (2001) reported an analgesic effect, with an NNT of 4.0. In HIV neuropathy, lamotrigine was shown to have a weak analgesic effect in only one of two trials (Simpson et al. 2003). In central post-stroke pain, lamotrigine has a weak effect (Vestergaard et al. 2001), but no effect in spinal cord injury pain (Finnerup et al. 2002).

Sodium valproate has several actions which indicate that it might exert an analgesic effect in neuropathic pain, including increased synthesis and release of GABA, sodium channel blockade, and reduced neuronal excitability to glutamate (Locher 1999). Although an analgesic effect was reported in three trials in diabetic neuropathy and postherpetic neuralgia, this was not found in another trial in painful poyneuropathy including patients with diabetic neuropathy, and valproate was ineffective in a trial in patients with spinal cord injury (see Finnerup et al. 2005).

Topiramate stabilizes neuronal membranes through various mechanisms, including anti-glutamate effects, sodium channel blockade, and enhancement of GABA-mediated inhibitory actions (Shank et al. 2000). Three out of four trials of topiramate in painful diabetic neuropathy showed no effect, and while the other trial did indicate an analgesic effect, NNT 7.4, all four trials were associated with high drop out rates due to adverse effects, producing a combined NNH of 6.3 (Raskin et al. 2004; Thienel et al. 2004).

Opioids. For many years, it was received wisdom that opioids were ineffective in neuropathic pain. This, combined with a reluctance to use opioids in patients with chronic non-malignant pain, resulted in a reappraisal of opioids in neuropathic pain only in recent years (Portenoy 1990). Finnerup et al. (2005) summarize the effectiveness of opioids in different neuropathic pain types. Morphine, oxycodone, or the weaker opioid tramadol have been shown to produce analgesia in peripheral neuropathic pain, including diabetic neuropathy, postherpetic neuralgia, and phantom limb pain, with combined NNT of 2.5–3.9, and NNH of 7.9–11.3.

NMDA antagonists. The action of N methyl Daspartate antagonists in blocking the afferent C fibre wind-up of dorsal horn neurons (Table 17.3) led to hopes that drugs of this class might be effective for neuropathic pain. Intravenous infusions of NMDA antagonists, including ketamine, do produce pain relief (Sang 2000), but often with a limited duration of minutes or hours. Moreover, the treatment is frequently associated with marked sedation and other unpleasant adverse effects including hallucinations. Two trials of dextromethorphan have shown an effect in painful polyneuropathy, including diabetic (Nelson et al. 1997; Sang et al. 2002). However, 11 other trials of dextromethorphan, memantine, or riluzole in painful neuropathy, postherpetic neuralgia, phantom limb pain, or mixed types of neuropathic pain have been negative (Finnerup et al. 2005).

Antarrhythmics. Lignocaine, a non-specific sodium channel blocker, given intravenously, produces short-lived relief of neuropathic pain, for example painful diabetic neuropathy (Kastrup et al. 1987). The oral analogue of lignocaine, mexiletine has been extensively trialed but found to be effective in only two studies, one in painful diabetic neuropathy (Dejgard et al. 1988) and the other in painful peripheral nerve injury (Chabal et al. 1992). Seven other studies have shown a lack of effect in mixed painful peripheral neuropathies, HIV neuropathy, and spinal cord injury (Finnerup et al. 2005).

Cannabinoids. The cannabinoid receptor CB₁ is widely distributed in the central nervous system. Of particular relevance to pain processing and the analgesic effects of cannabinoids is the expression of CB₁ receptors in the thalamus, periaqueductal grey, and rostroventromedial medulla. In the spinal cord, receptors are expressed in the superficial dorsal horn and dorsolateral funiculus. Several endogenous ligands are now known, the endocannabinoids. The history of development of cannabis as an analgesic and the analgesic mechanisms of cannabinoids are reviewed by Rice (2005).

In a study of pain related to brachial plexus avulsion, Berman et al. (2004), demonstrated a modest analgesic effect with two cannabis extracts. Using the synthetic cannabinoid CT-3, Karst et al. (2003) also showed a mild analgesic effect in a mixed group of patients with neuropathic pain. Pain was not a primary end-point of a large study of cannabinoids in patients with multiple sclerosis (Zajicek et al. 2003), but 30–50 per cent of patients reported some improvement in their pain. A preliminary study examined the effect of cannabinoids on a wide range of intractable neurogenic symptoms, finding a reduction of pain scores in the order of 30–35 per cent, though with a 21 per cent placebo response rate (Wade et al. 2003). Adverse effects led to 17 per cent of patients withdrawing from this study.

Comparison of systemic drug trial data is difficult for the reasons already outlined and it is thus hard to make firm recommendations. Trials have been far greater in number for peripheral than for central neuropathic pain. Finnerup et al. (2005) have drawn attention to the criteria relevant to setting an order of preference for prescribing drugs to patients. These include consistent outcome in high quality trials, low NNT and high NNH values, prolonged effectiveness, effect on quality of life measures, and low cost.

Not all the data necessary to satisfy these criteria are available for all classes of medication. Bearing this in mind, the following sequence for treating peripheral neuropathic pain is suggested:

For central neuropathic pain, recommendations are less securely evidence based, but the following order is suggested:

The place of epidural analgesia in acute pain management for childbirth, thoracic, and abdominal surgery is well established (Breivik 2005). For chronic pain, the indications are limited. Some patients with severe intractable cancer pain, mostly of non-neuropathic type, are effectively treated with chronic epidural infusions. The main neuropathic pain indication for epidural analgesia is severe spinal root pain that is resistant to all other measures. Injections of local anaesthetic and corticosteroid may relieve chronic root pain refractory to other treatments, sometimes for days or even weeks, and this may be a useful form of treatment in a few patients.

There is a synergistic analgesic action of local anaesthetic and opioids such as fentanyl given via the epidural route. The duration of action of this combination can be substantially prolonged by the addition of adrenaline, which reduces the absorption of the other two drugs. This occurs without causing any reduction in spinal cord blood flow. Epidural infusions can be very effective for the management of severe root pain, but over time, epidural fibrosis and adhesions develop, thus limiting the usefulness of the technique for long-term treatment (Breivik 2005).

Epidural treatment of chronic pain requires an experienced team with good facilities for monitoring the treatment. Accurate placement of the epidural catheter is critical to achieving good analgesia; loss of an initial good effect can be the result of catheter tip migration. Adverse effects include haemodynamic disturbances, sensory loss, and motor paralysis, loss of bladder function, respiratory problems due to rostral spread of the infused opioid, infection, granuloma formation, and intraspinal bleeding with haematoma formation leading to cord compression. For these reasons, epidural and intrathecal local anaesthetic and fentanyl are only rarely employed in the treatment of chronic severe non-malignant pain.

This topic has been considered earlier in relation to the pathophysiology and treatment of complex regional pain syndrome (Section 17.5.9).

Acupuncture, transcutaneous electrical nerve stimulation or TENS, and other forms of peripheral counter-stimulation, together with a variety of central nervous system stimulation techniques, are relatively poorly validated treatments for pain. The problems of adequate blinding of these modalities in clinical trials are obvious. However, neurophysiological mechanisms of analgesia have been established for all these treatments, and there is a limited evidence base for efficacy in patients with both nociceptive and neuropathic pains. The potential value of electrical stimulation and acupuncture has been rcognized since ancient times (Kane and Taub 1975; Lu and Needham 1980). The gate control theory (Melzack and Wall 1965), which proposed a neuromodulating effect of peripheral large fibre activity on the forward transmission of noxious inputs, in the dorsal horn of the spinal cord, reawakened interest in the scientific investigation of counter-stimulation as a means of pain control.

Transcutaneous electrical nerve stimulation, TENS. Physiological evidence indicates that the non-painful stimulation used therapeutically in TENS inhibits nociceptive transmission in the spinal cord by both pre- and post-synaptic mechanisms (Sluka and Walsh 2003), and possibly via long-range thalamic inhibition (Olausson et al. 2002). Recruitment of these inhibitions is dependent on stimulus frequency and intensity. High intensity, high frequency TENS activates A delta and C fibres, and is itself painful. It appears to be more effective in activating spinal meachanisms. Analgesia induced by this form of stimulation is probably mediated by endogenous opioids, and there is some evidence that low frequency TENS also exerts its effect partly through a similar mechanism (Han 2003). It is interesting to note that low intensity TENS, which selectively stimulates A beta fibres, can relieve pain associated with brush-evoked pain, allodynia, that is itself mediated by A beta fibres. However, in such patients, TENS usually cannot be tolerated within the area of allodynia but can be effective when applied to adjacent areas.

There is a body of evidence indicating efficacy of TENS in a variety of nociceptive pains, notwithstanding the methodological difficulties of clinical trials (Barlas and Lundeberg 2005). In patients with peripheral neuropathic pain, there are several investigations that strongly indicate a therapeutic action (Hansson and Lundeberg 1999). In central neuropathic pain, the evidence is weaker (Davis and Lentini 1975; Leijon and Boivie 1989).

From a practical point of view, a trial of TENS is worthwhile in many patients with neuropathic pain, particularly when the pain is relatively localized, which is more likely when the cause is related to peripheral nerve, plexus, or root pathology. The most common adverse effect is skin reaction to the electrodes, usually mild, but occasionally treatment-limiting. TENS should not be used in patients with cardiac pacemakers, and caution is advised in pregnancy, particularly during the first trimester.

An important consideration in the use of TENS is that it provides patients with active personal involvement and control in the treatment of their pain in a way that most other treatments do not. This helps to counter feelings of helplessness that are so frequently a feature of chronic neuropathic pain.

Acupuncture. This activates A delta and C fibres and is usually perceived as painful by patients. There is strong evidence that acupuncture analgesia is mediated via endogenous opioid mechanisms. Acupuncture is associated with an increase in CSF endorphin concentrations and can be blocked or reversed with naloxone (Pomeranz and Chiu 1976; Han and Terenius 1982). There have been few well-controlled clinical trials of acupuncture. Sham acupuncture is difficult but not impossible. Acupuncture may have a place in the treatment of tension type headache (Melchart et al. 2003). A Cochrane systematic review of acupuncture for the treatment of non-specific low back pain concluded that there was insufficient evidence of benefit (van Tulder et al. 1999). Acupuncture has been shown to have some effect in patients with painful diabetic neuropathy (Abuaisha et al. 1998), but there is otherwise, as yet, no evidence base to support its use in neuropathic pain.

Vibration. This activates superficial and deep mechanoreceptors and the primary endings of muscle spindles connected to large fibre afferents (Eklund and Hagbarth 1965; Vallbo and Hagbarth 1968). The best evidence of therapeutic efficacy of vibration derives from studies in nociceptive pain, particularly musculoskeletal (Hansson and Lundeberg 1999). For neuropathic pain, in a single randomized placebo-controlled trial, vibration was reported to relieve stump and phantom limb pain (Lundeberg 1985). Other counterstimulation methods, including hot and cold packs and massage are sometimes found to be helpful by patients with neuropathic pain, particularly cold packs in postherpetic neuralgia. However, none of these modalities has been subjected to rigorous assessment in clinical trials.

Peripheral nerve stimulation. This has been advocated for the treatment of neuropathic pain due to lesions of single peripheral nerves. It is associated with technical difficulties and is little used in clinical practice. Peripheral nerve stimulation can be effective in neuropathic pain resulting from mononeuropathies of limb nerves (Waisbrod et al. 1985) and for occipital neuralgia (Weiner and Reed 1999) (Section 19.2.6).

Spinal cord stimulation. This occurs on the dorsal aspect of the cord and produces both anterograde ascending impulse activity and retrograde descending activity in the dorsal columns. Both may be important in inducing analgesia (Simpson et al. 2005). In experimental studies, transection of the spinal cord rostral to the site of stimulation abolishes its effect (Roberts and Rees 1994). Descending impulses suppress activity in dorsal horn wide dynamic range neurons below the site of stimulation (Linderoth and Foreman 1999). The effectiveness of spinal cord stimulation is dependent upon GABA and possibly other neuromodulating substances, including 5-hydroxytryptamine, substance P, adenosine, and glycine (Linderoth and Foreman 1999).

There is evidence indicating an effect of spinal cord stimulation in symptom relief and restoration of the microcirculation in chronic limb ischaemia (Ubbink et al. 1999). It is now generally agreed that it has a role in the treatment of neuropathic pain, particularly the failed back surgery syndrome, in which there is usually a mixture of nociceptive and neuropathic pain components (North et al. 1991; Gybels et al. 1998). A recent study indicates that spinal cord stimulation is preferable to re-operation in failed back surgery syndrome (North et al. 2005). This has become the major indication for spinal cord stimulation, but robust evidence of efficacy is limited. A systematic review reached the conclusion that unequivocal evidence of efficacy was lacking, and was critical of the methods of many studies (Turner et al. 1995).

In clinical practice, spinal cord stimulation should only be considered for patients with failed back surgery syndrome and other intractable radicular pains, including postherpetic neuralgia, and stump and phantom limb pain, when all other measures have failed. There is a high incidence of technical problems with spinal cord stimulation, and loss of an initial analgesic effect after weeks or months is common, either for technical reasons or from presumed physiological adaptation.

Brain stimulation. Despite reports of pain relief from brain stimulation since the 1960s (Mazars et al. 1979), the development of therapeutic brain stimulation has been hampered by a paucity of well-controlled studies and by observer bias resulting from non-independent patient evaluations in many trials (see Simpson et al. 2005). Three target regions have attracted greatest interest: the sensory thalamus, the periaqueductal/periventricular grey, and most recently, the motor cortex.

Sensory thalamic stimulation. The mechanism of analgesia produced by stimulation of the main thalamic sensory nuclei remains uncertain, though experimental observations point to a type of supraspinal gating. For example, it has been shown that stimulation in the ventrobasal complex in monkeys reduces responses of dorsal horn neurones to noxious peripheral stimuli (Gerhart et al. 1983); and in a rat model of sciatic nerve injury, thalamic stimulation reduces hypersensitivity of the affected hind-paw (Kupers and Gybels 1993).

Thalamic sensory nucleus stimulation has been tried for many intractable pains, both nociceptive and neuropathic. A meta- analysis, based on reports of large series, indicates that neuropathic pain responds much better than nociceptive pain (Bendok and Levy 1998; and see also Hosobuchi 1986). However, central post-stroke pain usually does not respond. Suggested indications include failed back surgery syndrome when there is a major neuropathic component, stump and phantom pain, and facial anaesthesia dolorosa resulting most often from surgical deafferentation procedures performed for trigeminal neuralgia (Section 19.2.1). However, evidence of efficacy is weak. Although there are case reports of long-term pain relief, loss of effect in many patients after weeks or months is widely recognized.

Periaqueductal/perivantricular grey stimulation. Following the demonstration that periaqueductal grey stimulation and periventricular grey stimulation in the upper midbrain and medial thalamus exerts a powerful analgesic effect in experimental animals (Reynolds 1969), it was later found to be an endogenous opioid-mediated effect, associated with raised CSF concentrations of beta-endorphin, and reversible with naloxone (Richardson 1995). Such stimulation is more effective in nociceptive than in neuropathic pain. Periventricular stimulation is preferred to periaqueductal grey stimulation because of unwanted effects of the latter including dysphoria and diplopia (Hosobuchi 1986; Kumar et al. 1997).

Motor cortex stimulation. This has been reported to be effective for central post-stroke pain and trigeminal anaesthesia dolorosa (Nguyen et al. 2003). The mechanism is uncertain, but an inhibitory action in several areas of importance in pain perception seems likely. This is based on the finding in positron emission tomography studies that motor cortex stimulation leads to an increase in blood flow in the ipsilateral ventral lateral thalamus, cingulate gyrus, insula, and brainstem. Stimulation is effective at a threshold below that of motor activation, and tolerance does not develop (Brown 2004). The limited evidence available at the moment does not permit firm conclusions to be drawn about the place of motor cortex stimulation in the treatment of intractable pain.

The major comorbidities of chronic pain have already been discussed. Those distressed and disabled by their chronic pain, particularly when it fails to respond to standard treatments, are those most likely to benefit from psyschological interventions. However, it is important to emphasize that one should not wait until all other treatment options have been exhausted before considering psychological therapy. The latter can be invaluable either in combination with other treatment modalities or on its own.

A frequent sequence of events is that pain with limited responsiveness to treatment is accompanied by reduced physical activity, associated with fatigue, poor sleep, social and family isolation, depression, anger, frustration, and a fear of making the pain worse, particularly through physical exertion ‘catastrophizing’, and an increasing dependence on medical services. It is important to recognize this symptom constellation at an early stage, before it becomes entrenched, and consider psychological intervention. The aims of treatment must be tailored to the individual patient, but are likely to include improved physical activity and fitness, reduction in fear and catastrophizing, improved adaptive and coping behaviour, relief of depressive symptoms, and return to work.

Many psychological techniques have been employed, but the treatment of choice is cognitive behavioural therapy. A detailed description of the rationale and methods is beyond the scope of this chapter (see Eccleston et al. 2003). Sifting the published reports of cognitive behavioural therapy to assess evidence of efficacy presents a host of methodological difficulties (Eccleston et al. 2003). However, a meta-analysis of its use in chronic pain, without distinguishing the nociceptive, neuropathic, or mixed nature of the pain, yields some evidence of efficacy (Morley et al. 1999). Further studies are needed.

The value of decompressive surgery of peripheral nerves, plexuses, spinal roots, and sensory cranial nerves for the treatment of neuropathic pain and associated deficits is clearly established. The indications for ablative neurosurgery, on the other hand, are very limited. Neuropathic pain, both peripheral and central, is caused by lesions of the somatosensory system; thus surgical procedures designed to interrupt some part of this system are themselves at risk of leading to the development of neuropathic pain.

Periphal neurectomy. Resection of painful peripheral nerve neuromas is inevitably followed by regrowth of the neuroma. In nerve injury, nerve repair and grafting can alleviate pain, but regenerating nerves partially innervating their original territory are often associated with allodynia and hyperalgesia, so that while such surgery may improve motor and sensory function, it may also lead to the development of these additional painful symptoms, transiently or permanently, depending on the eventual success of peripheral tissue reinnervation.

Painful neuromas are often sited in positions where they are subject to repeated minor physical trauma due to tethering and traction with limb movement, or because they are subject to pressure. In such circumstances, surgery to relocate neuromas can be helpful, though it is by no means always successful, as it is difficult to fashion appropriate environments for exquisitely mechanically sensitive neuromas in many anatomical situations. Nonetheless, in highly selected patients the results can be excellent (Patil and Campbell 2005). Resection of the affected plantar digital nerve is routinely performed for the treatment of Morton’s neuralgia (Section 22.10.2). The nerve lesion has neuromatous elements, but is aetiologically an entrapment neuropathy, often histologically severe (Scadding and Klenerman 1987). Resection leads to neuroma formation, so that even with resection far enough proximally, away from the area of most intense pressure on the nerve end, 33 per cent of patients continue to experience pain in the long term (Johnson et al. 1988).

Dorsal rhizotomy and ganglionectomy. There is large overlap of the territories of sensory spinal segmental innervation, and sensory afferents entering through sensory roots at one level may terminate over several spinal segments (White and Kjellberg 1973). Thus in order to denervate a painful area, dorsal rhizotomy at several levels may be needed. Rhizotomy interrupts large as well as small nociceptive afferents, leading to unwanted sensory loss, including proprioceptive, and in the sacral segments also to impaired bladder and bowel function. In addition, the deafferentation produced surgically may cause the later development of central neuropathic pain. There have been numerous reports of rhizotomy and ganglionectomy for a wide range of pains, both nociceptive and neuropathic, with mixed results (Patil and Campbell 2005). Given the unpredictable pain relief and complications, these procedures have been largely abandoned, though there may be a limited place in the treatment of cancer pain in patients with a short prognosis.

Dorsal root entry zone lesioning. Of all the ablative operations for the treatment of chronic pain, dorsal root entry zone lesioning, the Nashold procedure (Nashold and Ostdahl 1979), has the best evidence base for efficacy. The operation ablates the dorsal root entry zone, including the dorsal horn containing the cells on which the majority of small afferent fibres terminate, and the more deeply situated wide dynamic range neurons, abnormal activity of which is an important component of central sensitization, and likely to contribute to the development of neuropathic pain. In deafferentation states, typified by brachial plexus avulsion, bursting activity of dorsal horn neurons develops in the deafferented spinal segments (Loeser and Ward 1967). Ablation of these cells, either by surgical section or by radio-frequency heat lesioning relieves pain in many patients with brachial plexus avulsion. The extent of lesioning is not easy to control, and other spinal cord structures are sometimes affected, including long tracts. This may lead to pyramidal deficits and impaired bladder function (Friedman and Bullitt 1988).

Indications for dorsal root entry zone lesioning include neuropathic pain, unresponsive to all medical measures, due to severe brachial plexus lesions, particularly avulsion and malignant pain, for example due to Pancoast tumours. In spinal cord injury, lesioning may relieve segmental pain at the upper extent of the lesion, but not myelopathic pain below the level of the lesion. Dorsal root entry zone lesioning has been used with variable success for amputation pain and for postherpetic neuralgia, for which it appears to be most effective when paroxysmal pain, allodynia, and hyperalgesia are prominent features. The procedure has also been used for the treatment of pain associated with disabling hyperspastic states (Patil and Campbell 2005).

Anterolateral cordotomy. Interruption of the spinothalamic tract in the cervical cord leads to contralateral loss of pain and temperature sensation. The only indication for the procedure is intractable cancer pain, though with improved methods of pain control it is now performed infrequently. It is contraindicated for the treatment of chronic pain of non-malignant origin for two reasons. First, because of the limited duration of analgesia of not more than 2–3 years (Nathan 1963), indicating a remarkable plasticity of the central nervous system. And second, because of the development of central neuropathic pain due to the surgical lesion itself, after months or years.

Anterolateral cordotomy performed at open operation was superseded by percutaneous radio-frequency heat lesioning at C1 and 2 (Rosomoff et al. 1965), a more acceptable procedure in terminally ill patients. Complications include Horner’s syndrome, mild pyramidal deficit, ataxia, and paraesthesiae. Bladder, sexual, and respiratory problems are more likely with bilateral cordotomies. Lipton (1989) reviewed a series of 300 cordotomies; 75 per cent of patients had complete pain relief and a further 8 per cent had partial relief. Transient weakness was common, but persistent at 1 month post-procedure in only 2 per cent. Lahuerta et al. (1985) reported complete pain relief in 64 per cent of patients and partial relief in 23 per cent. Mortality was 6 per cent in this series, due to respiratory complications.

Midline myelotomy. The rationale of midline myelotomy in the treatment of chronic pain is interruption of decussating spinothalamic fibres as they cross the midline in the anterior white commissure of the cord to form the anterolateral spinothalamic tract. The operation is performed at about three segmental levels above the level of the pain. Lesions produce bilateral hypoalgesia just below the level of the myelotomy (Sourek 1977). Intended to be highly selective, the adverse effects of this procedure can include sensory loss other than of spinothalamic type, paraesthesiae, ataxia, weakness, and sphincter disturbance. The operation has been recommended for bilateral, centrally situated pain of abdominal or pelvic visceral origin, usually due to malignant disease. It is not now commonly performed, partly because of improved medical methods of pain control in palliative cancer care, and partly because of the associated morbidity. However, recent case reports and a review indicate that the procedure probably still has a limited place (Hwang et al. 2004).

Mesencephalotomy. The rationale for making surgical lesions of ascending pathways in the midbrain is that quintothalamic fibres from the face and projections from the lower brainstem reticular formation to the thalamus can be interrupted, together with spinothalamic tract fibres. Mesencephalotomy has been performed for unilateral or bilateral pain caused by cancer of the head and neck, though there are reports of its use for a wide variety of intractable pains, including brachial plexus avulsion and other neuropathic pains (Nashold et al. 1977). The treatment target is extremely small: the spinothalamic tract occupies an area of about 0.65 mm² and the medially placed quintothalamic tract is smaller. Stimulation prior to lesioning is thus advised (Nashold et al. 1977). Despite this, mesencephalotomy is frequently associated with problems. Lasting analgesia is achieved in only 30–50 per cent of patients, mortality ranges from 3 to 10 per cent and morbidity is as high as 37 per cent (Nashold et al. 1977). Complications include ocular palsies, nystagmus, disabling contralateral dysaesthesiae, and occasionally contralateral hemiparesis. The operation is more effective for nociceptive than neuropathic pains (Tasker 1990). It is now rarely performed.

Thalamotomy and other supratentorial targets. Numerous supratentorial targets for ablative procedures have been proposed over many years, but evidence of efficacy is weak and these operations are now rarely performed. As brain lesioning has declined, so neuroaugmentation by brain stimulation has increased.

The thalamic region in which ablative surgery is most likely to produce selective analgesia is the intralaminar group of nuclei. Lesions at this site minimize the risk of accompanying loss of tactile and proprioceptive sensation. In a large systematic review of medial thalamotomy, Tasker (1990) found an overall pain relief rate of nociceptive pains of up to 57 per cent, but with a 50 per cent recurrence rate of pain. For neuropathic pains, up to 67 per cent of patients obtained some degree of pain relief. Complications occurred in up to 20 per cent of patients and included confusion, dysphasia, other cognitive deficits, ocular palsies, and dysaesthesiae. Stimulation and physiological recordings undertaken in awake patients during these procedures have provided important psychophysical insights into the perception of chronic pain (Gybels and Tasker 1999 , 2005).

Other brain areas lesioned for intractable pain include the dorsomedian nucleus, which projects to the cingulum, frontal lobes, and limbic system, and the medial and lateral pulvinar nuclei. Both produce only transient analgesia. Lesions of the frontothalamic connections and the frontal lobes themselves produces analgesia, but at the expense of a change in personality, albeit mild (Hitchcock 1977).