Numerous procedures have been described in the animal literature that involve surgical injury of peripheral sensory nerves, typically those innervating the hindpaw in rats or mice. Most if not all of these procedures involve surgical damage to dorsal spinal nerve roots that comprise a major sensory component of the sciatic nerve, of proximal branches of the sciatic nerve, or of varying degrees of injury to the sciatic nerve itself [4, 40, 45]. Regardless of the extent and location of the nerve injury, the primary behavioral outcome that has been used to validate these methods as animal models of neuropathic pain has been hindpaw hypersensitivity to mechanical or thermal stimuli. Typically, an increased response is observed to both nonnoxious stimuli (allodynia) and mildly noxious stimuli (hyperalgesia) [3, 45]. While hypersensitivity phenomena clearly exist in some patients with neuropathic pain, these endpoints are not typically used as the primary outcome measures for assessment of chronic pain in humans. Dysesthesias and paresthesias are as common, if not more common, in this patient population [24]. It has also been suggested that the presence of spontaneous pain at rest typically identified as having a dull, aching quality is the primary complaint of most patients with chronic neuropathic pain, and the symptom for which the majority of pain patients seek relief [2]. Perhaps more germane to this chapter, this is the symptom for which opioids appear to be particularly efficacious in neuropathic pain patients.

The use of hypersensitivity as a primary outcome measure for assessment of pain following peripheral nerve injury in rodents has provided a means to assess the relevance of a wealth of important physiological, biochemical, molecular, and pharmacological findings regarding the mechanisms through which nerve injury and the resulting plasticity produces such behavior. However, given the clinical presentation of neuropathic pain in most patients, perhaps it is not surprising that the use of hypersensitivity as the primary endpoint in preclinical studies has resulted in few novel medications that are efficacious in this population [36, 46]. For this reason, many investigators have sought to describe other behavioral phenomena that appear following peripheral nerve injury in rodents that might serve as surrogate measures for spontaneous pain in neuropathic pain patients. Paw incision results in guarding behavior that coincides with spontaneous electrophysiological activity of sensory afferents [65, 66]. However, similar ventroflexion of the hindpaw following spinal nerve injury in rats was found to be abolished by either dorsal or ventral rhizotomy of the nerve root, suggesting involvement of a motor component in addition to a sensory component [48]. Other behaviors that are altered or appear with moderate acute or subchronic pain stimuli include locomotion or exploration [37, 63], facial expression or grimacing [32, 61], food-maintained operant responding or feeding [38, 62], intracranial self-stimulation [54], and a variety of home cage behaviors [57]. Unfortunately, most if not all of these behaviors are not altered or do not arise in rodents following spinal or peripheral nerve injury [47]. Behaviors that have been found to be altered or produced in rodents following spinal nerve injury include conditioned place aversion to suprathreshold mechanical stimuli, conditioned place preference to certain analgesics administered spinally, conditioned place preference to deep brain stimulation of motor cortex, alteration of intravenous opioid self-administration, self-administration of spinal analgesics, and diminished opioid potentiation of rewarding electrical brain stimulation [17, 20, 27, 41, 42]. The last three of these behaviors are discussed further and the others are discussed within the context of the findings from studies using these paradigms.

A major concern with the treatment of neuropathic pain using opioids is addiction, both from patients and physicians alike. Anecdotal evidence suggests that the subjective effects of opioids differ in the presence of pain, and appropriate use of these drugs for pain relief is possible even with chronic nonmalignant pain. Although a considerable amount of research has identified mechanisms of opioid analgesia and addiction, the influence of chronic pain on the abuse liability of opioids has not been studied to a great extent. Induction of persistent inflammatory pain using Freund’s adjuvant in rats increases morphine intake through self-administration, an effect that is reversed by indomethacin [35]. Oral fentanyl self-administration is also increased in the presence of inflammation in rats [9, 10, 30]. These data suggest that the presence of acute or subchronic inflammatory pain alters opioid intake through self-administration in a manner consistent with titration of an analgesic effect.

In our laboratory, we sought to determine if animals with persistent neuropathic pain would self-administer opioids in order to reverse a subjective pain state, as opposed to normal animals that self-administer opioids to activate classical reward mechanisms. Our hypothesis was that only opioids that alleviated other behavioral measures of neuropathic pain, such as mechanical hypersensitivity, would be self-administered and only at effective doses. Further, we hypothesized that the rate of drug intake through self-administration would be consistent with the time course of reversal of neuropathic pain symptoms, and that alleviation of these behaviors by administration of adjuvant analgesics would decrease opioid intake selectively in rats with neuropathic pain. Using the nerve ligation model of Kim and Chung [26], we examined the ability of a number of opioids to maintain self-administration in rats with neuropathy compared to normal uninjured animals [41]. Most opioids maintain intravenous self-administration in rodents in a manner consistent with their potency, efficacy, and half-life when examined over the full dose-effect range. At low doses, little to no responding is maintained and the rate of responding or number of infusions increases as a function of increasing dose to a maximum. As dose is increased further the rate of responding or number of infusions decreases due to increased duration of the subjective reinforcing effect at higher doses, increased incidence of effects that are inconsistent with operant responding such as sedation or catalepsy, or some combination of both. This typical inverted U-shaped dose-effect was observed in normal animals with all opioids studied with typical potency and efficacy for each compound. However, not all of the opioids studied maintained robust self-administration in nerve-injured rats. Notably, both fentanyl and morphine maintained very low rates of responding at all doses examined. These drugs had the lowest relative intrinsic efficacy at mu-opioid receptors of all the opioids studied [58, 59]. Heroin and methadone, conversely, maintained robust responding in rats with neuropathic pain but only at higher doses sufficiently high to produce reversal of mechanical allodynia. Higher doses of these two opioids were required to maintain responding in rats with neuropathic pain compared to normal animals. These two opioids were also found to be the most efficacious of all opioids studied in reversing mechanical allodynia following nerve injury, and the time elapsed between self-administered infusions of either heroin or methadone was consistent with the duration of their antiallodynic effect when administered intravenously. Hydromorphone displayed a profile in between those of heroin and methadone, and those of morphine and fentanyl. The maximum rate of responding that could be obtained in nerve-injured rats was significantly diminished compared to normal rats with hydromorphone; however, this compound did maintain responding in a manner consistent with the duration of it is antiallodynic effect. These data collectively suggest that rats with neuropathic pain were self-administering opioids to maintain a different subjective state than normal animals, or that the subjective state was similar between these groups but diminished in the presence of neuropathic pain with lower doses of high efficacy opioids or at all doses of opioids with relatively lower intrinsic efficacy.

Theoretically, if rats with neuropathic pain were self-administering opioids to maintain a subjective state related to pain relief then administration of an adjuvant analgesic prior to opioid access should diminish opioid consumption through self-administration. There are examples from a variety of drug classes that alleviate hypersensitivity in laboratory animals and humans with neuropathic pain; however, relatively few are efficacious against the more troublesome aspects such as ongoing pain. Clonidine given spinally alleviates both spontaneous pain and hypersensitivity in patients; however, adenosine given by this route alleviates hypersensitivity only [14, 15]. When rats with spinal nerve injury are trained to self-administer opioids, clonidine given intrathecally significantly diminishes opioid intake in a dose–responsive manner [41]. However, intrathecal (i.t.) clonidine is without effect in normal rats self-administering opioids. Perhaps more interesting, i.t. administration of adenosine at a dose that is equiefficacious with the doses of clonidine given in reversing mechanical hypersensitivity has no effect on opioid self-administration in rats with nerve injury. Other investigators have found that clonidine, but not adenosine, induces a conditioned place preference selectively in rats with peripheral nerve injury [27]. These data have several implications. One is that rats with neuropathic pain appear to be titrating a different subjective state during opioid self-administration than normal rats. This is evidenced by i.t. clonidine reducing opioid consumption through self-administration only in rats with spinal nerve injury. Second, the subjective state does not appear to be strictly related to reversal of mechanical hypersensitivity. This idea is supported by the data showing that i.t. adenosine, while equiefficacious with clonidine in reversing mechanical hypersensitivity, has no effect on opioid intake. Lastly, these data demonstrate that spinal nerve ligation produces a subjective state in rats with a resultant behavioral pharmacology similar to that observed in chronic neuropathic pain patients. This last point is not trivial, as many investigators have suggested that these nerve injury models in rodents do not produce the full range of behaviors found with neuropathic pain in the clinic, but merely produce mechanical hypersensitivity that may have little to no relevance for treatment strategies in this population. One of the main primary outcome measures and goal for chronic pain treatment with novel analgesics is the reduction of opioid consumption. Opioid self-administration in rats following nerve injury appears to have significant face validity and displays pharmacology consistent with clinical data.

These data suggest that the neuronal mechanisms that are responsible for maintaining opioid consumption in the presence of neuropathic pain differ from those in the absence of chronic pain in rodents. Investigations into the mechanisms that contribute to opioid self-administration in rodents in a laboratory setting have utilized a variety of techniques to identify pertinent neurochemical, neurophysiological, and pharmacological mechanisms. A detailed discussion of these findings is beyond the scope of this chapter; however, many excellent reviews have been provided in the literature [28, 29]. Relatively few studies have specifically examined how the presence of pain might alter these mechanisms. In mice, peripheral nerve injury diminishes cFos activation in the ventral tegmental area [49]. This is accompanied by reduced mu-opioid receptor G-protein coupling in this region and diminished conditioned place preference to morphine [53]. Both mu-opioid and dopaminergic activity is reduced in the presence of acute and chronic pain [50]. These data are consistent with the opioid self-administration data reviewed above, namely that higher doses or more efficacious opioids would be hypothesized to be required to maintain self-administration in animals with nerve injury compared to normal. However, if this were the sole mechanism responsible for decreased self-administration in animals with nerve injury, then theoretically one might expect that alleviation of peripheral noxious input through intrathecal administration of an analgesic would restore activity of ventral tegmental neurons to normal, which would result in an increase in opioid self-administration. This clearly does not happen; however, suggesting that perhaps the conditioned stimulus that maintains operant behavior upon which opioid intake is made contingent differs between nerve-injured and normal animals. There are several brain regions that coordinate peripheral noxious input and activity within the limbic system. The amygdala has been extensively studied in this context. Once again, a thorough review of this body of work is beyond the scope of this chapter; however, several excellent reviews have been provided [1, 18]. To investigate the relevance of the amygdala in opioid self-administration in the presence of pain, we utilized an irreversible inhibitor of mu-opioid receptors beta-funaltrexamine [39]. We have previously demonstrated that beta-funaltrexamine irreversibly inhibits mu-opioid receptors when administered intracerebroventricularly or into discrete brain regions for approximately 7 days after administration [43]. This compound has several advantages over typical opioid antagonists. One advantage is that the effect last for several days rather than a few hours. This is a particular advantage with drug self-administration as the drug can be administered on 1 day, and behavior assessed the following day minimizing behavioral disruption due to the injection procedure itself. Additionally, determining modest changes or differences in self-administration within a relatively short time frame can be problematic, particularly if the rate of responding is relatively modest to begin with such as with higher doses of opioids. Determining the effect across several hours of self-administration and across several sessions renders the study more powerful statistically. Additionally, the irreversible action of beta-funaltrexamine allows a detailed anatomical determination of the location and extent of mu-opioid receptor inactivation in vitro using either receptor binding or opioid-stimulated GTPγS³⁵ binding by autoradiography. Utilizing these tools, we were able to target mu-opioid receptors in the lateral portion of the amygdala and reduce DAMGO-stimulated G-protein activation by 60 % following administration of beta-funaltrexamine. The effect of reduced mu-opioid receptor activation produced a modest increase in opioid self-administration in normal rats; however, the effect was approximately fivefold greater in rats with neuropathic pain. In the region of the opioid dose-effect curve that used for these experiments, an increase in rate of drug intake is typically interpreted as a decrease in the reinforcing effect of a single injection. The result is that animals must increase their rate of intake to achieve the same desired effect before the manipulation. Therefore, it appears that mu-opioid receptors within the lateral amygdala are more relevant for producing the subjective state motivating animals to self-administer opioids in the presence of neuropathic pain than in the absence of pain. There are several candidate neurons that could be differentially modulated by opioids and further investigation of how the presence of pain alters these inputs and the role of mu-opioid receptors merit further study. These data however further suggest that peripheral nerve injury models in rodents produce relevant effects on the behavioral pharmacology of opioids beyond production of hypersensitivity that can be measured using drug self-administration.

These findings however suggest that other drugs could potentially alter this negative subjective state and provide similar conditioned stimuli that would support self-administration. One drug in particular that would be hypothesized to maintain self-administration in the presence of neuropathic pain would be clonidine, particularly by an intrathecal route of administration. Clonidine would be expected to be self-administered intrathecally in rats with neuropathic pain if the effect on reducing opioid intake in these rats were due to production of a similar subjective state as intravenous opioids. Given that intrathecal administration of clonidine had no effect on opioid self-administration in normal rats, it would be expected that normal rats would not self-administer clonidine by this route. If clonidine failed to maintain self-administration in rats with neuropathic pain, then it might suggest that clonidine’s effect on opioid self-administration might be due to other effects not related to negative reinforcement, but rather production of effects that interfere with operant responding that are enhanced following nerve injury for unknown reasons. Therefore, demonstration that clonidine maintains self-administration intrathecally only in rats with nerve injury would add interpretive value to the studies described above. Studies in our laboratory indeed confirmed that intrathecal clonidine was selectively reinforcing in rats with nerve injury [42]. When both normal and nerve-injured rats were given access to intrathecal clonidine infusions through lever presses, the initial rates of responding were approximately equal. Over 2–3 days however, rate of responding increased in nerve-injured rats to a stable level, whereas the rate of responding decreased to only 3–4 lever presses daily in normal rats. In nerve-injured rats, decreasing the dose of clonidine per infusion increased responding, whereas increasing the unit dose decreased responding. These data support the hypothesis that clonidine was the salient reinforcer maintaining responding. When clonidine was replaced with saline, responding rapidly fell to three to four lever presses daily. Finally, rats with nerve injury failed to acquire intrathecal self-administration of a combination of clonidine and the α2 adrenergic antagonist idazoxan, consistent with the known antiallodynic and analgesic mechanism of clonidine’s pharmacological effect. Once clonidine intake through self-administration became stable in nerve-injured rats, the daily consumption was greater than would be expected given this drug’s maximal effect and duration of antiallodynic action using reflexive withdrawal measures. This increased intake occurred with a time course consistent with the development of tolerance to the antiallodynic actions of clonidine. However, the increased rate of intake could indicate that sufficient levels of drug were being achieved to have effects at peripheral or supraspinal sites. Indeed clinically, the pharmacological effects of clonidine that limit maximal dose are hypotension and sedation, which are not thought to be mediated by a spinal site of action. Following chronic intrathecal infusion of clonidine, we have found that α2 adrenergic receptors become desensitized in the central amygdala, suggesting that indeed sufficient levels of clonidine reach relevant supraspinal sites following intrathecal infusion to produce pharmacological effects (unpublished observations). To test the relevance of this finding, we found that administration of the nonselective receptor alkylating agent EEDQ into the amygdala reduced α2 adrenergic stimulated G-protein coupling by approximately 75 % and significantly decreased acquisition of intrathecal clonidine self-administration in nerve-injured rats (unpublished observations). These data support the idea that intrathecal clonidine provides a sufficient stimulus to serve as a reinforcer in rats with neuropathic pain, and only in rats with neuropathic pain, and that this stimulus is mediated through both spinal and supraspinal α2 adrenergic receptors.

The implications of these data go beyond the findings that intrathecal clonidine will maintain self-administration in the face of neuropathic pain however. As mentioned above, there is some controversy regarding the usefulness of peripheral nerve injury models in rodents as surrogates for clinical neuropathic pain. The concern is that reflexive paw withdrawal does not accurately reflect a clinically relevant painful stimulus, or that inhibition of this response is too nonspecific of a pharmacological effect to delineate between drugs that are likely to display clinical efficacy in patients with neuropathic pain from those that have relatively lower potential. The other primary concern and criticism is that these models produce only hypersensitivity but do not demonstrate the full range of behaviors found in the clinic, particularly ongoing or spontaneous pain. However, the finding that rats with peripheral nerve injury will self-administer intrathecal clonidine at a dose that produces an opioid-sparing effect in a self-administration model lends credence to the idea that these nerve injury models indeed produce a subjective state that is sufficient to result in selective reinforcement by administration of clonidine and results in a pharmacology of both clonidine and opioids that is consistent with clinical data. These data also support, and are supported by, findings from other investigators that intrathecal clonidine becomes selectively reinforcing in rats with neuropathic pain using the conditioned place preference model [27]. However, these data do support the idea that inhibition of reflexive withdrawal from mechanical or thermal stimuli is insufficient to delineate between drugs that have been found to be useful clinically for neuropathic pain treatment and from those that have been found to be ineffective or lacking sufficient efficacy relative to adverse dose-limiting effects. Self-administration methods become problematic for screening large numbers of potential therapeutics however due to the amount of time and resources required relative to more simple behavioral methods. However, clearly this technique has utility for identification or verification of potential novel therapeutic targets for clinical efficacy against neuropathic pain, as well as possess face validity for investigation of the basic neurophysiology, neurochemistry and pharmacology of nerve injury models of pain in laboratory animals.