Opioids and Neuropathic Pain
Experimental Approaches in the Rat
J. M. Besson, D. Besse, M. C. Lombard, N. Attal, J. Desmeules, and G. Guilbaud, Department of Physiopharmacology of the Nervous System, INSERM (U 161) and EPHE, 75014 Paris, France
Despite the fact that opioids have been used for many years to relieve intense pain, there is still a controversy as to whether these drugs may relieve neuropathic pain. Many clinical anecdotal observations have reported an absence of efficacy of opioids in cases of neuropathic pain. Few systematic investigations have been conducted; the conclusions were generally that neuropathic pain was insensitive to morphine. However, it has also been reported that patients with certain neuropathies may benefit from long-term opioid treatment. Thus this subject is highly controversial (3,36).
Clearly, it is of both theoretical and practical importance to determine if certain types of pain are indeed resistant to opiate treatment and if so, why?
In order to explain the lack of, or the reduction in, efficacy of morphine on neuropathic pain consecutive to peripheral lesions or plexus avulsion, three main hypotheses have been advanced (32).
The first hypothesis suggests that neuropathic pains are so severe that even morphine does not help much. In other words, are opiates incapable of providing significant relief for pain above a certain intensity level? In this respect, there are suggestions in the literature that the pain response to transient noxious stimuli may be less sensitive to morphine than responses to sustained noxious stimuli. If this is true, lancinating pain and other paroxysmal transients that are common features of neuropathic sensation will respond poorly to morphine.
The second hypothesis is based on the observation of Campbell and co-workers (28). These authors reported that sensory signals carried centrally along low-threshold Ab fibers could be responsible for pain in some neuropathic conditions. Unlike small-diameter nociceptive afferents, these fibers have few, if any, presynaptic opioid receptors (see below) and therefore the pain they evoke should be insensitive to morphine. This hypothesis is supported by an electrophysiological investigation we performed several years ago. In this experiment the effects of morphine were evaluated on the responses of dorsal neurons to sural nerve stimulation: the responses to large myelinated Aab afferents were not influenced, but there was a major reduction in responses due to the activation of both Ae and especially C fibers (54). The specificity of such effects of intravenous opiates has been demonstrated in terms of isomerism, dose dependency, and reversal of the depression by naloxone.
The third hypothesis is relevant to the presence of a high proportion of opioid receptors on thin primary afferent fibers entering the spinal cord (39,53,63). It is speculated that peripheral nerve lesions or posterior root avulsion induce a clear decrease in opioid binding sites. Thus, fewer opiate receptors might yield a weaker opioid response.
The first part of this chapter will consider the latter hypothesis, i.e., the decrease in opioid binding sites following rhizotomies of various extents. The second part will deal with the effects of opioid agonists on spontaneous and evoked pain in a model of mononeuropathy induced by loose ligation of one sciatic nerve in the rat (13). All the experiments were performed according to the ethical rules as proposed by the Committee for Research and Ethical Issues of the IASP (30).
Numerous studies have demonstrated the presence of n, e, and l opioid receptors within the superficial dorsal horn (which is the main termination site of thin nociceptive primary afferent fibers [see references in ref. 20]). As initially reported by Lamotte et al. (53), a clear decrease in binding has been described at this level after dorsal rhizotomy. In addition, several studies based on the administration of capsaicin in neonate rats (39,63) have shown that the opioid receptors are located on thin primary afferent fibers. However, with a few exceptions, the previous studies did not include data on the relative proportions of the three main types of receptors n, ;gd, and ;gk, and no quantitative data are available concerning their pre- and postsynaptic locations. Moreover, difficulties arise when trying to compare the results: a number of these binding studies have been performed on membrane preparations where it is impossible to locate the site of the reduced binding, and most of the ligands used were not selective and did not allow a clear distinction between each type of receptor. Finally, the extent of deafferentation lesions is, in many cases, rather restricted.
We used highly selective ligands, [3H]DAMGO for n receptors, [3H]DTLET for e receptors, and [3H]ethylketocylazocine (EKC) in the presence of DAMGO and DTLET for l receptors, quantitative autoradiography, and dorsal rhizotomies of various extents (14-19,22,89) to answer the following questions. What are the respective proportions of the three types of opioid receptors in the superficial layers of the dorsal horn? Are these proportions similar over the rostrocaudal axis of the spinal cord? What are their respective pre- and postsynaptic proportions? What is the rostrocaudal distribution of the thin primary afferent fibers arising from a single root? Is there a plasticity of opioid receptors in chronic deafferentation states?
As shown in
Figure 1
, the respective proportions of the three types of opioid receptors are remarkably homogeneous throughout the spinal cord (cervical, thoracic, lumbar, and sacral levels). There is a high percentage of n (70.4-74.3.1%), an intermediate percentage of e (18.4-20%), and a low percentage of l (7.3-9.5%) binding sites. These results are in good agreement with the recent data presented by Stevens et al. (79).
From these results several conclusions can be drawn:
The high density of n opioid receptors in the superficial layers of the dorsal horn is in good agreement with the fact that this region is one of the main sites of action of morphine on the transmission of nociceptive messages at the spinal level (see references in ref. 36). Both pre- and postsynaptic mechanisms have been proposed as a basis for the depressive effects of opioid agonists on the activity of nociceptive dorsal horn neurons.
As mentioned above, a high proportion of the opioid receptors are located on thin primary afferent fibers. However, there have been no quantitative available data on their pre- and postsynaptic location. This lack of information could be due to several reasons such as the various types (nerve section, rhizotomy) and extent of peripheral lesions and the survival delays following the lesion.
In order to define the optimal conditions to quantitatively evaluate pre- and postsynaptic opioid receptors, preliminary studies were carried out considering the influence of two parameters: the extent of the lesion and the postlesion (PL) delay. These studies were done using unilateral dorsal rhizotomies of 1, 3, 5, and 7 roots at the cervical level. Quantitative measurements were performed in the C7 segment.
As shown in
Fig. 2A
, for n binding sites, the decrease observed 1 week after the surgery in the side ipsilateral to the rhizotomy was clearly dependent on the extent of the lesion. This decrease in binding was significant when comparing the results after section of one root (C7) to three roots (C6-C8) and three roots to five roots (C5-T1). In contrast, no significant difference could be found when comparing the section of five to seven roots (C4-T2). These results indicate that after the section of seven roots, the C7 segment that corresponds to the center of the deafferented zone is totally deprived of primary afferent fibers. Similar data were obtained when considering e binding sites.
Taking into account the results obtained 1 week after the section of seven roots (C4-T2), we then considered the time-related modifications of n and e binding sites in the same lesion following various survival times from 1 to 90 days. As illustrated in
Figure 2B
, we observed a decrease in binding on the side ipsilateral to the lesion as early as the first day postrhizotomy, the maximal loss being attained at 8 days PL. After 8 days PL, the residual binding remains stable over the period of analysis (90 days); the loss of n receptors (71-74%) is significantly more pronounced than the loss of e receptors (57-62%). This loss of binding is due to a decrease in the number of binding sites, since the affinities of the remaining n and e sites are similar to those of the total receptor population in intact rats. We attribute the residual values to postsynaptic binding, whereas the decrease can be attributed to a loss of presynaptic sites.
Taken together, all these results indicate that a large unilateral lesion (C4-T2) and a survival delay equal to or greater than 8 days are necessary and sufficient conditions for assessing, with certainty, the relative proportions of pre- and postsynaptic opioid binding sites in the superficial dorsal horn. As illustrated in
Fig. 3
, the presynaptic components of n, e, and l receptors are 76%, 61%, and 53%, respectively, the postsynaptic components being 24%, 39%, and 47%, respectively. It must be noted for the proportion of pre- and postsynaptic l sites that, as these sites are relatively sparse (<10%) in the superficial dorsal horn, the accuracy of the measurement is poor.
The high proportion of n and e binding sites on primary afferent fibers favors a presynaptic action of opioids, and from a theoretical point of view one can speculate that the depressive effects of morphine on the activity of dorsal horn nociceptive neurons would be considerably reduced after a large deafferentation. To test this hypothesis we performed an electrophysiological investigation by considering the effect of morphine on the spontaneous hyperactivity of dorsal horn neurons in both intact and deafferented rats (57).
We first studied in the spinal decerebrate rat, the effects of the administration of systemic morphine (2 mg/kg, i.v.) on the spontaneous hyperactivity of dorsal horn neurons induced by a large chronic dorsal rhizotomy (C5-T1). A comparative study was performed, considering the effects induced by the same dose of morphine in the spinalized decerebrated arthritic rat in which dorsal horn neurons also display a high spontaneous activity (62). From these comparative electrophysiological studies, it appeared that morphine was only half as effective in the deafferented rat as in the normal arthritic rat. Thus, the removal of presynaptic opioid receptors diminishes by approximately 50% the depressive effects of morphine on the spontaneous activity of nociceptive dorsal horn neurons.
Taken together, this electrophysiological investigation and the presence of a high proportion of opioid receptors on thin primary afferent nociceptive fibers may explain why morphine is more effective in combating pain originating in nociceptors than pain arising from deafferentation.
However, these results do not support the classical clinical claims that morphine is ineffective against all neuropathic pains of peripheral origin. Indeed, the interpretation of our data must be extremely cautious for the following reasons:
To try to understand certain other clinical situations, additional experiments were performed in which the rostrocaudal distribution of thin afferent fibers belonging to a single dorsal root was evaluated in the superficial dorsal horn. In addition, this approach allowed us to specify the origin (1, 2, 3, ..., roots) of the presynaptic opioid receptors for a given spinal segment.
Taking into account the fact that a high proportion of n and e opioid receptors are presynaptically located, we postulated that opioid binding sites could be possible markers for evaluating the distribution of the thin afferent fibers in the superficial dorsal horn. In order to visualize and to quantify the distribution of these fibers arising from the C7 root, we used several experimental situations (Fig. 4A
): control rats with the dorsal roots intact (intact) and lesioned rats with a unilateral dorsal rhizotomy of (a) the seven roots C4-T2 (C4-T2 cut), (b) the three roots above and below C7 (C7 spared), and (c) the C7 root alone (C7 cut). Binding measurements were made 8 days after the various lesions. The spinal cord distribution of n and e opioid binding sites belonging to the C7 dorsal root was then calculated according to the following methods: (a) subtraction of the data of the C7 cut experimental situation from those in the ``intact'' one and (b) subtraction of the data of the C4-T2 cut experimental situation from those of the C7 spared one. The combination of the results obtained with these two methods allowed assessment of the spinal distribution of n and e receptors belonging to the C7 dorsal root.
As shown in Fig. 4B
, with [3H]DAMGO, the distribution of n sites belonging to the C7 root extends significantly in the segment of entry (C7), one segment caudal (C8) and two segments rostral (C6 and C5). More precisely, 40% of binding sites are found in the segment of entry and the proportion reaches 80% if the two adjacent segments (C6-C8) are included. A similar distribution was observed for e sites.
From these results, it appears that for a given spinal segment there is considerable overlapping of the projections arising from two or three adjacent roots, since in our experimental conditions 60% of the presynaptic receptors in the C7 segment did not belong to the C7 root.
The functional relevance of these data can be assessed by clinical observations in humans. Indeed, it is well known that, according to neurosurgical reports, there is minimal hypoaesthesia after section of a single dorsal root of the brachial or lumbosacral plexus (82). In contrast, at the time when dorsal rhizotomies were being proposed as a means to relieve pain originating from a localized peripheral area, it was reported that at least three consecutive dorsal roots must be sectioned to relieve pain (72). The clinical relevance of the preferential rostral projection of fine-diameter primary afferent fibers has not yet been established, but it could be important in consideration of neurosurgical procedures, namely, selective posterior rhizotomy (77) or the dorsal root entry zone lesion (64).
From a theoretical point of view, since the loss of opioid sites is relatively weak for a restricted lesion (after section of one root there is still 60% of presynaptic sites plus postsynaptic sites), one can speculate that morphine could still modulate the transmission of nociceptive messages at the spinal level. In addition, the loss of opioid binding sites could be submitted to compensatory regulation during chronic deafferentation states. To partly answer this question, an experimental investigation was performed to gauge an eventual plasticity of opioid receptors, varying the extent of unilateral dorsal rhizotomies and the PL delays.
To analyze the plasticity of opioid receptors, several experimental groups were investigated: control animals with intact dorsal roots and lesioned animals with a unilateral dorsal rhizotomy of 1 to 7 roots (DRh1 to DRh7). Four different PL delays were studied: 1, 2, 4, and 12 weeks for each lesion. The different lesions were performed so that all animals were the same age at the time of sacrifice.
Figure 5
ilustrates the changes in binding observed at 1 and 12 weeks PL. For the largest rhizotomies (DRh7 and DRh5), there is no significant difference in n opioid binding 1 and 12 weeks in the central C7 segment. For the more restricted rhizotomies (DRh3 and DRh1), there is a significant recovery of n opioid binding toward values of the side contralateral to the lesion. For the latter two lesions, the percentages of recovery of binding at 12 weeks over the 1-week level in the spinal segment central to the lesion (C7) were 18.1 (P
< 0.05) and 85% (P
< 0.01) for DRh3 and DRh1, respectively. Moreover, in the case of the most restricted rhizotomy (DRh1), a significant recovery of binding was already present at 4 weeks (45%; P
< 0.05). Similar observations were made for e sites. For n as for e sites, no difference in binding could be measured between 1 and 2 weeks PL, whatever the extent of the rhizotomy.
Since no variation of binding was detected between 1 and 12 weeks PL in the central segment (C7) in the case of the large C4-T2 rhizotomy (DRh7), our data suggest that postsynaptic sites are not significantly affected along the different PL delays. In contrast, the recovery of binding observed for the least extensive lesions (DRh1 and DRh3) might be imputed to variations of presynaptic n sites that are still present in a noticeable proportion in this segment 1 week after these lesions.
The recovery in binding could reflect either a simple receptor upregulation, producing an increase in receptor density in already existing terminals, or a sprouting of terminals of fine primary afferent fibers arising from intact roots. The hypothesis of collateral sprouting, which is in good agreement with the anatomical study of McNeill et al. (61) showing synaptogenesis in laminae I and IIo of primary afferent calcitonin gene-related peptide immunoreactive terminals after chronic partial denervation, is very attractive. This assessment is supported by the fact that the magnitude of the recovery of binding is closely related to the degree of deafferentation: the less deprived of afferents a spinal segment is, the more pronounced the recovery in this segment.
Our results are coherent with the idea that primary afferent fibers entering partially deafferented regions are capable, at least to a certain degree, of anatomical adjustments in response to injury. Moreover, the presence of an increased number of opioid binding sites in chronic lesions compared to acute ones suggests an increased capacity for opioid controls. Subsequently, this increase in binding sites could provide a basis for responsiveness to opioids in some particular circumstances of partial deafferentation or peripheral nerve lesions.
As mentioned before, the rhizotomy lesions we used mimic only a very particular syndrome of neuropathy, i.e., the deafferentation syndrome. In order to approach the most commonly encountered clinical situations, the modifications of opioid binding sites in the superficial dorsal horn of the spinal cord were studied in a model of mononeuropathy induced by loose sciatic nerve ligations (6,13).
In this model, the neuropathy is induced by placing four loose ligatures around one sciatic nerve. This procedure produces a chronic but reversible nerve lesion. Behavioral studies performed on this model have described a pain-related behavior, with mainly spontaneous symptoms and both hyperalgesia and allodynia in response to mechanical and thermal stimuli. According to the extensive investigation of Attal et al. (6), the time courses of pain-related disorders were similar whatever the behavioral test: following an early period of hypoalgesia at 5 days PL, hyperalgesia developed, being maximal at 2 weeks PL, with a recovery occurring around 2 months PL (see next section).
We have studied, in this model, the modification of n, e, and l opioid binding site at the following PL delays: 3 days and 2, 4, 8, and 15 weeks (18).
We have shown a 28% (P
< 0.01) and 24% (P
< 0.01) decrease in ipsilateral/contralateral side binding ratios for n and e sites, respectively, at 2 weeks PL (
Fig. 6
), which corresponds to the delay of maximal hyperalgesia and of maximal alteration of fine-diameter primary afferent fibers (11,12,40). Interestingly, a depletion of substance P (21%) and calcitonin gene-related peptide (19%) was observed (12) in laminae I and II ipsilateral to the nerve ligation at 20 days PL.
For longer survival delays (4,8, and 15 weeks PL), n and e binding ratios return toward control values, probably reflecting the occurrence of a long-term neuroplasticity (i.e., a new equilibrium in the metabolism of primary neurons or collateral sprouting from intact primary afferents) following loose nerve ligation. There was no significant modification of l sites (labeled with [3H]U-69593), whatever the PL delay.
Thus, from these results, there is apparently a relatively good relationship between behavioral observations (maximal hyperalgesia at 2 weeks) and the modifications in n and e opioid binding sites. However, the results are much more complex if we consider the specific binding concentration values in addition to the ipsilateral/contralateral side binding ratios.
Actually, as shown in Fig. 6
, a significant bilateral upregulation (30-50%) of n binding sites occurs in laminae I and II as early as 5 days PL as compared to intact rats. On the side contralateral to the ligation, this upregulation is permanent all along the period of observation (until 15 weeks PL). On the ipsilateral side, the up-[chregulation is also permanent, but it is reduced at 2 weeks when a transient downreg[chulation takes place. This transient downregulation corresponds to the decrease in the ipsilateral/contralateral side binding ratio. These modifications in binding densities reflect changes in the number of binding sites, since saturation studies failed to reveal any changes in receptor affinities. For e sites, no significant regulation could be measured, whatever the PL delay.
The bilateral increase [3H]DAMGO specific binding was a priori unexpected. Complex bilateral modifications have been reported in the same model by Stevens et al. (78) and in a model of ``unilateral footrot'' in sheep by Brandt and Livingstone (24,25).
To explain the contralateral increase in n binding, one hypothesis can be advanced. It could result from regulation of spinal endogenous opioid systems as a consequence of activation of segmental, heterosegmental, and/or supraspinal pain modulating systems. These various systems are known to be activated by nociceptive messages (see references in ref. 20), and it has been shown that, after sciatic nerve loose ligation, the afferent nociceptive message is increased. Actually, abnormal activities have been recorded in primary afferent fibers following chronic constriction of the sciatic nerve (45,46,86), indicating that these fibers are physiologically altered by 6 to 10 days postinjury (86). This abnormal spontaneous or evoked hyperactivity is also found in the spinal cord in the hyperalgesic phase after the sciatic nerve loose ligation (58,68; but see ref. 52). It is also found in supraspinal structures, since in the ventrobasal complex of the thalamus, neuronal responses to mechanical and thermal stimuli of the lesioned paw have been shown to be strongly increased 2 weeks after the sciatic nerve loose ligation (42). Thus, in response to the increased nociceptive information, descending control systems could be activated and affect both sides of the spinal cord, independently of the unilateral morphological changes observed after the lesion. This hypothesis could explain the unexpected bilateral increase in n binding sites observed after the unilateral sciatic nerve ligation.
The conclusion related to the repercussion of the opioid binding site modifications we observed in various experimental situations will be discussed in the general conclusion of this chapter. In the following section, we will consider the effects of opioids on the behavioral manifestations displayed in the model of mononeuropathy mentioned above.
We used the model described by Bennett and Xie (13; see also ref. 6). In brief, loose ligatures around the common sciatic nerve induce abnormal nociceptive reactions to thermal and mechanical stimuli, starting on day 5 after surgery, with a maximum occurring at week 2. The recovery begins after the 3rd week and is complete at 8 to 10 weeks.
These animals exhibit hyperalgesia (increased reaction to noxious stimuli) and allodynia (nociceptive reaction to normally nonnoxious stimuli) to both mechanical and thermal stimuli.
Hyperalgesia to mechanical stimulation was revealed by using pinch stimuli applied to the hindpaw. Hyperalgesia to thermal stimulation consisted of a marked decrease (by approximately 4 sec or 30%) in the mean immersion duration for a noxious temperature (46°C) to elicit a struggle reaction. In addition, qualitative modifications were also seen after this stimulation; indeed, rats could exhibit abnormal reactions lasting as long as several minutes after the stimulation. The latter observations may also relate to prolonged afterdischarges of the flexor electromyographic responses recorded by Bennett and Xie (13) from the ligated side after application of noxious temperature.
Evidence for allodynia to mechanical stimulation was provided using the vocalization threshold test to paw pressure (which showed a 100-g or 30% decrease in threshold values). In contrast to tonic pressure and pinch of the paw, brushing of the hindpaw did not elicit any abnormal reaction. Allodynia to nonnoxious thermal stimuli was seen clearly for both hot and cold stimulations (5). With hot temperature we observed a 4°C (or 30%) decrease in the mean temperature of the struggle threshold (from 45° to 41°C). With cold temperatures (10°C), most of the rats exhibited a reduction in the duration of hindpaw immersion to induce a withdrawal reaction (from 15 to 9 sec or 40%). These results are reminiscent of those of Bennett and Xie (13) using a cold or a hot plate as a stimulus.
Most tests employed to assess hyperalgesia and allodynia used phasic stimuli. Such stimuli do not reflect long-lasting or tonic pain, which is one of the most common features of neuropathic clinical pain. Since these animals exhibited abnormal positions of the hind paw after surgery, we decided to observe the animal's ``spontaneous'' behavior in a natural setting (5) (
Figure 7
) according to a procedure derived from that used for ``pain rating'' in the formalin test (34). One possible objection is that this behavior is not spontaneous, since the rat is placed on the floor, which could be considered as a pain-inducing stimulus. Thus, the use of the term spontaneous
here is based on the fact that the rat's behavior is observed without intervention from the experimenter, in contrast to the other behavioral tests.
These behavioral abnormalities are unlikely to be related to motor impairment (i.e., weakness of the hindpaw's dorsiflexors, commonly observed after the nerve ligation) or impairment of deep sensation, since, over a single observation session, the rat would often adopt different positions of the hindpaw and, more important, these positions were clearly modified by low doses of substances that are unlikely to affect motor activity or deep sensation. Therefore, the abnormal position of the paw seems to predominantly reflect a reluctance of the rat to press the hindpaw to the floor and this pain-relieving position could be interpreted as a pain-related behavior.
It has been suggested that the model of neuropathy studied here has features in common with human causalgias (13). Obviously one has to be very careful not to infer that the model exactly reproduces human pathology, which is clearly very complex, but one should consider the following points:
It is difficult to assert that the ``spontaneous'' pain-related behavior reported here may be related to the spontaneous pain observed in human causalgias, the typical characteristic of ``superficial burning pain'' seen in causalgias being obviously undetectable in an animal model.
The latency of occurrence of pain-related disorders (by day 5) is more prolonged than in human causalgia syndromes, which generally have a very rapid onset, although a delayed onset has also been reported (23).
However, trophic changes sometimes seen in patients with causalgia, such as asymmetries in skin temperature, have been reported in these animals and investigated using thermography (13). Last, it is clear that the hyperalgesia and allodynia to thermal stimuli exhibited by these rats, as well as the dramatic sensitization after nonnoxious stimulation, are consistent with human causalgia (23,29,56), and it has been reported that allodynia to cooling is a characteristic of certain ``reflex sympathetic dystrophies'' (37).
Thus, there are indeed common features between this experimental neuropathy and clinical causalgia in humans. However, other pathological situations may also provide a comparison with the model: in fact, some similarities have been discussed between the model and entrapment neuropathy (13), and certain features also resemble painful diabetic neuropathy (26,90).
Therefore, this rat model appeared to be relevant for the study of neuropathy mechanisms and, due to their consistent reproducibility, the various behavioral tests seemed to be suitable to gauge an eventual antinociceptive effect of opioids.
A significant effect of morphine has been obtained on the ``tonic'' pain related behavior due to neuropathy. The spontaneous pain rating can be modulated by relatively low doses of morphine. In particular (
Figure 7
), attention is drawn to the profound and prolonged decrease in the pain score after 1 mg/kg intravenous morphine, a dose that did not induce clinical signs of drowsiness, or electrocorticogram (ECoG) modification (44).
It can be noted that the morphine action observed here is of longer duration than that observed with the same dose in the ``phasic'' test (60-70 min only; see below). Interestingly, a more recent study in this model has revealed that ``scratching'' considered as a component of spontaneous pain was also depressed by morphine (51).
An initial study clearly indicates (4,5) that systemic morphine induces dose-[chdependent, naloxone-reversible antinociceptive effects in rats with mechanical allodynia due to peripheral mononeuropathy, using the measure of vocalization thresholds elicited by paw pressure (
Figure 8A
). These effects are roughly comparable to those observed under similar behavioral conditions in a model of inflammatory pain in the rat (Freund's adjuvant-induced arthritic rats) (47). The behavioral pain-related disorders exhibited by the animals clearly seem to relate to neuropathic pain and not to other sources of pain that might be independent of the ligatures. Nor are they due to the presence of inflammation, since we have previously shown that aspirin at a dose of 50 mg/kg intravenously did not alter vocalization threshold, whereas this substance clearly displayed an antinociceptive effect in an animal model of localized inflammation (7).
It must be also emphasized that the effects observed on the ligated side appear significantly more potent than in normal rats (
Fig. 8B
) whereas those observed on the sham-operated side are comparable to normal rats (
Fig. 8C
).
In contrast, the same dose of morphine did not modify the abnormal reactivity to thermal stimulus. As shown on Table 1, the struggle latency for paw immersion into a hot (44C) or cold (10°C) water bath was not enhanced by 1 mg/kg morphine administered intravenously.
Using the same phasic mechanical behavioral test as with morphine (measure of vocalization threshold to paw pressure), it was clearly established that systemic n, e, and l opioid agonists are able to attenuate mechanical pain-related disorders such as allodynia caused by peripheral sciatic mononeuropathy (
Figure 9
). Each agonist acts specifically, since its antinociceptive effect is prevented by the prior administration of its selective antagonist.
The prototype peptide DAMGO (49) shows a strong dose-related and naloxone-reversible antinociceptive effect. This confirms the analgesic effect obtained with morphine on mechanical allodynia as assessed in the initial studies reported above (4,6,44,65), and by electrophysiological studies performed at the thalamic level in this model (41). Taken together, these results mirrored the accumulating clinical analgesic observations of an effect of morphine in patients suffering from post-herpetic neuralgia (74) and a comprehensive range of neuropathic pains (36,50,69,70).
Chapter 9 table 1.Effects of morphine (1 mg/kg, i.v.) on struggle latency to 44° C| Time after injection | |||||||
| 44° C Morphine n=12 | 124 ± 12 | 100 ± 18 | 112 ± 13 | 96 ± 16 | 96 ± 2 | 95 ± 2 | |
| 44° C Saline n=6 | 93 ± 4 | 100 ± 4 | 101 ± 1 | 93 ± 2 | 101 ± 4 | 94 ± 3 | |
| 10° C Morphine n=8 | 103 ± 6 | 112 ± 9 | 102 ± 7 | 115 ± 10 | 97 ± 5 | 88 ± 6 | |
| 10° C Saline n=6 | 93 ± 4 | 96 ± 6 | 95 ± 5 | 93 ± 9 | 97 ± 5 | 92 ± 5 |
BUBU is an enkephalin-resistant peptide selectively acting at the e receptor site (38,73). An important characteristic of this hydrophobic compound, which displays a better affinity and selectivity for e receptors than DPDPE, DSLET, or DTLET, is its easy passage across the blood-brain barrier allowing systemic administration (31). The antinociceptive effect of BUBU on cutaneous mechanical pressure is clearly the consequence of the activation of e receptors at this dose range (1.5-6 mg/kg, i.v.). The antinociceptive effect of BUBU was completely prevented by the selective e opioid antagonist naltrindole (71), at a dose devoid of either antinociceptive or hyperalgesic effects (29a). From the theoretical point of view, BUBU has a potential clinical application, since we observed an increased effectiveness in mononeuropathic rats compared to normal rats and it remains active in morphine-tolerant animals (31a).
The l opioid selective agonist U-69593 displays an impressive and dose-related antinociceptive effect obtained with intravenous doses as low as 0.75 mg/kg on the mechanical cutaneous allodynia in the mononeuropathic syndrome. These potent antinociceptive effects are completely reversed by small doses of the l antagonist MR 2266 (0.4 mg/kg, i.v.). At higher dosages (3 mg/kg, i.v.) U-69593 provoked an aversive reaction already described for l agonists (see references in ref. 66), which could, of course, compromise its clinical application.
These comparative studies clearly demonstrate that the three main selective opioid agonists are able to modulate mechanical allodynia in mononeuropathic rats.
The first part of this chapter reported on the possible involvement of a decrease in spinal opioid binding sites densities in neuropathic pain. Our data, based on unilateral dorsal rhizotomies of various extent, demonstrated that in the superficial dorsal horn, the magnitude of decrease in opioid binding sites ipsilaterally to the lesion is clearly related to the degree of primary afferent fiber degeneration induced by the lesion. A massive loss in n and e binding sites is observed on the ipsilateral side as compared to the intact side after a large dorsal rhizotomy. This finding could explain the failure of morphine to relieve pain due to large deafferentation lesions such as brachial plexus avulsions (27). It is also reminiscent of our recent electrophysiological conclusion indicating that the depressive effect of morphine on the spontaneous activity of dorsal horn neurons in spinal deafferented rats was reduced by 50%.
In contrast, in the case of restricted unilateral dorsal rhizotomies, which induce only partial deafferentation, the decrease in opioid binding sites is not very pronounced. Similar results were obtained following loose sciatic nerve ligation. Thus, the sites for opiate action are still largely present in the spinal cord, and in these conditions, it is probable that opiates would still have an effect on pain induced by moderate peripheral lesions. Moreover, the capacity of adjustment in response to injury (increase in number of opioid binding sites) after long-term delays also provides a possible basis for responsiveness to opioids in some particular circumstances of partial deafferentation or peripheral nerve lesions.
It is reasonable to speculate that the decrease in opioid binding sites that occurs in most of the neuropathies of peripheral origin is rather weak and cannot alone account for an eventual lack of effect of opiates on these pain syndromes. In addition, as previously stressed, the spinal cord is not the only site of action of morphine and several supraspinal structures involved in the affective-emotional aspects of pain are rich in opioid receptors.
From a general point of view, our data do not exclude the other two hypotheses (neuropathic pains are so severe that even morphine does not help much [32] and Ab fibers could be responsible for pain in some neuropathic condition [28]) advanced to explain an eventual lack of effect of morphine. However, the results of the behavioral studies obtained in the mononeuropathic rat clearly demonstrate an antinociceptive effect of opioids on such a pain model.
In any case, our results testing opioid action on spontaneous pain behavior and abnormal reaction to mechanical stimulus appear to contradict the general clinical perception that morphine is ineffective in obtaining relief in neuropathic pain conditions. Curiously, and despite the long historical analgesic benefit of opioid agonists, in diverse pain conditions, their efficiency in alleviating neuropathic pain is still much debated (2,3,36,50,59,60,67,69,70,74,80). In fact, the belief of opioid inefficiency is principally based on anecdotal reports and may reflect the reluctance of using opioids to treat chronic benign pain (69). However, some clinical and experimental observations specifically related to this problem have shown an ineffectiveness of opioid agonists in a wide diversity of painful peripheral or central neuropathic situations (2,3,60,80). The precise nature of pain conditions underlying the neuropathic pain syndrome probably plays an important role in this controversial issue. It has been claimed that sensory symptoms observed in neuropathic syndromes, such as touch-evoked or temperature-evoked allodynia, hyperalgesia to mechanical or thermal stimulation, or spontaneous pain may each be produced by separate neuropathic mechanisms. These mechanisms can coexist and different mechanisms in different patients can evoke the same symptom (81). In fact, attempts to precisely correlate the signs of the abnormal behavior with pharmacological antinociception, performed in this animal model, demonstrated the ineffectiveness of morphine on thermal allodynia Table 1 (4,5). This finding could partly explain the apparent discrepancy regarding the clinical resistance to opioids in some neuropathic pain syndromes. In other experimental symptoms, such as touch-evoked allodynia following intrathecal strychnine, the therapeutic index of opioids could be reduced (87). Thus, the pharmacology of painful clinical neuropathic syndromes should be evaluated not only through global pain relief in specific clinical diagnosis but also by taking into account the selective drug effects on each sensory symptom as described for allodynia and lidocaine in clinical neuropathic pain conditions (74). Furthermore, the evaluation of drug response in terms of sensitive and cognitive factors is susceptible to reveal distinctive responses, as recently demonstrated in neuropathic pain by Kupers et al. (50).
The only way to assess the analgesic effects of morphine in neuropathic pain seems largely dependent on the elucidation of the underlying pathophysiological mechanisms, with a parallel appraisal of individual pharmacodynamic and pharmacokinetic factors. If all these factors are taken into account, they could undoubtedly clarify some of the apparent interindividual variabilities of opioid analgesic effects described in the literature.
Finally, there is a lack of control studies after chronic administration of opioid in such conditions and a rapid development of tolerance could disgrace their clinical use. However, animals studies performed on the mononeuropathic rats do not favor this hypothesis (65).
There is so far no clear physiopathological evidence that morphine or other opioid agonists would not be effective in peripheral neuropathic syndromes, which are by far the most important cause of pain after neurologic damage.
29aCattaneo I, Kayser V, Gacel G, Guilbaud G. Effects of a new e opioid antagonist naltrindole on the bidirectional effect of naloxone in arthritic rats. 14th Annual Meeting of the European Neuroscience Association, 1991;37(abstract 1167).
31aDesmeules JA, Kayser V, Gacel G, Guilbaud G, Roques BP. The highly selective e agonist BUBU induces an analgesic effect in normal and arthritic rat and this action is not affected by repeated administration of low doses of morphine. Brain Res (in press ).