CHAPTER 103 Chronic Pain
The Basic Science
In addition to the five senses, pain sensation is also an important physiologic means by which the organism is informed about the immediate external environment and internal bodily function. The pain “experience” is an integration of a number of neurologic functions. In response to a brief painful stimulus, a rapid unconscious reflex and subsequent purposeful movement steers the organism away from the source of pain. The painful experience may be paired with a cue and remembered to generate avoidance behavior and evoke profound emotions. Pain from internal organs may signal either imminent or ongoing tissue damage, which prompts the patient to seek medical attention. In a sense, acute pain assists survival.
However, chronic pain, pain that persists long after tissue healing due to injury or in tandem with disease, serves no useful biologic function.1 Chronic pain leads to heightened anxiety and diminished social functioning.2 The economic costs, in terms of both health care costs and lost productivity, are staggering and expected to rise even further as the population ages.3–6 Thus there is a vital need to understand the mechanism of chronic pain and to develop mechanism-based analgesic therapies.
Chronic pain following spinal cord injury (SCI) is estimated to occur in 65% to 81% of SCI patients, wherein about one third of patients rate it as severe.2,7 The addition of intractable pain to deficits in voluntary motor functions and autonomic dysfunction following SCI severely diminishes patient social and psychologic well-being.8 In order to improve patient quality of life, pain relief is an important consideration along with other SCI-related complications in an overall treatment and rehabilitation strategy.
Despite increasing understanding of the cellular and molecular processes that mediate chronic pain in general, effective treatment for SCI pain in particular is lacking. At least two impositions thwart attempts to effectively ameliorate SCI pain. First, SCI pain is a heterogeneous condition with a number of possible interacting neurologic and inflammation-mediated mechanisms. Secondly, although chronic SCI pain shares striking similarities in terms of symptoms with other chronic pain states such as peripheral nerve injury pain, it appears that distinct mechanisms distinguish chronic neuropathic SCI pain from that of other types of neuropathic pain. For example, there is a striking lack of parallelism between peripheral neuropathic pain and neuropathic SCI pain in terms of clinical pharmacology. The first steps toward effective analgesic treatment for SCI pain will involve not only involve careful clinical diagnosis but also consideration of SCI pain as a distinct chronic pain syndrome.
The International Association for the Study of Pain (IASP) has proposed a taxonomy that attempts to group SCI pain states to assist consistent clinical diagnosis and aid in designing effective pain management strategies (Table 103–1).9 In addition, classification also aids researchers in designing appropriate experiments that will uncover mechanisms underlying SCI pain and hopefully uncover new analgesic targets.
|Broad Type (Tier 1)
|Broad System (Tier 2)
|Specific structures/Pathology (Tier 3)
|Bone, joint, muscle trauma, or inflammation
|Secondary overuse syndromes
|Renal calculus, bowel, sphincter dysfunction, etc.
|Complex regional pain syndromes
|Nerve root compression (including cauda equina)
|Spinal cord trauma/ischemia (central dysesthesia syndrome, etc.)
|Spinal cord trauma/ischemia (central dysesthesia syndrome, etc.)
Reprinted with permission from Siddall PJ: Management of neuropathic pain following spinal cord injury: now and in the future. Spinal Cord 47:352-359, 2009.
Table 103–1 first divides pain into two general “types,” nociceptive and neuropathic, and further divides these pain types by possible underlying pathologies. The classification of the various pains (musculoskeletal, visceral, and neuropathic pain based on spinal lesion level) does not suggest definitive mechanisms, and the construct validity of the divisions has yet to be confirmed in clinical studies. However, such an uncomplicated taxonomy is a positive first step in systematically addressing the cause of SCI pain and developing useful therapies.
The first type of SCI pain, nociceptive pain, arises from stimulation of either somatic or visceral primary afferent nociceptors. Musculoskeletal pain has been characterized as dull aching pain that worsens with movement and eases with rest.7 In addition, this particular pain is localized to musculoskeletal structures. Sources of musculoskeletal pain include injury to muscles or ligaments related to the initial injury, overuse of the shoulders and arms of a wheelchair-bound patient, and vertebral instability and osteoporosis due to SCI.10 Given the initial challenges of novel self-propulsion and injury-associated pain, the onset of musculoskeletal pain is comparatively rapid, within weeks of the injury, and is the most commonly reported SCI-associated pain.11 Chronic musculoskeletal pain has been noted long after the injury (at least 5 years) and could be associated with the long-term skeletal changes in posture due to injury.7 Visceral pain has been described as spontaneous, dull, poorly localized, or cramping apparently originating from deep visceral structures.7 The occurrence of pain may or may not be coupled with visceral pathology and, unlike musculoskeletal pain, the onset of visceral pain may be months or years following SCI.7,12 Although the incidence of chronic visceral SCI pain is low, the pain is described as either severe or excruciating.7
The second type of SCI pain is neuropathic pain, which results from trauma or disease to the nervous system. Neuropathic pain has been described as unevoked pain characterized as sharp, burning, shooting, stabbing, and electric, occurring continuously or as paroxysms. In addition to spontaneous pain, there may also be exaggerated painfulness evoked by non-noxious stimulation (e.g., allodynia). Spontaneous and evoked neuropathic pain may occur at the level of injury due to a combination of damaged segmental spinal nerve roots or disinhibited spinal dorsal horn nociceptive neurons. About 40% of SCI patients experience at-level neuropathic pain.7 In contrast to at-level pain, below-level pain appears to have a delayed onset post-SCI and this could be due to dysfunction of brain regions postsynaptic to the spinothalamic tract.7 Below-level pain is as severe and persistent as at-level neuropathic SCI pain.7 Despite a lack of cutaneous thermal detection (either cold or heat) below the lesion, below-level neuropathic pain occurs in about one third of SCI patients. Interestingly, at-level and below-level cutaneous hypersensitivity correlates with the presence of below-level spontaneous pain, which suggests a common mechanism underlying these symptoms.13,14 A positive correlation between spontaneous pain and cutaneous mechanical hypersensitivity of the painful area has been reported for other neuropathic pain states.15 Such a correlation suggests that it may be possible to use defined stimuli to quantify spontaneous pain, beyond a subjective patient report, and objectively compare and contrast the efficacies of pain treatments.
A general outline of the normal pain “pathway” will be presented. Detailed neuroanatomic and neurochemical schemes have been elaborated elsewhere.16 Noxious cutaneous stimuli (either thermal or mechanical) stimulate myelinated or unmyelinated small diameter primary afferents, conducting the “noxious signal” to the spinal cord superficial dorsal horn (or in the face, to the brainstem trigeminal sensory nucleus). Within the dorsal horn, excitatory neurotransmitters released from primary afferent central terminals stimulate postsynaptic dorsal horn neurons. A number of neuroactive substances (e.g., adenosine 5′-triphosphate [ATP]), excitatory amino acids, and neuropeptides activate their respective receptors on the postsynaptic neuron. By contrast, non-nociceptive, large-diameter, primary afferents terminate in deep dorsal horn and also in the brainstem dorsal column nuclei.
Axons of dorsal horn nociceptive neurons ascend to the brain via a number of tracts. Axons of nociceptive neurons decussate in the spinal cord and ascend via the contralateral ventral funiculus (spinothalamic tract) and terminate in the ventroposterior lateral thalamus. The pain signal may be dispersed from the thalamic nucleus to various brain areas with diverse functions such as the sensory cortex, hypothalamus, limbic lobes, and motor nuclei. Although there are a number of indirect pathways between the spinal dorsal horn and brain, direct projections of dorsal horn neurons to a number of brain areas have also been reported.17 By virtue of the numerous direct and indirect connections between nociceptive neurons to higher brain areas, pain evokes multiple physiologic and psychologic responses. Given human genetic diversity, it is apparent that the pain experience, as well as potential treatment strategies, differs between individuals and groups.18,19
In the normally functioning pain system, the excitatory component is counterbalanced by endogenous inhibitory components such that the initial pain sensation is not permanently propagated or does not evoke an exaggerated response. Primary afferents not only synapse with dorsal horn nociceptive neurons but also inhibitory interneurons, which, in turn, synapse with spinal nociceptive neurons, thus moderating pain transmission.20 Also, axons of nociceptive spinal neurons that terminate in the brainstem activate serotonergic, catecholaminergic, and GABAergic (gamma-aminobutyrate) neurons, which in turn send axons down to the dorsal horn. Because terminals are found presynaptic to primary afferents and spinal nociceptive neurons, activation of these brainstem neurons leads to diminished pain perception. Direct application by intrathecal injection of inhibitory neurotransmitters and opioid neuropeptides reduce spinal nociceptive neuron responses to noxious peripheral stimulation and are antinociceptive in rats.21 Analgesia is also observed in humans, following intrathecal injection of similar substances (α-adrenergic, GABAergic, and opioid receptor agonists).22 These findings a considerable parallel between the human and rat spinal dorsal horn neurochemistry, which, furthermore, points out the utility of preclinical models in evaluating drugs for possible clinical use.
In contrast to the normal state, experimental evidence suggests that decreased inhibition, increased excitation, or a combination of both initiates and perpetuates a chronic pain state because of the pathology of the nervous system. For example, acute lumbar intrathecal injection of an antagonist to either the GABAA or GABAB receptor subtype or the glycine receptor leads to a transient yet robust hind paw hypersensitivity and vocalization in rats. This indicates that a tonic inhibition is present in the normal state, which “dampens” responses to peripheral stimulation. Similarly, intrathecal injection of an excitatory glutamate receptor agonist (e.g., N-methyl-D-aspartate) induces a long-lasting hind paw hypersensitivity.23 The sustained excitation of nociceptive neurons may lead to increased intracellular cation levels, upregulation of second messenger systems, and gene expression.16 These intracellular processes then lead to persistent hyperactivity and increased responsiveness to peripheral stimulation. Furthermore, such abnormal activity may be found throughout the pain neuraxis. Preclinical pain models have demonstrated considerable changes to normal neuroanatomy and neurochemistry following painful peripheral nerve injury or inflammation. For example, the central terminals of non-nociceptive primary afferents extend to spinal laminae normally receiving nociceptors, and these same afferents now express neuropeptides typically found in nociceptors.24,25 Changes in brain activity response to peripheral stimulation following an injury, including an SCI, have been observed.26
On the basis of findings in preclinical pain models, to attenuate clinical chronic pain states, it would be reasonable to increase inhibition (e.g., intrathecal GABAB receptor agonist baclofen) or decrease excitation (e.g., intrathecal [N-methyl-D-aspartate] NMDA receptor antagonist ketamine) at the level of the spinal cord, the first site of interaction between the peripheral and central nervous systems (see later).22,27 Even though such acute measures may prove efficacious, they are temporary. It is likely that a number of regions within the nervous system may express abnormal excitation and that these changes have been made permanent via genetic and structural mechanisms. Implantation of cells that continuously release endogenous analgesic substances and gene therapy may provide long-term pain relief.28–32
Most experimental pain studies have focused on neural function for obvious reasons. However, glial cells (e.g., microglia, astrocytes) vastly outnumber neurons in the central nervous system (CNS). Accumulating evidence indicates that glia have a key role in maintaining the neuropathic SCI pain state.33 In the normal state, glia appear to maintain the homeostasis of the extracellular milieu. Because glia express receptors and ion channels, similar to neurons, they may respond to neuroactive substances. Following exposure to these substances, activated glia may release, in turn, a number of neuroactive substances and proinflammatory cytokines. The glial response following injury has been intensely characterized because modulating the response is believed to be crucial to promoting motor and sensory recovery.34 Injury of the thoracic spinal cord leads to dramatic increases in microglia and astrocytes not only at the site of injury but at the level of the lumbar enlargement, several segments away.35,36 Interestingly, similar increases in spinal glial activity in the lumbar spinal cord are observed in models of painful peripheral neuropathies.37 Treatments designed to decrease glial function following SCI in order to improve motor function may also have a secondary effect of reducing SCI-induced pain. Such treatment studies should include sensory outcomes if the patient has SCI pain.
The majority of preclinical SCI pain experiments have been done in rodents, not only because of the convenient availability of near homogenous subjects but also because general clinical aspects of the histopathology following an SCI can be replicated in rats.38 In addition, numerous analgesic treatments initially screened in peripheral-injury chronic pain models have gone on to demonstrate clinical efficacy. Thus animal models of SCI pain may also be useful in both elucidating SCI pain mechanisms and developing novel clinical treatments.
A degree of controversy exists surrounding the clinical relevance of rodent models of pain.39 Such controversy plagues other fields of neurologic research as well.40 The main clinical diagnosis of pain is based on the patient’s verbal report, which would include pain severity, duration, and frequency. By contrast, in chronic pain models, a specific response to a given stimulus is interpreted as pain by experimenters. An exaggerated change in response to either non-noxious or noxious (hyperalgesia) cutaneous stimuli is interpreted to mean a pathologic alteration in the underlying pain mechanism. However, such changes in sensory perception may not always be reported by patients, who primarily present with unevoked spontaneous pain.
An additional controversy that arises concerning animal models is deciding which would be best to use for preclinical research such that the information obtained could be readily translated into clinical practice. In terms of evaluating new therapeutics, predictive validity and reliability are crucial.41 A model with predictive validity allows one to accurately foresee the effect of a treatment in humans.
If the animal displays behaviors that are analogous to clinical symptoms, the model is said to have face validity. In pain research, however, the outcome measures in the preclinical and clinical situations may not be identical. Most preclinical analgesic drug studies measure changes in response to stimulation, whereas few clinical drug trials have solely used quantitative sensory testing. Despite the stark difference in outcome measures, the concordance between the animal and clinical results is between 61% and 88%.42 In general, the high concordance suggests that the presence of cutaneous sensitivity (e.g., withdrawal threshold) is predictive of spontaneous pain (e.g., pain rating on a visual analog scale), but the wide range also indicates that some models may have better predictive value over others.
In an attempt to bridge the divide between clinical and preclinical outcome measures, basic scientists are working to quantify spontaneous pain-related behavior and changes in mood associated with chronic pain in animals.7,39 As mentioned earlier, changes in affect may impair SCI rehabilitation, further degrading overall well-being. Some of the well-defined experimental methods used in other behavioral research fields such as psychiatry and drug addiction have been adopted to detect and quantify spontaneous pain-related behavior, with varying degrees of success.43–45 Given the current concordance between preclinical and clinical results and the difficulty of quantifying affect in animals, it is not entirely clear how much more translational or clinical value will be gained with the addition of nonspontaneous measures to current testing procedures.40
If a model has face validity, the assumption that follows is that it also has construct validity. Although it is desirable to mimic the clinical pathology in the rat, this may not always be possible. Demonstrating construct validity requires knowledge of the clinical mechanism, which is often lacking and extremely difficult to obtain. Also, as Geyer points out, the theoretical basis of neurologic disorders is constantly evolving.40 Thus construct validity should not be the sole determinant to judge the usefulness of a model.
Finally, reliability refers to the “consistency and stability” of both the experimental procedure and the symptoms resulting from it.40 On the one hand, this quality is highly desirable in basic science research because a highly predictable outcome following a manipulation reduces the chance of error and the need to use large numbers of subjects. On the other hand, clinical SCI is not homogeneous, and although many SCI patients suffer from chronic pain, there are those who do not.
The first consideration in choosing a model should be the scientific purpose of the model. Then, on the basis of the three main criteria (face, predictive, construct validity), the experimenter can determine whether or not the model is appropriate for his or her objectives.
Three general SCI injury methods have been most frequently used: excitotoxin microinjection into the spinal cord parenchyma, acute trauma to spinal tissue, and photochemically induced ischemia of spinal vasculature. Each rat model displays distinct injury-induced, pain-related behaviors and histopathology—behaviors and pathologies that may not be observed in the other models. At the same time, a few general characteristics may be observed across models. The neurobiologic substrates that underlie SCI pain in one model could be specific to that model. On the one hand, this may complicate efforts to develop analgesic treatments to cover a heterogeneous population. On the other hand, the heterogeneity of the clinical population and a diversity of mechanisms suggest that that there is no “magic bullet” and that numerous treatments for pain need to be developed.
A series of microinjections of the nonsubtype selective glutamate receptor agonist quisqualate is placed into the deep dorsal horn of the thoraco-lumbar spinal cord.46 The lesion leads to a dramatic necrosis of the deep dorsal horn but spares the superficial dorsal horn, thus preserving ipsilateral sensory input. Furthermore, syrinx formation is observed in these rats. Significant bilateral hind paw hypersensitivity to both thermal and mechanical stimuli are observed as early as 8 days following microinjection. Also, spontaneous “grooming” behavior, excessive hind paw scratching directed within the injected dermatome, begins about 2 days postmicroinjection and may progressively expand and worsen over time. The grooming behavior is suggestive of at-level unevoked pain. Both the grooming behavior and cutaneous hypersensitivity persist for weeks following injection.
The excitotoxic SCI model is unique in that the spinal cord lesion is generally restricted to the dorsal horn. Because of this selective destruction, there is little or no hind paw motor dysfunction. Several novel treatments have been tried in this rat model that attenuated pain-related symptoms including grooming.28,47 The predictive validity of this model is not known because extensive evaluation of clinically available analgesic drugs has yet to be done.
Several methods of direct spinal injury lead to significant and long-lasting neuropathic pain-related symptoms. Following a laminectomy, the spinal cord at the thoracic level may be laterally hemisected, contused (e.g., New York University impactor), or compressed (e.g., vascular clip).
Following lateral hemisection of spinal cord at level T13, rats develop bilateral hind paw hypersensitivity to heat (decreased withdrawal latency) and mechanical (decreased withdrawal threshold) stimuli beginning 10 days after injury.48 Loss of ipsilateral hind limb function is also observed, but functional recovery begins about 7 days after injury. Despite mild-to-moderate hind paw dysfunction, as demonstrated by Basso-Beattie-Bresnahan (BBB49) locomotor rating scores, robust ipsilateral hind paw responses can be evoked in hemisected rats in order to evaluate therapeutics. In addition, a bilateral forepaw (above-level) hypersensitivity is observed beginning about 2 weeks after injury. A few clinically relevant analgesic drugs have been tested in these rats, along with novel treatments.30,50 Systemic and intrathecal (catheter terminating at the spinal level T13) injection of morphine ameliorated hind paw mechanical hypersensitivity.51 Intrathecal baclofen increased hind paw withdrawal thresholds in injured rats.52 By contrast, intrathecal injection of the selective serotonin reuptake inhibitor (SSRI) fluvoxamine attenuated forepaw mechanical hypersensitivity to a greater degree than hind paw hypersensitivity.53 The differing effect suggests a differential neuropharmacology between above- and below-level neuropathic pains. Hemisection of the spinal cord could be useful in comparing neurochemical and anatomic changes rostral or caudal to the lesion with the uninjured side. The lesion is relatively easy to generate and fairly consistent from rat to rat, which in part explains why this is the most commonly used SCI injury model.54 However, the relevance of hemisection to clinical injury is not entirely clear because there are few human cases of spinal hemisection injuries.
Most spinal cord injuries are due to acute extradural impact and compressive forces.38 These are traditionally modeled in animals using contusive impact devices in which the severity of the injury can be varied, resulting in locomotor dysfunction in a force-dependent manner.55 In these animals, the effect of various intervention strategies can be examined.54
Even with standardized procedures and devices, it is still difficult to uncover a correlation between the degree of tissue damage and symptoms of neuropathic pain. Following a “moderate” contusion of the thoracic dorsal spinal cord, the forepaws develop hypersensitivity to heat and mechanical stimuli, but the sensitivity of the hind paws to these stimuli are variable.56,57 A “moderate” contusion also yields long-lasting at-level hypersensitivity, whereas fewer rats with a “severe” injury displayed such hypersensitivity.58 Hind limb functionality scores in moderately injured rats were considerably decreased but recovered such that scores were eventually the same as uninjured rats. Cutaneous hypersensitivity persisted, however, long after functional recovery.58 Other studies have demonstrated significant at-level and below-level hypersensitivity with severe injury.59,60 Clinical studies have yet to find a correlation among the extent of spinal cord injury, specific loss of certain tracts, or gray matter areas and pain.61
Similar to the excitotoxic SCI model, there has been little pharmacologic validation of either the hemisection or contusion model with clinical analgesic drugs, except for the anticonvulsant gabapentin.62 The positive effect of an SSRI on above-level SCI pain in the hemisection suggests that this class of drug may be clinically useful. However, there is a lack of positive clinical evidence supporting the use of an SSRI in SCI pain.27 In fact, the use of SSRI and tricyclic antidepressants (TCAs) such as amitriptyline may enhance SCI-induced spasticity and TCAs may also be contraindicated because of side effects such as urinary retention and constipation.27,63 To increase confidence that novel analgesic interventions tested in these models will succeed, more pharmacologic validation studies are necessary.