Chapter 8. Pain

Pain is an important sign of illness, and it stands preeminent among all the sensory experiences by which humans judge the existence of disease within themselves. Indeed, pain is the most common medical symptom. Relatively few diseases do not have a painful phase and in many, pain is a characteristic without which diagnosis must be in doubt.

The painful experiences pose manifold problems in virtually every field of medicine; physicians must therefore be prepared to recognize disease in patients who have felt only the first rumblings of discomfort, before other symptoms and signs have appeared. Even more problematic are patients who seek treatment for pain that appears to have little or no structural basis; further inquiry may disclose that fear of serious disease, worry, or depression has aggravated some relatively minor ache or that the complaint of pain has become the means of seeking attention, drugs or monetary compensation. They must also cope with the "difficult" pain patients in whom no amount of investigation brings to light either medical or psychiatric illness. Finally, the physician must be prepared to manage patients who require relief from intractable pain caused by established and incurable disease. To deal intelligently with pain problems requires familiarity with the anatomy of sensory pathways and the sensory supply of body segments as well as insight into the psychological factors that influence the perception of and reaction to pain.

The ambiguity with which the term pain is used is responsible for some of our difficulty in understanding it. One aspect, the easier to comprehend, is the transmission of impulses along certain pathways in response to potentially tissue-damaging stimuli, i.e., nociception. Far more abstruse is its quality as a mental state intimately linked to emotion, i.e., the quality of anguish or suffering—"a passion of the soul," in the words of Aristotle—which defies definition and quantification. This duality (nociception and suffering) is of practical importance for certain drugs or surgical procedures, such as cingulotomy, may reduce the patient's reaction to painful stimuli, leaving awareness of the sensation largely intact. Alternatively, interruption of certain neural pathways may abolish all sensation in an affected part but the symptom of pain may persist (i.e., denervation dysesthesia or anesthesia dolorosa), even in an amputated limb ("phantom pain"). Finally, unlike most sensory modalities—which are aroused by a specific stimulus such as touch-pressure, heat, or cold—pain can be evoked by any one of these stimuli if it is intense enough.

It is apparent to us that in highly specialized medical centers, and often even in "pain centers," few physicians are capable of handling difficult and unusual pain problems in any comprehensive way. In fact, it is to the neurologist that other physicians regularly turn for help with these matters. Although much has been learned about the anatomy of pain pathways, their physiologic mechanisms, and which structures to ablate in order to produce analgesia, relatively little is known about which patients should be subjected to these destructive operations or how to manage their pain by medical means. The practice of pain medicine challenges every thoughtful physician, for it demands a high degree of skill in medicine, neurology, and psychiatry.

Historical Perspective

For more than a century, views on the nature of pain sensation have been dominated by two major theories. One, the specificity theory, was from the beginning associated with the name of von Frey. He asserted that the skin consisted of a mosaic of discrete sensory spots and that each spot, when stimulated, gave rise to one sensation—either pain, pressure, warmth, or cold; in his view, each of these sensations had a distinctive end organ in the skin and each stimulus-specific end organ was connected by its own private pathway to the brain. A second theory was that of Goldscheider, who abandoned his own earlier discovery of pain spots to argue that they simply represented pressure spots, a sufficiently intense stimulation of which could produce pain. According to the latter theory, there were no distinctive pain receptors, and the sensation of pain was the result of the summation of impulses excited by pressure or thermal stimuli applied to the skin. Originally called the intensivity theory, it later became known as the pattern or summation theory.

In an effort to conciliate the pattern and specificity theories, Head and colleagues, in 1905, formulated a novel concept of pain sensation, based on observations that followed his purposeful division of the cutaneous branch of the radial nerve in his own forearm. The zone of impaired sensation contained an innermost area in which superficial sensation was completely abolished. This was surrounded by a narrower ("intermediate") zone, in which pain sensation was preserved but poorly localized; extreme degrees of temperature were recognized in the intermediate zone but perception of touch, lesser differences of temperature, and two-point discrimination were abolished. To explain these findings, Head postulated the existence of two systems of cutaneous receptors and conducting fibers: (1) an ancient protopathic system, subserving pain and extreme differences in temperature and yielding ungraded, diffuse impressions of an all-or-none type; and (2) a more recently evolved epicritic system, which mediated touch, two-point discrimination, and lesser differences in temperature, as well as localized pain. The pain and hyperesthesia that follow damage to a peripheral nerve were attributed to a loss of inhibition that was normally exerted by the epicritic upon the protopathic system. This theory was used for many years to explain the sensory alterations that occur with both peripheral and central (thalamic) lesions. It lost credibility for several reasons but mainly because Head's original observations and deductions upon which they were based could not be confirmed (see Trotter and Davies; also Walshe). Nevertheless, both fast and slow forms of pain conduction were later corroborated (see below).

A later refinement of the pattern and specificity concepts of pain was made in 1965 when Melzack and Wall articulated their "gate-control" theory. They observed, in decerebrate and spinal cats, that peripheral stimulation of large myelinated fibers produced a negative dorsal root potential and that stimulation of small unmyelinated C (pain) fibers caused a positive dorsal root potential. They postulated that these potentials, which were a reflection of presynaptic inhibition or excitation, modulated the activity of secondary transmitting neurons (T cells) in the dorsal horn and that this modulation was mediated through inhibitory (I) cells. The essence of this theory was that the large-diameter fibers excited the I cells, which, in turn, caused a presynaptic inhibition of the T cells; conversely, the small pain afferents inhibited the I cells, leaving the T cells in an excitatory state. Melzack and Wall emphasized that pain impulses from the dorsal horn must also be under the control of a descending system of fibers from the brainstem, thalamus, and limbic lobes.

At first the gate-control mechanisms seemed to offer an explanation of the pain of ruptured disc and of certain chronic neuropathies (particularly those with large fiber out-fall) and attempts were made to relieve pain by subjecting the peripheral nerves and dorsal columns (presumably their large myelinated fibers) to sustained transcutaneous electrical stimulation. Such selective stimulation would theoretically "close" the gate. In some clinical situations these procedures have indeed given relief from pain, but not necessarily as a result of stimulation of large myelinated fibers alone (see Taub and Campbell). But in a number of other instances relating to pain in large- and small-fiber neuropathies, the clinical behavior has been quite out of keeping with what one would expect on the basis of the gate-control mechanism. As with preceding theories, flaws have been exposed in the physiologic observations on which the theory is based. These and other aspects of the gate-control theory of pain have been critically reviewed by Nathan.

During the last few decades there has been a significant accrual of information on cutaneous sensibility, demanding modification of earlier anatomic–physiologic and clinical concepts. Interestingly, much of this information is still best described and rationalized in the general framework of the oldest theory, that of specificity, as is evident from the ensuing discussion on pain and that on other forms of cutaneous sensibility in the next chapter.

Pain Receptors and Peripheral Afferent Pathways

In terms of peripheral pain mechanisms, there is indeed a high degree, though not absolute, specificity in the von Frey sense. In keeping with distinctions between nerve types, the sensory (and motor) fibers have been classified according to their size and function (Table 8-1). It is now well established that two types of afferent fibers in the distal axons of primary sensory neurons respond maximally to nociceptive (i.e., potentially tissue-damaging) stimuli. One type is the very fine, unmyelinated, slowly conducting C fiber (0.3 to 1.1μ in diameter); the other is the thinly myelinated, more rapidly conducting A-δ fiber (2 to 5μ in diameter). The peripheral terminations of both of these primary pain afferents or receptors are the free, profusely branched nerve endings in the skin and other organs; these are covered by Schwann cells but contain little or no myelin.

There is considerable evidence, based on their response characteristics, that a degree of subspecialization exists within these freely branching, nonencapsulated endings and their small-fiber afferents. Three categories of free endings or receptors are recognized: mechanoreceptors, thermoreceptors, and polymodal nociceptors. Each ending transduces stimulus energy into an action potential in the distal nerve membranes. The first two types of receptors are activated by innocuous mechanical and thermal stimulation, respectively; the mechanoeffects are transmitted by both A-δ and C fibers and the thermal effects mostly by C fibers. The majority of C fibers are polymodal and are most effectively excited by noxious or tissue-damaging stimuli, but they can respond to both mechanical and thermal stimuli and to chemical mediators such as those associated with inflammation. Moreover, certain A-δ fibers respond to light touch, temperature, and pressure as well as to pain stimuli and are capable of discharging in proportion to the intensity of the stimulus. The stimulation of single fibers by intraneural electrodes indicates that they can also convey information concerning the nature and location of the stimulus. These observations on the polymodal functions of A-δ and C fibers would explain the earlier observations of Lele and Weddell that modes of sensation other than pain can be evoked from structures such as the cornea, which is innervated solely by free nerve endings.

The manner in which painful stimuli are translated into electrical depolarizations in nerve endings is beginning to be understood. A number of specialized molecules, when activated by noxious stimuli, open cationic channels in membranes of the nerve ending. Opening of these channels, in turn, activates voltage-gated sodium channels and generates an action potential in the sensory axon. Mannion and Woolf have summarized the regulation and activation of these receptor molecules.

The peripheral afferent pain fibers of both A-δ and C types have their cell bodies in the dorsal root ganglia; central extensions of these nerve cells project, via the dorsal root, to the dorsal horn of the spinal cord (or, in the case of cranial pain afferents, to the spinal trigeminal nucleus, the medullary analogue of the dorsal horn). The pain afferents occupy mainly the lateral part of the root entry zone. Within the spinal cord, many of the thinnest fibers (C fibers) form a discrete bundle, the tract of Lissauer (Fig. 8-1A). That the tract of Lissauer is predominantly a pain pathway is shown (in animals) by the ipsilateral segmental analgesia that results from its transection but it contains deep sensory or propriospinal fibers as well. Although it is customary to speak of a lateral and medial division of the posterior root (the former contains small pain fibers and the latter, large myelinated fibers), the separation into discrete functional bundles is not complete, and in humans the two groups of fibers cannot be differentially interrupted by selective rhizotomy.

Dermatomic Distribution of Pain Fibers

(See Fig. 9.1) Each sensory unit (the sensory nerve cell in the dorsal root ganglion, its central and peripheral extensions, and cutaneous and visceral endings) has a unique topography that is maintained throughout the sensory system from the periphery to the sensory cortex. The discrete segmental distribution of the sensory units permits the construction of sensory maps, so useful to clinicians (see Fig. 9-1). This aspect of sensory anatomy is elaborated in the next chapter, which includes maps of the sensory dermatomes and cutaneous nerves. However, as a means of quick orientation to the topography of peripheral pain pathways, it is useful to remember that the facial structures and anterior cranium lie in the fields of the trigeminal nerves; the back of the head, second cervical; the neck, third cervical; the epaulet area, fourth cervical; the deltoid area, fifth cervical; the radial forearm and thumb, sixth cervical; the index and middle fingers, seventh cervical; the little finger and ulnar border of hand and forearm, eighth cervical–first thoracic; the nipple, fifth thoracic; the umbilicus, tenth thoracic; the groin, first lumbar; the medial side of the knee, third lumbar; the great toe, fifth lumbar; the little toe, first sacral; the back of the thigh, second sacral; and the genitoanal zones, third, fourth, and fifth sacral segments. The distribution of pain fibers from deep structures, although not fully corresponding to those from the skin, also follows a segmental pattern. The first to fourth thoracic nerve roots are the important sensory pathways for the heart and lungs; the sixth to eighth thoracic, for the upper abdominal organs; and the lower thoracic and upper lumbar, for the lower abdominal viscera. These areas of projection from visceral structures roughly correspond to the areas of adjacent root innervation, with some exceptions because of routing of sensory nerves to organs that migrate with development.

Neurologically relevant maps of pain projection from the bones, ligaments, and adjacent musculoskeletal structures have been termed sclerotomes; they differ slightly in their distribution from the dermatomes. A further discussion of referred pain and a figure comparing sclerotomes and dermatomes is given later in the chapter.

The Dorsal Horn

The afferent pain fibers, after traversing the tract of Lissauer, terminate in the posterior gray matter or dorsal horn, predominantly in the marginal zone. Most of the fibers terminate within the segment of their entry into the cord; some extend ipsilaterally to one or two adjacent rostral and caudal segments; and some project, via the anterior commissure, to the contralateral dorsal horn. The cytoarchitectonic studies of Rexed in the cat (the same organization pertains in primates and probably in humans) have shown that second-order neurons, the sites of synapse of afferent sensory fibers in the dorsal horn, are arranged in a series of six layers or laminae (Fig. 8-1B). Thinly myelinated (A-δ) fibers terminate principally in lamina I of Rexed (marginal cell layer of Waldeyer) and also in the outermost part of lamina II; some A-δ pain fibers penetrate the dorsal gray matter and terminate in the lateral part of lamina V. Unmyelinated (C) fibers terminate in lamina II (substantia gelatinosa). Yet other cells that respond to painful cutaneous stimulation are located in ventral horn laminae VII and VIII. The latter neurons are responsive to descending impulses from brainstem nuclei as well as segmental sensory impulses. From these cells of termination, second-order axons connect with ventral and lateral horn cells in the same and adjacent spinal segments and subserve both somatic and autonomic reflexes. The main bundle of secondary neurons subserving pain sensation projects contralaterally (and to a lesser extent ipsilaterally) to higher levels; this constitutes the spinothalamic tract, discussed below.

A number of important observations have been made concerning the mode of transmission and modulation of pain impulses in the dorsal horn and brainstem. Excitatory amino acids (glutamate, aspartate) and nucleotides such as adenosine triphosphate (ATP) are the putative transmitters at terminals of primary A-δ sensory afferents. Also, A-δ pain afferents, when stimulated, release several neuromodulators that play a role in the transmission of pain sensation. Slower neurotransmission by C neurons involves other substances, of which the most important is the 11-amino-acid peptide known as substance P ("P" for powder extracted from animal tissue and urine by von Euler in 1931). In animals, substance P excites nociceptive dorsal root ganglion and dorsal horn neurons; furthermore, destruction of substance P fibers produces analgesia. In patients with the rare condition of congenital neuropathy and insensitivity to pain, there is a marked depletion of dorsal horn substance P.

A large body of evidence indicates that opiates are important modulators of pain impulses as they are relayed through the dorsal horn and through nuclei in the medulla and midbrain. Thus, opiates have been noted to decrease substance P; at the same time, flexor spinal reflexes, which are evoked by segmental pain, are reduced. Opiate receptors of three types are found on both presynaptic primary afferent terminals and postsynaptic dendrites of small neurons in lamina II. Moreover, lamina II neurons, when activated, release enkephalins, endorphins, and dynorphins—all of which are endogenous, morphine-like peptides that bind specifically to opiate receptors and inhibit pain transmission at the dorsal horn level. The subject of pain modulation by opiates and endogenous morphine-like substances is elaborated further on.

Spinal Afferent Tracts for Pain

The (Anterolateral, or Lateral) Spinothalamic Tract

As indicated above, axons of secondary neurons that sub-serve pain sensation originate in laminae I, II, V, VII, and VIII of the spinal gray matter. The principal bundle of these axons decussates in the anterior spinal commissure and ascends in the anterolateral fasciculus of the opposite side of the cord as the spinothalamic tract to terminate in brainstem and thalamic structures (Fig. 8-2). It is of clinical consequence that the axons carrying pain impulses from each dermatome decussate one to three segments rostral to the level of root entry. For this reason, a discrete lesion of the lateral spinal cord creates a loss of pain and thermal sensation of the contralateral trunk, the dermatomal level of which is two to three segments below that of the spinal cord lesion. As the ascending fibers cross the cord, they are added to the inner side of the spinothalamic tract (the principal afferent pathway of the anterolateral fasciculus), so that the longest fibers from the sacral segments come to lie most superficially and fibers from successively more rostral levels occupy progressively deeper positions (Fig. 8-3). This somatotopic arrangement is of importance to the neurosurgeon performing an operation for pain relief, insofar as the depth to which the funiculus is cut will govern the level of analgesia that is achieved; for the neurologist, it provides an explanation of the pattern of "sacral sparing" of pain and thermal sensation created by centrally placed lesions of the spinal cord. The termination of this tract, mainly in the thalamus, is described further on.

Other Spinocerebral Afferent Tracts

In addition to the anterolateral spinothalamic tract—a fast-conducting pathway that projects directly to the thalamus—the anterolateral fasciculus of the cord contains several more slowly conducting, medially placed systems of fibers. One such group of fibers projects directly to the reticular core of the medulla and midbrain and then to the medial and intralaminar nuclei of the thalamus; these fibers are referred to as the spinoreticulothalamic or paleospinothalamic pathway. At the level of the medulla, these fibers synapse in the nucleus gigantocellularis; more rostrally, they connect with nuclei of the parabrachial region, midbrain reticular formation, periaqueductal gray matter, and hypothalamus. A second, more medially placed pathway in the anterolateral cord ascends to the brainstem reticular core via a series of short interneuronal links. It is not clear whether these spinoreticular fibers are collaterals of the spinothalamic tracts, as Cajal originally stated, or whether they represent an independent system, as more recent data seem to indicate. Probably both statements are correct. There is also a third, direct spinohypothalamic pathway in the anterolateral fasciculus.

The conduction of diffuse, poorly localized pain arising from deep and visceral structures (gut, periosteum, peritoneum) has been ascribed to these slow-conducting, indirect pathways. Melzack and Casey have proposed that this fiber system (which they refer to as paramedian), with its diffuse projection via brainstem and thalamus to the limbic and frontal lobes, subserves the affective aspects of pain, i.e., the unpleasant feelings engendered by pain. It is evident that these spinoreticulothalamic pathways continue to evoke the psychic experience of pain even when the direct spinothalamic pathways have been interrupted. However, it is the direct spinothalamic pathway, which projects to the ventroposterolateral (VPL) nucleus of the thalamus and thence to discrete areas of the sensory cortex, that subserves the sensory-discriminative aspects of pain, i.e., the processes that underlie the localization, quality, and possibly the intensity of the noxious stimulus. Also, the pathways for visceral pain from the esophagus, stomach, small bowel, and proximal colon are carried largely in the vagus nerve and terminate in the nucleus of the solitary tract (nucleus tractus solitarius [NTS]) before projecting to the thalamus, as described below. Other abdominal viscera still activate the NTS when the vagus is severed in animals, probably transmitting impulses through the splanchnic plexus.

It should be emphasized that the foregoing data concerning the cells of termination of cutaneous nociceptive stimuli and the cells of origin of ascending spinal afferent pathways have all been obtained from studies in animals (including monkeys). In humans, the specific cells of origin of the direct spinothalamic tract fibers have not been fully identified. Information about this pathway in humans has been derived from the study of postmortem material and from the examination of patients subjected to anterolateral cordotomy for intractable pain. What can be stated of clinical importance is that unilateral section of the anterolateral funiculus produces a relatively complete loss of pain and thermal sense on the opposite side of the body, extending to a level two or three segments below the lesion as noted earlier. After a variable period of time, pain sensibility usually returns, probably being conducted by pathways that lie outside the anterolateral quadrants of the spinal cord that gradually increase their capacity to conduct pain impulses. One of these is a longitudinal polysynaptic bundle of small myelinated fibers in the center of the dorsal horn (the dorsal intracornual tract); another consists of axons of lamina I cells that travel in the dorsal part of the lateral funiculus.

Thalamic Terminus of Pain Fibers

The direct spinothalamic fibers separate into two bundles as they approach the thalamus. The lateral division terminates in the ventrobasal and posterior groups of nuclei, the most important of which is the VPL nucleus. The medial contingent terminates mainly in the intralaminar complex of nuclei and in the nucleus submedius. Spinoreticulothalamic (paleospinothalamic) fibers project onto the medial intralaminar (primarily parafascicular and centrolateral) thalamic nuclei; i.e., they overlap with the terminations of the medially projecting direct spinothalamic pathway. Projections from the dorsal column nuclei, which have a modulating influence on pain transmission, are mainly to the ventrobasal and ventroposterior group of nuclei. Each of the four thalamic nuclear groups that receives nociceptive projections from the spinal cord has a distinct cortical projection and each is thought to play a different role in pain sensation (see below).

One practical conclusion to be reached from these anatomic and physiologic studies is that, at thalamic levels, fibers and cell stations transmitting the nociceptive impulses are not organized into discrete loci. In general, neurophysiologic evidence indicates that as one ascends from peripheral nerve to spinal, medullary, mesencephalic, thalamic, and limbic levels, the predictability of neuron responsivity to noxious stimuli diminishes. Thus it comes as no surprise that neurosurgical lesions that interrupt afferent pathways at progressively higher levels of the brainstem and thalamus become decreasingly successful.

Thalamocortical Projections

The ventrobasal thalamic complex and the ventroposterior group of nuclei project to two main cortical areas: the primary sensory (postcentral) cortex (a small number terminate in the precentral cortex) and the upper bank of the sylvian fissure. These cortical projections are described more fully in Chap. 9 but it can be stated here that they are concerned mainly with the reception of tactile and proprioceptive stimuli and with all discriminative sensory functions, including pain. The extent to which either cortical area is activated by thermal and painful stimuli is uncertain. Certainly, stimulation of these (or any other) cortical areas in a normal, alert human being does not produce pain. The intralaminar nuclei, which also project to the hypothalamus, amygdaloid nuclei, and limbic cortex, probably mediate the arousal and affective aspects of pain and autonomic responses.

The thalamic projection to the primary sensory cortex that is distributed mainly along the postcentral gyrus of the anterior parietal lobe is shown in Fig. 9-5 (the "sensory homunculus"). The cortical representation allows for accurate localization of the site of origin of a painful stimulus but the notion that thalamic projections terminate solely in this region is an oversimplification.

Thalamic and cerebral cortical localization of visceral sensation is not well known. However, cerebral evoked potentials and increased cerebral blood flow (by positron emission tomography [PET] studies) have been demonstrated in the thalamus and pre- and postcentral gyri of patients undergoing rectal balloon distention (Silverman et al; Rothstein et al).

Descending Pain-Modulating Systems

The discovery of a system of descending fibers and way stations that modulate activity in nociceptive pathways has proved to be a major addition to our knowledge of pain. The endogenous pain control system that has been studied most extensively emanates from the frontal cortex and hypothalamus and projects to cells in the periaqueductal region of the midbrain and then passes to the ventromedial medulla. From there it descends in the dorsal part of the lateral fasciculus of the spinal cord to the posterior horns (laminae I, II, and V; see further discussion under "Endogenous Pain-Control Mechanisms"). Several other descending pathways, noradrenergic and serotonergic, arise in the locus ceruleus, dorsal raphe nucleus, and nucleus reticularis gigantocellularis and are also important modifiers of the nociceptive response. The clinical significance of these pain-modulating pathways, still under study, is discussed further on.

The stimuli that activate pain receptors vary from one tissue to another. The adequate stimulus for skin is one that has the potential to injure tissue, i.e., pricking, cutting, crushing, burning, and freezing. These stimuli are ineffective when applied to the stomach and intestine, where pain is produced by an engorged or inflamed mucosa, distention or spasm of smooth muscle, and traction on the mesenteric attachment. In skeletal muscle, pain is caused by ischemia (the basis of intermittent claudication), necrosis, hemorrhage, and injection of irritating solutions as well as by injuries of connective tissue sheaths. Prolonged contraction of skeletal muscle evokes an aching type of pain. Ischemia is also the most important cause of pain in cardiac muscle. Joints are insensitive to pricking, cutting, and cautery, but pain can be produced in the synovial membrane by inflammation and by exposure to hypertonic saline. The stretching and tearing of ligaments around a joint can evoke severe pain. Injuries to the periosteum give rise to pain but probably not to other sensations. Blood vessels are a source of pain when pierced by a needle or involved in an inflammatory process. Distention of arteries or veins, as occurs with thrombotic or embolic occlusion, may be sources of pain; other mechanisms of headache relate to traction on arteries or inflammation of the meningeal structures by which they are supported. (The subject of headache and its origins is taken up in Chap. 10.) Pain from intraneural lesions probably arises from the sheaths of the nerves. Nerve root(s) and sensory ganglia, when compressed (e.g., by a ruptured disc), give rise to pain.

With damage to tissue, there is a release of proteolytic enzymes, which act locally on tissue proteins to liberate substances that excite peripheral nociceptors. These pain-producing substances—which include histamine, prostaglandins, serotonin, and similar polypeptides, as well as potassium ions—elicit pain when they are injected intraarterially or applied to the base of a blister. Other pain-producing substances such as kinins are released from sensory nerve endings or are carried there by the circulation. Local vascular permeability is also increased by these substances.

In addition, direct stimulation of nociceptors releases polypeptide mediators that enhance pain perception. The best studied of these is substance P, which is released from the nerve endings of C fibers in the skin during peripheral nerve stimulation. It causes erythema by dilating cutaneous vessels and edema by releasing histamine from mast cells; it also acts as a chemoattractant for leukocytes. This reaction, called neurogenic inflammation by White and Helme, is mediated by antidromic action potentials from the small nerve cells in the spinal ganglia and is the basis of the axon reflex of Lewis; the reflex is abolished in peripheral nerve diseases and can be studied electrophysiologically as an aid to clinical localization.

Perception of Pain

The threshold for perception of pain, i.e., the lowest intensity of a stimulus recognized as pain, is approximately the same in all persons. Inflammation lowers the threshold for perception of pain by a process called sensitization. This process, termed allodynia, allows ordinarily innocuous stimuli to produce pain in sensitized tissues. The pain threshold is, of course, raised by local anesthetics and by certain lesions of the nervous system as well as by centrally acting analgesic drugs. Mechanisms other than lowering or raising the pain threshold are important as well. Placebos reduce pain in about one-third of the groups of patients in which such effects have been recorded. Acupuncture at sites anatomically remote from painful operative fields also reduces the pain in some individuals. Distraction and suggestion, by turning attention away from the painful part, reduce the awareness of and response to pain but not the threshold for its perception. Strong emotion (fear or rage) suppresses pain, presumably by activation of the above-described descending noradrenergic system. The experience of pain appears to be lessened in manic states and enhanced in depression. Anxious patients in general have the same pain threshold as normal subjects but their reaction may be excessive or abnormal. The pain thresholds of frontal lobotomized subjects are also unchanged but they react to painful stimuli only briefly or casually if at all.

The conscious awareness or perception of pain occurs only when pain impulses reach the thalamocortical level. The precise roles of the thalamus and cortical sensory areas in this mental process are not fully understood. It was believed that the recognition of a noxious stimulus as such is a function of the thalamus and that the parietal cortex is necessary for appreciation of the intensity, localization, and other discriminatory aspects of sensation. This traditional separation of sensation (in this instance, awareness of pain) and perception (awareness of the nature of the painful stimulus) has evolved to the view that sensation, perception, and the various conscious and unconscious responses to a pain stimulus comprise an indivisible process. That the cerebral cortex governs the patient's reaction to pain cannot be doubted. It is also likely that the cortex can suppress or otherwise modify the perception of pain in the same way that corticofugal projections from the sensory cortex modify the rostral transmission of other sensory impulses from thalamic and dorsal column nuclei. It has been shown that central transmission in the spinothalamic tract can be inhibited by stimulation of the sensorimotor areas of the cerebral cortex, and, as indicated above, a number of descending fiber systems have been traced to the dorsal horn laminae from which this tract originates.

The functional imaging studies by Wager and coworkers has given insights into the ensemble of brain regions that are activated by painful stimuli. In addition to the expected thalamic and parietal sensory regions, the hypothalamus, and both insular and cingulate cortices, are prominently involved, in proportion to the intensity of the stimulus. These investigators have sought to develop an imaging "pain signature" that could, in the future, objectify the pain response. Moreover, physical pain in their experiments could be differentiated from social and emotional pain. Whether this reductionist approach to pain will find clinical use is discussed by Jaillard and Ropper.

Endogenous Pain-Control Mechanisms

An important contribution to our understanding of pain has been the discovery of a neuronal analgesia system that can be activated by the administration of opiates or by naturally occurring brain substances that share the properties of opiates. This endogenous system was first demonstrated by Reynolds, who found that stimulation of the ventrolateral periaqueductal gray matter in the rat produced a profound analgesia without altering behavior or motor activity. Subsequently, stimulation of other discrete sites in the medial and caudal regions of the diencephalon and rostral bulbar nuclei (notably raphe magnus and paragigantocellularis) was shown to have the same effect. Under the influence of such electrical stimulation, the animal could be operated on without anesthesia and move around in an undisturbed manner despite the administration of noxious stimuli. Investigation disclosed that the effect of stimulation-produced analgesia (SPA) is inhibition of the neurons of laminae I, II, and V of the dorsal horn, i.e., the neurons that are activated by noxious stimuli. In human subjects, stimulation of the midbrain periaqueductal gray matter through stereotactically implanted electrodes has also produced a state of analgesia, though not consistently. Other sites in which electrical stimulation is effective in suppressing nociceptive responses are the rostroventral medulla (nucleus raphe magnus and adjacent reticular formation) and the dorsolateral pontine tegmentum. These effects are relayed to the dorsal horn gray matter via a pathway in the dorsolateral funiculus of the spinal cord. Ascending pathways from the dorsal horn, conveying noxious somatic impulses, are also important in activating the modulatory network.

Opiates also act pre- and postsynaptically on the neurons of laminae I and V of the dorsal horn, suppressing afferent pain impulses from both the A-δ and C fibers. Furthermore, these effects can be reversed by the opioid antagonist naloxone. Interestingly, naloxone can reduce some forms of stimulation-produced analgesia. Levine and colleagues have demonstrated that not only does naloxone enhance clinical pain, but it also interferes with the pain relief produced by placebos. These observations suggest that the heretofore mysterious beneficial effects of placebos (and perhaps of acupuncture) may be a result of activation of an endogenous system that shuts off pain through the release of pain-relieving endogenous opioids, or endorphins (see below). Prolonged pain and fear are the most powerful activators of this endogenous opioid-mediated modulating system. The same system is probably operative under a variety of other stressful conditions; for example, some soldiers, wounded in battle, require little or no analgesic medication ("stress-induced analgesia"). The opiates also act at several loci in the brainstem, at sites corresponding with those producing analgesia when stimulated electrically and generally conforming to areas in which neurons with endorphin receptors are localized.

Soon after the discovery of specific opiate receptors in the central nervous system (CNS), several naturally occurring peptides, which proved to have a potent analgesic effect and to bind specifically to opiate receptors, were identified. These endogenous, morphine-like compounds are generically referred to as endorphins, meaning "the morphines within." The most widely studied are β-endorphin, a peptide sequence of the pituitary hormone β-lipotropin, and two other peptides, enkephalin and dynorphin. They are found in greatest concentration in relation to opiate receptors in the midbrain. At the level of the spinal cord, exclusively enkephalin receptors are found. Figure 8-4 illustrates a theoretical construct of the roles of enkephalin (and substance P) at the point of entry of pain fibers into the spinal cord. A subgroup of dorsal horn interneurons that are in contact with spinothalamic tract neurons also contains enkephalin.

Thus it appears that the central effects of a painful condition are determined by many ascending and descending systems using a variety of transmitters. A deficiency in a particular region would explain persistent or excessive pain. Some aspects of opiate addiction and also the discomfort that follows withdrawal of the drug might conceivably be accounted for in this way. Indeed, it is known that some of these peptides not only relieve pain but suppress withdrawal symptoms.

Finally it should be noted that the descending pain-control systems contain noradrenergic and serotonergic, as well as opiate, connections. A descending norepinephrine-containing pathway, as mentioned, has been traced from the locus ceruleus in the dorsolateral pons to the spinal cord, and its activation blocks spinal nociceptive neurons. The rostroventral medulla contains a large number of serotonergic neurons from which descending fibers inhibit dorsal horn cells concerned with pain transmission, perhaps providing a rationale for the use of certain antidepression medications that are serotonin agonists in patients with chronic pain.

Terminology (Table 8-2)

Several terms, related to the experience of altered sensations and pain, are often used interchangeably but each has specific meaning. Hyperesthesia is a general term for heightened cutaneous sensitivity. The term hyperalgesia refers to an increased sensitivity and a lowered threshold to painful stimuli. Inflammation and burns of the skin are common causes of hyperalgesia. The term hypalgesia, or hypoalgesia, refers to the opposite state—i.e., a decreased sensitivity and a raised threshold to painful stimuli. A demonstrable reduction in pain perception (i.e., an elevated threshold) associated with an increased reaction to the stimulus once it is perceived, is sometimes referred to as hyperpathia (subtly different from hyperalgesia). In this circumstance there is an excessive reaction to all stimuli, even those (such as light touch) that normally do not evoke pain, a symptom termed allodynia. The elicited allodynic pain may have unusual features, outlasting the stimulus and being diffuse, modifiable by fatigue and emotion, and often being mixed with other sensations. The mechanism of these abnormalities is not clear but both hyperpathia and allodynia are common features of neuropathic or neurogenic pain, such as the pain generated by peripheral neuropathy. These features are also exemplified by causalgia, a type of burning pain that results from interruption of a peripheral nerve (see "Causalgia and Reflex Sympathetic Dystrophy").

Skin Pain and Deep Sensibility

As indicated earlier, the nerve endings in each tissue are activated by different mechanisms, and the pain that results is characterized by its quality, locale, and temporal attributes. Skin pain is of two types: a pricking pain, evoked immediately on penetration of the skin by a needle point, or a stinging or burning pain that follows in a second or two. Together they constitute the "double response" of Lewis. Both types of dermal pain can be localized with precision. Compression of nerves by the application of a tourniquet to a limb abolishes pricking pain before burning pain because large fibers are more susceptible to pressure. The first (fast) pain is transmitted by the larger (A-δ) fibers and the second (slow) pain, which is somewhat more diffuse and longer lasting, by the thinner, unmyelinated C fibers.

Deep pain from visceral and skeletomuscular structures is usually aching in quality; if intense, it may be sharp and penetrating (knife-like). Occasionally visceral derangements cause a burning type of pain, as in the "heartburn" of esophageal irritation and rarely in angina pectoris. The pain is felt as being deep to the body surface. It is diffuse and poorly localized, and the margins of the painful zone are not well delineated, presumably because of the relative paucity of nerve endings in viscera. Visceral pain produces two additional sensations. First, there is tenderness at remote superficial sites ("referred hyperalgesia") and, second, an enhanced pain sensitivity in the same and in nearby organs ("visceral hyperalgesia"). This is a restatement of Head's early observations, discussed above, and the referred "Head zones," where somatic and visceral sensibility overlap as discussed below. The concept of visceral hyperalgesia has received considerable attention in a number of pain syndromes in reference to the transition from acute to chronic pain, particularly in headache.

Referred Pain

The localization of deep pain of visceral origin raises a number of problems. Deep pain has indefinite boundaries and its location is distant from the visceral structure involved. It tends to be referred not to the skin overlying the viscera of origin but to other areas innervated by the same spinal segment (or segments). This pain, projected to some fixed site at a distance from the source, is called referred pain. The ostensible explanation for the site of referral is that small-caliber pain afferents from deep structures project to a wide range of lamina V neurons in the dorsal horn, as do cutaneous afferents. The convergence of deep and cutaneous afferents on the same dorsal horn cells, coupled with the fact that cutaneous afferents are far more numerous than visceral afferents and have direct connections with the thalamus, is probably responsible for the phenomenon.

Because the nociceptive receptors and nerves of any given visceral or skeletal structure may project upon the dorsal horns of several adjacent spinal or brainstem segments, the pain from these structures may be fairly widely distributed. For example, afferent pain fibers from cardiac structures, distributed through segments T1 to T4, may be projected to the inner side of the arm and the ulnar border of the hand and arm (T1 and T2) as well as the precordium (T3 and T4). Once this pool of sensory neurons in the dorsal horns of the spinal cord is activated, additional noxious stimuli may heighten the activity in the whole sensory field ipsilaterally and, to a lesser extent, contralaterally.

The regions of projection of pain that originate in the bones and adjacent ligamentous structures have been called by Kellgren, "sclerotomes." His maps of pain referral patterns were established from studies of the injection of hypertonic saline into muscle and interspinous ligaments. Although dermatomes and sclerotomes overlap, the patterns are slightly different as shown in Fig. 8-5, which is taken from Inman and Saunders. These sclerotomatous projections are useful to neurologists in analyzing the origins of unusual pains of the cranium, spine, and limbs (see Chaps. 10 and 11).

Another peculiarity of localization is aberrant reference, explained by an alteration of the physiologic status of the pools of neurons in adjacent segments of the spinal cord. For example, cervical arthritis or gallbladder disease, causing low-grade discomfort by constantly activating their particular segmental neurons, may induce a shift of cardiac pain cephalad or caudad from its usual locale. Once it becomes chronic, any pain may spread quite widely in a vertical direction on one side of the body. On the other hand, painful stimuli arising from a distant site exert an inhibitory effect on segmental nociceptive flexion reflexes in the leg, as demonstrated by DeBroucker and colleagues. Yet another clinical peculiarity of segmental pain is the reduction in power of muscle contraction that it may cause (reflex paralysis, or algesic weakness).

Chronic Pain

One of the most perplexing issues in the study of pain is the manner in which chronic pain syndromes arise. Several theories have been offered, none of which satisfactorily accounts for all the clinically observed phenomena. One hypothesis proposes that in an injured nerve, the unmyelinated sprouts of A-δ and C fibers become capable of spontaneous ectopic excitation and afterdischarge and are susceptible to ephaptic activation. A second proposal derives from the observation that these injured nerves are also sensitive to locally applied or intravenously administered catecholamines because of an overabundance of adrenergic receptors on the regenerating fibers. Either this mechanism or ephapse (nerve-to-nerve cross-activation) is thought to be the basis of causalgia (persistent burning and aching pain in the territory of a partially injured nerve and beyond) and its associated reflex sympathetic dystrophy; either would explain the relief afforded in these conditions by sympathetic block. This subject is discussed in greater detail in relation to peripheral nerve injuries (see "Peripheral Nerve Pain" and Chap. 46).

Central sensory structures, e.g., sensory neurons in the dorsal horns of the spinal cord or thalamus, if chronically bombarded with pain impulses, may become autonomously overactive (being maintained in this state perhaps by excitatory amino acids) and may remain so even after the peripheral pathways have been interrupted. Peripheral nerve lesions have been shown to induce enduring derangements of central (spinal cord) processing (Fruhstorfer and Lindblom). For example, avulsion of nerves or nerve roots may cause chronic pain even in analgesic zones (anesthesia dolorosa or "deafferentation pain"). In experimentally deafferented animals, neurons of lamina V begin to discharge irregularly in the absence of stimulation. Later the abnormal discharge subsides in the spinal cord but can still be recorded in the thalamus. Consequently, painful states such as causalgia, spinal cord pain, and phantom pain are not abolished simply by cutting spinal nerves or spinal tracts.

Certainly none of these phenomena can adequately explain the entire story of chronic pain. It is likely that structural changes in the spinal cord, of the type alluded to above, are able to produce persistent stimulation of pain pathways. Indo and colleagues review the molecular changes in the spinal cord that may give rise to persistence of pain after the cessation of an injurious episode. It is an open question whether the early treatment of pain may prevent the cascade of biochemical events that allows for both spread and persistence of pain in conditions such as causalgia, but it has been the experience of most clinical pain experts that preemptive treatment of certain painful conditions (e.g., herpes zoster) may reduce the risk of a chronic pain syndrome.

Pain has several other singular attributes. It does not appear to be subject to negative adaptation—i.e., pain may persist as long as the stimulus is operative—whereas other somatic stimuli, if applied continuously, soon cease to be perceived. Furthermore, prolonged stimulation of pain receptors sensitizes them, so that they become responsive to even low grades of stimulation, even to touch (allodynia).

The Emotional Reaction to Pain

Another remarkable characteristic of pain is the strong feeling or affect with which it is endowed, nearly always unpleasant. Since pain embodies this element, psychologic conditions assume great importance in all persistent painful states. It is of interest that despite this strong affective aspect of pain, it is difficult to recall precisely, or to reexperience from memory, a previously experienced acute pain. Also, the patient's tolerance of pain and capacity to experience it without verbalization are influenced by culture and personality. Some individuals—by virtue of training, habit, and phlegmatic temperament—remain stoic in the face of pain, and others react in an opposite fashion. In other words, there are inherent variations among individuals that determine the limbic system's response to pain. In this regard, it is important to emphasize that pain may be the presenting or predominant symptom in a depressive illness (Chap. 52). Price reviews this subject of the affective dimension of pain in detail, but it must be acknowledged that the models offered are largely theoretical. It is noteworthy, however, that on functional imaging studies regions of the cerebrum that are activated by experimentally induced physical pain overlap with those for the experience of emotional pain, as reported by Wager and colleagues.

Finally, a comment should be made about the devastating behavioral effects of chronic pain. To quote from Ambroïse Paré, a sixteenth-century French surgeon, "There is nothing that abateth so much the strength as paine." Continuous pain increases irritability and fatigue, disturbs sleep, and impairs appetite. Patients in pain may seem irrational about their illness and make unreasonable demands on family and physician. Characteristic is an unwillingness to engage in or continue any activity that might enhance their pain. They withdraw from the main current of daily affairs as their thoughts and speech come to be dominated by the pain. Once a person is subjected to the tyranny of chronic pain, depressive symptoms are practically always added. A person's entire identity may be dominated by the mixture of pain and depression (l'homme doloureux). Determining the cause and effect is usually a futile exercise. Pain is depression and depression is pain.

One learns quickly in dealing with patients that not all pain is the consequence of serious disease. Every day, healthy persons of all ages have pains that must be taken as part of normal sensory experience. To mention a few, there are the "growing pains" of presumed bone and joint origin of children; the momentary shock-like pains over an eye or in the temporal or occipital regions ("ice-pick" pain), which strike with such suddenness as to raise the suspicion of a ruptured intracranial aneurysm; inexplicable split-second jabs of pain elsewhere; the more persistent ache in the shoulder, hip, or extremity that subsides spontaneously or in response to a change in position; the fluctuant precordial discomfort of gastrointestinal origin, which conjures up fear of cardiac disease; and the breathtaking "stitch in the side" caused by intercostal or diaphragmatic cramp during exercise. These "normal pains," as they may be called, tend to be brief and to depart as obscurely as they came. Such pains come to notice only when elicited by an inquiring physician or when experienced by a patient given to worry and introspection. They must always be distinguished from the pain of disease.

Whenever pain—by its intensity, duration, and the circumstances of its occurrence—appears to be abnormal or when it constitutes the chief complaint or one of the principal symptoms, the physician must attempt to reach a tentative decision as to its mechanism and cause. This is accomplished by a thorough interrogation of the patient, with the physician carefully seeking out the main characteristics of the pain in terms of the following:

• Location
• Mode and time of onset
• Associated features, e.g., nausea, muscle spasm
• Quality and time-intensity attributes
• Duration
• Severity
• Provoking and relieving factors

Knowledge of these factors in every common disease is the lore of medicine. The severity of pain is often difficult to assess. Extreme degrees of pain are betrayed by the patient's demeanor but lesser degrees can be roughly estimated by the extent to which the pain has interfered with the patient's sleep, work, and other activities, or by the patient's need for bed rest. Some physicians find it helpful, particularly in gauging the effects of analgesic agents, to use a "pain scale," i.e., to have the patient rate the intensity of his pain on a scale of zero (no pain) to 10 (worst pain) or to mark it on a line (the Visual Analog Pain Scale). It has been our experience that this effort to quantify pain is often unhelpful to the neurological analysis as patients rarely rate pain as trivial, when they have already decided to consult a physician about the problem. For most patients, pain that necessitates medical consultation is, by definition, severe. This general approach is put to use every day in the practice of general medicine. Together with the physical examination, including maneuvers designed to reproduce and relieve the pain and ancillary diagnostic procedures, it enables the physician to identify the source of most pains and the diseases of which they are a part. Whether the earlier mentioned functional imaging techniques will offer an additional tool to evaluate pain remains to be determined.

Once the pains caused by the more common and readily recognized diseases of each organ system are eliminated, there remain a significant number of chronic pains that fall into one of four categories: (1) pain from an obscure medical disease, the nature of which has not yet been disclosed by diagnostic procedures; (2) pain associated with disease of the central or peripheral nervous system (i.e., neurogenic, or neuropathic pain); (3) pain associated with psychiatric disease; and (4) pain of unknown cause.

Pain Caused by Undiagnosed Medical Disease

Here the source of the pain is usually in a bodily organ and is caused by a lesion that irritates and destroys nerve endings. Consequently, the term nociceptive pain is often used even though it is ambiguous. It usually means an involvement of structures bearing the origin of pain fibers. Cancer is the most frequent example. Osseous metastases, tumors of the kidney, pancreas, or liver, peritoneal tumor implants, invasion of retroperitoneal tissues or the hilum of the lung, and infiltration of nerves of the brachial or lumbosacral plexuses can be extremely painful, and the origin of the pain may remain obscure for a long time. Sometimes it is necessary to repeat all diagnostic procedures after an interval of a few months, even though at first they were negative. From experience one learns to be cautious about reaching a diagnosis from insufficient data. Treatment in the meantime is directed to the relief of pain, at the same time instilling in the patient a need to cooperate with a program of expectant observation.

Neurogenic, or Neuropathic Pain

These terms are generally used interchangeably to designate pain that arises from direct stimulation of nervous tissue itself, central or peripheral, exclusive of pain as a consequence of stimulation of C fibers by lesions of other bodily structures (i.e., the nociceptive pain described above). This category comprises a variety of disorders involving single and multiple nerves, notably trigeminal neuralgia and those caused by herpes zoster, diabetes, and trauma; neuromas and neurofibromas, a number of polyneuropathies of diverse type; root irritation, e.g., from a prolapsed disc; spinal arachnoiditis and spinal cord injuries; the thalamic pain syndrome of Dejerine-Roussy; and, rarely, parietal lobe infarction such as the ones described by Schmahmann and Leifer. As a rule, lesions of the cerebral cortex and white matter are associated not with pain but with hypalgesia. Schott (1995) reviewed the clinical features that characterize central pain. Particular diseases giving rise to neuropathic pain are considered in their appropriate chapters but the following remarks are of a general nature, applicable to all of the painful states that compose this group.

The sensations that characterize neuropathic pain vary and are often multiple—burning, gnawing, aching, and shooting or lancinating qualities are described. There is an almost invariable association with one or more of the symptoms of hyperesthesia, hyperalgesia, allodynia, and hyperpathia (see above). The abnormal sensations coexist in many cases with a sensory deficit and local autonomic dysfunction. Furthermore, the pain generally responds poorly to treatment, including the administration of opioid medications.

Peripheral Nerve Pain

Painful states that fall into this category are in most cases related to disease of the peripheral nerves, and it is to pain from this source that the term neuropathic is more strictly applicable. Pain states of peripheral nerve origin far outnumber those caused by spinal cord, brainstem, thalamic, and cerebral disease. Although the pain is localized to a sensory territory supplied by a nerve, plexus or nerve root, it often radiates to adjacent areas. Sometimes the onset of pain is immediate on receipt of injury; more often it appears at some point during the evolution or recession of the disorder. The disease of the nerve may be obvious, expressed by the usual sensory, motor, reflex, and autonomic changes, or these changes may be undetectable by standard tests. In the latter case, the term neuralgia is used.

The postulated mechanisms of peripheral nerve pain are diverse and differ from those of central diseases. Some of the major ideas were mentioned in the earlier section on chronic pain. One mechanism is denervation hypersensitivity, first described by Walter Cannon. He noted that when a group of neurons is deprived of its natural innervation, they become hyperactive. Others point to a reduced density of certain types of fibers in nerves supplying a causalgic zone as the basis of the burning pain but the comparison of the density of nerves from painful and nonpainful neuropathies has not proved to be consistently different. For example, Dyck and colleagues, in a study of painful versus nonpainful axonal neuropathies, concluded that there was no difference between them in terms of the type of fiber degeneration. Also, the occurrence of ectopic impulse generation all along the surface of injured axons and the possibility of ephaptic activation of unsheathed axons seem applicable particularly to some causalgic states. Stimulation of the nervi nervorum of larger nerves by an expanding intraneural lesion or a vascular change was postulated by Asbury and Fields as the mechanism of nerve trunk pain. The sprouting of adrenergic sympathetic axons in response to nerve injury has already been mentioned and is an ostensible explanation for the abolition of causalgic pain by sympathetic blockade. This has given rise to the term sympathetically sustained pain for some cases of causalgia, as discussed below.

Regenerating axonal sprouts, as in a neuroma, are also hypersensitive to mechanical stimuli. On a molecular level, it has been shown that voltage-gated sodium channels accumulate at the site of a neuroma and all along the axon after nerve injury, and that this gives rise to ectopic and spontaneous activity of the sensory nerve cell and its axon. Such firing has been demonstrated in humans after nerve injury. This mechanism is concordant with the relief of neurogenic pain by sodium channel-blocking anti-epileptic drugs. Spontaneous activity in nociceptive C fibers is thought to give rise to burning pain; firing of large myelinated A fibers is believed to produce dysesthetic pain induced by tactile stimuli. The abnormal response to stimulation is also influenced by sensitization of central pain pathways, probably in the dorsal horns of the spinal cord, as outlined in the review by Woolf and Mannion. Hyperalgesia and allodynia are thought to result from such a spinal cord mechanism. Several observations have been made regarding the neurochemical mechanisms that might underlie these changes but none provides a consistent explanation. Possibly more than one of these mechanisms is operative in a given peripheral nerve disease.

Evidence that the sodium channel can generate neural pain is given by the extraordinary disease "paroxysmal extreme pain disorder" also known as "familial rectal pain syndrome." Here, a mutation of the sodium channel gene, SCN9A, leads to the early onset of paroxysmal autonomic changes and attacks of excruciating deep burning pain in the rectum, eye, or jaw, or diffusely, as described by Fertleman and coworkers. Similar but more diffuse painful states such as erythromelalgia and paroxysmal extreme pain disorder are being uncovered that are predicated on similar voltage-gated sodium channel mutations and more impressively, by the congenital absence of the ability to experience pain due to a loss of function mutation in a sodium channel gene and a mutation in the tyrosine kinase receptor gene. Fischer and Waxman provide a summary of the mutations in the sodium channel gene and their clinical presentations.

Causalgia and Reflex Sympathetic Dystrophy (Complex Regional Pain Syndrome)

Causalgia is the name that Weir Mitchell applied to a rare (except in time of war) type of peripheral neuralgia consequent upon trauma, with partial interruption of the median or ulnar nerve and, less often, the sciatic or peroneal nerve (see also "Chronic Pain" further on and discussion in Chap. 46). It is characterized by persistent, severe pain in the hand or foot, most pronounced in the digits, palm of the hand, or sole. The pain has a burning quality and frequently radiates beyond the territory of the injured nerve. The painful parts are exquisitely sensitive to contact, so the patient cannot bear the pressure of clothing or drafts of air; even ambient heat, cold, noise, or emotional stimuli intensify the causalgic symptoms. The affected extremity is kept protected and immobile, often wrapped in a cloth moistened with cool water. Sudomotor, vasomotor, and, later, trophic abnormalities are usual accompaniments of the pain. The skin of the affected part is moist and warm or cool and soon becomes shiny and smooth, at times scaly, devoid of hair, and discolored.

A number of theories have been proposed to explain the causalgic syndrome. For many years it was attributed to a short-circuiting of impulses, the result of an artificial connection between efferent sympathetic and somatic afferent pain fibers at the point of the nerve injury. The demonstration that causalgic pain could be abolished by depletion of neurotransmitters at sympathetic adrenergic endings shifted the presumed site of sympathetic-afferent interaction to the nerve terminals and suggested that the abnormal cross-excitation is chemical rather than electrical in nature. Another possible explanation is that an abnormal adrenergic sensitivity develops in injured nociceptors and that circulating or locally secreted sympathetic neurotransmitters trigger the painful afferent activity. Another theory holds that a sustained period of bombardment by sensory pain impulses from one region results in the sensitization of central sensory structures. "True causalgia" of this type can be counted on to respond favorably, if only temporarily, to procaine block of the appropriate sympathetic ganglia and, for a longer time, to regional sympathectomy. Prolonged cooling and the intravenous injection of guanethidine, a sympathetic-blocking drug, into the affected limb (with the venous return blocked for several minutes) may alleviate the pain for days or longer. Epidural infusions, particularly of analgesics or ketamine, intravenous infusion of bisphosphonates, and spinal cord stimulators are other forms of treatment (see Kemler et al). The roles of the central and sympathetic nervous systems in causalgic pain have been critically reviewed by Schott and by Schwartzman and McLellan.

Increasingly, reflex sympathetic dystrophy has been reported after limb and bone injury but in the absence of evident damage to adjacent nerves. Oaklander and Fields have speculated that this is due to an induced small fiber neuropathy; limited confirmation of this notion is given by those authors.

Recent investigations have begun to define the molecular changes that occur in sensory neurons and the spinal cord in cases of chronic pain of this type. Alterations in N-methyl-D-aspartate (NMDA) receptors, induction of cyclooxygenase and prostaglandin synthesis, and changes in gabanergic inhibition in the dorsal horns are all implicated (Woolf).

The term causalgia is, in our view, best reserved for the syndrome described above—i.e., persistent burning pain and local abnormalities of autonomic innervation from trauma to a major nerve in an extremity. Some neurologists use "causalgia" to describe only the burning feature of pain due to partial nerve injury. Others have applied the term to a wide range of conditions that are characterized by persistent burning pain but have only an inconstant association with sudomotor, vasomotor, and trophic changes and an unpredictable response to sympathetic blockade. These latter states, which have been described under a plethora of terms (e.g., complex regional pain syndrome type 2, Sudeck atrophy of bone, minor causalgia, shoulder–hand syndrome, algodystrophy, or algoneurodystrophy), may follow non-traumatic lesions of the peripheral nerves or even lesions of the CNS ("mimocausalgia").

We have no explanation for the so-called causalgia–dystonia syndrome (Bhatia et al) in which a fixed dystonic posture is engrafted on a site of causalgic pain. The clinical features of both the causalgic and dystonic elements of the syndrome have been somewhat unusual in the cases reported. The degree of injury was often trivial or nonexistent and no signs of a neuropathic lesion were evident. Remarkably, both the causalgia and dystonia spread from their initial sites to widely disparate parts of the limbs and body. The syndrome did not respond to any form of treatment, although some patients recovered spontaneously. Another interesting type of causalgia and reflex sympathetic dystrophy follows deep venous thrombosis in a leg and had in the literature been recorded as "algodystrophy." It may be similar to the left shoulder and hand changes that come on months after a myocardial infarction ("shoulder–hand syndrome").

The treatment of reflex sympathetic dystrophy is largely unsatisfactory, although a certain degree of improvement can be expected if treatment is started early and the limb is mobilized. The options for treatment are discussed further on.

Central Neurogenic Pain

There are several configurations of central lesions that damage the sensory system and produce severe pain. Deafferentation of secondary neurons in the posterior horns or of sensory ganglion cells that terminate on them may cause the deafferented cells to become continuously active and, if stimulated by a microelectrode, to reproduce pain. In the patient whose spinal cord has been transected, there may be intolerable pain in regions below the level of the lesion. It may be exacerbated or provoked by movement, fatigue, or emotion and projected to areas disconnected from suprasegmental structures (akin to the phantom pain in the missing part of an amputated limb). Here, and in the rare cases of intractable pain with lateral medullary or pontine lesions, loss of the descending inhibitory systems seems a likely explanation. This may also explain the pain of the Dejerine-Roussy thalamic syndrome described in Chap. 9. Altered sensitivity and hyperactivity of central neurons are alternative possibilities.

Further details concerning the subject of neuropathic pain can be found in the older but still informative writings of Scadding and of Woolf and Mannion.

Pain in Association with Psychiatric Diseases

It is not unusual for patients with depression to have pain as a dominant symptom. As emphasized previously, most patients with chronic pain of all types are depressed. Wells and colleagues, in a survey of a large number of depressed and chronic pain patients, have corroborated this clinical impression. Fields has elaborated a theoretical explanation of the overlap of pain and depression. In such cases, one is faced with an extremely difficult clinical problem—that of determining whether a depressive state is primary or secondary. Complaints of weakness and fatigue, depression, anxiety, insomnia, nervousness, irritability, palpitations, etc., are woven into the clinical syndrome, attesting to the prominence of a psychiatric disorder. In some instances the diagnostic criteria for depression cited in Chap. 52 provide some insight, but in others it is impossible to make this determination and may not be necessary as depression and pain are so often coincident. Empiric treatment with antidepressant medication or, failing this, with electroconvulsive therapy is one way out of the dilemma.

Intractable pain may also be the leading symptom of both somatization and conversion reactions. Experienced physicians are familiar with the patient who has undergone multiple surgical procedures to address painful complaints (so-called Briquet disease). The recognition and management of this group of disorders are discussed in Chap. 51.

The desire for compensation (e.g. workman's compensation, disability status) is usually colored by persistent complaints of headaches, neck pain (whiplash injuries), low back pain, and other painful conditions. The question of ruptured disc is often raised, and laminectomy and spinal fusion may be performed (sometimes more than once) on the basis of dubious radiologic findings. Long delay in the settlement of litigation, allegedly to determine the seriousness of the injury, only enhances the symptoms and prolongs the disability. The medical and legal professions have no certain approach to such problems and often work at cross-purposes. We have found that a frank, objective appraisal of the injury, an assessment of any psychiatric problem, and encouragement to settle the legal claims as quickly as possible work in the best interests of all concerned. Although hypersuggestibility and relief of pain by placebos may reinforce the physician's belief that there is a prominent factor of hysteria or malingering (see Chap. 51), such data are difficult to interpret.

The possibility of drug addiction as a motivation for visiting the physician and reporting severe pain should be addressed. It is impossible to assess pain in addicted individuals, for their complaints are woven into their need for medication. Temperament and mood should be evaluated carefully; the physician must remember that the depressed patient often denies feeling dysphoric and may even occasionally smile. The use of alcohol to self-medicate for pain usually indicates a depressive illness or lifelong alcohol dependence. When no medical, neurologic, or psychiatric disease can be established, one may be resigned to managing the painful state by the use of nonnarcotic medications and periodic clinical reevaluations. Such a course, though not altogether satisfactory, is preferable to prescribing excessive opioids or subjecting the patient to ablative surgery.

Chronic Pain of Indeterminate Cause

Pain in the thorax, abdomen, flank, back, face, head, or other part that cannot be traced to any visceral abnormality can create challenging clinical problems. In most cases, obscure neurologic sources, such as a spinal cord tumor and neuroma, have been excluded by repeated examinations and imaging procedures. A psychiatric disorder to which the patient's symptoms and behavior might be attributed cannot be discerned. Yet the patient complains continuously of pain, is disabled, and spends a great deal of effort and resources seeking medical aid.

In such a circumstance, some physicians and surgeons, rather than concede their helplessness, may resort to extreme measures, such as exploratory thoracotomy, laparotomy, or laminectomy. Or they may injudiciously attempt to alleviate the pain and avoid drug addiction by severing roots and spinal tracts, often with the result that the pain moves to an adjacent segment or to the other side of the body.

This type of patient benefits from being seen more than once by the physician. All the medical facts should be reviewed and the clinical and laboratory examinations repeated if some time has elapsed since they were last done. Tumors in the hilum of the lung or mediastinum; in the retropharyngeal, retroperitoneal, and paravertebral spaces; or in the uterus, testicle, kidney, or prostate pose a special difficulty in diagnosis, often being undetected for many months. More than once, we have seen a patient for months before a kidney or pancreatic tumor became apparent. Neurofibroma causing pain in an unusual site, such as one side of the rectum or vagina, is another type of tumor that may defy diagnosis for a long time. Truly neurogenic pain is almost invariably accompanied by alterations in cutaneous sensation and other neurologic signs, the finding of which facilitates diagnosis; however, the appearance of the neurologic signs may be delayed—for example, in brachial neuritis.

Because of the complexity and difficulty in diagnosis and treatment of chronic pain, most medical centers have found it advisable to establish pain clinics. Here a staff of internists, anesthesiologists, neurologists, neurosurgeons, and psychiatrists can review each patient in terms of drug dependence, neurologic disease, and psychiatric problems. Success is achieved by treating each aspect of chronic pain, and addressing the individual's problem rather than treating it generically with emphasis on increasing the patient's tolerance of pain by means of biofeedback, meditation, and related techniques; by using special analgesic procedures (discussed later in the chapter); by establishing a regimen of pain medication that does not lead to a rebound exaggeration of pain between doses; and by controlling depressive illness.

Rare and Unusual Disturbances of Pain Perception

Lesions of the parietooccipital regions of one cerebral hemisphere sometimes have peculiar effects on the patient's capacity to feel and react to pain. Under the title of pain hemiagnosia, Hecaen and Ajuriaguerra described several cases of left-sided paralysis from a right parietal lesion, which, at the same time, rendered the patient hypersensitive to noxious stimuli. When pinched on the affected side, the patient, after a delay, became agitated, moaned, and seemed distressed but made no effort to fend off the painful stimulus with the other hand or to withdraw from it. In contrast, if the good side was pinched, the patient reacted normally and moved the normal hand at once to the site of the stimulus to remove it. The motor responses seemed no longer to be guided by sensory information from one side of the body.

There are also two varieties of rare individuals who from birth are totally indifferent to pain coupled with anhidrosis ("congenital insensitivity to pain") or are incapable of feeling pain ("universal analgesia"). The former have been found by Indo and colleagues to have a mutation in the a neural tyrosine kinase receptor, a nerve growth factor receptor; those in the second group suffer from either a congenital lack of pain neurons in dorsal root ganglia, or to a mutation in the sodium channel discussed earlier. A similar loss of pain sensibility is encountered in the Riley-Day syndrome (congenital dysautonomia, see Chap. 26).

The phenomenon of asymbolia for pain is another rare and unusual condition wherein the patient, although capable of distinguishing the different types of pain stimuli from one another and from touch, is said to make none of the usual emotional, motor, or verbal responses to pain. The patient seems totally unaware of the painful or hurtful nature of stimuli delivered to any part of the body, whether on one side or the other. The current interpretation of asymbolia for pain is that it represents a particular type of agnosia (analgognosia) or apractagnosia (see Chap. 22), in which the person loses his ability to adapt his emotional, motor, and verbal actions to the consciousness of a nociceptive impression. Pre-frontal lobe lesions from stroke, trauma, tumor, or in former times frontal lobotomy, can produce a version of this syndrome.

Treatment of Intractable Pain

Once the nature of the patient's pain and underlying disease has been determined, therapy must include some type of pain control. Initially, of course, attention is directed to the underlying disease with the idea of eliminating the source of the pain by appropriate medical, surgical, or radiotherapeutic measures. When the primary disease is not treatable, the physician should, if time and the circumstances permit, attempt to use the milder measures for pain relief first—for example, nonnarcotic analgesics and antidepressants or anti-epileptic drugs before resorting to narcotics, local nerve blocks or contemplating surgical approaches for pain relief. Not all situations allow this graduated approach, and large doses of narcotics may be required early in the course of illness—for example, to treat the pain of visceral and bone cancer. The same measured strategy is appropriate in the treatment of neuropathic pain and of pain of unclear origin except that one generally stops short of ablative procedures that irrevocably damage nerves or parts of the central nervous system.

The field of pain relief has been changed by the introduction of analgesic procedures that block nerves, alter neural conduction, or administer conventional medications in new ways. These have become the province of pain clinics and hospital pain services usually led by departments of anesthesiology. In addition, a number of special procedures or unique medications are highly effective for pain relief but are unique to specific situations. These include certain forms of headache and limb pain (temporal arteritis and polymyalgia rheumatica treated with corticosteroids, or migraine relief with "triptan" drugs); trigeminal neuralgia, which may be relieved by microvascular decompression of a branch of the basilar artery or by controlled damage of the gasserian ganglion; and painful dystonic disorders that are relieved by the injection of botulinum toxin. Special procedures that have been devised to treat various forms of spinal back pain fall into the same category. The following discussion provides some guidance for the physician who is asked to undertake or participate in the treatment of chronic pain or of neuropathic pain.

Narcotics (Opioids and Opiates)

A useful way in which to undertake the management of chronic pain that affects several parts of the body, as in the patient with metastases, is with codeine or oxycodone taken together with aspirin, acetaminophen, or another nonsteroidal antiinflammatory drug (NSAID), or tramadol. The analgesic effects of these types of drugs are additive, which is not the case when narcotics are combined with diazepam or phenothiazine. Antidepressants and antiepileptic drugs, as discussed further on, may have a beneficial effect on pain even in the absence of overt depression. This is true particularly in cases of neuropathic pain (painful polyneuropathy and some types of radicular pain). Sometimes these non-narcotic agents may, in themselves or in combination with these treatment modalities, be sufficient to control the patient's pain and the use of narcotics can then be kept in reserve.

Should the foregoing measures prove to be ineffective, one must turn to narcotic agents. Methadone and levorphanol are sometimes useful drugs with which to begin, because of their effectiveness by mouth and the relatively slow development of tolerance. Some pain clinics prefer the use of shorter-acting drugs such as oxycodone, given more frequently through the day. The oral route should be used whenever possible, as it is more comfortable for the patient than the parenteral route. Also, the oral route is associated with fewer side effects except for nausea and vomiting, which tend to be worse than with parenteral administration. Should the latter become necessary, one must be aware of the ratios of oral-to-parenteral dosages required to produce equivalent analgesia. The main medications used in the treatment of pain are summarized in Table 8-3.

If oral medication fails to control the pain, the parenteral administration of codeine or more potent opioids becomes necessary. One may begin with methadone, dihydromorphine (Dilaudid), or levorphanol, given at intervals of 4 to 6 h because of their relatively long duration of action (particularly in comparison to meperidine). Alternatively, one may first resort to the use of transdermal patches of drugs such as fentanyl, which provide relief for 24 to 72 h and which we have found particularly useful in the treatment of pain from brachial or lumbosacral plexus invasion by tumor and of painful neuropathies such as those caused by diabetes and systemic amyloidosis. Long-acting morphine preparations are useful alternatives.

Should long-continued injections of opiates become necessary, the optimal dose for the relief of pain should be established and the drug then given at regular intervals around the clock, rather than "as needed." The administration of morphine (and other narcotics) in this way represents a laudable shift in attitude among physicians. For many years it was widely believed that the drug should be given in the smallest possible doses, spaced as far apart as possible, and repeated only when severe pain reasserted itself. It has become clear that this approach results in unnecessary discomfort and, in the end, the need to use larger doses. Most physicians now realize that the fear of creating narcotic dependence and the expected phenomenon of increasing tolerance must be balanced against the overriding need to relieve pain. The most pernicious aspect of addiction, that of compulsive drug-seeking behavior with its attendant sociopathic behaviors, occurs only rarely in this setting and usually in patients with a previous history of addiction or alcoholism, with depression as the primary problem, or with certain characterologic disorders that have been loosely referred to as "addiction proneness." Even in patients with severe acute or postoperative pain, the best results are obtained by allowing the patient to determine the dose and frequency of intravenous medication, a method known as patient-controlled analgesia (PCA). Again, the danger of producing addiction is minimal.

Guidelines for the use of orally and parenterally administered opioids for cancer-related pain are contained in the article of Cherny and Foley and in the publication of the U.S. Department of Health and Human Services, which unfortunately, is no longer easily obtained.

The approach outlined above conforms to our understanding about pain-control mechanisms. Aspirin and other NSAIDs are believed to prevent the activation of nociceptors by inhibiting the synthesis of prostaglandins in skin, joints, viscera, and other structures in the peripheral nervous system. Morphine and meperidine given orally, parenterally, or intrathecally presumably produce analgesia by acting as "false" neurotransmitters at opiate receptor sites in the posterior horns of the spinal cord—sites that are normally activated by endogenous opioid peptides. The separate sites of action of NSAIDs and opioids provide an explanation for the therapeutic usefulness of combining these drugs. Opioids not only act directly on the central pain-conducting sensory systems but also exert a powerful action on the affective component of pain.

Supplemental Medications for the Treatment of Pain

Tricyclic antidepressants, especially the methylated forms (imipramine, amitriptyline, and doxepin), block serotonin reuptake and thus enhance the action of this neurotransmitter at synapses and putatively facilitate the action of the intrinsic opiate analgesic system. As a general rule, relief is afforded with tricyclic antidepressants in the equivalent dose range of 75 to 125 mg daily of amitriptyline, but little benefit accrues with higher amounts. The specific serotonin reuptake inhibitors (SSRI) antidepressants seem not to be as effective for the treatment of chronic neuropathic pain (see review by McQuay and colleagues) but these agents have not yet been extensively investigated in this clinical condition.

Antiepileptic drugs (AEDs) have a beneficial effect on many central and peripheral neuropathic pain syndromes but are generally less effective for causalgic pain caused by partial injury of a peripheral nerve. The mode of action of phenytoin, carbamazepine, gabapentin, levetiracetam, and other AEDs in suppressing the lancinating pains of tic douloureux and certain polyneuropathies, as well as pain after spinal cord injury and myelitis, is not fully understood, but they are widely used. Their action has been attributed to the blocking of sodium channels on axons, thereby reducing the evoked and spontaneous activity in nerve fibers. The full explanation is certainly more complex and related to separate central and peripheral sites, as summarized by Jensen. Often, large doses must be utilized—for example, more than 2,400 mg per day for gabapentin for full effect—but the soporific and ataxic effects may be poorly tolerated.

Most often a combination of medications is used for the treatment of intractable chronic pain. A common combination is the addition of gabapentin to an opioid such as morphine, and perhaps not surprisingly, this was superior to either drug alone in a crossover trial in patients with postherpetic neuralgia and diabetic neuropathy conducted by Gilron and colleagues but at the expense of side effects and lower tolerated doses of both drugs.

Table 8-3 summarizes the main analgesics (nonnarcotic and narcotic), antiepileptics, and antidepressant drugs in the management of chronic pain.

Treatment of Cancer Pain

If the patient is ridden with neoplastic disease and will not live longer than a few weeks or months and has widespread pain, surgical measures are usually not advisable. Pain from widespread osseous metastases, even in patients with hormone-insensitive tumors, may be relieved by radiation therapy or by hypophysectomy. If these are not feasible, opioid medications are required and are effective, but they must be prescribed in adequate doses. Many patients prefer transdermal fentanyl to oral or intravenous agents. Usually, nerve section is not a satisfactory way of relieving restricted pain of the trunk and limbs because the overlap of adjacent nerves prevents complete denervation. Other procedures to be considered are the regional delivery of narcotic analogues, such as fentanyl or ketamine by means of an external pump and a catheter that is implanted percutaneously in the epidural space in proximity to the dorsal nerve roots of the affected region; this device can be used safely at home.

Treatment of Neuropathic Pain

The treatment of pain induced by nerve root or intrinsic peripheral nerve disease is a challenge for the neurologist and employs several techniques that are generally administered by an anesthesiologist. One usually resorts first to one of the antiepileptic drugs discussed earlier and listed in Table 8-3. The next simplest treatments are topical; if the pain is regional and has a predominantly burning quality, capsaicin cream can be applied locally, care being taken to avoid contact with the eyes and mouth. The irritative effect of this chemical, which releases substance P, seems in some cases to mute the pain. We have also had success with several concoctions of "eutectic" mixtures of local anesthetic (EMLA) creams or the simpler lidocaine gel with ketorolac, gabapentin, and other medications; these are applied directly to the affected area, usually the feet, in the morning and evening. Concoctions such as topical Ketamine mixed in soy lecithin to produce a gel with drug concentration of 5 mg/ml, have been reportedly useful in treating post herpetic neuralgia according to Quan and associates in a small randomized trial. Aspirin mixed with chloroform in cold cream is said to be very effective in the topical treatment of post herpetic neuralgia, as suggested by King (see Chap. 10). These preparations may provide considerable relief in postherpetic neuralgia and some painful peripheral neuropathies, but they are totally ineffective in others.

Several types of spinal injections, including epidural, root, and facet blocks, have long been used for the treatment of pain. Injections of epidural corticosteroids or mixtures of analgesics and steroids are helpful in selected cases of lumbar or thoracic nerve root pain, and occasionally in painful peripheral neuropathy, but precise criteria for the use of this measure are not well established. Several studies do not support a beneficial effect but there is little doubt, in our view, that quite a few patients are helped, if only for several days or weeks (see Chap. 11). Nerve root blocks with lidocaine or with longer-acting local anesthetics are sometimes helpful in establishing the precise source of radicular pain. Their main therapeutic use in our experience has been for thoracic radiculitis from shingles, chest wall pain after thoracotomy, and diabetic radiculopathy. Similar local injections are used in the treatment of occipital neuralgia. Injection of analgesic compounds into and around facet joints and the extension of this procedure, radiofrequency ablation of the small nerves that innervate the joint, are as controversial as epidural injections, with most studies failing to find a consistent benefit. Despite these drawbacks, we have found both of these approaches very useful when pain can be traced to a derangement of these joints, as discussed in Chap. 11.

The intravenous infusion of lidocaine has a brief beneficial effect on many types of pain, including neuropathic varieties, localized headaches, trigeminal neuralgia, and other facial pains; it is said to be useful in predicting the response to longer-acting agents such as mexiletine, its oral analogue, although this relationship has been erratic in our experience (see Table 8-3). Mexiletine is given in an initial dose of 150 mg per day and slowly increased to a maximum of 300 mg three times daily; it should be used very cautiously in patients with heart block and has fallen very much out of favor in many centers, partly due to cardiac conduction abnormalities during and after the infusion.

Reducing sympathetic activity within somatic nerves by direct injection of the sympathetic ganglia in affected regions of the body (stellate ganglion for arm pain and lumbar ganglia for leg pain) has met with mixed success in neuropathic pain, including that of causalgia and reflex sympathetic dystrophy. A variant of this technique uses regional intravenous infusion of a sympathetic-blocking drug (bretylium, guanethidine, reserpine) into a limb that is isolated from the systemic circulation by the use of a tourniquet. This is known as a "Bier block," after the developer of regional anesthesia for single-limb surgery. These techniques, as well as the administration of clonidine by several routes and the intravenous infusion of the adrenergic blocker phentolamine, is predicated on the concept of "sympathetically sustained pain," meaning pain that is mediated by the interaction of sympathetic and pain nerve fibers or by the sprouting of adrenergic axons in partially damaged nerves. These forms of treatment have been under study for many decades and have given variable results but the most consistent responses to regional sympathetic blockade are obtained in cases of true causalgia resulting from partial injury of a single nerve.

A number of other treatments have proven successful in some patients with reflex sympathetic dystrophy and other neuropathic pains but the clinician should be cautious about their chances of success over the long run. A novel one of these has been the use of bisphosphonates (pamidronate, alendronate), which have been beneficial in painful disorders of bone, such as Paget disease and metastatic bone lesions. It is theorized that this class of drug reverses the bone loss consequent to reflex sympathetic dystrophy but how this relates to pain control is unclear (Schott, 1997). Electrical stimulation of the posterior columns of the spinal cord by an implanted device, as discussed below, has become popular. Another treatment of last resort is the intravenous or epidural infusion of drugs such as ketamine; sometimes this has a lasting effect on causalgic pain.

The approaches enumerated here are usually undertaken in sequence; a combination of drugs—such as gabapentin, narcotics, and clonidine—in addition to anesthetic techniques—is usually required. The ongoing attention and support of the neurologist often becomes the patient's mainstay. Further references can be found in the thorough review by Katz.

Ablative Surgery in the Control of Pain

It is our considered opinion that a program of medical therapy should always precede ablative surgical measures. Only when a variety of analgesic medications (including opioids) and only when certain practical measures, such as regional analgesia or anesthesia, have completely failed, should one turn to neurosurgical procedures. Also, one should be very cautious in suggesting a procedure of last resort for pain that has no established cause as, for example, limb pain that has been incorrectly identified as causalgic because of a burning component but where there has been no nerve injury.

The least-destructive procedure consists of surgical exploration for a neuroma if a prior injury or operation may have partially sectioned a peripheral nerve. Magnetic resonance imaging of the region should be performed first and will demonstrate most such lesions, but we are uncertain if all small neuromas are visualized, and it is this ambiguity that justifies exploration. Another nondestructive procedure is implantation of a spinal electrical stimulator, usually adjacent to the posterior columns. This procedure, in which there is now a resurgence of interest, has afforded only incomplete relief in our patients and may be difficult to maintain in place. However Kemler and colleagues found a sustained reduction in pain intensity and an improved quality of life in patients with intractable reflex sympathetic dystrophy, even after 2 years in a randomized trial. It is clear that careful selection of patients is the best assurance of a good outcome. We can add from experience with our patients that a temporary trial of the stimulator is advisable before committing to its permanent use. The ill-advised use of nerve section and dorsal rhizotomy as definitive measures for the relief of regional pain was discussed above under "Treatment of Intractable Pain."

Spinothalamic tractotomy, in which the anterior half of the spinal cord on one side is sectioned at an upper thoracic level, effectively relieves pain in the opposite leg and lower trunk. This may be done as an open operation or as a transcutaneous procedure in which a radiofrequency lesion is produced by an electrode. The analgesia and thermoanesthesia may last a year or longer, after which the level of analgesia tends to descend and the pain tends to return. Bilateral tractotomy is also feasible but with greater risk of loss of sphincteric control and, at higher levels, of respiratory paralysis. Motor power is nearly always spared because of the position of the corticospinal tract in the posterior part of the lateral funiculus.

Pain in the arm, shoulder, and neck is more difficult to relieve surgically. High cervical transcutaneous cordotomy has been used successfully, with achievement of analgesia up to the chin. Commissural myelotomy by longitudinal incision of the anterior or posterior commissure of the spinal cord over many segments has also been performed, with variable success. Dorsal root entry zone (DREZ) lesions may relieve pain in the distribution of one or two nerve roots. Lateral medullary tractotomy is another possibility but must be carried almost to the mid-line to relieve cervical pain. The risks of this latter procedure and also of lateral mesencephalic tractotomy (which may actually produce pain) are so great that neurosurgeons have abandoned these operations.

Stereotactic surgery on the thalamus for one-sided chronic pain is still used in a few centers and the results have been instructive. Lesions placed in the ventroposterior nucleus are said to diminish pain and thermal sensation over the contralateral side of the body while leaving the patient with all the misery or affective experience of pain; lesions in the intralaminar or parafascicular-centromedian nuclei relieve the painful state without altering sensation (Mark). Because these procedures have not yielded predictable benefits to the patient, they are now seldom used. The same unpredictability pertains to cortical ablations. Patients in whom a severe depression of mood is associated with a chronic pain syndrome have been subjected to bilateral stereotactic cingulotomy or the equivalent—subcaudate tractotomy. A considerable degree of success has been claimed for these operations but the results are difficult to evaluate. Orbito-frontal leukotomy has been discarded because of the personality change that it produces.

Non-Medical Methods for the Treatment of Pain

Included under this heading are certain techniques such as biofeedback, meditation, imagery, acupuncture, spinal manipulation, as well as transcutaneous electrical stimulation. Among the most intriguing treatments has been mirror therapy in which the patient is instructed to perform movements in the painful arm while watching the same moves in a mirror, made by the unaffected arm. The majority of patients in one blinded trial benefitted in terms of pain and mobility, but this study by Cacchio and coworkers included only patients with strokes and paretic limbs, not those with peripheral nerve injury. Each of these may be of value in the context of a comprehensive pain management program, usually conducted in a pain clinic as a means of providing relief from pain and suffering, reducing anxiety, and diverting the patient's attention, even if only temporarily, from the painful body part. Attempts to quantify the benefits of these techniques—judged usually by a reduction of drug dosage—have given mixed or negative results. Nevertheless, it is unwise for physicians to dismiss these methods, as well-motivated and apparently psychologically stable persons have reported subjective improvement with one or another of these methods and in the final analysis, this is what really matters. Conventional psychotherapy in combination with the use of medication and, at times, electroconvulsive therapy can be of benefit in the treatment of associated depressive symptoms, as discussed above (under "Pain in Association with Psychiatric Diseases") but it should not otherwise be expected to change the experience of pain.

Whatever treatment is undertaken, medical, procedural or surgical, the objective should be to allow and encourage increased use and mobilization of the affected limb or part, as success at this is most closely associated with relief of pain and reduced suffering.