Introduction

Top Abstract Introduction Methods Results Discussion References

Neuropathic pain is an important complication of diabetes and clinical studies have reported the difficulty of managing it even though antidepressants have been shown to be partially effective. The response of chronic neuropathic pain to opioid treatment is controversial and according to some authors, neuropathic pain is intrinsically nonresponsive to opioids (Arnér and Meyerson, 1988a, b), and to others, underdosing opioid medication may be responsible for this poor activity (Portenoy et al., 1990). Experimental models of diabetes have been described as relevant models of chronic pain with alterations of pain sensitivity (Calcutt et al., 1996; Courteix et al., 1993; Kamei et al., 1991; Lee and McCarty, 1990), less ability to depend physically on morphine (Shook and Dewey, 1986) and poor response to opioids administered systemically (Courteix et al., 1994; Raz et al., 1988; Simon and Dewey, 1981; Simon et al., 1981) or supraspinally (Kamei et al., 1992; Suh et al., 1996).

The mechanisms by which diabetes alters morphine potency are not clear, and few explanations have been proposed. Alteration of opiate receptor affinity for [³H]naloxone has been described in brain membranes from diabetic (db/db) mice (Morley et al., 1981). Dysfunction of the supraspinal mediation of opiate analgesia (Kamei et al., 1992) and decrease of the release of serotonin from the bulbospinal pathways (Suh et al., 1996) known to be activated by morphine (Widgor and Wilcox, 1987) have been reported in diabetic rats. However, other hypotheses, poorly or never investigated in the streptozocin-induced diabetic rat model, could account for the reduced efficacy of opiates. Reduced ability of opioid drugs to bind to mu and/or delta opiate receptors could occur, which may be revealed by a decrease in the binding characteristics of appropriate ligands to opiate receptors, or modifications of the pharmacokinetics of morphine could occur, such as reduced production of M6G which possesses a greater potency than morphine as a mu opioid agonist (Abbott and Palmour, 1988; Gong et al., 1991, 1992; Pasternak et al., 1987) or a lesser amount of the active drug because of an increase of its elimination.

First, this study compared the antinociceptive potency of morphine to relieve pain caused by mechanical (pressure), thermal (warm and cold) and chemical (formalin injection) stimuli in STZ-induced diabetic rats and normal rats. The use of various stimuli led to the determination of a complete spectrum of the antinociceptive effect of morphine taking into account that differences in the effect of opiates, according to the nature of the stimulation, have been described (Hill, 1994; Millan, 1986). Moreover, the different reactions assessed (i.e., paw withdrawal, tail withdrawal, paw elevation and paw licking) provided an opportunity to study the action of morphine at different sites of integration of pain (i.e., spinal or supraspinal). Second, to investigate possible opioid receptor changes induced by diabetes, the binding characteristics of [³H]DAMGO and [³H]DPDPE to mu and delta opiate receptors, respectively, were determined in diabetic and normal animals. Third, because the metabolite M6G is a long-lasting and powerful analgesic in animals (Gong et al., 1991) and humans (Abbott and Palmour, 1988; Portenoy et al., 1992; Sullivan et al., 1989) and binds to mu and delta opiate receptors (Hucks et al., 1992), and because M3G, another metabolite of morphine, may antagonize the antinociceptive effect of morphine (Ekblom et al., 1993), we compared the metabolism of morphine in diabetic and normal rats. Furthermore, the usual plasma pharmacokinetic parameters were determined.

	Methods

Top Abstract Introduction Methods Results Discussion References

Animals and Induction of Diabetes

Male Sprague-Dawley rats (Charles River France, Cléon, France) weighing 200 to 250 g were used. They were housed three per cage under standard laboratory conditions, and given food and water ad libitum. Because some suffering might result from this experiment, the IASP Committee for Research and Ethical Issues Guidelines (Zimmermann,1983) were followed.

The rats were rendered diabetic with an intraperitoneal injection of STZ (75 mg/kg) (Zanosar, Upjohn, St. Quentin en Yvelines, France) dissolved in distilled water. Diabetes was confirmed 4 weeks later by measurement of tail vein blood glucose levels with Ames Dextrostix and a reflectance colorimeter (Ames Division, Miles Laboratories, Puteaux, France). Only rats with a final blood glucose level 14 mM were included in the study.

Control (normal) rats received only distilled water.

Behavioral Study

Test procedures. Mechanical stimulus: the paw pressure test. The rats were submitted to the paw pressure test described previously by Randall and Selitto (1957). Nociceptive thresholds, expressed in grams, were measured with a Ugo Basil analgesimeter (Apelex; tip diameter of probe, 1 mm; weight, 30 g) by applying an increasing pressure to the left hind paw until withdrawal (cut-off was 750 g). The AUCs of the score variations were calculated by the trapezoidal rule to compare the global effect of morphine between normal and diabetic rats. The use of score variations (threshold values obtained for each experimental time-predrug threshold value) was justified by the difference in predrug threshold values between the two groups.

Treatment protocol and experimental design. Tests took place 4 weeks after the induction of diabetes. Only rats in which the reduction in pain scores at week 4 of diabetes was more than 15% of the value obtained in normal rats (except for the formalin test) were included. The animals were submitted to the nociceptive test (paw pressure or tail immersion) before drug injection. Once two stable threshold values were obtained, morphine or saline (NaCl, 0.9%) was injected i.v. in the caudal tail vein. Rats were arranged randomly in cages, each rat receiving either morphine or saline in the same volume (0.1 ml/100g b.wt.).

Statistical analysis. Statistical significance was assessed with a two-way analysis of variance followed, when the F value was significant, by a Dunnett's test to analyze the time course of the effect. Significant differences between normal and diabetic groups were determined by a Student's t test. The significance level was .05.

Binding Study

Preparation of membranes. Diabetic and normal rats were decapitated and whole brains (mu opioid receptors) or cerebral cortex (delta opioid receptors) were removed rapidly and placed on ice. The first preparation step is common to the two types of structure. The cerebral structures were homogenized in 10 volumes (w/v) of ice-cold 50 mM Tris-HCl buffer (pH 7.7) with a Ultra Turrax homogenizer. The homogenates were centrifuged at 48,340 × g (Beckman J2-21 M/E) for 15 min at 4°C, the pellets resuspended in 10 volumes of fresh Tris-HCl buffer and incubated at 37°C for 40 min to dissociate any receptor-bound endogenous opioid peptides. The homogenates were centrifuged again as described above. For the study of mu opioid receptors, the final pellet was used immediately and resuspended in 30 volumes of fresh Tris-HCl buffer to yield a final protein concentration of 0.5 mg/ml. For the study of delta opioid receptors, the preparation was frozen (70°C) until the day of use. After thawing, the pellet was resuspended in 50 volumes of fresh Tris-HCl buffer to yield a final protein concentration of 0.1 mg/ml.

Receptor binding assay. For the delta receptors, membrane suspension (0.5 mg protein/ml) was incubated with 1 to 60 nM [³H]DPDPE (25 Ci/mmol, Amersham, Les Ulis, France) in the absence and presence of 10 µM naloxone (Sigma, St. Quentin Fallavier, France) for determination of nonspecific binding.

Data analysis. Specific binding was defined as total binding minus nonspecific binding. Data of saturation experiments were analyzed with the programs EDBA (McPherson, 1983, 1985) and LIGAND (Munson and Rodbard, 1980).

Pharmacokinetic Study

Sample collection. Diabetic (n = 9) and normal (n = 9) rats received an i.v. injection of 4 mg/kg of morphine hydrochloride. Blood samples were collected at different times (5, 15, 30, 60, 90, 120 min after the injection) by intracardiac puncture under light anesthesia (halothane, 0.01% thymol). All samples were centrifuged (3000 × g at 4°C for 10 min) and stored at 40°C until the assay.

Morphine and its metabolites assay. Samples were assayed by high-performance liquid chromatography with an ion-pair formation that complied with the analytical recommendations of some authors (Derendorf and Kaltenbach, 1986; Zoer et al., 1986; Konishi et al., 1990). Extraction conditions and limits of quantification for morphine were improved markedly. Morphine, N-ethylnormorphine (internal standard) and glucuronides (M3G and M6G) were purchased from Asta Medical Laboratories (Mérignac, France) and used without further purification. The method used consisted of a solid-liquid extraction procedure requiring Bond Elut C18 columns which need pretreatment (Varian, St. Quentin en Yvelines, France) and a connection to an Alltech vacuum manifolds device (Alltech Chromatography, Templeuve, France). Plasma sample (0.5 ml) and internal standard (5.5 ng/ml in water, 50 µl) were added. The mobile phase, filtrated before use (HV 0.45-µm filters, Millipore, St. Quentin en Yvelines, France) was: 115 ml of acetonitrile and a dose of Pic B7 low UV reagent which were added to 885 ml of 0.01 M phosphate monobasic buffer (pH 2.1; adjusted with phosphoric acid). The samples were eluted at 30.2°C with heater equipment at the constant flow rate of 1.2 ml/min. A 300 mm × 3.9 mm µBondapack dimethyloctadecylsilyl (10 µm) amorphous silica was used (Waters-Millipore). The equipment used was a Shimadzu two-piston LC 9A pump high-performance liquid chromatograph (Touzard and Matignon, Vitry-sur-Seine, France) connected to a Perkin-Elmer ISS-100 autosampler (Perkin-Elmer, St. Quentin en Yvelines, France) combined, on the one hand, with an ESA Coulochem (model 5100A) electrochemical detector (detection of both morphine and M6G, Touzard and Matignon), and on the other hand, with a fluorescence spectrophotometer (detection of M3G) Merck F1050. To remove impurities in the mobile phase by electrolysis, the guard cell (ESA model 5020) was placed between the pump and the automatic injector; its potential was set at +0.50 V. Detections were performed, on the one hand, with the ESA Coulochem operating with an oxidation voltage at +0.35 V (ESA analytical cell) and with its sensitivity set at 100 nA full scale, and on the other hand, with the fluorescence spectrophotometer (_ex = 210 nm and _em = 350 nm). The response signals (height integrations, calculations and plotting of the chromatograms) were carried out by a data processor fitted with the I.C.S. PIC3 analytical software (Instrumentation Consommable Service, Toulouse, France).

Pharmacokinetic study. Data were analyzed by a noncompartmental method (Rowland, 1980). The programs used were Pharmac BD 1.0 and Siphar-Win. Terminal half-life (T_1/2, hours) was calculated by the equation: T_1/2 = ln2/. The area under the concentration-time curve from time zero to time 2 h of the last sample (AUC_02h; ng·h/ml) was calculated by the trapezoidal rule and was extrapolated to infinity. The mean residence time (M.R.T., h) was determined by the formula of Yamaoka et al. (1978). The clearance (Cl, ml/min/kg) of morphine was calculated by the following equation: Cl/F = Dose/AUC₀. The apparent volume of distribution (V_d, l/kg) was determined as follows: V_d = Cl/. T_max, C_max and C_2h were experimental values.

Result expression and statistical analysis. Plasma concentrations are expressed in nanograms per milliliter. Results are expressed as mean ± S.E.M. Statistical significance was assessed with a two-way analysis of variance, followed by a Dunnett's t test to analyze the time course of the effect when the F value was significant. Significant differences between normal and diabetic groups were determined by a Student's t test. The significance level was .05.

	Results

Top Abstract Introduction Methods Results Discussion References

Effect of Morphine on Nociceptive Tests

Mechanical stimulus. Before morphine injection, withdrawal thresholds of normal rats (from 123.3 ± 4.8 to 134.3 ± 6.6 g) were significantly higher (P < .01) than those of diabetic rats (from 80.0 ± 2.9 to 82.9 ± 1.8 g). Morphine (2 and 4 mg/kg) dose-dependently and significantly (F = 3.254, P = .0009 and F = 4.813, P < .0001, respectively) increased thresholds in normal rats (fig. 1A). The maximum increase at 40 min was +60 ± 21 g and +152 ± 32 g for 2 and 4 mg/kg morphine, respectively. This antinociceptive effect lasted 50 and 100 min after the injection of 2 and 4 mg/kg, respectively.

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Time course of the effect of several i.v. doses of morphine on mechanical pain thresholds in normal (A) and diabetic (B) rats. Paw withdrawal thresholds determined before (0) and after drug injection are expressed in grams. (C) Effect of morphine on pain thresholds in normal and diabetic rats. Results are expressed as the mean surface bounded by the time-course curves (AUC) of score variations (postdrug-predrug threshold values) calculated by the trapezoidal rule. Bars = S.E.; n = 7 for each morphine or saline-treated group. *P < .05 versus corresponding predrug values; P < .05 versus scores obtained in corresponding normal rats.

Thermal stimuli. Tail immersion in a 46°C water bath. Initial tail withdrawal delays were from 7.1 ± 0.5 to 8.7 ± 0.4 sec and from 5.3 ± 0.5 to 6.1 ± 0.4 sec in normal and diabetic rats, respectively. The significant difference (P < .01) between these scores confirms the diabetes-induced hyperalgesia (fig. 2, A and B).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Effect of several i.v. doses of morphine on tail withdrawal of normal (A) and diabetic (B) rats submitted to the tail immersion test in warm (46°C) water. Percentages of maximal possible effect (%MPE) are reported before (0) and after drug injections. Bars = S.E.; n = 7 for each morphine or saline-treated group. *P < .05, **P < .01, ***P < .001 versus corresponding predrug values.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Effect of several i.v. doses of morphine on tail withdrawal thresholds of diabetic rats submitted to the tail immersion test in cold (10°C) water. Percentages of maximal possible effect (%MPE) are reported before (0) and after drug injections. Bars = S.E.; n = 7 for each morphine or saline-treated group. *P < .05, *** P < .001 versus corresponding predrug values.

Chemical stimulus. Formalin-induced hyperalgesia was significantly (P < .05) more marked in diabetic (fig. 4B) than in normal (fig. 4A) rats. The scores of the first 3 min were 1.57 ± 0.16 and 1.93 ± 0.07 in normal and diabetic animals, respectively. The duration of the first phase was longer in diabetics (0-18 min) than in normal (0-9 min) rats. In diabetic animals, hyperalgesia of the second phase was significantly more intensive (P < .05) from the 18th to the 27th min than in normal rats.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Effect of i.v. morphine (2 mg/kg) on behavioral scores obtained after injection of formalin (5%) in the forepaw of normal (A) and diabetic (B) rats. Scores are the mean time (±S.E.M.) spent in the specific behavioral categories and affected by their category weights; bars = S.E.; n = 7 for each morphine- or saline-treated group. *P < .05, **P < .01, ***P < .001 versus scores obtained after NaCl (0.9%).

[³H]DAMGO and [³H]DPDPE Binding

No significant difference was observed in the binding of [³H]DAMGO to brain mu receptors between normal and diabetic rats. K_D values were 1.52 ± 0.26 nM and 1.54 ± 0.45 nM, respectively, and B_max values were 189.3 ± 9.4 fmol/mg protein and 163.3 ± 22.1 fmol/mg protein, respectively. The binding characteristics of [³H]DPDPE to delta receptors in diabetic rats were not significantly different from those obtained in normal animals. The K_D and B_max values were 5.71 ± 0.62 nM and 151.3 ± 9.8 fmol/mg protein, respectively, in diabetic rats, and 4.63 ± 0.29 nM and 129.7 ± 10.7 fmol/mg protein, respectively, in normal rats.

Pharmacokinetics of Morphine and Its Metabolites

The pharmacokinetic parameters for morphine and its metabolites are shown in table 1. No significant variation was observed between diabetic and normal rats for C_max, T_1/2, AUC and MRT values of either morphine or its metabolites. However, morphine Cl and V_d were increased significantly in diabetic rats, whereas no significant change was observed for its metabolites M3G and M6G between diabetic and normal animals.

	Discussion

Top Abstract Introduction Methods Results Discussion References

The present results in STZ-treated rats confirm that diabetes induces alterations of nociceptive thresholds and clearly indicate that morphine-induced analgesia is attenuated in diabetic animals, whatever the nature (mechanical, thermal, chemical) or the intensity (noxious or non-noxious) of the applied stimulation and the level of integration (spinal or supraspinal) of the response. This univocal result distinguishes this model from the model described by Bennett and Xie (1988). In this model of neuropathic pain in which a chronic constrictive nerve injury produces allodynia and hyperalgesia, the efficacy of opioids is increased (Attal et al., 1991) or reduced (Yamamoto and Yaksh, 1992) according to the stimuli (mechanical and thermal, respectively).

As reported previously on vocalization thresholds (Courteix et al., 1994), 2-fold higher doses of morphine are necessary to increase withdrawal thresholds in diabetic rats in comparison with normal rats. This demonstrates that both spinal (paw withdrawal) and supraspinal (vocalization) responses are depressed in diabetic rats. This alteration of both spinal and supraspinal responses to pain was reported also by Kamei et al. (1992) with use of the tail-pinch (supraspinal response) and tail-flick (spinal response) tests in STZ diabetic mice. In the same way, an alteration of content and/or release of endogenous opioid peptides at both spinal and supraspinal levels as well as changes in their interaction with opiate receptors has been suggested in diabetic mice (Ramabadran et al., 1989). Taken together, literature data and the present results show that diabetes induces a generalized alteration that modifies the activity of morphine at the spinal and supraspinal level.

The reduced efficacy of morphine was confirmed by use of a thermal nociceptive stimulation (tail immersion in a 46°C water bath). The dose of 1 mg/kg was devoid of activity in diabetic rats, whereas it increased tail withdrawal latency in normal rats. Furthermore, the MPE values obtained with 2 and 4 mg/kg morphine were significantly lower in diabetic (52 ± 12% and 69 ± 4%, respectively) than in normal rats (79 ± 12% and 88 ± 4%, respectively). The need to reach the dose of 2 mg/kg to alleviate thermal allodynia (cold stimulation) confirms the poor sensitivity to morphine of diabetic animals. For the dose of 2 mg/kg, the activity of morphine was more effective in countering allodynia (MPE = 92 ± 5%) than in reducing hyperalgesia caused by warm stimulation (MPE = 52 ± 12%). The impairment of morphine efficacy in thermal pain tests agrees with previous results of Simon and Dewey (1981) and Simon et al. (1981) which showed that acute or chronic hyperglycemia reduced the antinociceptive effect of morphine subcutaneously administered in the tail-flick test. Raz et al. (1988) by use of the hot plate test also showed a reduction of morphine analgesia in hyperglycemic rats. The intracerebroventricular injection of morphine resulted in an alteration of the response mediated by supraspinal mu opiate receptors measured by the tail-flick assay (Kamei et al., 1992; Suh et al., 1996). Finally, Ramabadran et al. (1989), with use of the same test but with a stimulus temperature of 50°C, demonstrated that the hyperalgesic response to naloxone was abolished in diabetic mice.

In the formalin test, the scores of the two phases were significantly greater in diabetic rats than in normal rats as described previously by Malmberg et al. (1993). Morphine significantly decreased responses in both the early and the late phase of the formalin test in diabetic rats. In normal rats, an antinociceptive effect was reached only on the late phase; the lack of effect of morphine on the early phase was inconsistent with literature data (Dubuisson and Dennis, 1977) even though morphine-induced score reduction of the first phase is usually lower than that of the second phase. Despite this discrepancy, the comparison between normal and diabetic rats clearly shows a significantly lower decrease in the second phase scores after morphine injection in diabetic rats than in normal animals.

To resolve whether prolonged hyperglycemia affects the affinity and the density of brain mu and delta receptors, we investigated the saturation curves of tritiated DAMGO and DPDPE to membrane preparations from diabetic and normal rats. Whole brain and total cortex were used to give a general approach of the effect of STZ treatment on cerebral opioid receptors, assuming that if alterations occurred, they would be revealed on total brain or cortex. Thus, the reduction of the analgesic activity of morphine in diabetic rats cannot be explained by a variation of K_D and B_max values for mu and delta opioid receptors, because no significant change is evident in comparison with K_D and B_max values for both opioid receptor subtypes in normal rats. However, although the differences in mu and delta receptors were not statistically significant, the density of the mu binding sites was reduced by approximately 14% by STZ treatment. On the other hand, B_max for DPDPE on delta receptors seemed to be (but not significantly) increased by approximately 17%. These data suggest that the opioid receptor populations are imbalanced after STZ treatment. The use of another methodology (i.e., membrane pretreatment with Na⁺ and GDP and binding in presence of GDP), as recently described by Liu-Chen et al. (1995), could have revealed small changes in receptor binding. However that may be, our conditions are the same as used by Adams et al. (1987), who successfully have reported alterations (a 20-31% reduction) in mu binding sites in rat brain induced by -funaltrexamine treatment. Finally, because our study carried out in whole-brain homogenate did not show changes of opioid receptor binding, it is also possible that uncoupling of G-protein with the mu opiate receptor and/or changes in mu opiate receptor-associated second messenger systems might result in the reduced effectiveness of morphine, as suggested by Mao et al. (1995) in an animal model of neuropathic pain.

Because the relationship between opioid concentration and analgesic efficacy is steep (Levine et al., 1983), the pharmacokinetic behavior of morphine, M3G and M6G, was considered in addition to opiate receptor exploration. In normal rats, the obtained morphine Cl, V_d and T_1/2 values are similar to those reported by Hasselström et al. (1996). The amount of M6G formed after the administration of morphine is negligible (Hasselström et al., 1996) or absent (Coughtrie et al., 1989) in rats, compared with the amount of M6G found in humans (Coughtrie et al., 1989; Stain et al., 1995). In our study, morphine was converted to M3G as the major metabolite. However, the M3G and M6G kinetic parameters reported (Hasselström et al., 1989) were obtained after injection of each metabolite and consequently cannot be compared with ours.

In diabetic rats, some pharmacokinetic parameters of morphine such as V_d and Cl are modified. The increase of V_d might be caused by glycosylation of plasma proteins, reported previously in diabetic patients (Gwilt et al., 1991), which can alter the protein binding and increase the unbound fraction of morphine. Because morphine and glucuronides are hydrophilic compounds, high amounts may be attracted in aqueous compartments. Thus, it is possible that the amount of morphine in blood crossing the brain-blood barrier and present in the central nervous system is reduced. The existence of a relationship between the intensity of analgesia and the levels of morphine and its glucuronide metabolites in the cortical extracellular fluid (Barjavel et al., 1995) has been demonstrated by use of the microdialysis procedure. Consequently, a much greater fraction of morphine might reach the biophase in normal rats than in diabetic rats, which explains the difference of potency of the drug.

Total clearance of morphine also is increased. This probably is not related to variations of hepatic metabolism despite the previously described impaired metabolic activity of hepatocytes from STZ rats toward xenobiotics (Favreau and Schenkman, 1988) but rather may be related to the well-known modifications of renal function described in humans (Gwilt et al., 1991), especially the increase of glomerular filtration occurring in early diabetes mellitus. The higher total clearance observed in diabetic rats than in normal rats clearly means that frequent doses will be required to maintain the same blood concentration in the two groups of animals as it is the rule in humans (Mather, 1987).

Thus, both the increase of morphine Cl, which results in a nonsignificant reduction of morphine AUC by approximately 37%, and the increase of morphine V_d, which results in a high diffusion of morphine in the aqueous compartment, can lead to a reduction of available morphine and to a decrease of its amount in the target organs, i.e., in the central nervous system. These two process may explain the reduced effect of morphine, despite the unmodified T_1/2. The examination of experimental values obtained from the paw pressure test and plasma morphine levels tended to show that higher morphine levels would be necessary in diabetic rats to obtain the same pharmacological effect as observed in normal rats.

The present study fails to show any change in morphine metabolism. For morphine, phase II reactions are the most important metabolic reactions, and glucuronidation is the major reaction in humans, but this process is species dependent (Xu et al., 1993). Taking into account (1) the increased -glucuronidase activity observed in diabetes (Rao et al., 1989) and (2) the previously reported increase of glucuronidation of the hydroxylated metabolites of amitriptyline in STZ-induced diabetic rats (Coudoré et al., 1996), increases of M3G and M6G were expected, and the unchanged pharmacokinetics of morphine glucuronides observed in this study was surprising. Nevertheless, M3G/morphine and M6G/morphine AUC ratios obtained in diabetic rats (5.10 and 0.20, respectively) tends to be higher than those obtained in normal rats (2.93 and 0.16, respectively). It is possible that an increase of glucuronidation may occur but that it is compensated by the increase of morphine and/or glucuronide elimination. Because glucuronide pharmacokinetic parameters remain unchanged, this phenomenon does not influence morphine analgesic activity. However, in rats, despite contradictory reports, morphine glucuronidation, especially in the 6-position seems to be deficient (Oguri et al., 1990; Kuo et al., 1991) and eventual modifications may be not observed. Specific in vitro studies on morphine metabolism with hepatocytes from STZ-diabetic rats and in vivo behavioral studies with direct administration of M6G and M3G would be necessary.

Study of phase I metabolism was impossible because levels of oxided metabolites, N-demethylated morphine and morphine-N-oxide, or of the O-methylated metabolite, codeine, were not assessed. However, the oxidation reaction minor pathways of morphine metabolism and the activity of the specific cytochrome P450 involved in the formation of codeine, i.e., cytochrome P2D6, were unchanged in human diabetes mellitus (Bechtel et al., 1988), contrary to those observed with other cytochrome P450 isoenzymes in STZ-diabetic rats (Favreau and Schenkman, 1988).

To conclude, the data reported herein confirm that diabetes-induced hyperglycemia alters pain sensitivity by inducing mechanical, thermal and chemical hyperalgesia and thermal allodynia. They show that this metabolic trouble reduced the antihyperalgesic and antiallodynic activity of morphine whatever the nature of stimulus applied and the level of integration of pain reaction. This reduced efficacy is not caused by significant changes in the binding parameters for brain mu and delta opiate receptors. The meaningful explanation is in the pharmacokinetic alteration of morphine (increase in V_d and Cl) which could lead to the reduction of morphine levels in central nervous system. However, further investigations are needed to determine the involvement of some endogenous systems that are involved in opiate analgesia. Then, the recent demonstration of the decrease in morphine antinociception caused by activation of N-methyl-D-aspartate receptors (Mao et al., 1995) and of the modulatory role of antiopioid peptides (neuropeptide FF or cholecystokinin) in acute and chronic pain states (Gouardères et al., 1996; Stanfa et al., 1994, respectively) would justify specific studies in diabetic rats.