- Email: [email protected]
The necessity of animal models in pain research
Ò
PAIN 151 (2010) 12–17
www.elsevier.com/locate/pain
Comprehensive review
The necessity of animal models in pain research Jeffrey S. Mogil a,*, Karen D. Davis b,c, Stuart W. Derbyshire d a
Department of Psychology and Alan Edwards Centre for Research on Pain, McGill University, Montreal, QC, Canada H3A 1B1 Division of Brain, Imaging and Behaviour-Systems Neuroscience, Toronto Western Research Institute, University Health Network, Toronto, Canada c Department of Surgery and Institute of Medical Science, University of Toronto, Toronto, ON, Canada M5T 2S8 d School of Psychology, University of Birmingham, Birmingham B15 2TT, UK b
a r t i c l e
i n f o
Article history: Received 22 June 2010 Accepted 15 July 2010
a b s t r a c t There exists currently a fair degree of introspection in the pain research community about the value of animal research. This review represents a defense of animal research in pain. We discuss the inherent advantage of animal models over human research as well as the crucial complementary roles animal studies play vis-à-vis human imaging and genetic studies. Finally, we discuss recent developments in animal models of pain that should improve the relevance and translatability of findings using laboratory animals. We believe that pain research using animal models is a continuing necessity–to understand fundamental mechanisms, identify new analgesic targets, and inform, guide and follow up human studies–if novel analgesics are to be developed for the treatment of chronic pain. Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Basic scientific understanding of pain processing and modulation has greatly increased over the past few decades. However, our knowledge of the intricate molecular, cellular, and systems organization of nociception remains substantially incomplete. Furthermore, the management of pain in both acute and especially chronic settings remains far from optimal. Although the pharmaceutical industry has made substantial investments in analgesic drug development, a paucity of analgesics acting at novel molecular targets have been approved. Also lacking are new, more effective surgical targets and behavioral strategies for pain control, despite the clear need to improve upon the relatively modest efficacy of current treatments for many chronic pain conditions. This unfortunate state of affairs – whether accurate or simply ‘‘looking at the glass half empty” – has engendered considerable cynicism in the value of the animal models of pain that are currently at the core of the research and drug development enterprise. Simultaneously, new options have presented themselves in pain research using humans as subjects. The number of pain-related neuroimaging studies on human volunteers and patients has exploded, and there is increased interest in complementary human experimental techniques including quantitative sensory testing, microdialysis, epidemiology, physiology (e.g., nerve conduction studies), ex vivo studies of human cells and tissues, and both DNA- and RNA-based genetic
* Corresponding author. Address: Department of Psychology and Alan Edwards Centre for Research on Pain, McGill University, 1205 Drive Penfield Avenue, Montreal, QC, Canada H3A 1B1. Tel.: +1 514 398 6085; fax: +1 514 398 4896. E-mail address: [email protected] (J.S. Mogil).
studies. The perceived failure of animal studies for analgesic drug development and the increasing interest in human-based techniques has led some to call for the replacement of animal pain experiments with human volunteer studies [34]. We believe this would be a grievous mistake. The following represents a case for the continued reliance on, and necessity of, animal models in pain research. It is instructive to begin with a consideration of the current state of pain research. Of all primary research papers published in the journal Pain, from its inception in 1975 until 2007 [46], approximately two-thirds were human-based, studying either patients with pain (47.8%) or healthy human volunteers (17.2%) as experimental subjects. The remaining third of published papers were based on studies in laboratory animals. In almost two-thirds of those, the subjects were awake and behaving, with the overwhelming majority testing rats and mice [40]. Thus, humans, not animals, have always represented the lion’s share of pain research. Most of these studies, however, are primarily aimed at characterizing pain states; only a small percentage of human studies directly test the anatomical, biochemical or physiological mechanisms of pain [46]. Two quantitative trends observed in this analysis [46] are of interest here. First, the percentage of studies published in Pain with awake animals as subjects has increased steadily over 30+ years, from less than 10% to more than 30%. Closer analysis reveals that studies including a behavioral pharmacology component underlie most of this increase. Correspondingly, the percentage of human studies has decreased, significantly in the case of human patient studies. However, the percentage of imaging studies has increased, such that they represented almost 10% of published studies in Pain by 2007.
0304-3959/$36.00 Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2010.07.015
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
2. The value of animal models in understanding pain Over a decade ago Pat Wall lambasted neuroscientists for their attempts to understand pain as a result of activity within dedicated line systems because of his view of pain as far too complex for such simplicity [61]. Bud Craig and colleagues responded by explaining that approaching the mammalian nervous system as being functionally and anatomically well-organized has allowed for notable progress in understanding both the brain and pain [10]. Wall and Craig were both correct. Pain is psychologically complex and, as well as being a proper subject matter for basic scientists, will become fully comprehensible only when subjected to psychological and sociological interrogation. If you want to know what pain feels like then you must interrogate a conscious person and not a nerve fiber. But if you want to know how a degenerate sciatic nerve creates an afferent barrage – often experienced as pain in a conscious person – then it is a better idea to interrogate the sciatic nerve and not the patient. This Cartesian divide may be unsatisfactory for many reasons but it continues to serve us well and, critically, neuroimaging, genotyping, tissue databanks and epidemiological statistics do not challenge that divide. Conscious persons can no more be found in colorful brain images, DNA, tissue fragments or population statistics than they can be found in animal models. Technically, animal models offer extremely fine characterization of neurochemistry and anatomy, and the excellent temporal and spatial resolution and direct recording of electrophysiology. Animal models also have obvious advantages over human subjects with respect to standardization of genetic and environmental backgrounds, safety, and economy. Although DNA can be as easily obtained from humans as from animals, mRNA from pain-relevant tissues can usually only be obtained from animals except in unusual circumstances. Most importantly, animal studies allow controlled investigation of chronic pain conditions that are simply impossible to perform in humans. A feature of several animal models of peripheral neuropathic pain, for example, is partial denervation. Loose ligation of the whole peripheral nerve [2], tight ligation of a part of a large peripheral nerve [57], or a tight ligation of an entire spinal segmental nerve [28] are all common techniques to a create a mixture of intact and injured fibers. Human models of neuropathic pain can, obviously, not be created in this way. Is capsaicin-induced sensitization in human ‘‘normals” really a better model of neuropathic pain? The subject (human vs. rodent) is more appropriate, of course, but the assay (capsaicin vs. actual nerve damage) is a mere proxy, only as appropriate as the assumptions that underlie it. One could use actual patients as subjects instead of human normals, but this introduces difficult to control secondary factors that greatly complicate interpretation of experimental data. One way in which humans and animals are identical with respect to pain research is that pain is a subjective phenomenon in both species. We might never be able to know what pain feels like in a rat, but we will never know what pain feels like in you either. In both humans and animals we infer pain based on behavior. Humans, however, have the unique behavior of speech, and speech provides relatively rapid and direct access to subjective experience via introspective ratings and descriptors. The ability of human subjects to reflect on their subjective experience is of considerable advantage when investigating chronic pain conditions that have no obvious cause, or where the pain exceeds what is typically expected from a given cause [35]. These functional pain syndromes can be accompanied by complex personal and autobiographical content that cannot be easily captured in animal experiments (although attempts are being made; see Section 6). In human subjects, the subjective experience can at least be associated with objective measures such as those obtained from neuroimaging, and it is obvious that any objective correlate of the pain experience will more closely model the human condition when studied in hu-
13
mans. A further, and relatively new, advantage of human subjects over rodents is that they can be persuaded to stay motionless in scanning devices without training or anesthesia; human subjects can generally follow complex instructions quickly and readily.
3. Do animal models predict analgesic efficacy in humans? The advantages of animal models in the exploration of basic physiological mechanisms of pain is only one of two major reasons animals have been employed in pain research; the other is the prediction of analgesic efficacy leading to clinical drug development. One can explore the relevance of animal models towards this second aim empirically. Despite the impressive achievements of human imaging studies, this approach has not yet discovered specific brain regions not already implicated in pain based on animal experiments, although the organization of such brain areas into networks can be derived from human neuroimaging in a manner not previously possible using electrophysiology. Neither has a molecule or pain-related phenomenon (e.g., counter-irritation, stress-induced pain modulation) ever been found in humans that did not have a rodent counterpart. Nonetheless, there are celebrated examples of failed ‘‘translation,” where efficacy in animal models predicted efficacy in human clinical trials, but no efficacy was found. The most definitive example of this failure to translate is with neurokinin-1 (NK1) antagonists (i.e., MK-869), which failed despite demonstrations of adequate exposure, penetration, and occupancy [21]. There are other less-openly reported failures as well, including glycine-site antagonists [63] and sodium channel blockers [62] for neuropathic pain, and neuronal gap-junction blockers for migraine [19]. In 2010 alone, analgesic drugs failing to show efficacy in phase 2 or phase 3 clinical trials include ralfinamide (a mixed Nav1.7 blocker and NMDA antagonist) and indantadol (a mixed monoamine oxidase inhibitor and NMDA antagonist). This problem of translation, however, is not specific to the pain field. In fact, from 1991 to 2000 there was a 17% success rate from ‘‘first-in-man” (i.e., clinical trial phase 1) to drug registration for pain-related pharmaceuticals compared to 11% for all drug categories [29], although it should be noted that many of the successful pain drugs are simply new formulations or congeners of old compounds. More worrisome is the fact that as of 2000 the primary reason for attrition was lack of efficacy, responsible for over 25% of clinical failures [29]. In other words, these are drugs that worked in animal models, with no intractable pharmacokinetic or toxicity issues, but were subsequently demonstrated as no more effective than placebo in pain patients. Are animal models to blame? It’s difficult to know for sure. First, virtually all clinically effective compounds have been ‘‘back-translated” to show efficacy, at appropriate exposure levels, in existing animal models [30,65]. Second, at least one compound represents an obviously successful ‘‘forward” translation, the snail conopeptide, ziconotide (PrialtÒ) (see [54]). The drug is a synthetic form of the Conus magus peptide, x-MVIIA. The peptide was found to be neuroactive after intracranial injection, and later found to bind with high-affinity to N-type voltage sensitive calcium channels, which themselves are upregulated in the dorsal horn after inflammation or nerve damage. Intrathecal ziconotide was then found to produce strong analgesic effects in a broad array of animal models, and after three positive randomized control trials is now approved by the U.S. Food and Drug Administration (FDA) and European Medicines Agency (EMEA) for the treatment of severe chronic pain. Ziconotide is no panacea, requiring intrathecal injection and featuring serious adverse effects, but it does represent a clear example of successful ‘‘rational” analgesic drug development. It should also be noted, however, that although animal models successfully predicted the efficacy of ziconotide, they failed to predict its side effects.
14
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
Ultimately, enthusiasm among stakeholders in analgesic drug development regarding the value of animal models in drug development at any point in time seems very much to depend on a running ‘‘tally” of successes and failures. In terms of recent developments, at the time of writing it appears that TRPV1 antagonists have failed (due to side effects), but tanezumab (a humanized monoclonal nerve growth factor antibody) has just successfully passed a phase 3 knee osteoarthritis pain study.
4. The value of animal models in complementing human neuroimaging One might question whether animal models are still needed to study pain given the advancement of human neuroimaging. We argue that they are, given limitations imposed by human experimentation and technical limitations of neuroimaging to reveal the function and structure at the cellular level. There are several technical limitations of current human neuroimaging technologies that impact specificity and sensitivity of the findings. A review of some of these issues has been previously discussed [11,12] and so only three main factors will be mentioned here. First and most critical are those limitations that preclude studies of individual neuronal events that provide the basic information code about pain. For example, it is not always possible to distinguish the precise location of where pain functions reside compared to other related functions. Electrophysiological studies in humans have revealed that neurons that encode pain, and neurons that encode various cognitive/attention or emotional salient information, are located within the same area of the cingulate cortex [13,14,22]; this co-localization cannot be resolved with human neuroimaging. The specificity and sensitivity of electrophysiology provide a unique window into the workings of the elemental components that form the basis of the nociceptive system. Although there are a few laboratories that have had unique opportunities to record from neurons in the human brain during neurosurgical procedures, these studies are rare and data are only obtainable from patients being treated for a serious intractable condition. Thus, our knowledge of the normal physiological properties of single neurons must be derived from animal studies. The second main limitation of neuroimaging with functional magnetic resonance imaging (fMRI) is that the blood oxygen level-dependent (BOLD) signal can reach a ceiling, and so it may not be able to distinguish very high levels of neuronal activity. A third limitation of human neuroimaging is that the technology is not yet able to interrogate very small areas of the central nervous system – such as specific lamina of the dorsal horn, the dorsal root ganglia, or the peripheral nervous system – with any precision. This limitation is unlikely to be resolved because most neuroimaging techniques are based on indirect proxies of neuronal activity, or integrate electrical signals across large sampling regions. Clearly, an essential role of animal models of pain is to guide, inform and complement imaging and other types of human-based studies. First, animal studies are important in focusing human studies to relevant questions. The ability to interrogate the entire human brain in a functional or structural imaging study can provide a wealth of information. However, this can also lead to ‘‘information overload” and statistical multiple comparison problems because whole-brain analysis typically involves over one million data points (i.e., there can be >1 million imaging voxels in a given brain). A hypothesis-driven approach to test reasonable predictions is often used to avoid this problem. The vast knowledge accumulated from animal studies of subcortical nociceptive pathways and spinal and thalamic neuronal responses to noxious stimuli provides the fundamental basis from which we can interpret neuroimaging findings [4,11,66]. However, we know very little
about cortical nociceptive neurons and their anatomical connections. Thus, we propose that animal studies now need to focus on the function and organization of cortical nociceptive neurons and pathways. Indeed, although much animal research continues to be focused on the dorsal root ganglion, spinal cord, and brain stem nuclei, increasingly rodent studies of cortical pain processing are being successfully implemented (e.g., [24,25,33,67]). In tandem with this effort human studies should then ask reasonable questions based on strong fundamental knowledge of neurons and pathways implicated in pain based on the animal models. Second, animal studies can help develop appropriate neuroimaging experimental designs, tasks, stimuli and controls. Although there is a segregation of nociceptive and non-nociceptive neurons at the spinal level, and some clustering at brainstem and thalamic areas, the few electrophysiological studies that have recorded from cortical neurons (e.g., in S1, S2, insula, cingulate) indicate that these brain regions likely contain more non-nociceptive neurons than nociceptive neurons (e.g., [27]). A good experimental design requires strong knowledge about such neuronal response properties, the stimuli that can influence their activity and even the basal (‘‘resting state”) activity of neurons contained within each imaging data voxel. These crucial factors are not clearly understood at the cortical level and so impede interpretation of imaging data derived from a sampling space (i.e., voxel) in the order of several square millimeters that contain, at a minimum, thousands of neurons (see [11]). Therefore, a greater understanding of these fundamental issues must be derived from animal studies to inform neuroimaging design to yield specific and relevant information about the human pain system. Third, animal studies can inform the interpretation of neuroimaging findings in light of the spatial and temporal limitations of imaging technologies. For example, data in each functional MRI voxel reflect information derived from hemodynamic changes over several seconds presumably due to activity in thousands of neurons. Similarly, structural MRI measures of brain density/volume and cortical thickness derive from voxels containing an unknown mix of neurons, axons, glia and other elements. Thus, interpretation of the meaning of functional and structural imaging data requires an understanding of the physiological and anatomical properties of the elements within the voxel (e.g., excitatory and inhibitory neurons, synaptic activity, dendritic spines, glia) (see [11,12]). The potential sources of neuronal influences on the elements within that voxel are also needed for a complete picture of the mechanisms underlying MRI data. A detailed understanding of neuronal circuitry is also needed to fully understand white matter structural imaging data obtained from diffusion tensor imaging and functional connectivity findings obtained from task-free ‘‘resting state” functional MRI studies. Decades ago, pioneers in the pain field such as Pat Wall, Ron Melzack and Ken Casey all recognized that the pain experience included an emotional/motivational/affect component and could be shaped by attention and cognitive factors [37–39]. There is also electrophysiological evidence for attentional and multisensory effects on brainstem and cortical nociceptive responses (e.g., [17,23]). Clearly, more animal studies of this nature are needed to understand the location, preponderance and capacity of neurons to modulate their response to noxious stimuli and thus to properly design experiments that take these cognitive effects into consideration.
5. The value of animal models in complementing human genetics Another new technique that some would suggest be used to replace animal studies is investigation into the human genomics of pain. Indeed, DNA can be obtained from patients or normals via
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
blood or buccal swab, and both linkage and association methodologies are in common use in humans. Linkage mapping in mice has proven its translational value, as pain-relevant genes identified in mice [45,47] (at a fraction of the cost of human genetics studies) have been shown to play analogous roles in humans [47, W. Maixner, unpublished data]. Linkage mapping works in humans as well, as virtually all known monogenic disorders of pain have their responsible gene currently identified (see [32]). However, all such disorders are extraordinarily rare, and genes affecting common disorders can only be identified in humans by association study or large-scale sequencing projects. As has happened in other fields, candidate gene-based association studies of pain have thus far largely failed to replicate [32,41]. Hugely expensive genome-wide association studies (GWASs) have not yet been performed on pain-related traits, and even their utility has also been questioned [20,36]. Eventually, costs will decrease to the point where pain will be the subject of both GWASs and sequencing efforts. But what then? The great advantage of the genetic approach is that novel trait-related molecules can be identified via their genomic position alone, without any prior evidence for their involvement. When such genes are eventually identified by GWAS or sequencing, how will we ever know how they function as mediators of pain without follow-up studies in animals (e.g., transgenic knockout mice)? What otherwise would be the point of their identification?
6. New developments in animal models of pain Whatever one’s thoughts are about the value of classical animal models of pain, it is obvious that major developments have occurred in recent years (see [40]). There are essentially six main criticisms of the dominant paradigms in animal pain testing. The first is that too much emphasis has been placed on reflexive withdrawal from mechanical or thermal (hot and cold) stimuli as a dependent measure [60]. The measurement of reflexive withdrawals is thought to be suboptimal both in terms of its poor match with human symptoms and also the considerable experimenter bias involved [5,8]. There is some irony here, because at the same time the trend in human testing has been to move away from rating scales and questionnaires to quantitative sensory testing paradigms that use very similar psychophysical approaches [52]. Recently, a number of conditioning paradigms have been developed, based either on classical conditioning (i.e., place aversion to environments paired with pain or place preference to environments paired with pain relief), training rodents to escape from painful stimuli (i.e., operant conditioning), or using motivational conflicts between pain and other drives (e.g., thirst) (see [40]). The use of such paradigms often leads to different or even opposite conclusions compared to reflex models; for example, low doses of morphine attenuated rats’ tendency to escape from a heated plate (44–50 °C) but increased reflexive guarding responses [59]. The second criticism is that all the emphasis has been placed on measuring pain itself, with almost none placed on important states that accompany pain, either as sequelae or comorbidities. Chronic pain patients experience disability, anxiety, depression, cognitive dysfunction, sleep loss, loss of libido, social withdrawal, and any number of other symptoms that could be modelled in animals experiencing chronic inflammatory or neuropathic states. Thus far the results are decidedly equivocal, but there is definitely some evidence at least that similar changes may occur in rodents (see [40]). There have been tremendous recent advances in the ability to automatically track not only animal position and locomotion but gross behavior (e.g., grooming, drinking, eating, social approach) [53], and any number of studies are ongoing that aim to provide a full ‘‘ethogram” of the 24-h-a-day behavior of rats and mice with chronic pain.
15
The third criticism is that the existing models are too artificial, with inflammatory mediators such as formalin, carrageenan and Freund’s adjuvant standing in for arthritis, and surgical nerve damage standing in for painful diabetic neuropathy and post-herpetic neuralgia. Great strides are being made here too, for example with the development of post-operative (incision) pain [6], cancer pain [56] and chemotherapy-induced neuropathic pain [49,58] that have far more obvious face validity to human (patho)physiological conditions featuring pain. A fourth criticism is that whereas the primary symptoms of chronic pain in humans are spontaneous pain (either continuous or paroxysmal), numbness and dysesthesia [1,55], basic scientists overwhelmingly measure thermal (in 48% of studies) and mechanical (in 42% of studies) hypersensitivity [43]. For both neuropathic and non-neuropathic pain, mechanical and especially thermal hypersensitivity are symptoms of limited prevalence, importance and correlation with global ratings of pain severity [1,55]. There are exceptions to this generalization, of course; disorders like post-herpetic neuralgia and vestibulodynia feature mechanical hypersensitivity almost exclusively. As reviewed previously, any number of dependent measures of spontaneous pain in rodents have been proposed, but most are problematic in terms of their specificity, frequency, practicality, and susceptibility to confound [43]. A recent study specifically evaluated the utility of a number of these measures in the context of nerve injury in mice. We found that hypolocomotion and decreased rearing, licking, lifting, flinching and shaking are simply not present in nerve-damaged mice at levels exceeding sham-operated or unoperated controls, and that although changes in dynamic weight bearing (i.e., gait abnormalities) were often present, the severity of the gait changes could not be used as a measure of any pain-related phenomenon [44]. In contrast, in certain pain states mice present with reliably and accurately scored facial expressions of pain [33], and these of course are spontaneously emitted behaviors. Although previous investigations were negative [26,64], a new study suggests that with improvements in design and habituation, chronic pain states are accompanied by increased ultrasonic vocalizations in mice [31]. Fifth, there exists an obvious mismatch between the epidemiological realities of chronic pain prevalence in the human population and the usual choices of animal model subjects. That is, although chronic pain sufferers are overwhelmingly female [3], the prevalence of chronic pain is higher in the middle-aged and elderly than in young adults [18], and humans feature considerable genetic variability [48], pain research is overwhelmingly performed on young adult, male rats and mice of a single strain [40,42]. This choice is based on convenience and inertia, and on (empirically false [42]) expectations of higher variability in populations other than young male animals. There are any number of examples of conclusions from animal studies later found to be utterly dependent on the choice of subject sex or strain (e.g., [7,45,47]). As researchers become more sensitized to the robust influence of such factors on the results of animal pain experiments, the utility and predictiveness of findings obtained should improve. Finally, it has been argued [51] that design issues and reporting standards in animal experiments are greatly inferior to those currently prevailing in human clinical trials. Specifically, details regarding blinding, randomization, and data dropouts are rarely reported in animal studies of pain, likely leading to high experimental bias. Of course, only a minority of rodent pain studies are specifically designed to assess analgesic efficacy, but especially in those more methodological detail would only enhance transparency and efforts to replicate. Many journals now have online methods (or supplementary methods) sections with no practical word limits. Overall, then, we contend that despite any perceived limitations of current animal models of pain, we have at the present time the
Ò
16
J.S. Mogil et al. / PAIN 151 (2010) 12–17
ability to greatly improve their implementation, and the near future will feature animal pain research of ever greater relevance to clinical pain in humans. 7. Ethical objections to pain research We understand that many people object to animal research, and especially animal pain research, on principle [50]. A detailed discussion of such objections belongs elsewhere [9,15,16]. Our aim here is to provide a scientific rationale for the continuation of animal research as an important mechanism in advancing our understanding of pain. Those that object to animal research on principle will, understandably, be unmoved by the scientific advances animal research provides. 8. Conclusions The need for greater understanding of the fundamental physiology underlying pain will persist at least as long as treatment of chronic pain patients remains suboptimal. As a scientific matter, we see no scenario in the short-to-medium term whereby true advances are possible without the participation of animal models. That being said, we believe that the combination of human imaging (and other human) studies along with ever-more-relevant animal studies will lead to far more effective and translatable science than has been the case thus far. Thus, the time is ripe for true translational (‘‘bedside-to-bench-to-bedside”) pain research. Conflict of interest statement None of the authors have a conflict of interest. References [1] Backonja M-M, Stacey B. Neuropathic pain symptoms relation to overall pain rating. J Pain 2004;5:491–7. [2] Bennett GJ, Xie Y-K. A peripheral mononeuropathy in rat that produces disorders of pain sensation like those seen in man. Pain 1988;33:87–107. [3] Berkley KJ. Sex differences in pain. Behav Brain Sci 1997;20:371–80. [4] Besson J-M, Chaouch A. Peripheral and spinal mechanisms of nociception. Physiol Rev 1987;67:67–186. [5] Bove G. Mechanical sensory threshold testing using nylon monofilaments: the pain field’s ‘‘Tin Standard”. Pain 2006;124:13–7. [6] Brennan TJ, Vandermeulen EP, Gebhart GF. Characterization of a rat model of incisional pain. Pain 1996;64:493–502. [7] Chanda ML, Mogil JS. Sex differences in the effects of amiloride on formalin test nociception in mice. Am J Physiol Regul Integr Comp Physiol 2005;291: R335–42. [8] Chesler EJ, Wilson SG, Lariviere WR, Rodriguez-Zas SL, Mogil JS. Influences of laboratory environment on behavior. Nat Neurosci 2002;5:1101–2. [9] Cohen C. The case for the use of animals in biomedical research. N Engl J Med 1986;315:867–8. [10] Craig AD, Zhang ET, Bushnell MC, Blomqvist A. A reply. Pain 1995;62: 391–3. [11] Davis KD. Neurophysiological and anatomical considerations in functional imaging of pain. Pain 2003;105:1–3. [12] Davis KD. Recent advances and future prospects in neuroimaging of acute and chronic pain. Future Neurol 2006;1:202–13. [13] Davis KD, Hutchison WD, Lozano AM, Tasker RR, Dostrovsky JO. Human anterior cingulate cortex neurons modulated by attention-demanding tasks. J Neurophysiol 2000;83:3575–7. [14] Davis KD, Taylor KD, Hutchison WD, Dostrovsky JO, McAndrews MP, Richter EO, Lozano AM. Human anterior cingulate cortex neurons encode cognitive and emotional demands. J Neurosci 2005;25:8402–6. [15] Derbyshire SWG. Why animal rights are wrong. In: Lee E, editor. Animal experiments: good or bad? London: Hodder & Stoughton; 2002. [16] Derbyshire SWG. Time to abondon the three Rs. Scientist 2006;20:23. [17] Dong WK, Chudler EH, Sugiyama K, Roberts VJ, Hayashi T. Somatosensory, multisensory, and task-related neurons in cortical area 7b (PF) of unanesthetized monkeys. J Neurophysiol 1994;72:542–64. [18] Gagliese L, Melzack R. Chronic pain in elderly people. Pain 1997;70:3–14. [19] Goadsby PJ, Ferrari MD, Csanyi A, Olesen J, Mills JG. Randomized, double-blind, placebo-controlled, proof-of-concept study of the cortical spreading depression inhibiting agent tonabersat in migraine prophylaxis. Cephalalgia 2009;29:742–50.
[20] Goldstein DB. Common genetic variation and human traits. N Engl J Med 2009;360:1696–8. [21] Hill R. NK1 (substance P) receptor antagonists – why are they not analgesic in humans? Trends Pharmacol Sci 2000;21:244–6. [22] Hutchison WD, Davis KD, Lozano AM, Tasker RR, Dostrovsky JO. Pain-related neurons in the human cingulate cortex. Nat Neurosci 1999;2:403–5. [23] Iwata K, Kamo H, Ogawa A, Tsuboi Y, Noma N, Mitsuhashi Y, Taira M, Koshikawa N, Kitigawa J. Anterior cingulate neuronal activity during perception of noxious thermal stimuli in monkeys. J Neurophysiol 2005;94:1980–91. [24] Jasmin L, Rabkin SD, Granato A, Boudah A, Ohara PT. Analgesia and hyperalgesia from GABA-mediated modulation of the cerebral cortex. Nature 2003;424:316–20. [25] Jeon D, Kim S, Chetana M, Jo D, Ruley HE, Lin S-Y, Rabah D, Kinet J-P, Shin H-S. Observational fear learning involves affective pain system and Cav1.2 Ca2+ channels in ACC. Nat Neurosci 2010;13:482–8. [26] Jourdan D, Ardid D, Eschalier A. Analysis of ultrasonic vocalisation does not allow chronic pain to be evaluated in rats. Pain 2002;95:165–73. [27] Kenshalo DR, Iwata K, Sholas M, Thomas DA. Response properties and organization of nociceptive neurons in area 1 of monkey primary somatosensory cortex. J Neurophysiol 2000;84:719–29. [28] Kim SH, Chung JM. An experimental model for peripheral neuropathic produced by segmental spinal nerve ligation in the rat. Pain 1992;50:355–63. [29] Kola I, Landis J. Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov 2004;3:711–6. [30] Kontinen VK, Meert TF. Predictive validity of neuropathic pain models in pharmacological studies with a behavioral outcome in the rat: a systematic review. In: Dostrovsky JO, Carr DB, Koltzenburg M, editors. Proceedings of the 10th world congress on pain. Seattle: IASP Press; 2002. p. 489–98. [31] Kurejova M, Nattenmuller U, Hildebrandt U, Selvaraj D, Stosser S, Kuner R. An improved behavioural assay demonstrates that ultrasound vocalizations constitute a reliable indicator of chronic cancer pain and neuropathic pain. Mol Pain 2010;6:18. [32] LaCroix-Fralish ML, Mogil JS. Progress in genetic studies of pain and analgesia. Annu Rev Pharmacol Toxicol 2009;49:97–121. [33] Langford DL, Bailey AL, Chanda ML, Clarke SE, Drummond TE, Echols S, Glick S, Ingrao J, Klassen-Ross T, LaCroix-Fralish ML, Matsumiya L, Sorge RE, Sotocinal SB, Tabaka JM, Wong D, van den Maagdenberg AMJM, Ferrari MD, Craig KD, Mogil JS. Coding of facial expressions of pain in the laboratory mouse. Nat Methods 2010;7:447–9. [34] Langley CK, Aziz Q, Bountra C, Gordon N, Hawkins P, Jones A, Langley G, Nurmikko T, Tracey I. Volunteer studies in pain research – opportunities and challenges to replace animal experiments: the report and recommendations of a focus on alternatives workshop. Neuroimage 2008;42:467–73. [35] Mayer EA, Bushnell MC. Functional pain syndromes: presentation and pathophysiology. Seattle: IASP Press; 2009. [36] McClellan J, King M-C. Genetic heterogeneity in human disease. Cell 2010;141:210–7. [37] Melzack R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975;1:277–99. [38] Melzack R, Casey KL. Sensory, motivational, and central control determinants of pain: a new conceptual model. In: Kenshalo DR, editor. The skin senses. Springfield, IL: Chas C. Thomas; 1968. p. 423–39. [39] Melzack R, Wall PD. Pain mechanisms: a new theory. Science 1965;150:971–9. [40] Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci 2009;10:283–94. [41] Mogil JS. Are we getting anywhere in human pain genetics? Pain 2009;146:231–2. [42] Mogil JS, Chanda ML. The case for the inclusion of female subjects in basic science studies of pain. Pain 2005;117:1–5. [43] Mogil JS, Crager SE. What should we be measuring in behavioral studies of chronic pain in animals? Pain 2004;112:12–5. [44] Mogil JS, Graham AC, Ritchie J, Hughes SF, Austin J-S, Schorscher-Petcu A, Langford DL, Bennett GJ. Hypolocomotion, asymmetrically directed behaviors (licking, lifting, flinching, and shaking) and dynamic weight bearing (gait) changes are not measures of neuropathic pain in mice. Mol Pain 2010;6:34. [45] Mogil JS, Meirmeister F, Seifert F, Strasburg K, Zimmermann K, Reinold H, Austin J-S, Bernardini N, Chesler EJ, Hoffman HA, Hordo C, Messlinger K, Nemmani KVS, Rankin AL, Ritchie J, Siegling A, Smith SB, Sotocinal SB, Vater A, Lehto SG, Klussmann S, Quirion R, Michaelis M, Devor M, Reeh PW. Variable sensitivity to noxious heat is mediated by differential expression of the CGRP gene. Proc Natl Acad Sci USA 2005;102:12938–43. [46] Mogil JS, Simmonds K, Simmonds MJ. Pain research from 1975 to 2007: a categorical and bibliometric meta-trend analysis of every Research Paper published in the journal – Pain. Pain 2009;142:48–58. [47] Mogil JS, Wilson SG, Chesler EJ, Rankin AL, Nemmani KVS, Lariviere WR, Groce MK, Wallace MR, Kaplan L, Staud R, Ness TJ, Glover TL, Stankova M, Mayorov A, Hruby VJ, Grisel JE, Fillingim RB. The melanocortin-1 receptor gene mediates female-specific mechanisms of analgesia in mice and humans. Proc Natl Acad Sci USA 2003;100:4867–72. [48] Ng PC, Levy S, Huang J, Stockwell TB, Walenz BP, Li K, Axelrod N, Busam DA, Strausberg RL, Venter JC. Genetic variation in an individual human exome. PLoS Genet 2008;4:e1000160. [49] Polomano RC, Mannes AJ, Clark US, Bennett GJ. A painful peripheral neuropathy in the rat produced by the chemotherapeutic drug, paclitaxel. Pain 2001;94:293–304.
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17 [50] Regan T. Defending animal rights. Champaign, IL: University of Illinois Press; 2001. [51] Rice ASC, Cimino-Brown D, Eisenach JC, Kontinen VK, LaCroix-Fralish ML, Machin I, Mogil JS, Stohr T. Animal models and the prediction of efficacy in clinical trials of analgesic drugs: a critical appraisal and call for uniform reporting standards. Pain 2008;139:241–5. [52] Rolke R, Baron R, Maier C, Tolle TR, Treede R-D, Beyer A, Binder A, Birbaumer N, Birklein F, Botefur IC, Braune S, Flor H, Huge V, Klug R, Landwehrmeyer GB, Magerl W, Maihofner C, Rolko C, Schaub C, Scherens A, Sprenger T, Valet M, Wasserka B. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006;123:231–43. [53] Roughan JV, Wright-Williams SL, Flecknell PA. Automated analysis of postoperative behaviour: assessment of HomeCageScan as a novel method to rapidly identify pain and analgesic effects in mice. Lab Anim 2009;43:17–26. [54] Schmidtko A, Lotsch J, Freynhagen R, Geisslinger G. Ziconotide for treatment of severe chronic pain. Lancet 2010;375:1569–77. [55] Scholz J, Mannion RJ, Hord DE, Griffin RS, Rawal B, Zheng H, Scoffings D, Phillips A, Guo J, Laing RJ, Abdi S, Decosterd I, Woolf CJ. A novel tool for the assessment of pain: validation in low back pain. PLoS Med 2009;6:e1000047. [56] Schwei MJ, Honore P, Rogers SD, Salak-Johnson JL, Finke MP, Ramnaraine ML, Clohisy DR, Mantyh PW. Neurochemical and cellular reorganization of the spinal cord in a murine model of bone cancer pain. J Neurosci 1999;19:10886–97. [57] Seltzer Z, Dubner R, Shir Y. A novel behavioral model of causalgiform pain produced by partial sciatic nerve injury in rats. Pain 1990;43:205–18. [58] Tanner KD, Reichling DB, Levine JD. Nociceptor hyper-responsiveness during vincristine-induced painful peripheral neuropathic in the rat. J Neurosci 1998;18:6480–91.
17
[59] Vierck CJ, Acosta-Rua A, Nelligan R, Tester N, Mauderli A. Low dose systemic morphine attenuates operant escape but facilitates innate reflex responses to thermal stimulation. J Pain 2002;3:309–19. [60] Vierck CJ, Hansson PT, Yezierski RP. Clinical and pre-clinical pain assessment: are we measuring the same thing? Pain 2008;135:7–10. [61] Wall PD. A pain in the brain and lower parts of the anatomy. Pain 1995;62:389–91. [62] Wallace MS, Rowbotham M, Bennett GJ, Jensen TS, Pladna R, Quessy S. A multicenter, double-blind, randomized, placebo-controlled crossover evaluation of a short course of 4030W92 in patients with chronic neuropathic pain. J Pain 2002;2:227–33. [63] Wallace MS, Rowbotham M, Katz NP, Dworkin RH, Dotson RM, Galer BS, Rauck RL, Backonja M, Quessy SN, Meisner PD. A randomized, double-blind, placebocontrolled trial of a glycine antagonist in neuropathic pain. Neurology 2002;59:1694–700. [64] Wallace VCJ, Norbury TA, Rice ASC. Ultrasound vocalisation by rodents does not correlate with behavioural measures of persistent pain. Eur J Pain 2005;9:445–52. [65] Whiteside GT, Adedoyin A, Leventhal L. Predictive validity of animal pain models? A comparison of the pharmacokinetic–pharmacodynamic relationship for pain drugs in rats and humans. Neuropharmacology 2008;54:767–75. [66] Willis WD, Westlund KN. Neuroanatomy of the pain system and of the pathways that modulate pain. J Clin Neurophysiol 1997;14:2–31. [67] Xu H, Wu L-J, Wang H, Zhang X, Vadakkan KI, Kim SS, Steenland HW, Zhuo M. Presynaptic and postsynaptic amplifications of neuropathic pain in the anterior cingulate cortex. J Neurosci 2008;28:7445–53.
PAIN 151 (2010) 12–17
www.elsevier.com/locate/pain
Comprehensive review
The necessity of animal models in pain research Jeffrey S. Mogil a,*, Karen D. Davis b,c, Stuart W. Derbyshire d a
Department of Psychology and Alan Edwards Centre for Research on Pain, McGill University, Montreal, QC, Canada H3A 1B1 Division of Brain, Imaging and Behaviour-Systems Neuroscience, Toronto Western Research Institute, University Health Network, Toronto, Canada c Department of Surgery and Institute of Medical Science, University of Toronto, Toronto, ON, Canada M5T 2S8 d School of Psychology, University of Birmingham, Birmingham B15 2TT, UK b
a r t i c l e
i n f o
Article history: Received 22 June 2010 Accepted 15 July 2010
a b s t r a c t There exists currently a fair degree of introspection in the pain research community about the value of animal research. This review represents a defense of animal research in pain. We discuss the inherent advantage of animal models over human research as well as the crucial complementary roles animal studies play vis-à-vis human imaging and genetic studies. Finally, we discuss recent developments in animal models of pain that should improve the relevance and translatability of findings using laboratory animals. We believe that pain research using animal models is a continuing necessity–to understand fundamental mechanisms, identify new analgesic targets, and inform, guide and follow up human studies–if novel analgesics are to be developed for the treatment of chronic pain. Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Basic scientific understanding of pain processing and modulation has greatly increased over the past few decades. However, our knowledge of the intricate molecular, cellular, and systems organization of nociception remains substantially incomplete. Furthermore, the management of pain in both acute and especially chronic settings remains far from optimal. Although the pharmaceutical industry has made substantial investments in analgesic drug development, a paucity of analgesics acting at novel molecular targets have been approved. Also lacking are new, more effective surgical targets and behavioral strategies for pain control, despite the clear need to improve upon the relatively modest efficacy of current treatments for many chronic pain conditions. This unfortunate state of affairs – whether accurate or simply ‘‘looking at the glass half empty” – has engendered considerable cynicism in the value of the animal models of pain that are currently at the core of the research and drug development enterprise. Simultaneously, new options have presented themselves in pain research using humans as subjects. The number of pain-related neuroimaging studies on human volunteers and patients has exploded, and there is increased interest in complementary human experimental techniques including quantitative sensory testing, microdialysis, epidemiology, physiology (e.g., nerve conduction studies), ex vivo studies of human cells and tissues, and both DNA- and RNA-based genetic
* Corresponding author. Address: Department of Psychology and Alan Edwards Centre for Research on Pain, McGill University, 1205 Drive Penfield Avenue, Montreal, QC, Canada H3A 1B1. Tel.: +1 514 398 6085; fax: +1 514 398 4896. E-mail address: [email protected] (J.S. Mogil).
studies. The perceived failure of animal studies for analgesic drug development and the increasing interest in human-based techniques has led some to call for the replacement of animal pain experiments with human volunteer studies [34]. We believe this would be a grievous mistake. The following represents a case for the continued reliance on, and necessity of, animal models in pain research. It is instructive to begin with a consideration of the current state of pain research. Of all primary research papers published in the journal Pain, from its inception in 1975 until 2007 [46], approximately two-thirds were human-based, studying either patients with pain (47.8%) or healthy human volunteers (17.2%) as experimental subjects. The remaining third of published papers were based on studies in laboratory animals. In almost two-thirds of those, the subjects were awake and behaving, with the overwhelming majority testing rats and mice [40]. Thus, humans, not animals, have always represented the lion’s share of pain research. Most of these studies, however, are primarily aimed at characterizing pain states; only a small percentage of human studies directly test the anatomical, biochemical or physiological mechanisms of pain [46]. Two quantitative trends observed in this analysis [46] are of interest here. First, the percentage of studies published in Pain with awake animals as subjects has increased steadily over 30+ years, from less than 10% to more than 30%. Closer analysis reveals that studies including a behavioral pharmacology component underlie most of this increase. Correspondingly, the percentage of human studies has decreased, significantly in the case of human patient studies. However, the percentage of imaging studies has increased, such that they represented almost 10% of published studies in Pain by 2007.
0304-3959/$36.00 Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2010.07.015
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
2. The value of animal models in understanding pain Over a decade ago Pat Wall lambasted neuroscientists for their attempts to understand pain as a result of activity within dedicated line systems because of his view of pain as far too complex for such simplicity [61]. Bud Craig and colleagues responded by explaining that approaching the mammalian nervous system as being functionally and anatomically well-organized has allowed for notable progress in understanding both the brain and pain [10]. Wall and Craig were both correct. Pain is psychologically complex and, as well as being a proper subject matter for basic scientists, will become fully comprehensible only when subjected to psychological and sociological interrogation. If you want to know what pain feels like then you must interrogate a conscious person and not a nerve fiber. But if you want to know how a degenerate sciatic nerve creates an afferent barrage – often experienced as pain in a conscious person – then it is a better idea to interrogate the sciatic nerve and not the patient. This Cartesian divide may be unsatisfactory for many reasons but it continues to serve us well and, critically, neuroimaging, genotyping, tissue databanks and epidemiological statistics do not challenge that divide. Conscious persons can no more be found in colorful brain images, DNA, tissue fragments or population statistics than they can be found in animal models. Technically, animal models offer extremely fine characterization of neurochemistry and anatomy, and the excellent temporal and spatial resolution and direct recording of electrophysiology. Animal models also have obvious advantages over human subjects with respect to standardization of genetic and environmental backgrounds, safety, and economy. Although DNA can be as easily obtained from humans as from animals, mRNA from pain-relevant tissues can usually only be obtained from animals except in unusual circumstances. Most importantly, animal studies allow controlled investigation of chronic pain conditions that are simply impossible to perform in humans. A feature of several animal models of peripheral neuropathic pain, for example, is partial denervation. Loose ligation of the whole peripheral nerve [2], tight ligation of a part of a large peripheral nerve [57], or a tight ligation of an entire spinal segmental nerve [28] are all common techniques to a create a mixture of intact and injured fibers. Human models of neuropathic pain can, obviously, not be created in this way. Is capsaicin-induced sensitization in human ‘‘normals” really a better model of neuropathic pain? The subject (human vs. rodent) is more appropriate, of course, but the assay (capsaicin vs. actual nerve damage) is a mere proxy, only as appropriate as the assumptions that underlie it. One could use actual patients as subjects instead of human normals, but this introduces difficult to control secondary factors that greatly complicate interpretation of experimental data. One way in which humans and animals are identical with respect to pain research is that pain is a subjective phenomenon in both species. We might never be able to know what pain feels like in a rat, but we will never know what pain feels like in you either. In both humans and animals we infer pain based on behavior. Humans, however, have the unique behavior of speech, and speech provides relatively rapid and direct access to subjective experience via introspective ratings and descriptors. The ability of human subjects to reflect on their subjective experience is of considerable advantage when investigating chronic pain conditions that have no obvious cause, or where the pain exceeds what is typically expected from a given cause [35]. These functional pain syndromes can be accompanied by complex personal and autobiographical content that cannot be easily captured in animal experiments (although attempts are being made; see Section 6). In human subjects, the subjective experience can at least be associated with objective measures such as those obtained from neuroimaging, and it is obvious that any objective correlate of the pain experience will more closely model the human condition when studied in hu-
13
mans. A further, and relatively new, advantage of human subjects over rodents is that they can be persuaded to stay motionless in scanning devices without training or anesthesia; human subjects can generally follow complex instructions quickly and readily.
3. Do animal models predict analgesic efficacy in humans? The advantages of animal models in the exploration of basic physiological mechanisms of pain is only one of two major reasons animals have been employed in pain research; the other is the prediction of analgesic efficacy leading to clinical drug development. One can explore the relevance of animal models towards this second aim empirically. Despite the impressive achievements of human imaging studies, this approach has not yet discovered specific brain regions not already implicated in pain based on animal experiments, although the organization of such brain areas into networks can be derived from human neuroimaging in a manner not previously possible using electrophysiology. Neither has a molecule or pain-related phenomenon (e.g., counter-irritation, stress-induced pain modulation) ever been found in humans that did not have a rodent counterpart. Nonetheless, there are celebrated examples of failed ‘‘translation,” where efficacy in animal models predicted efficacy in human clinical trials, but no efficacy was found. The most definitive example of this failure to translate is with neurokinin-1 (NK1) antagonists (i.e., MK-869), which failed despite demonstrations of adequate exposure, penetration, and occupancy [21]. There are other less-openly reported failures as well, including glycine-site antagonists [63] and sodium channel blockers [62] for neuropathic pain, and neuronal gap-junction blockers for migraine [19]. In 2010 alone, analgesic drugs failing to show efficacy in phase 2 or phase 3 clinical trials include ralfinamide (a mixed Nav1.7 blocker and NMDA antagonist) and indantadol (a mixed monoamine oxidase inhibitor and NMDA antagonist). This problem of translation, however, is not specific to the pain field. In fact, from 1991 to 2000 there was a 17% success rate from ‘‘first-in-man” (i.e., clinical trial phase 1) to drug registration for pain-related pharmaceuticals compared to 11% for all drug categories [29], although it should be noted that many of the successful pain drugs are simply new formulations or congeners of old compounds. More worrisome is the fact that as of 2000 the primary reason for attrition was lack of efficacy, responsible for over 25% of clinical failures [29]. In other words, these are drugs that worked in animal models, with no intractable pharmacokinetic or toxicity issues, but were subsequently demonstrated as no more effective than placebo in pain patients. Are animal models to blame? It’s difficult to know for sure. First, virtually all clinically effective compounds have been ‘‘back-translated” to show efficacy, at appropriate exposure levels, in existing animal models [30,65]. Second, at least one compound represents an obviously successful ‘‘forward” translation, the snail conopeptide, ziconotide (PrialtÒ) (see [54]). The drug is a synthetic form of the Conus magus peptide, x-MVIIA. The peptide was found to be neuroactive after intracranial injection, and later found to bind with high-affinity to N-type voltage sensitive calcium channels, which themselves are upregulated in the dorsal horn after inflammation or nerve damage. Intrathecal ziconotide was then found to produce strong analgesic effects in a broad array of animal models, and after three positive randomized control trials is now approved by the U.S. Food and Drug Administration (FDA) and European Medicines Agency (EMEA) for the treatment of severe chronic pain. Ziconotide is no panacea, requiring intrathecal injection and featuring serious adverse effects, but it does represent a clear example of successful ‘‘rational” analgesic drug development. It should also be noted, however, that although animal models successfully predicted the efficacy of ziconotide, they failed to predict its side effects.
14
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
Ultimately, enthusiasm among stakeholders in analgesic drug development regarding the value of animal models in drug development at any point in time seems very much to depend on a running ‘‘tally” of successes and failures. In terms of recent developments, at the time of writing it appears that TRPV1 antagonists have failed (due to side effects), but tanezumab (a humanized monoclonal nerve growth factor antibody) has just successfully passed a phase 3 knee osteoarthritis pain study.
4. The value of animal models in complementing human neuroimaging One might question whether animal models are still needed to study pain given the advancement of human neuroimaging. We argue that they are, given limitations imposed by human experimentation and technical limitations of neuroimaging to reveal the function and structure at the cellular level. There are several technical limitations of current human neuroimaging technologies that impact specificity and sensitivity of the findings. A review of some of these issues has been previously discussed [11,12] and so only three main factors will be mentioned here. First and most critical are those limitations that preclude studies of individual neuronal events that provide the basic information code about pain. For example, it is not always possible to distinguish the precise location of where pain functions reside compared to other related functions. Electrophysiological studies in humans have revealed that neurons that encode pain, and neurons that encode various cognitive/attention or emotional salient information, are located within the same area of the cingulate cortex [13,14,22]; this co-localization cannot be resolved with human neuroimaging. The specificity and sensitivity of electrophysiology provide a unique window into the workings of the elemental components that form the basis of the nociceptive system. Although there are a few laboratories that have had unique opportunities to record from neurons in the human brain during neurosurgical procedures, these studies are rare and data are only obtainable from patients being treated for a serious intractable condition. Thus, our knowledge of the normal physiological properties of single neurons must be derived from animal studies. The second main limitation of neuroimaging with functional magnetic resonance imaging (fMRI) is that the blood oxygen level-dependent (BOLD) signal can reach a ceiling, and so it may not be able to distinguish very high levels of neuronal activity. A third limitation of human neuroimaging is that the technology is not yet able to interrogate very small areas of the central nervous system – such as specific lamina of the dorsal horn, the dorsal root ganglia, or the peripheral nervous system – with any precision. This limitation is unlikely to be resolved because most neuroimaging techniques are based on indirect proxies of neuronal activity, or integrate electrical signals across large sampling regions. Clearly, an essential role of animal models of pain is to guide, inform and complement imaging and other types of human-based studies. First, animal studies are important in focusing human studies to relevant questions. The ability to interrogate the entire human brain in a functional or structural imaging study can provide a wealth of information. However, this can also lead to ‘‘information overload” and statistical multiple comparison problems because whole-brain analysis typically involves over one million data points (i.e., there can be >1 million imaging voxels in a given brain). A hypothesis-driven approach to test reasonable predictions is often used to avoid this problem. The vast knowledge accumulated from animal studies of subcortical nociceptive pathways and spinal and thalamic neuronal responses to noxious stimuli provides the fundamental basis from which we can interpret neuroimaging findings [4,11,66]. However, we know very little
about cortical nociceptive neurons and their anatomical connections. Thus, we propose that animal studies now need to focus on the function and organization of cortical nociceptive neurons and pathways. Indeed, although much animal research continues to be focused on the dorsal root ganglion, spinal cord, and brain stem nuclei, increasingly rodent studies of cortical pain processing are being successfully implemented (e.g., [24,25,33,67]). In tandem with this effort human studies should then ask reasonable questions based on strong fundamental knowledge of neurons and pathways implicated in pain based on the animal models. Second, animal studies can help develop appropriate neuroimaging experimental designs, tasks, stimuli and controls. Although there is a segregation of nociceptive and non-nociceptive neurons at the spinal level, and some clustering at brainstem and thalamic areas, the few electrophysiological studies that have recorded from cortical neurons (e.g., in S1, S2, insula, cingulate) indicate that these brain regions likely contain more non-nociceptive neurons than nociceptive neurons (e.g., [27]). A good experimental design requires strong knowledge about such neuronal response properties, the stimuli that can influence their activity and even the basal (‘‘resting state”) activity of neurons contained within each imaging data voxel. These crucial factors are not clearly understood at the cortical level and so impede interpretation of imaging data derived from a sampling space (i.e., voxel) in the order of several square millimeters that contain, at a minimum, thousands of neurons (see [11]). Therefore, a greater understanding of these fundamental issues must be derived from animal studies to inform neuroimaging design to yield specific and relevant information about the human pain system. Third, animal studies can inform the interpretation of neuroimaging findings in light of the spatial and temporal limitations of imaging technologies. For example, data in each functional MRI voxel reflect information derived from hemodynamic changes over several seconds presumably due to activity in thousands of neurons. Similarly, structural MRI measures of brain density/volume and cortical thickness derive from voxels containing an unknown mix of neurons, axons, glia and other elements. Thus, interpretation of the meaning of functional and structural imaging data requires an understanding of the physiological and anatomical properties of the elements within the voxel (e.g., excitatory and inhibitory neurons, synaptic activity, dendritic spines, glia) (see [11,12]). The potential sources of neuronal influences on the elements within that voxel are also needed for a complete picture of the mechanisms underlying MRI data. A detailed understanding of neuronal circuitry is also needed to fully understand white matter structural imaging data obtained from diffusion tensor imaging and functional connectivity findings obtained from task-free ‘‘resting state” functional MRI studies. Decades ago, pioneers in the pain field such as Pat Wall, Ron Melzack and Ken Casey all recognized that the pain experience included an emotional/motivational/affect component and could be shaped by attention and cognitive factors [37–39]. There is also electrophysiological evidence for attentional and multisensory effects on brainstem and cortical nociceptive responses (e.g., [17,23]). Clearly, more animal studies of this nature are needed to understand the location, preponderance and capacity of neurons to modulate their response to noxious stimuli and thus to properly design experiments that take these cognitive effects into consideration.
5. The value of animal models in complementing human genetics Another new technique that some would suggest be used to replace animal studies is investigation into the human genomics of pain. Indeed, DNA can be obtained from patients or normals via
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17
blood or buccal swab, and both linkage and association methodologies are in common use in humans. Linkage mapping in mice has proven its translational value, as pain-relevant genes identified in mice [45,47] (at a fraction of the cost of human genetics studies) have been shown to play analogous roles in humans [47, W. Maixner, unpublished data]. Linkage mapping works in humans as well, as virtually all known monogenic disorders of pain have their responsible gene currently identified (see [32]). However, all such disorders are extraordinarily rare, and genes affecting common disorders can only be identified in humans by association study or large-scale sequencing projects. As has happened in other fields, candidate gene-based association studies of pain have thus far largely failed to replicate [32,41]. Hugely expensive genome-wide association studies (GWASs) have not yet been performed on pain-related traits, and even their utility has also been questioned [20,36]. Eventually, costs will decrease to the point where pain will be the subject of both GWASs and sequencing efforts. But what then? The great advantage of the genetic approach is that novel trait-related molecules can be identified via their genomic position alone, without any prior evidence for their involvement. When such genes are eventually identified by GWAS or sequencing, how will we ever know how they function as mediators of pain without follow-up studies in animals (e.g., transgenic knockout mice)? What otherwise would be the point of their identification?
6. New developments in animal models of pain Whatever one’s thoughts are about the value of classical animal models of pain, it is obvious that major developments have occurred in recent years (see [40]). There are essentially six main criticisms of the dominant paradigms in animal pain testing. The first is that too much emphasis has been placed on reflexive withdrawal from mechanical or thermal (hot and cold) stimuli as a dependent measure [60]. The measurement of reflexive withdrawals is thought to be suboptimal both in terms of its poor match with human symptoms and also the considerable experimenter bias involved [5,8]. There is some irony here, because at the same time the trend in human testing has been to move away from rating scales and questionnaires to quantitative sensory testing paradigms that use very similar psychophysical approaches [52]. Recently, a number of conditioning paradigms have been developed, based either on classical conditioning (i.e., place aversion to environments paired with pain or place preference to environments paired with pain relief), training rodents to escape from painful stimuli (i.e., operant conditioning), or using motivational conflicts between pain and other drives (e.g., thirst) (see [40]). The use of such paradigms often leads to different or even opposite conclusions compared to reflex models; for example, low doses of morphine attenuated rats’ tendency to escape from a heated plate (44–50 °C) but increased reflexive guarding responses [59]. The second criticism is that all the emphasis has been placed on measuring pain itself, with almost none placed on important states that accompany pain, either as sequelae or comorbidities. Chronic pain patients experience disability, anxiety, depression, cognitive dysfunction, sleep loss, loss of libido, social withdrawal, and any number of other symptoms that could be modelled in animals experiencing chronic inflammatory or neuropathic states. Thus far the results are decidedly equivocal, but there is definitely some evidence at least that similar changes may occur in rodents (see [40]). There have been tremendous recent advances in the ability to automatically track not only animal position and locomotion but gross behavior (e.g., grooming, drinking, eating, social approach) [53], and any number of studies are ongoing that aim to provide a full ‘‘ethogram” of the 24-h-a-day behavior of rats and mice with chronic pain.
15
The third criticism is that the existing models are too artificial, with inflammatory mediators such as formalin, carrageenan and Freund’s adjuvant standing in for arthritis, and surgical nerve damage standing in for painful diabetic neuropathy and post-herpetic neuralgia. Great strides are being made here too, for example with the development of post-operative (incision) pain [6], cancer pain [56] and chemotherapy-induced neuropathic pain [49,58] that have far more obvious face validity to human (patho)physiological conditions featuring pain. A fourth criticism is that whereas the primary symptoms of chronic pain in humans are spontaneous pain (either continuous or paroxysmal), numbness and dysesthesia [1,55], basic scientists overwhelmingly measure thermal (in 48% of studies) and mechanical (in 42% of studies) hypersensitivity [43]. For both neuropathic and non-neuropathic pain, mechanical and especially thermal hypersensitivity are symptoms of limited prevalence, importance and correlation with global ratings of pain severity [1,55]. There are exceptions to this generalization, of course; disorders like post-herpetic neuralgia and vestibulodynia feature mechanical hypersensitivity almost exclusively. As reviewed previously, any number of dependent measures of spontaneous pain in rodents have been proposed, but most are problematic in terms of their specificity, frequency, practicality, and susceptibility to confound [43]. A recent study specifically evaluated the utility of a number of these measures in the context of nerve injury in mice. We found that hypolocomotion and decreased rearing, licking, lifting, flinching and shaking are simply not present in nerve-damaged mice at levels exceeding sham-operated or unoperated controls, and that although changes in dynamic weight bearing (i.e., gait abnormalities) were often present, the severity of the gait changes could not be used as a measure of any pain-related phenomenon [44]. In contrast, in certain pain states mice present with reliably and accurately scored facial expressions of pain [33], and these of course are spontaneously emitted behaviors. Although previous investigations were negative [26,64], a new study suggests that with improvements in design and habituation, chronic pain states are accompanied by increased ultrasonic vocalizations in mice [31]. Fifth, there exists an obvious mismatch between the epidemiological realities of chronic pain prevalence in the human population and the usual choices of animal model subjects. That is, although chronic pain sufferers are overwhelmingly female [3], the prevalence of chronic pain is higher in the middle-aged and elderly than in young adults [18], and humans feature considerable genetic variability [48], pain research is overwhelmingly performed on young adult, male rats and mice of a single strain [40,42]. This choice is based on convenience and inertia, and on (empirically false [42]) expectations of higher variability in populations other than young male animals. There are any number of examples of conclusions from animal studies later found to be utterly dependent on the choice of subject sex or strain (e.g., [7,45,47]). As researchers become more sensitized to the robust influence of such factors on the results of animal pain experiments, the utility and predictiveness of findings obtained should improve. Finally, it has been argued [51] that design issues and reporting standards in animal experiments are greatly inferior to those currently prevailing in human clinical trials. Specifically, details regarding blinding, randomization, and data dropouts are rarely reported in animal studies of pain, likely leading to high experimental bias. Of course, only a minority of rodent pain studies are specifically designed to assess analgesic efficacy, but especially in those more methodological detail would only enhance transparency and efforts to replicate. Many journals now have online methods (or supplementary methods) sections with no practical word limits. Overall, then, we contend that despite any perceived limitations of current animal models of pain, we have at the present time the
Ò
16
J.S. Mogil et al. / PAIN 151 (2010) 12–17
ability to greatly improve their implementation, and the near future will feature animal pain research of ever greater relevance to clinical pain in humans. 7. Ethical objections to pain research We understand that many people object to animal research, and especially animal pain research, on principle [50]. A detailed discussion of such objections belongs elsewhere [9,15,16]. Our aim here is to provide a scientific rationale for the continuation of animal research as an important mechanism in advancing our understanding of pain. Those that object to animal research on principle will, understandably, be unmoved by the scientific advances animal research provides. 8. Conclusions The need for greater understanding of the fundamental physiology underlying pain will persist at least as long as treatment of chronic pain patients remains suboptimal. As a scientific matter, we see no scenario in the short-to-medium term whereby true advances are possible without the participation of animal models. That being said, we believe that the combination of human imaging (and other human) studies along with ever-more-relevant animal studies will lead to far more effective and translatable science than has been the case thus far. Thus, the time is ripe for true translational (‘‘bedside-to-bench-to-bedside”) pain research. Conflict of interest statement None of the authors have a conflict of interest. References [1] Backonja M-M, Stacey B. Neuropathic pain symptoms relation to overall pain rating. J Pain 2004;5:491–7. [2] Bennett GJ, Xie Y-K. A peripheral mononeuropathy in rat that produces disorders of pain sensation like those seen in man. Pain 1988;33:87–107. [3] Berkley KJ. Sex differences in pain. Behav Brain Sci 1997;20:371–80. [4] Besson J-M, Chaouch A. Peripheral and spinal mechanisms of nociception. Physiol Rev 1987;67:67–186. [5] Bove G. Mechanical sensory threshold testing using nylon monofilaments: the pain field’s ‘‘Tin Standard”. Pain 2006;124:13–7. [6] Brennan TJ, Vandermeulen EP, Gebhart GF. Characterization of a rat model of incisional pain. Pain 1996;64:493–502. [7] Chanda ML, Mogil JS. Sex differences in the effects of amiloride on formalin test nociception in mice. Am J Physiol Regul Integr Comp Physiol 2005;291: R335–42. [8] Chesler EJ, Wilson SG, Lariviere WR, Rodriguez-Zas SL, Mogil JS. Influences of laboratory environment on behavior. Nat Neurosci 2002;5:1101–2. [9] Cohen C. The case for the use of animals in biomedical research. N Engl J Med 1986;315:867–8. [10] Craig AD, Zhang ET, Bushnell MC, Blomqvist A. A reply. Pain 1995;62: 391–3. [11] Davis KD. Neurophysiological and anatomical considerations in functional imaging of pain. Pain 2003;105:1–3. [12] Davis KD. Recent advances and future prospects in neuroimaging of acute and chronic pain. Future Neurol 2006;1:202–13. [13] Davis KD, Hutchison WD, Lozano AM, Tasker RR, Dostrovsky JO. Human anterior cingulate cortex neurons modulated by attention-demanding tasks. J Neurophysiol 2000;83:3575–7. [14] Davis KD, Taylor KD, Hutchison WD, Dostrovsky JO, McAndrews MP, Richter EO, Lozano AM. Human anterior cingulate cortex neurons encode cognitive and emotional demands. J Neurosci 2005;25:8402–6. [15] Derbyshire SWG. Why animal rights are wrong. In: Lee E, editor. Animal experiments: good or bad? London: Hodder & Stoughton; 2002. [16] Derbyshire SWG. Time to abondon the three Rs. Scientist 2006;20:23. [17] Dong WK, Chudler EH, Sugiyama K, Roberts VJ, Hayashi T. Somatosensory, multisensory, and task-related neurons in cortical area 7b (PF) of unanesthetized monkeys. J Neurophysiol 1994;72:542–64. [18] Gagliese L, Melzack R. Chronic pain in elderly people. Pain 1997;70:3–14. [19] Goadsby PJ, Ferrari MD, Csanyi A, Olesen J, Mills JG. Randomized, double-blind, placebo-controlled, proof-of-concept study of the cortical spreading depression inhibiting agent tonabersat in migraine prophylaxis. Cephalalgia 2009;29:742–50.
[20] Goldstein DB. Common genetic variation and human traits. N Engl J Med 2009;360:1696–8. [21] Hill R. NK1 (substance P) receptor antagonists – why are they not analgesic in humans? Trends Pharmacol Sci 2000;21:244–6. [22] Hutchison WD, Davis KD, Lozano AM, Tasker RR, Dostrovsky JO. Pain-related neurons in the human cingulate cortex. Nat Neurosci 1999;2:403–5. [23] Iwata K, Kamo H, Ogawa A, Tsuboi Y, Noma N, Mitsuhashi Y, Taira M, Koshikawa N, Kitigawa J. Anterior cingulate neuronal activity during perception of noxious thermal stimuli in monkeys. J Neurophysiol 2005;94:1980–91. [24] Jasmin L, Rabkin SD, Granato A, Boudah A, Ohara PT. Analgesia and hyperalgesia from GABA-mediated modulation of the cerebral cortex. Nature 2003;424:316–20. [25] Jeon D, Kim S, Chetana M, Jo D, Ruley HE, Lin S-Y, Rabah D, Kinet J-P, Shin H-S. Observational fear learning involves affective pain system and Cav1.2 Ca2+ channels in ACC. Nat Neurosci 2010;13:482–8. [26] Jourdan D, Ardid D, Eschalier A. Analysis of ultrasonic vocalisation does not allow chronic pain to be evaluated in rats. Pain 2002;95:165–73. [27] Kenshalo DR, Iwata K, Sholas M, Thomas DA. Response properties and organization of nociceptive neurons in area 1 of monkey primary somatosensory cortex. J Neurophysiol 2000;84:719–29. [28] Kim SH, Chung JM. An experimental model for peripheral neuropathic produced by segmental spinal nerve ligation in the rat. Pain 1992;50:355–63. [29] Kola I, Landis J. Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov 2004;3:711–6. [30] Kontinen VK, Meert TF. Predictive validity of neuropathic pain models in pharmacological studies with a behavioral outcome in the rat: a systematic review. In: Dostrovsky JO, Carr DB, Koltzenburg M, editors. Proceedings of the 10th world congress on pain. Seattle: IASP Press; 2002. p. 489–98. [31] Kurejova M, Nattenmuller U, Hildebrandt U, Selvaraj D, Stosser S, Kuner R. An improved behavioural assay demonstrates that ultrasound vocalizations constitute a reliable indicator of chronic cancer pain and neuropathic pain. Mol Pain 2010;6:18. [32] LaCroix-Fralish ML, Mogil JS. Progress in genetic studies of pain and analgesia. Annu Rev Pharmacol Toxicol 2009;49:97–121. [33] Langford DL, Bailey AL, Chanda ML, Clarke SE, Drummond TE, Echols S, Glick S, Ingrao J, Klassen-Ross T, LaCroix-Fralish ML, Matsumiya L, Sorge RE, Sotocinal SB, Tabaka JM, Wong D, van den Maagdenberg AMJM, Ferrari MD, Craig KD, Mogil JS. Coding of facial expressions of pain in the laboratory mouse. Nat Methods 2010;7:447–9. [34] Langley CK, Aziz Q, Bountra C, Gordon N, Hawkins P, Jones A, Langley G, Nurmikko T, Tracey I. Volunteer studies in pain research – opportunities and challenges to replace animal experiments: the report and recommendations of a focus on alternatives workshop. Neuroimage 2008;42:467–73. [35] Mayer EA, Bushnell MC. Functional pain syndromes: presentation and pathophysiology. Seattle: IASP Press; 2009. [36] McClellan J, King M-C. Genetic heterogeneity in human disease. Cell 2010;141:210–7. [37] Melzack R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975;1:277–99. [38] Melzack R, Casey KL. Sensory, motivational, and central control determinants of pain: a new conceptual model. In: Kenshalo DR, editor. The skin senses. Springfield, IL: Chas C. Thomas; 1968. p. 423–39. [39] Melzack R, Wall PD. Pain mechanisms: a new theory. Science 1965;150:971–9. [40] Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci 2009;10:283–94. [41] Mogil JS. Are we getting anywhere in human pain genetics? Pain 2009;146:231–2. [42] Mogil JS, Chanda ML. The case for the inclusion of female subjects in basic science studies of pain. Pain 2005;117:1–5. [43] Mogil JS, Crager SE. What should we be measuring in behavioral studies of chronic pain in animals? Pain 2004;112:12–5. [44] Mogil JS, Graham AC, Ritchie J, Hughes SF, Austin J-S, Schorscher-Petcu A, Langford DL, Bennett GJ. Hypolocomotion, asymmetrically directed behaviors (licking, lifting, flinching, and shaking) and dynamic weight bearing (gait) changes are not measures of neuropathic pain in mice. Mol Pain 2010;6:34. [45] Mogil JS, Meirmeister F, Seifert F, Strasburg K, Zimmermann K, Reinold H, Austin J-S, Bernardini N, Chesler EJ, Hoffman HA, Hordo C, Messlinger K, Nemmani KVS, Rankin AL, Ritchie J, Siegling A, Smith SB, Sotocinal SB, Vater A, Lehto SG, Klussmann S, Quirion R, Michaelis M, Devor M, Reeh PW. Variable sensitivity to noxious heat is mediated by differential expression of the CGRP gene. Proc Natl Acad Sci USA 2005;102:12938–43. [46] Mogil JS, Simmonds K, Simmonds MJ. Pain research from 1975 to 2007: a categorical and bibliometric meta-trend analysis of every Research Paper published in the journal – Pain. Pain 2009;142:48–58. [47] Mogil JS, Wilson SG, Chesler EJ, Rankin AL, Nemmani KVS, Lariviere WR, Groce MK, Wallace MR, Kaplan L, Staud R, Ness TJ, Glover TL, Stankova M, Mayorov A, Hruby VJ, Grisel JE, Fillingim RB. The melanocortin-1 receptor gene mediates female-specific mechanisms of analgesia in mice and humans. Proc Natl Acad Sci USA 2003;100:4867–72. [48] Ng PC, Levy S, Huang J, Stockwell TB, Walenz BP, Li K, Axelrod N, Busam DA, Strausberg RL, Venter JC. Genetic variation in an individual human exome. PLoS Genet 2008;4:e1000160. [49] Polomano RC, Mannes AJ, Clark US, Bennett GJ. A painful peripheral neuropathy in the rat produced by the chemotherapeutic drug, paclitaxel. Pain 2001;94:293–304.
Ò
J.S. Mogil et al. / PAIN 151 (2010) 12–17 [50] Regan T. Defending animal rights. Champaign, IL: University of Illinois Press; 2001. [51] Rice ASC, Cimino-Brown D, Eisenach JC, Kontinen VK, LaCroix-Fralish ML, Machin I, Mogil JS, Stohr T. Animal models and the prediction of efficacy in clinical trials of analgesic drugs: a critical appraisal and call for uniform reporting standards. Pain 2008;139:241–5. [52] Rolke R, Baron R, Maier C, Tolle TR, Treede R-D, Beyer A, Binder A, Birbaumer N, Birklein F, Botefur IC, Braune S, Flor H, Huge V, Klug R, Landwehrmeyer GB, Magerl W, Maihofner C, Rolko C, Schaub C, Scherens A, Sprenger T, Valet M, Wasserka B. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006;123:231–43. [53] Roughan JV, Wright-Williams SL, Flecknell PA. Automated analysis of postoperative behaviour: assessment of HomeCageScan as a novel method to rapidly identify pain and analgesic effects in mice. Lab Anim 2009;43:17–26. [54] Schmidtko A, Lotsch J, Freynhagen R, Geisslinger G. Ziconotide for treatment of severe chronic pain. Lancet 2010;375:1569–77. [55] Scholz J, Mannion RJ, Hord DE, Griffin RS, Rawal B, Zheng H, Scoffings D, Phillips A, Guo J, Laing RJ, Abdi S, Decosterd I, Woolf CJ. A novel tool for the assessment of pain: validation in low back pain. PLoS Med 2009;6:e1000047. [56] Schwei MJ, Honore P, Rogers SD, Salak-Johnson JL, Finke MP, Ramnaraine ML, Clohisy DR, Mantyh PW. Neurochemical and cellular reorganization of the spinal cord in a murine model of bone cancer pain. J Neurosci 1999;19:10886–97. [57] Seltzer Z, Dubner R, Shir Y. A novel behavioral model of causalgiform pain produced by partial sciatic nerve injury in rats. Pain 1990;43:205–18. [58] Tanner KD, Reichling DB, Levine JD. Nociceptor hyper-responsiveness during vincristine-induced painful peripheral neuropathic in the rat. J Neurosci 1998;18:6480–91.
17
[59] Vierck CJ, Acosta-Rua A, Nelligan R, Tester N, Mauderli A. Low dose systemic morphine attenuates operant escape but facilitates innate reflex responses to thermal stimulation. J Pain 2002;3:309–19. [60] Vierck CJ, Hansson PT, Yezierski RP. Clinical and pre-clinical pain assessment: are we measuring the same thing? Pain 2008;135:7–10. [61] Wall PD. A pain in the brain and lower parts of the anatomy. Pain 1995;62:389–91. [62] Wallace MS, Rowbotham M, Bennett GJ, Jensen TS, Pladna R, Quessy S. A multicenter, double-blind, randomized, placebo-controlled crossover evaluation of a short course of 4030W92 in patients with chronic neuropathic pain. J Pain 2002;2:227–33. [63] Wallace MS, Rowbotham M, Katz NP, Dworkin RH, Dotson RM, Galer BS, Rauck RL, Backonja M, Quessy SN, Meisner PD. A randomized, double-blind, placebocontrolled trial of a glycine antagonist in neuropathic pain. Neurology 2002;59:1694–700. [64] Wallace VCJ, Norbury TA, Rice ASC. Ultrasound vocalisation by rodents does not correlate with behavioural measures of persistent pain. Eur J Pain 2005;9:445–52. [65] Whiteside GT, Adedoyin A, Leventhal L. Predictive validity of animal pain models? A comparison of the pharmacokinetic–pharmacodynamic relationship for pain drugs in rats and humans. Neuropharmacology 2008;54:767–75. [66] Willis WD, Westlund KN. Neuroanatomy of the pain system and of the pathways that modulate pain. J Clin Neurophysiol 1997;14:2–31. [67] Xu H, Wu L-J, Wang H, Zhang X, Vadakkan KI, Kim SS, Steenland HW, Zhuo M. Presynaptic and postsynaptic amplifications of neuropathic pain in the anterior cingulate cortex. J Neurosci 2008;28:7445–53.