Chapter 89. Mechanisms of Chronic Pain

1. Understanding the mechanisms underlying chronic pain requires knowledge of the neuroanatomy, neurochemistry, and neurophysiology of nociception and central pain processing.

2. Generation of pain hypersensitivity results from changes in the function, chemistry, and structure of both the peripheral and central nervous systems.

3. Pain can be categorized into the following broad etiologic groups: nociceptive pain is associated with an ongoing nociceptive stimulus and reflects minimal central modulation of the painful stimulus; inflammatory pain describes pain due to tissue inflammation; neuropathic pain results from injury of the peripheral or central nervous system; dysfunctional pain refers to pain due to abnormal functioning of the nervous system, despite the absence of an identified insult.

4. Nociception serves a protective function that is important from an evolutionary perspective. However, chronic pain—especially neuropathic pain and dysfunctional pain—do not serve such a protective role and result instead from a pathologic condition of the nervous system. In these circumstances, chronic pain is a disease.

5. As an injury heals, pain can dissipate with resolution of tissue injury and inflammation; alternatively, pain can persist and become independent of peripheral stimulation. The independence of ongoing stimulus reflects the marked pathologic changes that persistent pain effects on the nervous system.

6. Different mechanisms are responsible for different types of pain, and therefore the rational and successful treatment of pain requires diagnostic and therapeutic modalities that reflect the specific molecular mechanisms.

The major function of the pain system is to protect the body from injury. Its evolutionary importance is underscored by the disfiguring injuries and frequent early death of people who have the rare recessive condition congenital insensitivity to pain. The importance of pain from a clinical perspective led the Joint Commission on Accreditation of Healthcare Organizations to declare that pain level constitutes a fifth vital sign. Unlike other diagnostic signs, the pain score is subjective, thus making pain difficult to study clinically. The initial step in pain signaling, nociception, was defined by Sir Charles Sherrington a century ago as the sensory detection of a noxious event or a potentially harmful environmental stimulus.1 However, pain represents the integration of a wide range of inputs and modulations that span the neuraxis, from nociception in the distal periphery to complex processing of pain in the brain (Fig. 89-1). Consistent with this, psychologic trauma can cause pain that is perceived much in the same way as that associated with an injured extremity. And yet despite the similar experience of pain, the mechanisms underlying these 2 conditions must be different, at least in all but the highest-level representations of the pain. Indeed, we now know that virtually all chronic pain is fundamentally different from "nociception" in the sense of the simple detection of an acute noxious stimulus and its unmodulated transmission to the brain. Rather, chronic pain is the result of amplified and broadened inputs into a sensitized central system. This chapter focuses on mechanisms that generate and maintain pain.

In studying pain, clinicians and scientists have established taxonomies based on clinical presentation and animal models (Fig. 89-2). Nociceptive pain refers to pain in response to an ongoing high-threshold, noxious stimulus. Nociceptive pain ends when the stimulus ends, and does not involve additional short- or long-term changes in the pain receptors, higher-order signaling neurons, or other modulators. Inflammatory pain results from tissue damage and the associated release of inflammatory mediators that activate and sensitize nociceptors, and subsequently effect central changes that enhance pain transmission. Because of the sensitization of nociceptors, stimuli that are normally innocuous can elicit pain. The International Association for the Study of Pain (IASP) defines neuropathic pain as "pain initiated or caused by a primary lesion or dysfunction in the nervous system." We prefer to restrict the definition of neuropathic pain to a specific lesion in the nervous system that produces a sensitized pain state and loss of function, most often decreased sensation. As in inflammatory pain, neuropathic pain involves the sensitization of the afferent signaling machinery so that non-noxious stimuli elicit pain. However, despite the facilitated or enhanced nature of both nociceptive and neuropathic pain, as we will see, the mechanisms of the sensitization processes are different.2

Although inflammatory pain can play beneficial physiologic roles by preventing further tissue injury during healing, and neuropathic pain is associated with injury to the nervous system, dysfunctional pain refers to pain without a benefit to the organism and without a precipitant. Examples of dysfunctional pain conditions include fibromyalgia, interstitial cystitis, and irritable bowel syndrome. Although properties such as hyperalgesia may be present, there is no noxious stimulus that initiates the condition. Rather, the process is thought to result from a primary dysfunction of the nervous system.

One confounding issue arises from how the words that categorize pain conditions are used in the clinical and research arenas. Clinically, neuropathic pain refers to pain that has certain characteristic features (Table 89-1): allodynia, hyperalgesia, stimulus-independence (spontaneous pain), and evidence of sensory loss.3 Patients report allodynia, pain upon non-noxious stimulation such as light stroking of the skin or wind blowing on the face. Hyperalgesia, an increased response to a normally painful stimulus, is also often present. In many cases, the "revving up" of the central pain signaling machinery occurs to such an extent so as to produce spontaneous pain, that is, pain that occurs without any stimulus. There is also often evidence of a sensory deficit, and frequently the painful area extends beyond any region for which pain might be expected, based on the region of tissue damage or distribution innervated by an injured or severed nerve. Neuropathic pain can result from nerve injury in the periphery or central nervous system. The most common peripheral causes of neuropathic pain include metabolic disease such as diabetic neuropathy, infectious or postinfectious processes such as postherpetic neuropathy or human immunodeficiency virus (HIV) neuropathy, toxic neuropathy such as in chemotherapy-induced neuropathy, and paraprotein-related processes such as in monogammopathy of unknown significance, paraneoplastic neuropathy, or cryoglobulin-related neuropathy most often seen in hepatitis C.4,5 Central causes of neuropathic pain include spinal cord injury, ischemic or hemorrhagic stroke, multiple sclerosis and other causes of myelitis, tumors and abscesses, and structural abnormalities such as vascular malformations, syringomyelia, and syringobulbia.

In basic science research using animal models, neuropathic pain refers to pain due to an inflicted nerve injury, and inflammatory pain has been defined as pain resulting from an inflammatory stimulus. The benefit of such a distinction is that fundamentally different molecular mechanisms have been identified in each of the animal models of these 2 major forms of pain sensitization. However, the translation of this animal research to the clinical setting has been problematic: the degree to which mechanisms of nociceptive sensitization in animal models translate to mechanisms underlying chronic pain in humans remains unclear. This is highlighted by the virtual absence of novel human therapeutics for pain that have been derived from fundamental animal research.6 In addition, the clinical classification of a chronic pain condition into inflammatory or neuropathic is often ambiguous. For example, should a painful inflammatory neuropathy be considered neuropathic pain or inflammatory pain? At least some overlap exists among taxonomical categories, reflecting a degree of common underlying mechanisms of pain sensitization (although we focus on the distinctions) as well as an incomplete understanding of disease processes. For example, although the pathologic process in dysfunctional pain conditions such as fibromyalgia is presumed to reside in the central nervous system, a currently unidentified abnormality in the peripheral nervous system may contribute to the syndrome.

Although it is essential to understand the underlying mechanisms of basal and sensitized pain states, having a taxonomy of pain types can be helpful (Fig. 89-3). Several points can be made about the different types of pain. Nociceptive pain results solely from activation of nociceptors by high-threshold, noxious stimuli, and the sensation does not continue beyond the duration of the stimulus. No short- or long-term changes in the pain signaling machinery persist beyond the termination of the painful stimulus. In contrast, facilitated pain, whether due to inflammation associated with tissue damage (inflammatory pain), nerve injury (neuropathic pain), or primary nervous system dysfunction (dysfunctional pain), is enhanced; that is, the extent of the pain, whether with regard to intensity, duration, quality, or location, is heightened and expanded. As we will see, different molecular mechanisms underlie the facilitation in the different pain states.

We begin with a review of the anatomy of the nociceptive system.

Primary Afferents

Primary afferent neurons have peripheral terminals that detect external stimuli, cell bodies located in the dorsal root ganglia for perception from the body, and central terminals in the spinal cord. Aside from the unusual exception of trigeminal proprioceptive neurons, which have their cell bodies in the central nervous system, primary afferent neurons have their cell bodies in the peripheral nervous system: first-order neurons conveying perception from the face have cell bodies in the trigeminal ganglia and project to the brainstem, and first-order neurons carrying information from the body have cell bodies in the dorsal root ganglia and project to the spinal cord.

Nociceptors are high-threshold stimulus detectors and include the thinly myelinated A-δ fibers and unmyelinated C-fibers.7 Nociceptors represent one of several functional groups of sensory fibers in peripheral nerves (Table 89-2), including those fibers involved in mediating proprioception, low-threshold mechanoreception, vibration sense, and innocuous thermal stimuli (Table 89-3). For both A-δ fibers and C-fibers, the distal dendritic inputs are simply bare nerve endings, as opposed to, for example, the more elaborate pacinian corpuscles and muscle spindles used to detect vibration and proprioception. The A-δ fibers mediate acute, precisely localized initial pain. A-δ fibers fall into 2 groups based on electrophysiological studies of the stimuli required for their activation: Type 1 or high-threshold mechanical A-δ nociceptors have high heat thresholds (>50°C) and respond to mechanical stimuli as well. Type 2 or low-threshold mechanical A-δ fibers have a much lower heat threshold but a very high mechanical threshold. Type 1 A-δ fibers most likely mediate pain due to pinprick, and Type 2 A-δ fibers convey the initial response to noxious heat.

C-fibers convey poorly localized, delayed pain. Most C-fibers are also polymodal, although a subset of mechanically insensitive, heat-responsive neurons develops mechanical sensitivity following tissue injury. These neurons are known as "silent nociceptors" and are thought to be particularly important in the development of facilitated pain states.8 Interestingly, distinct sets of C-fibers appear to be involved in itch9 as well as pleasant touch.10

C-fibers can be divided into nonpeptidergic and peptidergic neurons. The nonpeptidergic population uses only the traditional excitatory neurotransmitter glutamate and expresses the c-Ret neurotrophin receptor, which binds the ligand glial-derived neurotrophic factor (GDNF).11 Peptidergic C-fibers use both glutamate and neuropeptides, such as substance P and calcitonin gene-related peptide (CGRP), as neurotransmitters. They express the TrkA neurotrophin receptor, which binds the ligand nerve growth factor (NGF).

Knowledge about peripheral fibers involved in nociception allows us to formulate key concepts that relate not only to the encoding and transmission in the periphery but to the entire afferent system. First, there is a high degree of segregation of afferent information. This is reflected in the labeled lines hypothesis,12 in which different receptors on different types of neurons encode information specific to different stimuli, such as cold, hot, or high-threshold mechanosensation. We see that this spatial segregation of the structures responsible for particular types or components of the painful experience is preserved in the spinal cord, thalamus, and higher brain regions as well.

Second, and in contrast, many nociceptors and other higher-order neurons are polymodal; the convergence and integration of particular types of inputs determine the properties of a response. Thunberg's thermal grill illusion in 1896 demonstrates an interesting example of the complexity of convergence: the combination of innocuous cool and warm stimuli applied in a grid produces a noxious sensation of heat,13 despite the fact that neither component was noxious. The proposed explanation is that first-order neurons sensitive to innocuous cool converge on the same higher-order neurons as those activated by noxious cold; however, there is basal inhibition of the innocuous signal. The simultaneous warm stimulus in the grid removes that inhibition, allowing innocuous cool stimuli to cause pain.

Spinal Cord

The central axons of nociceptors enter the gray matter of the spinal cord via the dorsal roots and synapse on second-order neurons in the dorsal horn of the spinal cord (Fig. 89-4). Some primary afferent fibers traverse caudally or rostrally 1 to 2 segments in the tract of Lissauer prior to synapsing. The gray matter of the spinal cord is organized into 10 laminae (of Rexed). Laminae I to V compose the dorsal horn of the spinal cord (Fig. 89-5). Terminals of different afferent fiber types are separated spatially in the spinal cord. C-fibers terminate superficially within laminae I and II of the dorsal horn (lamina II is termed the substantia gelatinosa, based on its appearance); however, the segregation of fiber type is even more detailed. Most peptidergic C-fibers project to lamina I and the dorsal or outer part of lamina II, and nonpeptidergic C-fibers tend to project more ventrally to the midregion of lamina II. A-δ fibers project to lamina I as well as to lamina V. The large myelinated A-β fibers that mediate light touch project to the deeper laminae (III, IV, and V) as well as to a group of neurons in the ventral-most area of lamina II that express the γ isoform of protein kinase C (PKC-γ).

Thus neurons in lamina I generally respond to noxious stimulation and neurons in layers III and IV respond to innocuous stimuli. Layer V neurons receive dual noxious and non-noxious stimuli and are therefore known as wide dynamic range (WDR) neurons.

Recall that large A-α and A-β sensory fibers that mediate vibration, proprioception, and touch ascend ipsilaterally in the dorsal columns after sending branches that synapse within the deep laminae of the spinal cord. Indeed, some evidence suggests a role for the dorsal columns as well in pain signaling. Inputs from the large fibers may reduce the ascending transmission of nociceptive inputs.14 Indeed, the dorsal columns themselves may transmit pain signals, as suggested by the reduction in visceral pain after dorsal column lesions.15

It is important to point out that most neurons in the dorsal horn of the spinal cord are interneurons, both inhibitory and excitatory, rather than the second-order projection neurons mentioned previously. Equally important is the recognition of the role of non-neuronal cells in the dorsal horn, such as microglia, in modulating neurotransmission. Thus the dorsal horn plays a primary role in sensory integration and modulation, as well as in transmission of nociceptive information. As will be covered, these functions are critical in modulating the gain of the nociceptive system and generating states of sensory sensitization.

Ascending Pathways

The major projection neurons conveying the output from the dorsal horn reside in laminae I and V. These cells comprise the second-order neurons in the spinothalamic (anterolateral) tract, which carries sensory and discriminative information about painful stimuli to the thalamus. Projections to the posterolateral thalamus, near the terminals from the dorsal column/medial lemniscus pathway, transmit discriminative information about the nature of a noxious stimulus. In contrast, connections to the medial thalamic nuclei convey information associated with the affective component of pain. Separation of information from the superficial and deep layers of the spinal cord appears to be preserved in the thalamus.

Substantial nociceptive information passes from the dorsal horn to the brainstem via the spinobulbar tracts. This projection system terminates in 4 areas of the brainstem: the parabrachial region of the dorsolateral pons, the periaqueductal gray, regions of catecholamine cell groups, and the brainstem reticular formation. The parabrachial region of the pons provides a rapid connection to the amygdala, which mediates the aversive properties of pain; the periaqueductal gray balances ascending noxious inputs with descending pain-suppressing signals.

The Brain

The integration and processing of nociceptive information (ie, pain) in the brain is perhaps the most difficult to study—largely due to its high degree of complexity. Indeed, much debate has transpired with regard to pain as a discriminative sense as opposed to an emotional experience.16 Pain has not only a sensory dimension but an affective one as well. The affective component of pain is comprised of the unpleasant emotional feelings generated by pain and the associated desires to end or reduce pain.17 Thus the intensity of perceived pain depends not only on the nature of a painful stimulus, but also on several other independent factors: emotional state—what emotions will a painful stimulus evoke in an individual person, and how will these additional emotions balance against the current emotional state; cognitive state—has the person learned to anticipate the pain itself or a reward associated with the pain; and level of arousal—how does the overall responsiveness of the person modulate a painful stimulus? In this sense, even the simplest pain, nociceptive pain, undergoes modulation by the brain. Perhaps the clearest demonstrations of modulation of pain are the suppression of pain by strong emotional stimuli. For example, intense emotions surrounding battle and athletics have been shown to suppress the response to pain. The soldier needs to focus on surviving and the athlete on performance.18 Indeed, soldiers injured in battle "entirely denied pain,"19 at least to some extent due to the positive experience associated with survival and perhaps also due to the positive expectation of returning home. Similar responses occur in animals, in which expectation of a reward completely suppressed response to pain, known as the nocifensive response, in dogs.20 Such mechanisms can also clearly serve an evolutionary purpose, in that an animal's response to pain or injury can be suppressed in the presence of a predator or other danger.

Much has been learned about the different components of pain, suggesting that different regions of the brain are responsible for different aspects of pain processing, and that there are complex interactions between these different centers.21 Moreover, pain is tightly linked to the reward pathways. Indeed, there is substantial overlap between regions of the brain activated by reward signals and those activated by pain signals.22

The ascending tracts terminate in both the lateral and medial thalamic nuclei. The lateral nuclei and their projections to primary sensory cortex transmit discriminatory pain, the perceptual identification of a precisely specified location, nature, and intensity of a noxious stimulus. The projections to the medial thalamus, in contrast, and the wide projections to limbic and frontal brain regions mediate the affective components of the pain experience.

One of the most expansive tools for studying pain has been functional imaging, including positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). The assumption underlying these 2 techniques is that visualized blood flow and metabolic changes localized to precise regions reflect neuronal activity in the same locations. Although the metabolic effects of inhibitory neurons need to be considered in interpreting the imaging studies on a cellular level, brain imaging has consistently demonstrated regional effects of pain on the brain and has supported a structural segregation of discriminatory and affective pain processing.21 For example, heat pain in patients with somatoform pain disorder was associated with a hyperactive state of several limbic structures including the parahippocampal gyrus, amygdala, and anterior insula.23 Similar results were obtained in fibromyalgia patients during pain catastrophizing and inflammatory bowel patients during rectal distention. Additionally, the anterior insula and rostral anterior cingulate cortices were activated by both painful stimuli to volunteers and by observation of painful stimuli in their loved ones; in contrast, the posterior insula, sensorimotor cortex, and caudal anterior cingulated cortex were only activated by direct painful stimulation.24

Furthermore, these identified regions have been consistent with what might be expected, given prior neurologic knowledge from lesion-based studies. Thus pain asymbolia, the absence of the emotional or affective component of pain with preserved discriminative pain sensation, has been shown to occur after insular lesions.25

In conditioned place aversion tests, an animal associates a particular location with a noxious stimulus and subsequently learns to avoid that location. This anticipatory component of the affective pain response occurs in regions of the brain that are becoming more and more well defined. Indeed, animals with a lesion in the anterior cingulate do not undergo conditioned place aversion learning; that is, the rats do not avoid the place associated with a noxious stimulus.26 Furthermore, microinjection of an excitatory neurotransmitter into the anterior cingulate provided a substitute for the noxious stimulus and an antagonist blocked the development of the process.27 Thus excitatory stimulation of the anterior cingulate was sufficient to produce aversion to a particular location. Importantly, in all of these experiments, the nocifensive response was unaffected, suggesting that the affective dimension of pain—but not the sensory one—was selectively modulated.

The widespread changes in brain function associated with chronic pain conditions revealed by functional imaging studies support the presence of major effects of pain on cognition and mood, especially with changes in the frontal and limbic regions, respectively. In fact, degree of depression in patients with fibromyalgia has been correlated with amygdala and anterior insular activity. Although such studies do not "prove" that pain causes depression, they do highlight the connection and support a careful regard for a patient's mood in treating pain.

Chronic low back pain has been associated with reduced gray matter mass in the thalamus and the lateral prefrontal cortex.28 The effects of pain itself are difficult to separate from secondary effects due to drug use or lack of normal activity and use-dependent structural reinforcements or changes; nonetheless, similar imaging studies have documented decreased gray matter among patients with chronic headache, fibromyalgia, irritable bowel syndrome, and chronic regional pain syndrome. Although the regions identified in the different studies are not identical, they tend to involve regions used during learning and during painful stimulation applied to healthy subjects.

On the cellular level, animal data also support changes in the brain associated with chronic pain, in particular with neuropathic pain. Investigation of cortical neurons after nerve injury shows increased length and branching of layer II/III pyramidal neurons in the contralateral motor cortex.29

Another important but confounding consideration in studying pain is the placebo effect.30 Just as intense emotional experiences such as battle can suppress the sensation of painful stimuli, the anticipation of reward also can increase the pain threshold, suggesting that affective modulation of pain and the placebo response—pain relief due to the expectation of pain relief—operate via a common mechanism.31 The widespread influence of the placebo effect is indeed extraordinary; placebo analgesia has even been shown to correlate with changes in fMRI signals within the dorsal horn of the spinal cord.32 Similar brain regions are involved in the placebo effect and in opioid analgesia.33 Opioid antagonists can block the placebo effect,34 and opioid agonists can prevent increases in fMRI signal in response to a noxious stimulus.35 Thus the placebo mechanism may involve activation of the descending pain modulatory system, to which we will now turn our attention.

Descending Pathways

The power of descending modulation of the pain system was first recognized by Reynolds, who in 1969 found that electrical stimulation of the midbrain periaqueductal grey (PAG) produced analgesia sufficient for surgery in rats.36 Since then, the multiple descending projections from several brainstem centers have been investigated.

The best-characterized system originates in the periaqueductal grey and sends fibers to the rostral ventromedial medulla (RVM), which in turn projects via the dorsal part of the lateral funiculus to both superficial and deep laminae of the dorsal horn (see Fig. 89-1). The RVM is comprised of the serotonergic nucleus raphe magnus and adjacent reticular formation that are located in the midline near the pontomedullary junction. Within the RVM, cells are grouped as on-cells, off-cells, or neutral-cells, with regard to their firing pattern during a nocifensive withdrawal.37 Although the initial emphasis was on the inhibition of pain transmission, recent studies have led to a greater appreciation of both inhibition and enhancement of pain pathways by descending fibers. The inhibitory descending fibers primarily target and blunt the nociceptive inputs of C-fibers.38

The descending pathways of the PAG and RVM receive extensive inputs from higher brain centers, including the hypothalamus, amygdala, and prefrontal cortex. Inputs from the amygdala are associated with intense fear-behaviors, and opioid injection into the basolateral nucleus of the amygdala activates off-cells in the RVM, presumably reducing pain during times of extreme fear.39 In contrast, stimulation of the ventrolateral orbital cortex and anterior cingulate cortex mediates hyperalgesia by stimulating on-cells in the RVM.

In addition to the PAG-RVM pathways, the dorsal reticular nucleus and caudal ventrolateral medulla have also been demonstrated to mediate descending pain signal modulation that occurs at the level of the spinal cord.

Descending inputs can also be categorized by transmitter type. The major transmitters involved in the descending pain modulatory pathways are the endogenous opioids, noradrenalin, and serotonin (5-hydroxytryptamine, 5-HT). The 3 classes of endogenous opioids are enkephalins, β-endorphin, and dynorphins, and are found within neurons of different regions of the central nervous system. Enkephalins and dynorphins are found predominantly in the PAG, RVM, and dorsal horn of the spinal cord, and β-endorphin is found primarily in neurons within the hypothalamus. There are 3 major classes of opioid receptors, μ, δ, and κ, and a less well-understood orphan-type receptor. The role of the μ-opioid has been most securely established: knockout of the μ-opioid receptor abolishes the analgesic effect of morphine.40 The contributions of the other receptor types have not been as clearly documented. The δ-receptor agonists may produce analgesia, with fewer side effects than the μ-receptor agonists. The δ-receptor seems to be involved in the development of opioid tolerance, as well as in anxiety. The κ-receptor may be involved in stress-induced dysphoria. The G-protein–coupled opioid receptors exert inhibitory effects by 2 major mechanisms: they decrease calcium conductances and increase potassium conductance, which can act either presynaptically to decrease transmitter release or postsynaptically to inhibit signal propagation.

Noradrenergic bulbospinal fibers that originate in the locus ceruleus mediate antinociceptive properties41 and act through the G-protein–coupled adrenoceptors. α1-Adrenoceptors are coupled to Gq, leading to an activation of phospholipase C, increased intracellular calcium, and transmitter release; α2-adrenoceptors are coupled to Gi/o, which causes an increased potassium conductance and therefore membrane hyperpolarization and decreased transmitter release. The descending noradrenergic effects occur through the inhibitory α2-adrenoceptor effects on nociceptive terminals and through the activation of presynaptic α1-adrenoceptors on inhibitory interneurons.42,43 Thus noradrenergic inhibition of the nociceptive pathway occurs either through inhibition of the nociceptors or stimulation of the antinociceptive inhibitory interneurons.

Serotonergic descending fibers from the RVM can mediate both antinociceptive and pronociceptive effects, depending on the receptor types activated.42 Of the 14 subtypes of serotonin receptors, most of which are coupled to G-proteins, predominantly the 5-HT1A, 5-HT1B, 5-HT1D, and 5-HT3 receptor subtypes are found in the spinal cord. Serotonin hyperpolarizes second-order neurons in the dorsal horn via Gi/o-coupled 5-HT1A receptors, which act by opening potassium channels and closing calcium channels. Unlike other serotonin receptors, the 5-HT3 receptor is ionotropic, and the binding of serotonin leads to the opening of a nonselective, cation-permeable channel also found in the superficial layers of the dorsal horn. It is located on the afferent fibers of nociceptors and functions to enhance transmitter release from these neurons.

Detection

Acute pain refers to pain due to the detection and transmission of an ongoing noxious stimulus, which can be due to heat, cold, or pressure (mechanical) (Fig. 89-6). Different receptors are involved not only in detecting different stimulus modalities, but also different stimulus intensities (Table 89-4). As mentioned earlier, this baseline, unenhanced pain is termed nociceptive pain.

We perceive temperatures greater than around 43°C as painful. This corresponds to temperature necessary to activate most C-fibers and Type 2 A-δ fibers, whereas minorities of these neurons have a threshold of 50°C. Nociceptors respond to heat through activation of a group of nonselective cation channels called transient receptor potential (TRP) channels. Among the earliest and best characterized of the thermoreceptors is TRPV1, which was identified as the molecular target of capsaicin, the primary pungent component of hot peppers. In support of a primary role for TRPV1 in sensing hot temperatures, TRPV1-deficient animals have a significant impairment in their ability to react to noxious thermal stimuli.44,45 Indeed, TRPV1 activation occurs at temperatures above 42°C, correlating well with the onset of a painful sensation.

A number of other TRPV channels may transduce other heat signals. TRPV2 channels have an activation threshold of about 52°C, and TRPV3 and TRPV4 activate at mildly elevated temperatures (between 25°C and 35°C). Mice that lack TRPV3 and TRPV4 display changes in temperature preference when placed on a surface with graded temperatures.46 Thus the receptors present on a particular C-fiber determine the temperature range to which it is sensitive.

TRP channels were also suspected as possible cold detectors after their role in heat detection was identified. Just as most temperature-sensing neurons respond to capsaicin, most cold-detecting neurons respond to menthol. TRPM8 activates in response to both menthol and cooling to about 28°C. Mice that lack TRPM8 show dual deficiencies in their responses to menthol and cold. Remaining cold sensation may be due to TRPA1, which is activated by higher concentrations of menthol and icillin, as well as a wide range of other stimuli discussed below.

The identity of receptors responsible for detecting noxious mechanical stimuli is not clear. Based on studies in the worm Caenorhabditis elegans, a group of epithelial sodium channels known as acid-sensing ion channels (ASICs) have been hypothesized to be involved. Deletion of 1 or even 2 of the 3 isoforms, however, has not revealed marked deficits, thus raising questions regarding their importance in mechanotransduction. TRP channels may also play a role in mechanotransduction, as TRPV2 responds to osmotic stretch (in addition to heat). Some studies of TRPA1-deficient mice have found changes in the response to mechanical stimulation,47 and others have not.48,49

Receptors for a number of other noxious stimuli have been identified. For example, the TRPV1 receptor responds to protons (ie, acid). TRPA1 has been found to a wide range of agonists that share the structural property of forming covalent adducts with cysteines in the N-terminal region of the receptor. Activating molecules include the pungent ingredients in wasabi (allyl isothiocyanate) and garlic (allicin), as well as environmental toxins such as acrolein, an irritant found in tear gas and smoke.

The active pungent agent in Szechuan pepper, hydroxy-α-sanshool, produces a sensation "akin to the experience of touching one's tongue to the terminals of a 9-V battery."50 This compound has been shown to exert its effect by closing the 2-pore potassium channels, KCNK3, KNCK9, and KCNK18. Interestingly, hydroxy-α-sanshool closes these channels, thereby producing an excitatory effect, in subgroups of both C-fibers and large-diameter myelinated fibers.

Conduction

Like all neurons, electrical signals are transmitted through action potentials formed by tight regulation of the opening and closing of different ion channels. However, nociceptors have a number of unique properties in their assembly of ion channels (Table 89-5). Voltage-gated sodium channels are major determinants of the firing pattern of neurons. The pore of a sodium channel is encoded by a single gene that contains 4 subunits connected by intracellular linkers. Each subunit has 6 transmembrane segments, of which the fourth contains a series of positively charged residues that function as a voltage sensor, and a helix between the fifth and sixth segments forms the pore. Of the 9 voltage-gated sodium channels, nociceptors express NaV1.1, 1.6, and 1.7, which are sensitive to very low concentrations of the pufferfish toxin tetrodotoxin (TTX), and NaV1.8 and 1.9, which are tetrodotoxin resistant. Although NaV1.1 and NaV1.6 are expressed in a wide range of cells, NaV1.7 is found predominantly in nociceptors and sympathetic neurons, but also in other sensory neurons and neuroendocrine cells. NaV1.8 is expressed only in nociceptors, whereas NaV1.9 is expressed predominantly in nociceptors.51

Although the development of TTX-resistant sodium channel expression in the muscles of snakes that pray on TTX-producing newts provides an evolutionary advantage,52 the explanation for the expression of toxin-resistant sodium channels in sensory neurons in humans is less clear. The TTX-resistant sodium channels have unusual gating properties that shape the nociceptive signal. NaV1.8 produces the most current during the action potential upstroke and supports repetitive firing, a key property of nociceptors. NaV1.9 has a much lower activation threshold and therefore amplifies and prolongs small depolarizations.

Nociceptors also have an array of calcium channels involved in modulating and conducting the electrical signal and in releasing transmitter.53 The pore of voltage-gated calcium channels is formed by a tetramer of 4 α1-subunits, each of which is analogous to 1 of 4 subunits encoded by a sodium channel protein, as well as modulatory α2δ-, β-, and γ-subunits. The gabapentinoids, gabapentin and pregabalin, are thought to act by blocking calcium channels containing the α2δ-subunit.54 Subtypes of calcium channels can be identified pharmacologically as well as by their voltage-dependence and kinetics of gating. L-type, or low-voltage activated, calcium channels are expressed in many types of cells including muscle, neurons, and osteoblasts. T-type, transient, channels are most often found in neurons and osteocytes. N-type, neuronal, channels are found predominantly in neuronal tissue. P/Q-type channels are expressed at synaptic terminals in the superficial dorsal horn and other central nervous system synapses where they mediate transmitter release. R-type channels were first identified in cerebellar granule cells and are found predominantly in neurons, including dorsal root ganglion neurons.

Transmission

The dorsal horn is a critical region for the processing, modulation, and projection of nociceptive information to the brain (Tables 89-6 and 89-7). Invasion of the central terminals of nociceptors in the spinal cord by action potentials arriving from the periphery results in activation of voltage-dependent calcium channels and calcium entry, which lead to transmitter release. Enough transmitter may be released to produce depolarization of the postsynaptic cell sufficient to initiate an action potential in the second-order neuron. Alternatively, the invading action potential may result in minimal release of transmitter. Between these extremes lies a range of graded synaptic efficacies from strong to almost ineffective.

Although the dorsal horn is the site of primary afferent transmission, it is also the site of modulation by an extensive network of interneurons, as well as descending axonal projections. Indeed, primary connections between first-order neurons and projection neurons comprise only a small percentage of spinal cord neurons. Numerous transmitters and synaptic neuromodulators are involved in the transmission and modulation of nociceptive information in the dorsal horn (Fig. 89-7).

Synaptic transmission between primary afferent nociceptors (predominantly C-fibers) and second-order neurons has both fast and slow components. As in the case of the nervous system in general, the excitatory amino acid glutamate is the primary fast excitatory neurotransmitter used by nociceptors. Glutamate activates α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA)/kainate ionotropic glutamate receptors, leading to a short-lasting depolarization in second-order neurons (a few milliseconds), the fast excitatory postsynaptic potential. As discussed in more detail later on, the N-methyl-d-aspartate (NMDA) ionotropic glutamate receptor is another ionotropic glutamate receptor that is normally blocked by magnesium at resting membrane potentials and requires prolonged depolarization to relieve that block. The NMDA ion channel allows both calcium and sodium ion influx into the cell; calcium then can act as a second messenger and produce long-lasting changes in the cell.

In addition to glutamate, peptidergic C-fibers have a variety of other cotransmitters (eg, substance P, calcitonin gene-related peptide [CGRP], neuropeptide Y [NPY], and brain-derived neurotropic factor [BDNF]). Whereas glutamate activates ionotropic AMPA/kainate receptors to produce a fast synaptic potential, the coreleased neuropeptides activate G-protein–coupled receptors, resulting in a delayed and longer-lasting slow postsynaptic potential that persists for several seconds.

Glycine and γ-aminobutyric acid (GABA) are the predominant mediators of inhibitory neurotransmission in the dorsal horn. Glycine and GABA activate glycine and GABAA ionotropic receptors to produce inward chloride currents that hyperpolarize and therefore inhibit neuronal activity. Several inhibitory metabotropic (ie, G-protein–coupled) receptors are present on afferent central terminals and dorsal horn neurons, including GABAB, opioid, and cannabinoid receptors.

Control of synaptic efficacy is one of the major driving forces of neuronal plasticity in general and of pain hypersensitivity in particular. As we will see in the next section, changes in the synaptic efficacy comprise major mechanisms for pain facilitation.

In acute nociceptive pain, noxious stimuli produce pain that is restricted temporally to the stimulus duration and geographically to the site of the stimulus, with minimal change exerted on the nociceptive system. Other types of pain involve profound modulation of the receptive, synaptic transmissive, and transductive pathways. Persistent noxious stimuli cause tissue damage, which results in the release and accumulation of a wide range of molecules, known as "inflammatory soup" released from the injured tissue. These factors are released from immune cells, neurons, and non-neuronal cells and include peptides (substance P, CGRP, bradykinin), eicosanoids (products of the cyclo-oxygenase and lipoxygenase pathways such as thromboxane, leukotrienes, and prostaglandins), neurotrophins (such as nerve growth factor and brain-derived neurotrophic factor), cytokines, and chemokines.

Animal models of persistent pain are often divided into inflammatory pain and neuropathic pain based on the method of pain induction. In inflammatory pain models, proinflammatory molecules such as Complete Freund's Adjuvant (CFA) or specific components of the inflammatory soup are injected into, for example, a mouse paw, and the resulting hypersensitivity can be quantified using behavioral tests. Neuropathic models involve the direct injury to a nerve via pressure or ligation.

Of key importance, different mechanisms are involved in the different types of pain models. For example, in the NaV1.9-deficient animals thermal hypersensitivity induced by CFA injection is reduced and mechanical hypersensitivity is unaffected. In contrast, hypersensitivity in 2 neuropathic pain models is unchanged.55 In a separate study, ablation of the population of nociceptors that express NaV1.8 eliminates pain due to inflammatory stimuli but maintains pain due to nerve injury.56

Interestingly, C-fibers expressing the VGLUT3 type of glutamate transporter seem to be important in the facilitation of mechanosensitivity in both inflammatory and neuropathic pain models. These neurons are activated by light touch and cold but not heat under normal conditions. However, in models of persistent pain using inflammatory substance or nerve injury, mice lacking this type of C-fiber exhibit sensitization to heat but not mechanical stimuli.57

Thus there is a clear separation in mechanisms effecting different types of hypersensitivity, thermal or mechanical, and different pain models, inflammatory or neuropathic. This thesis forms the basis of the labeled-line study of pain mechanisms.

Peripheral Sensitization

Peripheral sensitization refers to a decrease in nociceptor threshold and an increase in output, and many inflammatory mediators promote such changes. The initial sensitization process occurs by direct or indirect activation of the detectors that respond to noxious stimuli and the transmitters that conduct the signal (Figure 89-8 and Table 89-8). These modulatory effects occur early by mechanisms such as phosphorylation and ion channel insertion, and later through gene modulation by transcription factors activated at the end of the various pathways.

The signaling pathways that lie between the inflammatory mediators and their modulatory targets are complicated. Most of the pathways start with activation of a G-protein–coupled receptor (GPCR) or serine/threonine receptor kinase. For example, in the protein kinase A pathway, activation of adenyl cyclase by a GPCR produces increased levels of adenosine 3′,5′-cyclic monophosphate (cAMP), which heightens the activity of protein kinase A (PKA) and facilitates phosphorylation of targets. Another G-protein–coupled receptor activates phospholipase C, which hydrolyzes phosphatidylinositol into inositol triphosphate (IP3) and diacylglycerol (DAG), the latter pairing with increased calcium to activate protein kinase C (PKC). Other factors work by activating receptor tyrosine kinases that have their own signaling cascades.

Injection of the inflammatory mediators into animals and the subsequent demonstration of reduced threshold to noxious stimuli have formed the basis for inflammatory pain models. A number of key modulatory targets have been identified. Heat hypersensitivity follows treatment with several inflammatory factors such as bradykinin, ATP, or NGF. In TRPV1-deficient animals, heat hypersensitivity does not develop, supporting the role of TRPV1 in thermal hyperalgesia.44,45 Prostaglandin E2 causes an enhancement of the capsaicin-induced current.58 This sensitization of the channel and resulting increased current has been shown to result from phosphorylation of the TRPV1 receptor, as mutation of the phosphorylation site abrogates the modulatory effect of PKC, which is a downstream effecter of prostaglandin E2.59 Activators of the PKA pathway have been shown to cause insertion of TRPV1 channels into the plasma membrane.60,61

Modulation of TTX-resistant sodium channels provides another mechanism for sensitization. Indeed, prostaglandin E2 has also been shown to increase the amplitude of TTX-resistant sodium channels and decrease the voltage necessary for activating the channels.62 The inflammatory cytokine interleukin-1β also increases TTX-resistant currents by relieving slow inactivation of these channels.63 Furthermore, in NaV1.9-deficient mice there is a reduction in pain hypersensitivity in response to a number of inflammatory mediators including prostaglandin E2, interleukin-1β, bradykinin, and capsaicin.

More subtle sensitization may occur through "priming" of the primary afferent nociceptor. In this case, nociceptive thresholds are unchanged. Rather, the cell has an increased sensitivity to inflammatory mediators. Entry into and maintenance of the primed state depends on PKC.64 In the primed state, cross-activation among pathways increases, so that, for example, protein kinase C, which is normally activated by calcium and DAG, can be activated by cAMP.

Central Sensitization

Central sensitization refers to changes in the spinal cord and brain that result in an increase in the gain of the system with subsequent increased pain perception.65 In contrast to peripheral sensitization, in which nociceptive signals to the brain still depend on continuous peripheral stimuli, albeit weaker ones, central sensitization can render pain stimulus-independent by uncoupling the pain response from the presence and duration of nociceptive inputs (see Fig. 89-3). Instead, the pathologic state of the central nervous system is sufficient in itself to cause pain. Central sensitization is thought to be critical to spontaneous pain, a key feature of clinical neuropathic pain. There are several major components to central sensitization: augmentation of existing inputs, recruitment of new inputs, loss of normal inhibitory signals, and changes in the glial and inflammatory milieu that affect neuronal signaling.

One feature of central sensitization, or perhaps more accurately the process that leads to central sensitization, is temporal windup, the progressive increases in the responses elicited by repeated identical stimuli. Repeated low-frequency stimulation of C-fiber stimulation leads to increased action potential firing in the second-order spinal cord neuron.66 This electrophysiological phenomenon correlates with the development of pain reported by patients upon repeated stimulation with what is initially an innocuous stimulus, but one that becomes noxious on repeated stimulation and results in a prolonged pain following the end of the stimulation. In humans this clinical finding is termed "summation."

The initial experimental observations leading to the understanding of a central component to pain hypersensitivity took place in 1983.67,68 Under normal conditions, a noxious mechanical or thermal stimulus to the skin of the paw is required to stimulate the biceps femoris α-motor neurons and thereby elicit the flexor withdrawal reflex. Non-noxious stimuli produce only subthreshold activity in the motor neurons. However, after repeated noxious stimuli, pain sensation is enhanced so that previously innocuous low-threshold stimuli such as light touch became sufficient to activate the reflex. Electrophysiological evidence supported a central as opposed to a peripheral mechanism. First, after noxious heat application, electrical stimulus of A-β sensory fibers becomes capable of eliciting responses in the α-motor neurons, whereas prior to the noxious heat the electrical stimulation was insufficient. Second, the receptive fields for the A-β sensory fibers are expanded, and local anesthetic block of afferents from the initial peripheral noxious stimulus do not decrease the enlarged receptive fields. Third, repeated low-frequency electrical stimulation, which is sufficient to stimulate only C-fibers, recapitulates the key findings of expanded receptor fields and the ability of previously non-noxious stimuli to activate the nociceptive system. Subsequent experiments clarified the findings by recording from the sensory neurons themselves (Fig. 89-9).

Having established that sustained peripheral nociceptor activation, whether through selective temperature or chemical stimulation, produces central sensitization, we now turn to the mechanisms that underlie the different components of central sensitization. As with peripheral sensitization, the mechanisms fall into 2 groups: early changes involve transcription-independent processes such as phosphorylation (termed post-translational modification); later, more persistent changes result from changes in gene expression (ie, post-transcriptional modification) and synaptic structure. However, in comparison to peripheral sensitization, the changes due to central sensitization tend to be much larger in magnitude and longer in duration.

Synaptic Plasticity

Glutamate, the fast excitatory neurotransmitter at the primary afferent synapse, binds to several types of ionotropic and metabotropic receptors. The ionotropic receptors are tetramers and include those of the amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA), N-methyl-d-aspartate (NMDA), and kainate types. Under basal conditions, AMPA receptors in excitatory neurons tend to be calcium impermeable, due to their inclusion of the GluR2 subunit. In contrast, inhibitory interneurons preferentially express the GluR1 subunit and not the GluR2 subunit, and thus are mostly calcium permeable. The distinguishing property of the NMDA receptor is that at rest the receptor is blocked by external magnesium ions. However, persistent and strong depolarization, such as those produced by inflammation or injury, relieves the block and allows the channels to open and conduct calcium into the cell. Thus the NMDA receptor is an "incident" receptor as it is only activated when there is both the presence of ligand (glutamate) and cellular depolarization. This quality of the NMDA receptor is essential for central sensitization: when a nociceptive signal is sufficiently strong and long acting, NMDA receptor activation effects a marked amplification, and prolongation of the afferent signal occurs.

Activation of the NMDA receptor is critical for central sensitization, as conditional deletion of an essential NMDA receptor subunit abrogates the development of central sensitization.70 The key initial step downstream of NMDA receptor activation is calcium entry, and other signaling molecules and receptors that contribute to central sensitization also depend on calcium entry.

The 6 different metabotropic glutamate receptors act through 2 different types of G-proteins and are arranged in a lamina-specific manner. Group I metabotropic glutamate receptors (mGluR1 and mGluR5) are coupled to Gq-proteins, whose activation causes an increase in intracellular calcium. Increase in the intracellular calcium concentration appears to be the key initiating step in the development of acute central sensitization and occurs through calcium-dependent phosphorylation of transporters and receptors. The effect is amplification of the pain signal. Phosphorylation of AMPA and NMDA receptors increases the conducted currents and affect channel trafficking so that more channels are delivered to the membrane and fewer are removed.

The augmented inputs stimulate the signaling pathways that sustain central sensitization (Fig. 89-10). This process involves multiple signaling pathways including those of map kinase (MAPK), PKA, PKC, and Src. Many of the pathways have multiple downstream targets, and multiple pathways can activate the same target molecules.

A key target of many of the pathways is activation of the extracellular signal-related kinases (ERKs). ERK phosphorylates NMDA receptors, leading to increased activity. Phosphorylation of AMPA receptors by ERK increases trafficking of the receptors to the membrane. Additionally, ERK phosphorylation of the voltage-gated potassium channel Kv4.2 decreases the activity of that channel and thus leads to an increase in neuronal excitability.

Src interacts directly with the NMDA receptor. Injection of a peptide fragment of Src disrupts this interaction and thereby decreases hypersensitivity following injury. In contrast, the response to acute pain is unchanged, illustrating the mechanistic separation between acute and facilitated pain.71

The post-translational modifications lead to transcriptional changes, due to activity of the cAMP responsive element binding (CREB) protein and other transcription factors. These transcription factors increase the expression of several genes, including c-Fos and Cox-2, and receptors such as NK1 and TrkB.

Peripheral inflammation leads to a variety of changes in nociceptive neurons that underlie their sensitization. Large DRG neurons, which normally do not express the neuropeptides substance P and BDNF, begin to express these inflammatory mediators. Thus, non-nociceptive afferents develop the ability to release nociceptive signaling peptides.72

Cox-2 activation leads to the production of prostaglandin E2 (PGE2), which binds to G-protein–coupled receptors on neurons to potentiate excitatory glutamatergic signaling73 and reduce inhibitory glycinergic neurotransmission.74 Elimination of Cox-2 from neurons does not affect pain due to peripheral inflammation and heat hyperalgesia, but almost completely eliminates mechanical hypersensitivity, thus again confirming the role of central sensitization in mechanical pain hypersensitivity.75

Nerve injury, the sine qua non for neuropathic pain, produces massive changes in the genes expressed in dorsal root ganglia neurons. Whereas peripheral inflammation affects only a small number of transcripts, nerve injury affects the expression of approximately 1000 transcripts, including ion channels, receptors, transmitters, and structural proteins involved in axonal regeneration.76 Peripheral injury results in a degeneration of C-fiber terminals in lamina II. The combination of the loss of afferent stimuli together with the proregenerative environment after injury produces sprouting of myelinated A-β fibers, which normally only synapse into deeper laminae II to IV, into laminae I to II and the formation of new nociceptive inputs. Although there is some controversy in the interpretation of axonal labeling experiments, immunostaining and electrophysiological experiments have been convincing.77,78 c-Fos activation in the superficial dorsal horn occurs only following noxious stimuli in noninjured animals; and A-β fiber stimulation is sufficient to induce c-Fos activation after nerve injury. The importance of A-β fibers in the development of nerve injury pain is underscored by the finding that ablation of most nociceptive inputs does not affect neuropathic pain,79 and selective block of the large A-β fibers eliminates dynamic allodynia, a key component of neuropathic pain.80 Furthermore, the elimination of PKC-γ–expressing neurons, which lie in the ventral-most part of lamina II and receive input only from A-β fibers, decreases pain in response to nerve injury.81

An additional component of pain enhancement is suggested by the observation that although norepinephrine administration to the skin of healthy subjects does not cause pain, norepinephrine administration does cause pain in the setting of preexisting nerve injury or inflammation. α2-Adrenoreceptors have been found to be upregulated following nerve injury in both large- and small-fiber primary. That sympathetic fibers sprout onto primarily large non-nociceptive neurons in the DRG after nerve injury suggests that sympathetic pain modulation may occur through the nocicepive actions acquired by these large fibers as part of central sensitization.82,83

Disinhibition

Inhibitory neurons, which are densely distributed in the superficial layers of the dorsal spinal cord, use the neurotransmitters GABA and glycine to inhibit their postsynaptic targets. Application of blockers of these transmitters enhances the electrophysiological responses to activation of primary nociceptors84 and produces behavioral sensitivity similar to that seen following peripheral nerve injury.85,86

Indeed, the very ability of inhibitory neurons to decrease activity is disturbed. BDNF is increased after nerve injury, and following BDNF treatment, the K-Cl transporter KCC2 is downregulated.87 The decreased transporter expression disrupts the normal potassium and chloride gradient such that the internal chloride concentration in cells is increased. Consequently, after nerve injury, opening of chloride channels by the inhibitory neuropeptides then causes an outward flux of chloride, leading to cell depolarization and increased excitatory activity.88 Thus normally inhibitory inputs are transformed into excitatory ones during the development of persistent pain.

Immune and Glial Modulation

Under basal conditions, microglia are the only immunocompetent cells of the central nervous system. After peripheral nerve injury, microglia migrate to the superficial dorsal horn and undergo activation: they change their shape and gene expression. P38 MAPK upregulation leads to the production and release of several proinflammatory cytokines, including IL-1β and tumor necrosis factor α (TNF-α). As these require nerve injury and not just prolonged nerve activity, the activation of the microglia must be able to detect damage to the neuron. One potential signaling molecule for this role is adenosine triphosphate (ATP). Injection of ATP is sufficient to reproduce the pain behavioral phenotype produced by nerve injury. ATP receptors are divided into 2 types, ionotropic P2X receptors and metabotropic P2Y receptors.89 Each of the 7 types of P2X genes encodes a single subunit of a nonselective cation channel, with a presumed stoichiometry of 3 subunits per channel. There is also coassembly of certain isomers. In immune cells, the most commonly expressed P2X genes include P2X1, P2X4, and P2X7. Indeed, in macrophages and microglia, P2X7 activation leads to caspase-1 activation and the release of IL-1β. In contrast, BDNF release seems to be due to P2X4 receptors. When microglia release BDNF, as discussed above, the result is a local reversal of the chloride gradient, thus shifting the hyperpolarizing effect of inhibitory neurons to a depolarizing and activating one. The 8 P2Y isoforms are coupled to heterotrimeric G-proteins. The P2Y12 isoform seems particularly important for microglial activation.90

A number of other ligands and receptors appear to be involved in the activation of microglia following nerve injury. Fractalkine is expressed by both primary afferent and spinal cord neurons, and its receptor, CX3CR1L, is upregulated on microglia following nerve injury.91 Interfering with fractalkine signaling has been shown to reduce injury-induced pain. Additionally, interfering with signaling through the toll-like receptors affects both microglial activation and nerve injury–induced pain.

Although the activation of microglial cells is perhaps best characterized, these cells are not the only types of cells recruited and activated during peripheral nerve injury. T cells infiltrate the dorsal horn after nerve injury, and cytokines produced by T cells, such as interferon (IFN)-γ, are involved in pain following nerve injury.92 Both Schwann cells in the periphery and astrocytes in the central nervous system are also activated by peripheral nerve injury.

Notably, the glial activation changes occur not only at the level of the spinal cord, but also at the level of the brainstem, where they are thought to interfere with the descending pain inhibitory systems.93

Evidence for Central Mechanisms of Pain in Humans

Several studies in humans suggest that prolonged pain does indeed involve central sensitization. Injection of capsaicin in human volunteers produces nested circles of hypersensitivity of different sizes and types: in the center is a small circle of hyperalgesia to heat stimulus. Outside that is a circle of allodynia to light touch, and the outermost region shows hyperalgesia to noxious punctate stimuli. The innermost region exemplifies primary hyperalgesia, in that the symptoms are mediated by C-fibers with decreased thresholds. However, allodynia in the middle region and mechanical hyperalgesia in the outer area involve pain resulting from non-nociceptive fibers, thus requiring central changes that enable non-nociceptive afferents to cause pain. Thus these hypersensitivities mediated by A-β fibers are secondary in contrast to the nociceptor-mediated primary hyperalgesia.

In an elegant experiment, injection of capsaicin into anesthetized skin fails to produce hyperalgesia and mechanical allodynia in the surrounding unanesthetized skin.94 Electrical intraneural stimulation that initially produces a non-noxious tactilelike sensation produces a noxious sensitization after capsaicin is injected adjacent into the adjacent skin region.95

Consistent with a mechanistic separation for peripheral and central pain enhancement, NDMA-receptor antagonists such as ketamine block secondary hyperalgesia but not primary hyperalgesia. Thus peripheral changes are insufficient to account for findings of allodynia; instead, central changes allow the activation of large mechanical fibers, normally unable to cause pain, to do so. As we discussed, this effect is neatly explained by the sprouting of afferents from these fibers into the superficial dorsal horn. Although some effects of central sensitization appear relatively local, other effects are impressively global. Indeed, capsaicin has been found to cause some degree of contralateral hyperalgesia and allodynia,96 and this has been termed "tertiary hyperalgesia."

Not only does central sensitization underlie the development of allodynia in experiments on human volunteers, but it also is important in human diseases and their treatment.63 Patients with chronic pain due to a wide range of diseases, including rheumatoid and osteoarthritis, fibromyalgia, headache, visceral pain syndromes such as inflammatory bowel syndrome, and headache, exhibit decreased thresholds to noxious stimuli, increased temporal summation, and increased pain. The changes in threshold and pain can be documented not only by patient report, but also by increased fMRI signal.97 Both NMDA antagonists and the gabapentinoids block secondary allodynia and mechanical hyperalgesia induced by intradermal capsaicin. The importance of controlling postoperative pain in the healing process and the benefits of treatments that block central sensitization underscore the clinical relevance of central sensitization.

Research in human pain genetics has used 2 approaches. The first strategy requires the identification of rare single-gene mutations. Although very rare, these conditions can provide insights into pain pathways and suggest novel therapeutic possibilities. The second approach involves analysis of more common pain phenotypes or experimental models using genetic association.

The study and understanding of rare disorders has highlighted the importance of a number of molecules in the pain sensory transduction. Perhaps the best example of this involves mutations in the NaV1.7 ion channel, which is preferentially expressed in nociceptors, but is also found in other dorsal root ganglion neurons, sympathetic neurons, and Schwann cells (Table 89-9). A striking phenotype known as congenital insensitivity to pain was identified in members of 2 cosanguineous families who carry mutations in both copies of the gene.98 Despite completely intact sensation and transmission of all sensory modalities as demonstrated on quantitative sensory testing and nerve conduction tests, these people remarkably do not feel pain. In fact, they perform "stunts" such as driving knives through their arms. The disfiguring effects on the jaws and limbs underscore the importance of pain from a self-preservation point of view and explain why pain is so well enforced by evolution. One can only wonder about how the inability to experience physical pain affects the psyche of such people.

Affected family members were found to have homozygous nonfunctioning mutations in the NaV1.7 ion channel. This study is extremely encouraging from a research perspective, because it shows that pain is not essential to the human condition, and that there is no a priori reason why pain cannot be blocked in the greater population.

Interestingly, mutations in the same ion channel, NaV1.7, have been associated with 2 rare dominant familial conditions characterized by increased levels of pain.99 In the first, primary erythromelalgia, affected individuals suffer from episodic redness and severe burning pain in the extremities. The age of disease onset can be as early as 1 year. Genetic linkage analysis of these patients also implicated the NaV1.7 gene. Nearly 10 different point mutations have been identified in such families, and these mutations have been shown to facilitate opening of the channel and slow deactivation of the channel. These mutations lower the threshold necessary for the generation of action potentials (Fig. 89-11).

The second disease, paroxysmal extreme pain disorder, has a distinct phenotype with pain in the rectal, ocular, and mandibular areas. Again, genetic studies identified NaV1.7 as the responsible gene, and over 5 distinct point mutations have been identified in affected families. Like those that cause inherited erythromelalgia, mutations responsible for paroxysmal extreme pain disorder also cause increased excitability in nociceptors. However, they do so by inhibiting channel inactivation, the process by which the channels rapidly close milliseconds after opening. This effect is quantified as a depolarizing shift in the voltage dependence of inactivation.

Another rare monogenic dominant pain syndrome has been associated with mutation in TRPA1.100 Episodes of severe pain primarily localized to the upper body are triggered by conditions of fatigue, decreased food intake, and decreased temperature. The responsible mutation was showed to have an enhancing effect on TRPA1 currents.

Despite the meticulous detail in which researchers now understand the mutations responsible for these conditions, much still remains unknown. In particular, the reasons for the episodic nature of the diseases, as well as the restriction of symptoms to particular locations in the body remain mysteries.

In addition to the NaV1.7-associated channelopathies, a number of other mutations have been associated with striking pain-related phenotypes. Another cause of congenital insensitivity to pain is hereditary sensory and autonomic neuropathy (HSAN), which is distinguished by the presence of peripheral neuropathy. Two of 5 types are due to mutations in genes encoding the NGF β protein and the NGF receptor (TRKA).101

Gene association studies have identified several genes that may modulate pain. For example, polymorphisms in catechol-O-methyltransferase (COMT) have been found to be associated with pain on injection of hypertonic saline and in pain originating from the temporomandibular joint.102,103 Other such studies have linked polymorphisms in the melanocortin-1 receptor (MCR1) to μ-opioid–induced analgesia104 and in the κ-opioid receptor 1 (OPRM1) to morphine sensitivity (Table 89-10).105-108 Given the known importance of NaV1.7 in pain insensitivity and hypersensitivity, a search for polymorphisms in NaV1.7 revealed an allele with a frequency of approximately 10% that was associated with increased pain perception and was found to impede sodium channel slow inactivation.109 Such polymorphisms in NaV1.7 may underlie common variation in pain thresholds.110

Other approaches have used animal models to identify genes. Utilizing single nucleotide polymorphism analysis, the enzyme guanosine triphosphate cyclohydrolase (GCH1) was found to be increased in peripheral sensory neurons following axonal injury in mice.111 By investigating haplotypes in humans, researchers identified a pain protective haplotype (frequency 15.4%), for which homozygotes have reduced postsurgical pain and lower pain thresholds.

Nociception and pain perception are critical components of the nervous system that when functioning properly serve to protect an organism from injury. Multiple types of nociceptive inputs are processed through both segregated and convergent pathways, and are amplified and expanded during periods of prolonged pain. We now understand that chronic pain results from changes in gene expression and in neuronal and supporting cell architecture primarily within the central nervous system. Simply put, chronic pain results from changes in the spinal cord and brain that promote both sensitization of the nociceptive system and the continued and augmented sensation of pain. Through better understanding of the precise molecular mechanisms of how chronic pain develops and the genetic factors that predispose particular individuals to pain, tantalizing new targets for analgesics have been identified. In the future, pain treatments will be tailored to the specific underlying mechanisms responsible for pain in particular diseases and in individual patients.

• Basbaum AI, Bautista DM, Scherrer G, Julius D. Cellular and molecular mechanisms of pain. Cell. 2009;139(2):267-284.
• Costigan M, Scholz J, Woolf CJ. Neuropathic pain: a maladaptive response of the nervous system to damage. Annu Rev Neurosci. 2009;32:1-32.
• Perl ER. Ideas about pain, a historical view. Nat Rev Neurosci. 2007;8(1):71-80.
• Woolf CJ. Central sensitization: implications for the diagnosis and treatment of pain. Pain. 2011;152(3 Suppl):S2-S15.