Chapter 37. Polyneuropathy

Painful polyneuropathy is a debilitating neurologic problem and frequently a challenging therapeutic management issue. Difficulties in managing patients are too often the result of poor understanding of their problem on the part of the treating physician. Many physicians assume that there is no need to work up neuropathy because the final outcome is likely to be an idiopathic, axonal disorder for which there is no effective therapy. In fact, many neuropathies are responsive to immunosuppressive therapies. Although responses to such therapy constitute the minority, they should be vigorously sought before telling patients there is no treatment for their progressive disorder. In many cases, treatment of the polyneuropathy also leads to improved pain control; however, pain is often a primary issue in and of itself and must be treated irrespective of the potential for improvement of the underlying polyneuropathy. In these cases, pain management specialists may work in concert with neurologists to provide a comprehensive treatment approach.

Clinical Issues in Polyneuropathy: Introduction

The first step in developing a rational approach to patient management is obtaining a working knowledge of the underlying disorders that fall under the category of neuropathy. Although it is often used loosely to refer to polyneuropathy, the term neuropathy is actually not specific and implies any peripheral nerve lesion, focal or diffuse. Classification schemes used widely among peripheral neurologists are based on anatomic and physiologic characteristics of the various disorders affecting peripheral nerves. The use of these classifications is not just an academic exercise but creates a basis for rational decision making in the evaluation and management of patients. The workup and treatment of individual patients with neuropathy must be approached with a basic understanding of the clinical behavior, including the anatomic and pathophysiologic characteristics, of the various neuropathic disorders.

Classification and Clinical Course

The term polyneuropathy is used to describe a condition that is fairly symmetric and generalized, as opposed to focal neuropathy (mononeuropathy) or multifocal neuropathy (mononeuropathy multiplex). This chapter focuses on the diffuse disorders, including polyneuropathy and multifocal mononeuropathies. These two groups of disorders may be indistinguishable clinically and are frequently accompanied by severe and disabling pain.

Once a disorder of peripheral nerves is suspected, an attempt should be made to characterize the clinical features based on the time course, anatomic distribution, and physiology. Using this information, a reasonable differential diagnosis can be developed, which will determine appropriate further workup and management. This section further discusses the clinical and physiologic features of the diffuse neuropathies; their diagnostic evaluation is covered in the next section.

Polyneuropathy

Time Course

An important clue as to the etiology of a particular polyneuropathy (PN) is its time course. Generally accepted guidelines would classify a neuropathy as acute (< 3 weeks), subacute (weeks to months), or chronic (> 4–6 months). Notably, neuropathies in each of these categories may be associated with debilitating pain. The typical clinic patient presenting with chronic, insidious PN is probably the most easily diagnosed; the more acute neuropathies may be difficult to differentiate from central nervous system disease, particularly spinal cord compression.

Fiber Type and Distribution

PN may involve motor, sensory, or autonomic fibers. There is a tendency in some neuropathies for selective involvement of fibers of the same general size distribution. Thus, a neuropathy may involve predominantly large-diameter sensory fibers (mediating vibration and proprioception) in addition to intermediate-sized motor fibers. Conversely, PN may primarily involve small-diameter sensory fibers (mediating pain and temperature) with or without involvement of autonomic fibers. Typically, pain is a prominent feature of these so-called small-fiber neuropathies. Certainly, a neuropathy may be generalized in terms of the fiber type involvement (as commonly seen in diabetes); however, careful examination often reveals a predominance of one group of fibers over another.

Pathophysiology

A detailed description of the pathophysiologic mechanisms underlying PN in various disorders is beyond the scope of this chapter; instead, we focus on the relevant clinical-pathologic correlates as well as the electrodiagnostic characteristics of the different neuropathies, depending on the primary site of pathologic change.

The two primary sites of pathologic involvement in neuropathy are the axon and the myelin sheath, or Schwann cell; the former being more common in PN. In general, in axonopathies, the longest and larger diameter fibers tend to be involved first, with degeneration originating in distal portions of individual axons and proceeding proximally. This creates a length-dependent pattern, which can be demonstrated both clinically and electrophysiologically. This generalized “dying-back” is theorized to result from metabolic derangement in the cell bodies or diffusely within axons. Axonopathies tend to be quite symmetric in terms of side-to-side involvement, a feature that distinguishes them from the multiple mononeuropathies. These are the most common type of PN; almost all toxic and metabolic insults to the peripheral nervous system result in axonal degeneration.

Less frequently, the axon is largely spared and demyelination is the primary pathologic change. Although myelinopathies may be the result of abnormal Schwann cell development or metabolism, these situations are rare and the most frequent clinical situation is one in which segmental demyelination, or loss of myelin between the nodes of Ranvier, occurs. Segmental demyelination usually is the result of an autoimmune attack on peripheral nerves and nerve roots (as in Guillain-Barré syndrome), with the clinical pattern being somewhat variable; the limbs are involved proximally as well as distally, although usually fairly symmetrically side-to-side.

Etiology and Prognosis

Tables 37-1 and 37-2 list the most common causes of PN, based on their physiology (axonal versus demyelinating) and time course.

Although the clinical course of various neuropathies is highly variable, depending on etiology, there are a few generalizable rules regarding prognosis. Any disorder involving significant axonal injury will be less likely to recover than one in which the primary physiology is segmental demyelination. Thus, the differentiation between these two is of practical importance for the clinician. In a disorder characterized purely by the latter, recovery occurs by remyelination and usually occurs over 6 to 8 weeks. In generalized neuropathies, even of primary demyelinating type (e.g., Guillain-Barré syndrome), there is nearly always some accompanying axonal injury and, ultimately, prognosis is dependent on its severity. Thus, recovery may be complete but take months or even years.

In primary axonal neuropathies, the prognosis is dependent on the nature and severity of the axonal injury. For example, in typical mononeuropathy multiplex, in which nerve injury results from ischemic insult, recovery occurs largely through axonal regrowth. Individual regrowth of axons occurs from the proximal nerve stump, at a rate of about 1 inch per month. This form of recovery is very slow and nearly always incomplete. In neuropathies characterized by a dying-back type of physiologic change, there usually is very little regrowth of individual axons. Instead, functional recovery occurs through reinnervation of muscle fibers by nearby healthy axons, a mechanism that probably only limits the severity of the deficit related to axonal loss, rather than allowing any improvement in function.

Mononeuropathy Multiplex

Clinical Course

Mononeuropathy multiplex (MM) is a diffuse neuropathic disorder, similar to PN, but it is distinguished by involvement of multiple individual nerves. MM may be impossible to distinguish from PN based on history and examination alone, as the cumulative involvement of multiple nerves can produce a generalized and fairly symmetric picture. A high index of suspicion for MM is important in the appropriate clinical setting. MM is almost invariably the result of ischemic insult to nerves, the most common etiology being small- and medium-vessel vasculitis. MM occurs in the majority of cases of systemic vasculitis and may be its presenting symptom; in rare cases, vasculitis is restricted to the peripheral nervous system (nonsystemic vasculitic neuropathy).1

Regardless of etiology, MM typically presents with the acute onset of severe pain and numbness in the involved limb; motor and sensory deficit develop over days. Nerves that are often involved early are those at so-called watershed zones of the vascular tree (e.g., the sciatic nerve in the thigh, the ulnar nerve in the forearm). Progression to other nerves eventually produces a picture suggestive of severe, axonal PN; the progression may be subacute or, rarely, chronic. In all cases, pain remains a prominent feature of the disorder.

The prognosis is dependent on the underlying etiology of the MM. In vasculitic neuropathy, aggressive treatment of the underlying disease is aimed at preventing further ischemia. Recovery of existing lesions occurs by means of axonal regrowth, at a pace of about 1 inch per month from the site of the injury. An aggressive therapeutic approach to pain is appropriate, particularly early in the course of the disorder, when immunosuppressive therapy has not reached maximum effectiveness. Over time, if the vasculitis can be adequately treated, it may be possible to reduce or withdraw pain therapies.

Etiology

Table 37-3 lists the most common etiologies of MM, based on time course.

Diagnostic Evaluation

History and Physical Examination

History

The first step in the diagnostic evaluation of any neuropathy is the history and physical examination of the patient. Several important historical points should be reviewed. One should determine the time course of the illness—acute, subacute, or chronic. Next, involvement of nerve fiber types should be ascertained—sensory, motor, and autonomic, with regard to both positive and negative symptoms. Specific inquiry should be made regarding the presence or absence of pain. Positive symptoms refer to abnormal spontaneous sensory or motor phenomena (e.g., pins and needles, fasciculations), whereas negative symptoms describe a loss of function (e.g., weakness, numbness). Inquiry should be made regarding the symptom distribution and symmetry (i.e., stocking glove versus individual nerve territories), as well as the progression (i.e., slow and insidious versus acute-onset deficits with plateaus). Finally, one should obtain a family history, with an eye to excluding a congenital form of neuropathy.

Characteristic features of neuropathic pain are useful in differentiating it from pain from any other source. It is vital that these symptoms and signs are sought during the history and physical examination as they are often primary evidence for a diagnosis of PN. In some neuropathies, such as those limited to involvement of small fibers, normal laboratory studies are the rule, and diagnosis is based on clinical grounds alone. The presence of positive symptoms is typical of the neuropathic disorders.

Positive symptoms that are typical of PN include (1) paresthesias—nonpainful, spontaneous sensory phenomena such as pins and needles or tingling; (2) dysesthesias—unpleasant spontaneous or evoked sensory phenomena such as burning; (3) hyperesthesia—increased sensitivity to stimuli, often with an unpleasant quality; (4) allodynia—pain created by a normally nonpainful stimulus, such as the bedcovers; and (5) hyperpathia or hyperalgesia—exaggerated pain response created by a normally painful stimulus.

The presence of these symptoms should be sought specifically, in addition to allowing the patient to describe the precise nature of his or her pain. Also, the effect of pain on quality of life and functional status is extremely important. Specific pain measures, such as the Neuropathic Pain Scale,2 may be used to quantify the patient’s pain as well as its effect on the quality of life. Such scales are particularly helpful for patients involved in clinical therapeutic trials and may be used to assess efficacy of treatment regimens outside of experimental trials.

Physical Examination

The physical examination should be guided by the patient history. For example, the suggestion of asymmetric onset of symptoms should prompt a careful search for evidence of individual nerve involvement, as opposed to a stocking-glove distribution of sensory loss.

A complete neurologic examination is required. One cannot adequately localize the problem to the peripheral nerve without a careful physical examination to rule out myelopathy, polyradiculopathy, or myopathy, which may mimic or complicate PN. The details of the neurologic examination are not reviewed here, but a few points are worth emphasizing. The goal of the physical examination is to characterize the pattern, symmetry, and distribution of abnormalities and to determine which modalities are involved (motor, sensory, autonomic); the distribution with regard to fiber type should also be demonstrable. The typical pattern to look for is bilaterally symmetric, usually distally predominant. The proximal lower extremities tend to be involved before the distal upper extremities, although this is variable; the anterior thoracic region is often involved in more severe cases. However, there may be proximal predominance, and upper extremities may be involved disproportionately. In MM, the pattern is usually multiple nerve involvement, although it may likely be impossible to differentiate from PN, other than subtle asymmetry in an otherwise stocking-glove distribution. The deep tendon reflexes are part of the overall pattern, as well, and are typically reduced or absent in a distribution consistent with the underlying pathophysiology. For example, patients with distal axonopathies typically have absent ankle jerks, whereas those with chronic demyelinating neuropathies are areflexic.

In patients with painful positive symptoms, there often are correlative signs on the physical examination. Allodynia may be elicited by lightly stroking the involved area (mechanical stimulus) or by testing with a cold instrument (thermal stimulus). Hyperalgesia or hyperpathia may be elicited during pinprick testing. A single, painful stimulus may be reported as a sensory deficit, whereas repeated stimuli in the same area produce an exaggerated pain; this phenomenon is called summation.3 Abnormal sensations may last for several seconds or minutes after discontinuation of the stimulus, a phenomenon called after sensations.3 These examination findings are important, because they are unique to patients with neuropathic pain.

Clinical Neurophysiology

Electrophysiologic Studies

The second logical step in the diagnostic evaluation is the electrodiagnostic examination, specifically, nerve conduction studies (NCS) and electromyography (EMG). The utility of these studies is several. Firstly, they usually clarify the diagnosis of PN. Although important, there is a limit to the localizing value of physical examination, even when carefully performed. For example, a detailed physical examination cannot differentiate multiple root involvement from PN or MM in most cases; coexisting neurologic problems may also significantly alter the physical examination. It is important to note that NCS cannot assess the integrity of small-diameter sensory fibers (i.e., those mediating pain and temperature) and so will be normal in patients with pure small-fiber neuropathies. However, these neuropathies are rare, and in those patients with involvement of larger fibers, NCS are more sensitive than physical examination for diagnosing PN.

In addition to establishing a diagnosis with accuracy, electrodiagnostic studies provide several other types of information—the predominant pathophysiology (i.e., axonal or demyelinating), time course and severity of the disorder, and whether motor or sensory fibers, or both, are involved. NCS can differentiate hereditary from acquired forms of PN and is much more sensitive than physical examination for identifying MM. Chapter 9 discusses the use of electrodiagnostic testing in detail. Once the underlying pathophysiology and the time course are understood, the differential diagnostic possibilities are narrowed considerably.

Quantitative Sensory Testing

Quantitative sensory testing (QST) is a specialized technique for measuring the intensity of a given stimulus required to elicit specific sensory perceptions.4 Specifically, QST assesses a sensory detection threshold to various stimuli, including touch-pressure, vibration, heat, and coolness. QST, when properly performed, provides a quantitative, noninvasive means of assessing sensory function. One major advantage is that QST for thermal thresholds allows some quantifiable measure of small-fiber function, which is not possible with routine NCS. Several commercial systems exist for QST. However, a major problem with QST is that it is not widely available for clinical use, and its reliability is highly operator dependent. Notably, its sensitivity in patients with pure small-fiber neuropathies, the group in whom it is potentially of the greatest diagnostic importance, has been reported in several studies to be around 60%.5,6 Nonetheless, it can be quite helpful in confirming the physical examination findings and substantiating the clinical suspicion of neuropathy. In small-fiber neuropathies, in particular, QST is recommended, as it may provide the only objective means for establishing a diagnosis and can be used to measure efficacy of various therapeutic modalities.4 QST is typically pursued in the evaluation of PN if the routine NCS are nondiagnostic.

Laboratory Studies

The differential diagnostic considerations in any given neuropathy are dependent on the physiologic characteristics and time course of the underlying disorder (see Tables 37-1, 37-2, and 37-3). Thus, the diagnostic evaluation should proceed initially with electrophysiologic studies, as discussed earlier, and laboratory workup should be guided by the differential diagnostic considerations raised by these findings. The laboratory workup will nearly always include blood studies and occasionally urine studies. Cerebrospinal fluid evaluation is no longer routinely necessary; it is most commonly obtained in fairly acute neuropathies in which Guillain-Barré is suspected.

Table 37-4 outlines the appropriate laboratory workup for PN and MM based on time course and electrophysiology.

Biopsy

Nerve and Muscle

Nerve biopsy is required in a small proportion of patients with neuropathy, and should be performed only in situations in which the indication is clearly defined. Only rarely is biopsy necessary to establish a diagnosis of PN. Nerve biopsy is most useful in patients with MM, as a means of determining the causative disorder, which is usually inflammatory in nature and has a high morbidity and mortality if untreated. Other disorders appropriately diagnosed by nerve biopsy include sarcoidosis, amyloidosis, and, rarely, leprosy. In the cases of vasculitis and sarcoidosis, a muscle biopsy is usually obtained simultaneously. Nerve biopsy is infrequently performed in cases of progressive PN in which exhaustive workup has failed to reveal an underlying diagnosis. In these cases, the aim of the biopsy is to determine the presence or absence of a potentially treatable disorder.

Skin

In recent years, a technique has been developed for quantitative assessment of cutaneous innervation in punch biopsies of the skin. The epidermis contains free nerve endings, which are the terminals of small-caliber, unmyelinated fibers. Using control values obtained from a healthy cohort, investigators have been able to identify abnormal patterns of intraepidermal nerve fiber (IENF) density in patients with small-fiber sensory neuropathies (SFSNs).5,7 Examination of patients with idiopathic, human immunodeficiency virus (HIV)—associated, and diabetic painful sensory neuropathies indicates a correlation between IENF density and clinical estimates of small-fiber sensory dysfunction.7 Studies of patients with idiopathic SFSNs suggest that IENF is a more sensitive diagnostic indicator of pure, small-fiber neuropathies than either QST or sural nerve biopsy.5 Although IENF is still available only in specialized centers, it holds promise as an important tool for routine use in patients with painful sensory neuropathies, whose routine workup may be entirely normal, especially when small-fiber involvement predominates.

Overview

The painful neuropathies are clinically heterogeneous. Even if one restricts analysis to the diffuse neuropathies, the frequency with which pain will occur in a given setting is impossible to predict based on etiology or physiology. Pain may be present early or late in the clinical course of a given neuropathy. It is important for the clinician to evaluate pain issues in an individual patient with knowledge of the underlying disorder and its clinical course, understanding that the course of the pain may be independent of the disease. To that end, it is worth reviewing the neuropathies most commonly associated with pain.

The most commonly encountered painful PN syndromes occur in the setting of diabetes, HIV, and acquired immunodeficiency syndrome (AIDS) and toxic-metabolic disorders. Aside from these, Guillain-Barré syndrome is probably seen most frequently. These disorders are discussed individually in the following sections. Table 37-5 lists diffuse neuropathies commonly associated with pain as a prominent feature. MM syndromes, although significantly less frequent, are worth noting as they are almost invariably associated with pain. MM is usually the result of ischemic vascular injury, regardless of the underlying etiology (see Table 37-3). Pure SFSNs are classically painful, as well, although they are a rare form of neuropathy.

Diabetic Neuropathies

The diabetic neuropathies are clinically diverse and involve focal, multifocal, and diffuse disorders of varying types. Pain, although not universal, is a characteristic feature of neuropathy in diabetic patients, and the most common cause of painful PN is diabetes. It is worth noting that the majority of diabetic patients suffer from chronic, distal, predominantly sensory PN with involvement of all fiber types; this syndrome is frequently painless for much of its course. However, small sensory fibers may be involved disproportionately or exclusively in diabetes mellitus and, in these syndromes, pain is invariably a feature. Painful diabetic neuropathies include several clinical syndromes, including PN. The acute syndromes include lumbar radiculoplexopathy (i.e., diabetic amyotrophy), acute thoracic radiculopathy, and acute distal sensory PN. Rarely, patients may develop an acute, distal sensory PN shortly after initiation of insulin therapy, this has been termed insulin neuritis.

The PN syndromes encountered in diabetics include acute-subacute and chronic forms. Typical, chronic, axonal PN may be associated with pain at any point during its course, usually in proportion to the severity of the overall neuropathy; however, the most important factor appears to be the severity of small-fiber injury, and this may be disproportionate in an otherwise typical axonal PN. Diabetes is the most common etiology underlying pure SFSN, in which small-diameter sensory fibers are almost exclusively involved. These patients typically present with distal lower extremity burning pain as the chief complaint. Clinically, pain and temperature sensation are reduced, with sparing of large-fiber sensory modalities (vibration, proprioception); reflexes are usually intact. Autonomic dysfunction may be prominent in these patients. Nerve biopsies demonstrate a predominant loss of small fibers.8 The clinical course in these patients is variable, and pain may be a chronic problem.

Acute diabetic PN, although less common, is much more likely to be associated with neuropathic pain. The SFSNs associated with diabetes may be acute; in some of these cases, there may be a mixed pattern of involvement, with large fibers being less affected than small. In rare cases, the onset follows the initiation of insulin therapy (so-called insulin neuritis). A similar picture may be seen following precipitous weight loss in diabetic patients (so-called diabetic cachexia). The relationship to glycemic control is unclear; a similar syndrome has been reported following episodes of ketoacidosis as well as establishment of tight control. These cases appear to be largely self-limited, with slow resolution of symptoms as glycemic control and normal weight are established and appropriately maintained.8

HIV/AIDS-Related Neuropathies

The neuropathies associated with HIV manifest a wide spectrum of disorders, both clinically and pathophysiologically. The most common HIV-associated neuropathy in which pain is a characteristic feature is distal, primarily sensory, symmetric polyneuropathy (DSPN). Other painful polyneuropathies include those produced by toxins (particularly antiretroviral agents) and Guillain-Barré syndrome. Painful MM occurs in this population but is relatively rare.

DSPN is the most common HIV-related form of neuropathy. A predominantly sensory neuropathy, it has been estimated to affect 10% to 30% of patients with AIDS.9,10 Although DSPN is a relatively uncommon entity in early HIV infection, electrophysiologic studies have shown that up to one third of patients with AIDS have DSPN.11 As immunologic status worsens, clinical manifestations of DSPN increase in incidence. Clinical features include symmetric numbness; burning paresthesias and dysesthesias in the distal lower extremities, with decreased pinprick, temperature, and vibratory sense in a stocking distribution; and depressed ankle reflexes.

Several of the antiretroviral medications have dose-dependent peripheral nerve toxicities, resulting in a clinical syndrome identical to HIV-associated DSPN. These include the nucleoside analogues didanosine (ddI), zalcitabine (ddC) and the pyrimidine analogue, stavudine (d4T). Didanosine is a now well-known cause of a painful peripheral neuropathy. Early clinical trials found 8 of 37 patients receiving ddI developed a dose-related painful neuropathic syndrome, which resolved within 8 weeks of the removal of the drug.12 Likewise, ddC and d4T have been clearly shown to cause dose-dependent neurotoxicity, which improves with drug withdrawal.13–15 Notably, when the offending agent is removed, intensified neuropathic symptoms may persist for several weeks, a phenomenon known as coasting.13 DSPN is the most important side effect limiting the use of thalidomide in the treatment of painful aphthous ulcers in patients with AIDS.16 PN is also a well-described complication of therapy with isoniazid, an antituberculous agent used in patients with AIDS.17 Chemotherapeutic agents such as vincristine and paclitaxel, used in the treatment of Kaposi’s sarcoma and lymphoma, also have been associated with DSPN.

Rarely, in early HIV disease, patients may develop MM involving cranial as well as peripheral nerves, which responds either spontaneously or with immunomodulating therapy.18 In late HIV disease, a more fulminant, progressive MM may develop. In both cases, pain is a prominent feature. Etiologic agents identified include toxoplasmosis, herpes zoster, cryptococcus,19 cryoglobulinemia,20 and lymphoma. However, the most prevalent etiologic agent is cytomegalovirus; in these cases, marked improvement in symptoms is expected with ganciclovir or foscarnet therapy.21 MM in HIV patients also has been associated with vasculitis; in these patients, pain is a presenting symptom and may take months to abate.22

Diffuse infiltrative lymphocytosis syndrome (DILS) has been recognized as a rare complication of HIV disease for more than a decade; only recently has it been found to be associated with a characteristic peripheral neuropathy. DILS-associated neuropathy is a painful, symmetric, axonal sensorimotor PN of acute-subacute onset.23 Immunosuppressive and antiviral therapy are helpful in improving symptoms of the neuropathy.24

Otherwise-typical Guillain-Barré syndrome (see later discussion) may occur in patients with early HIV, particularly at the time of seroconversion.25

Guillain-Barré Syndrome

Acute inflammatory demyelinating polyradiculoneuropathy, or Guillain-Barré syndrome, is the most common cause of acute-subacute PN. Guillain-Barré syndrome is an immune-mediated PN, which is characterized physiologically by segmental demyelination. The typical clinical picture develops over days to weeks, with limb paresthesias developing in a distal-to-proximal fashion, accompanied by weakness and loss of tendon reflexes in a similar distribution. Weakness may spread to involve craniobulbar and respiratory muscles. A classic laboratory finding is a cerebrospinal fluid protein with relatively few white cells (albuminocytologic dissociation). Electrophysiologic studies reveal evidence of primary demyelination with conduction block. Spontaneous recovery, in variable degrees, occurs over weeks to months. Death from respiratory complications occurs in a small number of patients. Pain has been a relatively underappreciated aspect of the disease but is actually a prominent feature in the majority of cases, as pointed out by Ropper and colleagues.26

Patients with Guillain-Barré syndrome describe deep, aching muscular pain involving large muscles of the thighs, buttocks, and back, and, less often, sciatica or painful distal limb paresthesias. Pain is usually worst at night and interferes with sleep. Some patients may respond to simple analgesic agents. There is no published experienced regarding the use of tramadol, a non-narcotic analgesic, in Guillain-Barré syndrome, but its use should be considered before resorting to narcotic analgesic agents. The pain, even dysesthetic limb pain, generally responds poorly to antidepressants and anticonvulsants, therapeutic agents that are considered first-line therapy in chronic PN.26 The experience to date indicates that narcotics have been the most effective treatment and should be used appropriately to control pain, particularly when it is interfering with sleep. In patients whose respiratory status is so marginal that there is a concern of decompensation with the use of narcotics, intubation should be considered.26

Toxic Neuropathies

The largest number of toxic neuropathies is related to the use of pharmaceutical agents in appropriate clinical situations. Environmental and occupational toxins are much less frequently seen. Table 37-6 lists pharmaceutical agents that are commonly associated with neuropathy. The typical clinical scenario is a distal axonopathy that develops after sustained use of the medication, although the time course to onset of symptoms is highly variable. Sensory and motor symptoms and signs develop in a length-dependent fashion, in a time course ranging from weeks to months, or even years, depending on the dose and neurotoxic properties of the drug. Painful paresthesias and dysesthesias are common complaints. Prognosis is variable, but in most cases, significant resolution of clinical signs and symptoms occurs with drug withdrawal; however, recovery is variable and related to the severity of the underlying axonal injury.

Proposed Mechanisms of Pain in Polyneuropathy

Peripheral Mechanisms

Anatomic and Physiologic Considerations

The perception of pain is the result of a complex interplay among cells at multiple levels of the nervous system. When an insult results in damage to a peripheral nerve, the transmission and sensation of pain involves the axon, the dorsal root ganglion, and the central nervous system, as well. In this section, we review theories concerned with the role of the peripheral nervous system in the generation and propagation of pain. These include a number of possible pathophysiologic mechanisms, such as ectopic generation, afferent sensitization and hyperexcitability, amplification of neural pacemakers, and electrical cross-talk. We also focus on the pathophysiology of neuronal damage in common illnesses associated with peripheral neuropathies, including diabetes mellitus and HIV infection.

Neuroma Formation and Ectopic Generation

When an axon is injured and disconnected from its cell body, it undergoes wallerian degeneration, or so-called dying-back. Subsequently, nerve regeneration occurs through axonal sprouting and elongation from the proximal terminal. If there is some obstacle to forward progress, the axonal sprouts grow in a disorganized fashion and may form a neuroma, a mass of tangled neuronal tissue.27 Several studies have found the neuroma to function as the nerve’s sensory terminal,28,29 as well a being a source of ectopic discharge in both myelinated and unmyelinated axons.27 Thus, axonal injury may result in positive symptoms (e.g., paresthesias), as well as negative symptoms (e.g., numbness).

The function of a sensory axon is to propagate information from the sensory receptors to the central nervous system. In normal nerve, a temporary noxious stimulus results in transient pain, because the axon ceases its firing once the stimulus is removed.30 The development of ongoing pain and paresthesias following the removal of an external stimulus implies a fundamental change in the properties of the axon. Instead of functioning purely as an impulse conductor, it has become an impulse generator.27 It has been proposed that this metamorphosis following nerve injury results from the formation of a neuroma that has spontaneous pacemaker activity. The neuroma can depolarize and fire continuously in response to various different types of stimulation and may depolarize in the absence of a stimulus; this phenomenon is termed ectopic generation.27 An electrically active neuroma is demonstrated in Figure 37-1.

The development of electrogenetic potential in nerve fibers is one of the key changes that underlie the generation of neuropathic pain. Once an ectopic generator has been established, various stimuli affecting membrane excitability can induce repeated nerve depolarization. Examples of such stimuli include ischemia, mechanical stimuli, and neuroactive substances, including histamines, prostaglandins, and catecholamines.31

Unlike peripheral afferents, which develop pacemaker capability as a result of injury, there are a small number of cells in the dorsal root ganglia (DRG) that normally discharge spontaneously. This intrinsic rhythmogenicity is amplified by chronic nerve damage.32 It has been postulated that enhanced pacemaker activity in the DRG, as well as changes in central pathways, might explain why peripheral nerve blocks may fail to relieve sensory symptoms associated with peripheral nerve lesions.27Figure 37-2 demonstrates a prolonged afterdischarge of an afferent axon following mechanical stimulation of the DRG.

Sensitization and Hyperexcitability

Several studies have shown that, following injury, intact nerve fiber endings adjacent to injured nerve fibers start to grow and exhibit collateral sprouting. It appears that nociceptive afferent fibers proliferate preferentially. Thus, following nerve injury, an area of denervation may become surrounded by an area that is partially reinnervated but has increased sensitivity to noxious stimuli.33

Other changes that take place as a result of nerve injury include axonal membrane remodeling, with an accumulation of voltage-sensitive sodium channels in the end bulb of the axon.34 These channels are normally in a constant state of turnover, and there is continuous transport of sodium channels downstream.35 However, if the axon is disrupted, the sodium channels accumulate in the end bulb and render the cell membrane hyperexcitable. Even minor injury, resulting in the loss of myelin without any damage to the axon, triggers the deposition of sodium channels in areas of the axon where they are not usually present.36

Amplification

Initiation of an action potential by an undamaged neuron is an all-or-none phenomenon, which occurs when a threshold potential is reached. However, the attainment of a certain threshold potential by a damaged neuron can result in amplification, or repetitive firing. For example, ectopic neuronal pacemakers that are depolarized to their threshold potentials will fire repetitively, despite a single stimulus. Further increase in intensity of a stimulus can cause a linear, proportional increase in the rate of firing. As ectopic neuronal pacemakers and DRG pacemaker cells have resting membrane potentials near the repetitive-firing threshold potential, even a weak stimulus may initiate repetitive firing, resulting in symptoms of ongoing pain or parasthesiae.37

Electrical Cross-Talk

Experiments performed in the 1940s by Granit and Skoglund38 suggest that if the glial insulation between adjacent axons is disrupted, current may flow from severed axons to neighboring fibers. This electrical cross-talk initially subsides following acute injury but may reappear during the process of nerve regeneration and neuroma formation. Because different fiber types lie adjacent to one another (e.g., large, myelinated IA afferent fibers next to nociceptive C fibers), a low-threshold large fiber may be stimulated distally but, because of electrical cross-talk at the site of nerve damage, current might be transferred to a proximal nociceptive afferent.39

Central Mechanisms

Individuals with pain syndromes associated with both peripheral and central nervous system lesions may have altered firing in thalamic neurons, changes in the somatotopic organization of the ventrobasal complex of the thalamus, and altered responses to electrical stimulation of the central nervous system.40 Thus, both the peripheral and central nervous systems play an important pathophysiologic role with respect to pain in peripheral nerve disorders.

Mechanisms of Nerve Injury in Specific Disorders

The two disorders most commonly associated with painful PN are diabetes and HIV/AIDS. Although mechanisms of pain in these specific disorders are not well studied, the pathophysiology of nerve injury in these diseases may shed light on the underlying process that ultimately leads to some of the peripheral mechanisms discussed in the previous section.

Diabetes

The diabetic neuropathies are clinically heterogeneous. Pain related to diabetic neuropathies occurs in several clinical syndromes, including DSPN (predominantly involving small fibers), diabetic amyotrophy, truncal radiculopathy, and mononeuritis multiplex.

Several theories have been proposed to explain the pathophysiology of diabetic neuropathies; the most widely accepted are those invoking either metabolic derangement or microangiopathy as primary mechanisms. In spite of intensive study, it remains unclear how metabolic and vascular abnormalities affect the excitability of nerve cell membranes to result in the clinical manifestations of diabetic PN, including pain.

Metabolic Theories

It is generally agreed that chronic hyperglycemia plays a key role in complications of diabetes, including neuropathy. In 1993, the Diabetes Control and Complications Trial (DCCT) provided evidence that intensive glycemic control delayed the onset and slowed the progression of the long-term complications of insulin-dependent diabetes mellitus, including neuropathy.41,42 Additionally, several studies have shown improvement of diabetic PN following pancreatic or pancreatic-renal transplants.43–45 Two proposed mechanisms for the induction of neuropathy by hyperglycemia are (1) the aldose reductase-sorbitol model, and (2) the glycation model.

In the first model, investigators note that, in the setting of hyperglycemia, glucose is converted to sorbitol by aldose reductase within the nerve. Sorbitol competitively inhibits myoinositol uptake into nerve cells; theoretically, reduced uptake of myoinositol into nerve results in decreased Na+/K+-ATPase activity at the nodes of Ranvier. Ultimately, sodium accumulates within the axon, leading to loss of sodium channels and potassium current leak.46 Loss of sodium channels is, in turn, associated with disruption of the junctional complexes between myelin and axons in the paranodal region. Subsequent paranodal demyelination and axonal shrinkage are the predecessors to clinical neuropathy.

Although attractive, this hypothesis has been challenged based on incomplete metabolic, electrophysiologic, and morphometric evidence.47,48 Additionally, aldose reductase inhibitors and supplementation with myoinositol have failed to demonstrate therapeutic efficacy in clinical trials of patients with diabetic neuropathy.49

According to the glycation model, the presence of excessive glucose in the extracellular matrix results in the formation of so-called advanced glycation end products (AGE). Theoretically, these compounds interfere with several cell functions; DNA and nuclear proteins also may be modified by AGE.48 In animal studies, diabetic complications have been reduced by the administration of aminoguanidine, an inhibitor of AGE formation.50,51

Vascular Theory

Another putative mechanism for the development of painful diabetic neuropathy is through microangiopathic changes in the vasi nervorum. Theoretically, nerve injury results directly from ischemia, or from ischemia and inflammation associated with immune-mediated vasculitis. Various studies have demonstrated pathologic evidence in nerve biopsy samples for both inflammation and immune complex deposition in vessel walls in patients with various painful forms of diabetic neuropathy.52,53

In addition, significant structural changes are seen in blood vessels of patients with diabetes, including basement membrane thickening and deposition of cellular debris. A significant increase in vascular mural area resulting from to deposition of cellular debris is described in diabetes, even in the absence of clinical neuropathy.54 These microvascular changes appear to precede the development of clinical neuropathy and their severity correlates with duration of diabetes and with the severity of the neuropathy.55 These pathologic findings, although well described, are of unclear significance in the development of neuropathy in diabetes.

Nitric Oxide

Endoneurial blood flow has been shown to be reduced in experimental animal models of diabetes.56 The pathogenesis of this is unclear; however, it appears that the endothelium-derived relaxing factor nitric oxide (NO) may be involved. Depletion of NO or reduced smooth muscle sensitivity to NO is thought to decrease endothelium-dependent relaxation of smooth muscle, a function that has been found to be reduced in diabetic animal models and humans.56–58 Alternatively, NO may act indirectly on endoneurial blood by modulating sympathetic tone; depletion of NO may result in nerve ischemia as the result of increased vasoconstriction.59–61

The metabolic and vascular mechanisms proposed as underlying diabetic neuropathy may converge with the NO theory.62 Studies have demonstrated a metabolic competition for NADPH by aldose reductase, the enzyme that converts glucose to sorbitol, and NO synthetase, the enzyme that converts L-arginine to NO. It has been shown that aldose reductase inhibitors improve nerve blood flow and restore endothelium-dependent relaxation to normal.63,64Figure 37-3 demonstrates the theoretical mechanism by which this metabolic defect might lead to endoneurial ischemia and abnormalities in nerve conduction.

HIV/AIDS-Related Neuropathies

The HIV-associated neuropathies manifest a wide spectrum of disorders, both clinically and pathophysiologically. The most common HIV-associated neuropathy in which pain is a characteristic feature is DSPN. Of the HIV-related neuropathies, this one is the best studied with regard to pathophysiology. The leading theories that have undergone rigorous investigation propose either direct HIV infection or immune-mediated nerve injury as the primary mechanisms underlying DSPN.

Several studies have demonstrated perivascular inflammatory infiltrates in peripheral nerves of patients with HIV-associated DSPN, consisting of macrophages and T lymphocytes, suggesting an immune mechanism underlying the disorder.10,65,66 It appears unlikely that immune complex deposition is involved in the pathogenesis of this neuropathy, based on the absence of electron microscopic evidence for immune globulin or complement deposition in nerves.65,67 However, HIV could lead to a cell-mediated immune attack on the nerves, with resultant demyelination and axonal degeneration.65 Evidence of cytokine activation and T cells and macrophages at all levels in the peripheral nervous system suggests that the damage to the peripheral nervous system in AIDS is a multifocal, immunologic process.10,66

The isolation of HIV, HIV-like particles and HIV-associated mRNA in peripheral nerves of patients with DSPN led some investigators to postulate that neuropathy resulted from direct invasion of peripheral nerve by the virus.67,68 Others have suggested direct infection by the virus of lumbosacral dorsal root ganglion cells, with subsequent central-peripheral distal axonal degeneration as a possible mechanism of DSPN.69 Subsequent pathologic studies have not been able to confirm HIV particles or antigens in peripheral nerves, nerve roots, or DRG cells.69,70 Furthermore, in the patients in which HIV was isolated from peripheral nerve in the original studies, the diagnosis of DSPN was subsequently revised to other types of neuropathy.

Approach to Pain in Polyneuropathy

The pharmacologic approach to patients with PN who suffer from positive symptoms, particularly pain, is the same, regardless of the underlying cause of the neuropathy. To date, most clinical studies looking at pharmacologic efficacy in painful neuropathy have been done in patients with diabetic neuropathies, and only a limited number of studies have used double-blind, controlled methodology. Nonetheless, management of pain in PN is often successful using standard agents, particularly when the appropriate principles are considered. Often, patients fail to achieve pain control because of inadequate dosing rather than inefficacy of a particular agent.

In neuropathy, amelioration of pain may be achieved with the use of agents that block pain transmission centrally or peripherally; the role of agents that act on the sympathetic nervous system is less clear. Certainly, some patients with PN may have an element of sympathetically maintained pain. Often, patients require trials with various agents and dosages before optimal therapeutic effect is achieved. The choice of therapy must be tailored to the individual patient (i.e., with regard to age, medical illnesses, previous medication record, and concomitant medications). Drug dosages must be titrated slowly to achieve relief of pain without intolerable side effects. One of the primary principles of pharmacologic management should be the use of single drug therapy whenever possible.

Several classes of drugs are currently used to treat patients with painful peripheral neuropathies. These include tricyclic antidepressants, anticonvulsants, antiarrhythmics, analgesics, and narcotic agents. For an extensive discussion of pharmacologic therapies for pain, including specific pharmacologic agents, see Part VI (Chaps. 58, 59, 60, 61, 62, 63, and 64). A brief review of the efficacy of the most commonly used agents for painful PN is presented next.

Pharmacologic Agents

Antidepressants

Tricyclic antidepressants have been shown to be safe and effective in alleviating the pain of peripheral neuropathy. Commonly used agents include amitriptyline, nortriptyline, imipramine, and desipramine. These agents have been studied in double-blind, randomized controlled trials with results suggesting that each of them reduces pain independent of their effect on depression.71–74 In one study, 74% of patients who received 75 to 150 mg of amitriptyline attained moderate or greater relief of their pain compared with 41% of patients receiving placebo.71 Similar results were seen with desipramine.

The tricyclic antidepressants are thought to exert their analgesic effect by inhibiting norepinephrine and serotonin reuptake in the central nervous system. They also affect cholinergic, histaminergic, and adrenergic transmission, resulting in some limiting side effects. These include sedation, orthostasis, cardiac arrhythmia, and urinary retention. Nortriptyline and desipramine, secondary amine tricyclics, tend to have fewer of these adverse effects, which is preferable for many patients, especially the elderly. Side effects may be minimized by slow titration, starting with 10 to 25 mg at night and increasing the dose by 10 to 25 mg once or twice weekly. There is no specific target dose; however the dose usually associated with pain relief is between 50 and 150 mg/day.75

Anticonvulsants

The classic anticonvulsant agents phenytoin and carbamazepine have been used in treating neuropathic pain since the 1960s. It is postulated that anticonvulsants modulate pain by suppressing neuronal firing.76 These agents have been examined in controlled trials and found to be efficacious in treating pain associated with diabetic neuropathy. A systematic review of these agents found them to be efficacious in treating some neuropathic pain syndromes, including trigeminal neuralgia and diabetic neuropathy, but noted that the risk of adverse effects was equal to the likelihood of significant benefit.77 Other agents such as clonazepam and valproate have been used anecdotally without much success in pain relief. Overall, the clinical experience suggests that these older anticonvulsant agents are less beneficial in treatment of pain in PN than tricyclic antidepressants or the newer anticonvulsants.3

In 1994, gabapentin was approved for use in the United States as an anticonvulsant. It has subsequently enjoyed broad use, including off-label use in the treatment of neuropathic pain. A recent randomized, controlled trial of gabapentin for treatment of pain in diabetic neuropathy found that patients treated with gabapentin had clinically significant alleviation of daily pain severity and improvement of quality-of-life measures.78 Patients were started on 900 mg/day and the dose increased by 900 mg/day every week for a total of 4 weeks. Sixty-seven percent of patients achieved the maximum dose of 3,600 mg/day, although patients reported therapeutic effect at doses of 900 to 1800 mg/day. Reported side effects included mild dizziness and somnolence. It was concluded that gabapentin monotherapy was efficacious and safe in treating the pain of diabetic neuropathy.

Antiarrhythmics

Studies in animal models have shown that antiarrhythmic agents modulate pain pathways by decreasing spontaneous discharges associated with pain generation in injured nerves.79 Oral mexiletine has been evaluated in several controlled studies; the general consensus is that mexiletine is safe and effective in treating neuropathic pain.80–82 The dose found to be efficacious in the studies has been 450 to 675 mg/day. It is recommended that the drug be started at 150 mg/day and titrated upward for effective pain relief. Side effects in the studies were mild and included nausea and dizziness. Electrocardiogram changes were not demonstrable in patients without cardiac disease.

Analgesics

Non-Narcotic Agents

Another recent addition to the pain therapy armamentarium is tramadol, a centrally acting, synthetic, non-narcotic analgesic that has been available in the United States since 1995. Its two mechanisms of action are (1) low affinity binding to μ-opioid receptors, and (2) weak inhibition of norepinephrine and serotonin reuptake.83 A recent double-blind, randomized, placebo-controlled trial of patients with painful diabetic neuropathy found that patients receiving an average dose of 210 mg/day of tramadol had significant pain relief with better physical and social functioning compared with patients receiving placebo.84 The most frequently occurring side effects were headache, constipation, nausea, and somnolence.

Narcotic Agents

Clinicians often prescribe narcotics when other therapies have failed. As with other therapeutic agents, it is recommended that a low dose be begun initially and titrated upward to achieve either effective pain relief or intolerable side effects. Patients should be informed of the risk of tolerance. When prescribing an opiate for painful neuropathies, it is recommended that a long-acting agent, such as extended-release morphine, oxycodone, or methadone, be used.3

Topical Agents

Capsaicin is an alkaloid found in capsicum peppers. When applied topically, it produces desensitization to noxious stimuli.85 It has been studied in patients with painful diabetic neuropathies with mixed results. One double-blind comparison of topical capsaicin and oral amitriptyline found topical capsaicin to be equally efficacious as amitriptyline but without any systemic side effects.86 Another randomized, controlled trial of the drug in painful diabetic neuropathy found it to be of no benefit.87 Although the literature appeared optimistic originally, clinical experience has been disappointing. Patients find the drug difficult to use because care must be taken not to allow contamination of unaffected body parts. Even in affected body parts, the drug causes significant burning pain on application, which most patients find an intolerable side effect.

Other Agents

Dextromethorphan, a low-affinity N-methyl-d-aspartate (NMDA) receptor antagonist and dextro isomer of the codeine analogue levorphanol, has been studied in a randomized, double-blind, placebo-controlled crossover trial of patients with a painful diabetic neuropathy and patients with postherpetic neuralgia.88 In diabetic neuropathy, patients who received dextromethorphan had their pain decrease by an average of 24% relative to patients who received placebo. Dextromethorphan did not alleviate pain for patients with postherpetic neuralgia. The average dose received by the diabetic patients was 381 mg/day. The limiting side effects of treatment with dextromethorphan include ataxia and sedation. Although this study suggests that NMDA receptor antagonists may useful in the treatment of painful neuropathies, the clinical experience has been disappointing.3