Combination Therapy with Flupirtine and Opioid: Studies in Rat Pain Models

PAIN MEDICINE Volume 9 • Number 7 • 2008

TRANSLATIONAL RESEARCH SECTION

Original Research Article Combination Therapy with Flupirtine and Opioid: Studies in Rat Pain Models Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021

Colin S. Goodchild, MA, MBBChir, PhD, FRCA, FANZCA, FFPMANZCA,*† Anton Kolosov, BSc,* Adam P. Tucker, MB, BS, PhD,* and Ian Cooke, PhD† *Department of Anaesthesia and Perioperative Medicine, Monash Institute of Medical Research, Monash University, Clayton, Victoria; †CNSBio Ltd Pty Ltd, Melbourne, Victoria, Australia

ABSTRACT

Objectives. Flupirtine is an established clinical analgesic for mild to moderate musculoskeletal pain states. It has recently been shown to be a KCNQ2–3 potassium channel opener. These experiments were performed to see if this property could be useful in treating pain states characterized by central sensitization with the drug either given alone or in combination with morphine. Design. Experiments were performed in rats in an observer-blinded fashion with vehicle controls. Nonsedating doses of flupirtine, morphine, and combinations containing both drugs were defined using the rotarod test and open-field activity monitoring. Dose–response relationships were determined for nonsedating doses of both drugs given alone and together in combination in causing antinociception in two nociception paradigms: carrageenan paw inflammation and streptozotocin- induced diabetic neuropathy. Results. Flupirtine and morphine, when given alone at the highest nonsedating doses, caused slight to moderate antinociception in both paradigms. Flupirtine also caused significant increases in morphine antinociception in both models. In carrageenan paw inflammation, complete reversal of carrageenan-induced hyperalgesia was caused by 10 mg/kg flupirtine in combination with 0.4 mg/kg morphine. These doses of the two drugs were ineffective when given alone, but the combination caused complete antinociception in this model of inflammatory pain. In the diabetic neuropathy model, morphine 3.2 mg/kg given alone caused no significant antinociception. However, a lower dose of morphine (1.6 mg/kg shown to be ineffective when it was given alone) given in combination with flupirtine 10 mg/kg caused highly significant antinociceptive effects causing complete reversal of hyperalgesia caused by diabetic neuropathy (P < 0.001, one-way analysis of variance). This combination of drugs was not sedating. Conclusions. Flupirtine increases morphine antinociception without causing an increase in sedation. Flupirtine should be investigated as an adjunct analgesic with opioids for the management of patients with pain states involving central sensitization.

Key Words. Flupirtine; Morphine; Inﬂammation; Antinociception; Animal Models; Neuropathic Pain; Diabetes; Streptozotocin; Carrageenan

Reprint requests to: Colin S. Goodchild, MA, MBBChir, Introduction PhD, FRCA, FANZCA, FFPMANZCA, 38 Somers euronal potassium channels [KCNQ2–5 or Avenue, Malvern, Victoria, 3144 Australia. Tel: +61 (4) 1456 1401; Fax: +61 (3) 9824 6322; E-mail: N Kv7.2–7.5]) were ﬁrst described in 1998 [email protected]. after it was discovered that the channel proteins

© American Academy of Pain Medicine 1526-2375/08/$15.00/928 928–938 doi:10.1111/j.1526-4637.2008.00514.x Flupirtine–Morphine Combination Therapy: Preclinical Studies 929 shared common sequences with potassium chan- retina [10]. However, binding studies have failed nels found in cardiac tissues [1]. Neuronal KCNQ to show that flupirtine has an affinity for NMDA channels have been found to be molecular sub- recognition sites. The controversy has finally been strates of M channels, and several molecules have solved by the discovery that flupirtine and the been identified as openers or activators of these structurally related analog retigabine bind with channels. The M current has been originally iden- KCNQ receptors and act as potassium channel tified as a membrane hyperpolarizing current cau- openers. As such, these compounds act to increase sed by muscarine [2]. It is the only hyperpolarizing the M current, decreasing neuronal excitability, current that operates around resting membrane and thus decreasing the effect of a number of neu-

potential and thus has the possibility of drastically rotransmitters including glutamate (at NMDA Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 altering neuronal excitability [3]. Involvement of receptors) [11]. KCNQ channels with the M current suggests that Although flupirtine has been available for use drugs acting on neuronal KCNQ channels could as an analgesic in humans for many years, its use be useful in treating a variety of clinical conditions has been limited and confined to monotherapy; it caused by neuronal hyperexcitability such as has not been described in combination therapies. epilepsy, acute pain, and neuropathic pain [4]. It is reported to be less effective for severe pain Although this theoretical property has been sug- compared with morphine or pethidine but as gested, it has not been tested by experiment. effective an analgesic as codeine or pentazocine However, there are experimental observations that with a similar side-effect profile [7]. It is the dose- might suggest how this property might be used in related side effects of somnolence, dizziness, and antinociception and analgesia. Activation of the M hallucinations that have limited its use (as mono- current by openers of KCNQ channels decreases therapy) to mild or moderate musculoskeletal the release of a number of neurotransmitters pain syndromes. known to be involved with such neuronal hyper- In the light of the new evidence on its mode of excitability such as excitatory amino acids [5]. It is action, it is appropriate to consider flupirtine again well established that glutamate is involved with but for the treatment of pain states that involve central sensitization states such as neuropathic neuronal hyperexcitability. Furthermore, a drug pain in which it targets the N-methyl-d-aspartate that controls neuronal hyperexcitability via the M (NMDA) subtype of excitatory amino acid recep- current, and thus NMDA receptor activation, is tor. Compounds inhibiting this action, such as likely to have synergistic or additive interactions NMDA antagonists, are active as antinociceptive with other analgesics, particularly opioids such as agents to some extent when given alone but more morphine. The latter provides for the possibility potently when combined with opioids such as of a more effective treatment for these pain states morphine. Unfortunately, the compounds devel- using smaller doses of the constituent drugs in a oped as NMDA antagonists have an unfavorable combination and thus reducing the likelihood of therapeutic effect/side-effect profile. Perhaps dose-limiting side effects. This would only be another method of invoking this mechanism, for useful, of course, if there was no similar synergy example, KCNQ channel openers, might have less between them with respect to sedating properties problems with therapeutic ratio. or other undesirable side effects. Flupirtine is a centrally acting nonopioid anal- This article describes experiments performed gesic that has been used clinically in Germany using two rat models of inflammatory and neuro- since 1984. It has been used for the treatment of a pathic pain investigating the antinociceptive ac- variety of pain states, but it has not been advocated tivity of flupirtine given alone and also in for the treatment of neuropathic pain. Despite its combination with morphine at low doses, less than long history of clinical use, the exact mode of those causing side effects of central nervous system action of flupirtine as an analgesic has remained depression. unknown until recently. In early studies, it has been found that flupirtine does not appear to Methods interact with serotonin, dopamine, nicotine, or adrenoceptors [6,7]. Subsequently, flupirtine has All experiments were carried out with the approval been found to decrease spinal polysynaptic reflexes of Monash University Standing Committee on mediated by NMDA receptors [8] and a2 adreno- Animal Experimentation and in all experiments, ceptors [9], and to antagonize NMDA-induced attention was paid to ethical guidelines for the changes on g-aminobutyric acid (GABA) in the investigation of experimental pain in conscious 930 Goodchild et al. animals [12]. All experiments reported in this the minimum speed for two training sessions of article were performed on male Wistar rats 1–2 minutes separated by an interval of 30–60 (weight 150–200 g for experiments investigating minutes. After this conditioning period, an intra- sedation and antinociception using the carrag- peritoneal injection of vehicle, drug, or drug com- eenan inflammation paradigm and weight 65–80 g bination was given. Five minutes later, the animals for experiments evaluating sedation and antinoci- were placed onto the rotarod at a constant speed of ception using the diabetic neuropathy model). 4 rpm. As the animal took grip of the drum, the The experiments were performed in an observer- accelerator mode was selected on the treadmill, blinded fashion with parallel saline vehicle treat- i.e., the rotation rate of the drum was increased

ment controls. All drug solutions and vehicle were linearly at the rate of 20 rpm every minute there- Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 given intraperitoneally in a volume of 1.0 mL. after. The time was measured from the start of the acceleration period until the rat fell off the drum; Materials this was the control (pretreatment) performance Morphine sulphate (Mayne Pharma Pty Ltd), flu- time for each rat. The cutoff or maximum run time pirtine maleate (AstaMedica) and saline for injec- for the test was 2 minutes. Nonsedated rats can all tion (Pfizer) were obtained from the Monash run for 2 minutes. This test was performed on Medical Centre pharmacy (Victoria, Australia). each rat three times with intervals between tests of Streptozotocin (STZ) was obtained from Sapphire 10 minutes. The shortest run time measured after Bioscience Pty Ltd (Redfern, NSW, Australia). drug injection was identified during the 30-minute Type I carrageenan was purchased from Sigma- test period for each rat. These values were com- Aldrich Chemical Company (Castle Hill, NSW, bined for all rats treated with each drug at each Australia). Male Wistar rats (specific pathogen- dose to calculate means Ϯ standard error of the free HarLan: Wis strain) were supplied by Monash mean (SEM). The data from saline-treated vehicle University Animal Services. controls were compared with the data following intraperitoneal drug injections using one-way Test of Sedation analysis of variance (anova) with Dunnett post hoc Prior to investigation of the antinociceptive effects test. These comparisons allowed definition of drug of morphine and flupirtine, the doses of each drug doses that caused sedation in normal rats, the given alone and in combination that did not cause maximum dose that could be used without causing any signs of sedation or CNS depression were sedation, as defined by a run time equal to vehicle determined using the rotarod test [13]. Further, controls. bearing in mind that diabetic rats may be more Groups of rats were tested with the rotarod as susceptible to the sedating effects of drugs, some mentioned earlier with the following intraperito- of the doses of morphine and flupirtine and the neal treatments: two drugs in combination that were the highest • saline; nonsedating doses when assessed by the rotarod • morphine at doses of 0.4, 0.8, 1.6, 3.2, 6.4, and test were also given to rats made diabetic by treat- 12.8 mg/kg; ment with STZ (vide infra). Any sedation that • flupirtine at doses of 5, 10, and 20 mg/kg; might be caused by the drug treatment was • a combination of flupirtine at 5 mg/kg with assessed in these rats using the open-field activity morphine at 0.2 and 0.4 mg/kg; monitor. In addition to the possibility of increased • a combination of flupirtine at 10 mg/kg with sensitivity to CNS depressant drugs, there were morphine at 0.4 and 1.6 mg/kg. two reasons for using this test: Open-Field Activity Monitor • Diabetic rats have poor muscle power and are The rats were naive to the drugs, with no previous incapable of walking on the rotating drum, and exposure to the open-field activity box. They were thus sedation could not be assessed this way. placed individually in an open-field activity • It is a different test that looks for sedating effects monitor (Ugo Basile), which was a darkened box in in a nonstimulating environment. which the movement of the rat can be monitored Rotarod Test remotely by the frequency and the number of The rats were naive to the drugs, with no previous interruption of infrared beams directed across the exposure to the rotarod test. They were placed on box in a grid. The activity in the monitor was the rotarod accelerator treadmill (7,650 accelera- measured for a period of 20 minutes in each rat tor rotarod, Ugo Basile, Comerio VA, Italy) set at after an intraperitoneal injection of drug or saline. Flupirtine–Morphine Combination Therapy: Preclinical Studies 931

Figure 1 Time–response curves for the carrageenan paw inﬂammation paradigm. After three stable measurements of paw withdrawal latency, carrageenan was injected into one hind paw at time 0. This caused hyperalgesia as shown by the fall in the paw withdrawal latency readings. At time

120 minutes, saline, a drug, or drug Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 combination was injected intraperitoneally (ip). All drug and saline responses had achieved maximum onset by 20 minutes and remained stable for a further 20 minutes after drug injection. Thus, readings taken at these time points were combined and were recalculated as presented in Table 2 for statistical analysis.

If a rat was sedated by a drug treatment, the move- Time was allowed for the induction of inflamma- ments recorded would be less than the saline tion (2 hours). Paw withdrawal latencies (time controls. As the rats became habituated to the measured in seconds taken to withdraw the open-field monitor, only one experiment was per- inflamed paw) were measured using noxious heat formed on each rat with this test. Groups of rats from an infrared beam focused onto the plantar made diabetic and neuropathic as described next surface of the right hind paw in freely moving were tested with the open-field activity monitor as animals using apparatus from Ugo Basile. mentioned earlier with the following intraperito- Paw withdrawal latencies were measured every neal treatments: 10 minutes before the induction of inflammation with carrageenan injections until three stable read- • saline; ings were obtained (-20, -10, and 0, as shown in • morphine at a dose of 3.2 mg/kg; Figure 1). Once an inflammatory reaction was • flupirtine at a dose of 10 mg/kg; induced, paw withdrawal thresholds were mea- • a combination of flupirtine at 5 mg/kg with sured 60, 110, and 120 minutes after the carrag- morphine at 3.2 mg/kg; eenan injection to confirm the development of • a combination of flupirtine at 10 mg/kg with hyperalgesia, a decrease in paw withdrawal latency morphine at 3.2 mg/kg. typically from control pre-carrageenan level of 12 The rest times measured in seconds were com- seconds down to 6 seconds. A test drug or drug bined for all rats in each group and were expressed combination was injected and paw pressure values as means Ϯ SEM for each treatment group. Sta- were measured at 10-minute intervals for the fol- tistical comparisons between the drug treatment lowing 40 minutes. Replicate values of paw with- groups and the saline-treated controls were made drawal latencies for each time of measurement and with one-way anova with Dunnett post hoc cor- drug treatment were combined to calculate means. rection for multiple comparisons. These were plotted as time–response curves to assess the time of onset and the period of stability Antinociception Testing of the response to drug treatment (time of stable Two tests for antinociception were used: the car- response). The time of stable response involved rageenan paw inflammation test and the STZ- three readings after the drug injections: 140, 150, induced diabetic neuropathy model. and 160 minutes (see Figure 1). For each rat, these three readings were averaged (postdrug) as were Carrageenan Paw Inflammation the three pretreatment readings. All the pretreat- Inflammation of the right hind paw was induced ment values from rats in each treatment group by an intraplantar injection of carrageenan were combined as were the corresponding post- (100 mL of a 2% carrageenan solution in saline). drug values to calculate means and SEM. 932 Goodchild et al.

The following drug intraperitoneal treatments paw pressure test, more extensive nociceptive were given to separate groups of rats: testing paradigms were carried out in diabetic neuropathic animals and in weight-matched controls • saline controls; (weight: 130–200 g); the control rats were 1–2 • ﬂupirtine at doses of 5 and 10 mg/kg alone; weeks younger. The reason for performing these • morphine at doses of 0.4, 0.8; and 1.6 mg/kg measurements on normal rats was to establish the alone; normal paw withdrawal thresholds that might be • combinations of ﬂupirtine at 5 and 10 mg/kg expected from rats of this size that have smaller with morphine at 0.4 mg/kg. amounts of subcutaneous tissues that might affect

All means Ϯ SEM (pretreatment and postdrug) the measurement. In this way, a comparison of the Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 calculated for the saline and for each drug treat- rats made diabetic with these normal weight- ment group were compared statistically using matched controls allowed a clear demonstration of anova followed by Tukey–Kramer post hoc test. hyperalgesia. This allowed a comparison of the starting paw Paw pressure (PP) was measured by the method withdrawal latencies (pretreatment) for all groups described by Randall and Selitto using a Ugo to make sure that all groups were comparable prior Basile algesimeter (Apelex; probe 1 mm, weight: to treatment. The analysis also allowed compari- 10 g) [16]; increasing pressure was applied to the sons of pretreatment and postdrug values for each left hind paw until vocalization or sharp paw with- group to assess if there was a significant antinoci- drawal was elicited. Paw withdrawal thresholds ceptive effect. Finally, the statistical analysis also were measured for a group of weight-matched compared the postdrug values for each treatment controls and also in drug treatment groups of rats, to determine if one was greater than another. 20 and 10 minutes before, immediately before (time 0), and also at 20, 30, and 40 minutes after STZ-Induced Diabetic Neuropathy intraperitoneal injections of STZ-induced diabetic rats were prepared following the method of Courteix et al. [14,15]. Young • saline (controls); male Wistar rats (65–80 g at induction of diabetes) • flupirtine 5 mg/kg alone; were used in these experiments. • flupirtine 10 mg/kg alone; • morphine 1.6 mg/kg alone; Induction of Diabetes and Hyperalgesia • morphine 3.2 mg/kg alone; Rats were injected ip with STZ (150 mg/kg total • flupirtine 5 mg/kg plus morphine 3.2 mg/kg dose) dissolved in 0.9% sodium chloride solution. together; The 150 mg dose was given in two 80 mg/kg • flupirtine 10 mg/kg plus morphine 1.6 mg/kg injections on consecutive days. The induction of together. diabetes was confirmed 1 week after injection of STZ by measurement of tail vein blood glucose Values of paw withdrawal thresholds measured levels with AccuCheck Active test strips and a for individual rats in each treatment group were reflectance colorimeter (AccuCheck Glucometer, combined for each testing time to calculate the Roche, Castle Hill, NSW, Australia). Only animals means, which were plotted on time–response with final blood glucose levels Ն15 mM were curves as shown in Figure 2. These time–response deemed to be diabetic. The rats were retested for curves revealed that the responses to drugs, if hyperglycaemia once per week to confirm con- present, were apparent at 20 minutes after the tinued high blood glucose readings. Hyperalgesia injection of the drug or the drug combination, was assessed using the paw pressure test described and the response was constant and stable between previously by Randall and Selitto [16]. 20 and 40 minutes after the injection, which was Tests took place 5 weeks after the first injection given at time 0. For each rat, the three readings of STZ. Animals that had paw pressure nocicep- made at 20, 30, and 40 minutes after saline or tive thresholds below 30 g (60% of the value in drug treatment injection were averaged (post- normal weight-matched rats) were deemed to have drug) as were the three pretreatment readings at developed hyperalgesia/neuropathic pain and thus times -20, -10, and 0. All the pretreatment values were used in further experiments. from the rats in each treatment group were combined as were the corresponding postdrug values Nociceptive Test to calculate means and SEM. All means Ϯ SEM After the successful documentation of the devel- (pretreatment and postdrug) calculated for the opment of hyperalgesia in diabetic animals by the saline and for each drug treatment group were Flupirtine–Morphine Combination Therapy: Preclinical Studies 933 Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021

Figure 2 Time–response curves for the streptozotocin (STZ)-induced diabetic neuropathy paradigm. Paw withdrawal thresholds are shown in grams for 20 minutes prior to injection of saline, drug, or drug combination at time 0. Normal withdrawal thresholds in nondiabetic weight-matched rats are shown at time -20 minutes (mean 44.7 g). Compared with this datum point, all starting values in the other groups were reduced and were similar to each other indicating hyperalgesia was caused by STZ-induced diabetes. All drug and saline responses had achieved maximum onset by 20 minutes and remained stable for a further 20 minutes after drug injection. Thus, readings taken at these time points were combined and were recalculated as presented in Table 3 for statistical analysis. ip = intraperitoneally. compared statistically using anova followed by and 12.8 mg/kg ip, all remained walking on the Tukey–Kramer post hoc test. This allowed a rotating drum for periods that were not signifi- comparison of the starting paw withdrawal laten- cantly different compared with the saline controls cies (pretreatment) for all groups to make sure and the maximum run time of 120 seconds. It can that all groups were comparable prior to treat- be concluded from these experiments that sedation ment and that they were significantly less than the is caused by doses of flupirtine greater than values for weight-matched controls, meaning that 10 mg/kg and of morphine greater than 3.2 mg/ hyperalgesia had been caused by STZ-induced kg. In addition, the combination of 10 mg/kg diabetic neuropathy. The analysis also allowed flupirtine with 3.2 mg/kg morphine was also comparisons of pretreatment and postdrug values nonsedating. for each group to assess if there was a significant antinociceptive effect. Finally, the statistical Open-Field Activity Monitor analysis also compared the postdrug values for Table 1B shows the results in STZ-induced dia- each treatment to determine if one was greater betic neuropathic rats injected with saline, flupir- than another. tine, morphine, and combinations of flupirtine with morphine. The times spent resting, i.e., not moving and interrupting the infrared beams, were Results not significantly different for any of the drug treat- Test of Sedation ments compared with the saline controls (one-way Rotarod anova with Dunnett post hoc correction). Thus, Table 1A shows the results of the rotarod test for we may conclude that the maximum nonsedating normal animals injected with saline, morphine, drug doses are the same for diabetic neuropathic flupirtine, or a combination of flupirtine and mor- rats as for normal rats. phine. These results were analyzed using one-way anova with the Dunnett post-hoc correction com- Antinociception Testing paring all drug treatments with saline controls. Carrageenan Paw Inflammation The rats treated with all drug treatments, except The results of the carrageenan experiments are flupirtine 20 mg/kg ip and morphine at doses 6.4 shown in Figure 1. It can be seen that the effect 934 Goodchild et al.

Table 1 Results of sedation studies

A) Rotarod Performance (seconds): Normal Rats Mean SEM N Saline controls 119.6 0.23 59 Flupirtine 5 mg/kg ip 119.4 0.88 18 Flupirtine 10 mg/kg ip 115.9 3.87 26 Flupirtine 20 mg/kg ip 80.07* 15.77 10 Flupirtine 5 mg/kg ip + morphine 0.2 mg/kg ip 119.8 0.3 10 Flupirtine 5 mg/kg ip + morphine 0.4 mg/kg ip 119.8 0.25 10 Flupirtine 10 mg/kg ip + morphine 0.4 mg/kg ip 116.4 1.6 25 Flupirtine 10 mg/kg ip + morphine 1.6 mg/kg ip 118.5 0.99 10 Morphine 0.4 mg/kg ip 120.0 0 10

Morphine 0.8 mg/kg ip 120.0 0 10 Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 Morphine 1.6 mg/kg ip 111.0 6 10 Morphine 3.2 mg/kg ip 99.0 14.0 10 Morphine 6.4 mg/kg ip 77.0* 13.0 10 Morphine 12.8 mg/kg ip 48.0* 18.0 10

B) Open-Field Rest Times (seconds): Diabetic Rats Saline controls 960.0 45.0 10 Flupirtine 10 mg/kg ip 993.8 26.4 10 Flupirtine 5 mg/kg ip + morphine 3.2 mg/kg ip 982.6 17.5 10 Flupirtine 10 mg/kg ip + morphine 3.2 mg/kg ip 1,071.3 24.4 6 Morphine 3.2 mg/kg ip 943.6 31.9 9

Table 1A shows the rotarod run times of normal rats injected with saline or flupirtine and morphine given alone at a range of doses as well as in four dose combinations. Data for each treatment group are shown are means and SEM. Those items shown in bold and marked with * denote those drug treatments that caused sedation as assessed by this test compared with saline-treated controls (P < 0.001, one-way analysis of variance [ANOVA] with Dunnett post hoc correction). Table 1B shows the rest times of diabetic rats in the open-field activity monitor. None of the drug treatments caused an increase in rest times compared with saline-treated controls (P > 0.05, one-way ANOVA with Dunnett post hoc correction). ip = intraperitoneally; SEM = standard error of the mean. of the ip drug injection reaches a plateau from post hoc correction for multiple comparisons 140 to 160 minutes. In order to assess statistically revealed the following: the differences between the antinociception caused by the different treatments, the withdrawal • There were no significant differences in pre- latencies were averaged in each individual rat for treatment values in any of the groups. testing times -20, -10, and 0 (pretreatment) and • Flupirtine 5 and 10 mg/kg or morphine 0.4 and also for 140-, 150-, and 160-minute readings 0.8 mg/kg alone had no effect on carrageenan- (postdrug). The pretreatment and postdrug values induced hyperalgesia (postdrug values com- (means and SEM for each treatment group) are pared with saline controls and with their shown in the Table 2. One-way anova with Tukey replicate pretreatment values).

Table 2 Results of experiments with the carrageenan paw inﬂammation paradigm

Predrug Postdrug

Summary of Results with Carrageenan Paw Inﬂammation Mean SEM N Mean SEM

Saline controls 11.0 0.4 29 6.0 0.3 Flupirtine 5 mg/kg ip alone 10.9 0.9 10 5.8 0.5 Flupirtine 10 mg/kg ip alone 11.0 0.9 8 5.5 0.6 Morphine 0.4 mg/kg ip alone 12.1 0.7 12 5.8 0.8 Morphine 0.8 mg/kg ip alone 9.9 0.4 12 4.8 0.4 Morphine 1.6 mg/kg ip alone 10.3 0.9 8 8.9* 1.0 Flupirtine 5 mg/kg and morphine 0.4 mg/kg ip together 11.6 0.4 26 8.8* 0.6 Flupirtine 10 mg/kg and morphine 0.4 mg/kg ip together 9.7 0.3 18 10.3*† 0.9

Paw withdrawal latencies (seconds) are shown as means and SEM for each treatment group, prior to the ip injection of saline or drug (pretreatment) and 20–40 minutes after drug or saline injection (postdrug). The results of the statistical analysis (one-way analysis of variance with Tukey–Kramer post hoc test) are • Flupirtine 5 and 10 mg/kg or morphine 0.4 and 0.8 mg/kg alone had no effect on carrageenan-induced hyperalgesia (P > 0.05). • The combination of flupirtine 5 mg/kg and also 10 mg/kg with morphine 0.4 mg/kg caused significant reversal of carrageenan-induced hyperalgesia, and this was equal to the effect of 1.6 mg/kg morphine given alone (* P < 0.001). • Flupirtine 10 mg/kg in combination with morphine 0.4 mg/kg led to a greater effect compared with either drug alone at those doses (* P < 0.001). • Complete reversal of carrageenan-induced hyperalgesia was caused by 10 mg/kg flupirtine in combination with 0.4 mg/kg morphine. These doses of the two drugs were ineffective when given alone but when given together, they caused complete antinociception in this model of pain († P > 0.05). ip = intraperitoneally; SEM = standard error of the mean. Flupirtine–Morphine Combination Therapy: Preclinical Studies 935

Table 3 Results of experiments with the streptozotocin (STZ)-induced diabetic neuropathy model

Pretreatment Postdrug Summary of Results with STZ-Induced Diabetic Neuropathy Mean SEMN Mean SEM Saline controls 28.54 0.93 16 30.94 0.89 Flupirtine 5 mg/kg ip alone 28.25 0.97 21 31.9 1.16 Flupirtine 10 mg/kg ip alone 26.9 1.13 21 38.89* 2.8 Morphine 1.6 mg/kg ip alone 28.1 1.56 14 31.9 1.38 Morphine 3.2 mg/kg ip alone 26.67 1.72 8 35.0 3.09 Flupirtine 5 mg/kg and morphine 3.2 mg/kg ip together 26.67 1.35 8 36.88 4.2 Flupirtine 10 mg/kg and morphine 1.6 mg/kg ip together 28.82 1.24 17 49.41*† 3.25

Weight-matched nondiabetic controls 44.71 0.83 63 Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021

Paw withdrawal thresholds (grams weight) are shown as means and SEM for normal weight-matched controls and for each treatment group, prior to the ip injection of saline or drug (pretreatment) and 20–40 minutes after drug or saline injection (postdrug). The results of the statistical analysis (one-way analysis of variance with Tukey–Kramer post hoc test) are • When compared with weight-matched controls, all drug treatment groups were hyperalgesic prior to drug treatment, and the extent of this hyperalgesia was the same in all groups (P < 0.001). • Highly significant antinociception (* P < 0.001; comparison of replicate pretreatment and postdrug values) occurred with complete reversal of streptozotocin- induced diabetic hyperalgesia caused by flupirtine 10 mg/kg given alone and also flupirtine 10 mg/kg + morphine 1.6 mg/kg together († P > 0.05; comparison of postdrug values with weight-matched diabetic controls); i.e., the paw withdrawal thresholds after the drug treatment were not statistically different from thresholds for normal nondiabetic weight-matched controls. • No significant antinociception occurred after treatment with saline, flupirtine 5 mg/kg, morphine 1.6 or 3.2 mg/kg, or with the combination of flupirtine 5 mg/kg with morphine 3.2 mg/kg (P > 0.05; comparison of replicate pretreatment and postdrug values). • Finally, flupirtine 10 mg/kg in combination with morphine 1.6 mg/kg caused greater antinociception than flupirtine 10 mg/kg alone (P < 0.05). ip = intraperitoneally; SEM = standard error of the mean.

• The combination of flupirtine 5 mg/kg and also drug and saline treatment groups of diabetic rats, flupirtine 10 mg/kg with morphine 0.4 mg/kg and these values were significantly below those for both caused significant reversal of carrageenan- normal weight-matched controls, thus confirming induced hyperalgesia (P < 0.001), and this was that the diabetes caused hyperalgesia. equal to the effect of 1.6 mg/kg morphine given It can also be seen that the responses to drugs, if alone (P > 0.05); that means that flupirtine present, were apparent at 20 minutes after the increased the antinociceptive effect of morphine injection of the drug or the drug combination, and fourfold. the response to any particular drug or drug com- • Flupirtine 5 mg/kg in combination with mor- bination was constant and stable between 20 and phine 0.4 mg/kg led to significant reversal of 40 minutes after the injection. In order to assess carrageenan-induced hyperalgesia (P < 0.001, statistically the differences between the antinoci- compared with saline), and this was greater than ception caused by the different treatments, the either drug given alone (P < 0.001). withdrawal thresholds were averaged in each indi- • Finally, complete reversal of carrageenan- vidual rat for testing times -20, -10, and 0 (pre- induced hyperalgesia was caused by 10 mg/kg treatment) and also for 20-, 30-, and 40-minute flupirtine in combination with 0.4 mg/kg mor- readings (postdrug). Thus, each rat experiment phine (P > 0.05; comparison of replicate pre- contributed one pretreatment value and one post- treatment and postdrug values for that group drug value to the statistical calculations. The and also of all other pretreatment values). These pretreatment and postdrug values (combined to doses of the two drugs that were ineffective calculate means and SEM for each treatment when given alone caused complete antinocicep- group) are shown in Table 3. A one-way anova was tion in this model of inflammatory pain when applied to the values in this table to compare the given together (P > 0.05). postdrug values with the values for paw withdrawal thresholds in weight-matched nondiabetic rats; a None of these doses or combinations of drugs return of no significant difference indicates that caused sedation in either the rotarod or the open- the drug or the drug combination had reversed field activity tests. completely the diabetes-induced hyperalgesia. In STZ-Induced Diabetic Neuropathy addition, a one-way anova (with Tukey–Kramer The time–response curves for each mode of treat- post hoc test for multiple comparisons) was ment in rats with hyperalgesia caused by diabetic applied to the data in Table 3 to assess whether any neuropathy are shown in Figure 2. It can be seen of the drug treatments led to any antinociception; that the values of the paw withdrawal thresholds i.e., was there a significant increase in paw with- measured at -20, -10, and 0 were the same for all drawal thresholds after the drug treatment com- 936 Goodchild et al. pared with the paw withdrawal thresholds before because of dose-related side effects limiting dose the treatment and also saline controls? Finally, this escalation toward more effective larger doses. statistical analysis allowed comparison of the mag- For example, the treatment of neuropathic pain nitude of the different drug treatment effects. states, including painful diabetic neuropathy in This analysis reported the following: humans, is frequently unsatisfactory. Current pharmacological regimens consist of tricyclic anti- • There was no statistical difference between any depressants [17–19], other antidepressants such as of the pretreatment paw withdrawal thresholds duloxetine, anticonvulsants, systemic local anaes- for saline or drug treatment groups, and these thetics (lignocaine) and mexiletine, and, more were all significantly lower than the thresholds

recently, GABApentin and pregabalin. All have Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 for weight-matched nondiabetic controls limited success [20–23]. (P < 0.001); STZ-induced diabetes caused sig- It is accepted generally that human neuropathic nificant hyperalgesia, which was equal for all pain states are resistant to opioid treatment [20]. drug treatment groups and saline controls. The analgesic response in patients with neuro- • No significant antinociception occurred after pathic pain treated with opioids is typically limited treatment with saline, flupirtine 5 mg/kg, mor- to a maximum of 30–50% by dose-limiting side phine 1.6 or 3.2 mg/kg, or with the combination effects [24]. Some researchers have found that of flupirtine 5 mg/kg with morphine 3.2 mg/kg opioids may produce antinociceptive effects in (P > 0.05; comparison of replicate pretreatment neuropathic pain models in animals but at higher and postdrug values). than normal doses that also cause sedation • Flupirtine 10 mg/kg given alone and also flupir- revealed by tests such as open-field activity moni- tine 10 mg/kg + morphine 1.6 mg/kg together toring and the rotarod test. This indicates a shift of caused highly significant antinociception the dose–response curve to the right, beyond the (P < 0.001; comparison of replicate pretreat- normal therapeutic range [25]. Thus, it would ment and postdrug values) that occurred with seem to date that there is no really efficacious complete reversal of streptozotocin-induced treatment for pain states such as neuropathic pain diabetic hyperalgesia (P > 0.05; comparison of without the administration of doses of drugs that postdrug values with weight-matched diabetic cause side effects. Faced with this problem of uni- controls); i.e., the paw withdrawal thresholds modal therapy, most clinicians have adopted a after the drug treatment were not statistically multimodal approach, including the use of drugs different from thresholds for normal nondia- of different classes in an attempt to increase anal- betic weight-matched controls. gesia without causing side effects [26–29]. This is • Finally, flupirtine 10 mg/kg in combination the approach used in the experiments described with morphine 1.6 mg/kg caused greater anti- herein. nociception than flupirtine 10 mg/kg alone Although flupirtine is an analgesic drug that has (P < 0.01). been available in some parts of the world for over 20 years [7], it has not been suggested that it should be used as an adjunctive treatment or as Conclusions part of a multimodal treatment regimen for neu- Tworat pain models have been used here to inves- ropathic pain states. Clinical trials have been con- tigate the antinociceptive properties of flupirtine fined to straight comparisons of unimodal and morphine given alone and in combinations. therapies. In such studies, flupirtine is reported to Sedating doses and dose combinations were cause equivalent or slightly superior analgesia excluded to be sure that the observed changes in compared with paracetamol, or opioids such as nociception thresholds were due to antinocicep- codeine, dihydrocodeine, or pentazocine [7]. Side tive rather than sedation or effects on attention in effects such as nausea, dizziness, and somnolence these behavioral models. The results demonstrate are common but are reported to be minor. clearly that the antinociception caused by using However, the new evidence concerning the action the opioid and KCNQ channel opener together of flupirtine as a potassium channel opener and causes more antinociception than could be thus as an activator of the M current [5,11] leads us achieved with either drug given alone. This result to reconsider the position of this drug. Its side- is potentially very useful clinically because many effect profile limits dose escalation in pain states types of pain cannot be treated to an effective caused by CNS sensitization. However, control of degree with single agents such as morphine the M current opens the possibility of useful drug Flupirtine–Morphine Combination Therapy: Preclinical Studies 937 interactions. Animal experiments using modern References paradigms mimicking human persistent pain 1 Yang WP, Levesque PC, Little WA, et al. Func- states, such as the carrageenan and STZ diabetic tional expression of two KvLQT1-related po- neuropathy models, have not been used to evaluate tassium channels responsible for an inherited the antinociceptive properties of flupirtine. idiopathic epilepsy. J Biol Chem 1998;273(31): However, there is one isolated report of antinoci- 19419–23. ception in the carrageenan model following reti- 2 Brown DA, Adams PR. Muscarinic suppression of a gabine, a structurally related analog of flupirtine novel voltage-sensitive K+ current in a vertebrate neurone. Nature 1980;283(5748):673–6. and also a KCNQ channel opener [30]. 3 Marrion NV. Control of M-current. Annu Rev We can conclude from the results of the experi- Physiol 1997;59:483–504. Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 ments reported in this article that nonsedating 4 Gribkoff VK. The therapeutic potential of neuronal doses of flupirtine can increase the overall anti- KCNQ channel modulators. Expert Opin Ther nociceptive effect of morphine without causing Targets 2003;7(6):737–48. sedation in two animal models of pain: inflamma- 5 Martire M, Castaldo P, D’Amico M, et al. M chan- tory and neuropathic. It is important in such nels containing KCNQ2 subunits modulate nore- studies to confine the doses used in the nocicep- pinephrine, aspartate, and GABA release from tion paradigms to those that do not cause gener- hippocampal nerve terminals. J Neurosci 2004;24(3): alized depression in the CNS. There are two 592–7. reasons for this. The first is that one cannot be 6 Bleyer H, Carlsson KH, Erkel HJ, Jurna I. Flupir- tine depresses nociceptive activity evoked in rat sure that a decreased response is an antinocicep- thalamus. Eur J Pharmacol 1988;151:259–65. tive effect of the drug if all responses in the subject 7 Friedel HA, Fitton A. Flupirtine. A review of its are diminished. The second reason is one of cli- pharmacological properties, and therapeutic efficacy nical applicability, the aim being to develop a in pain states. Drugs 1993;45(4):548–69. therapy that enhances antinociception but without 8 Schwarz M, Block F, Pergande G. N-methyl-D- causing sedation. The results shown here indicate aspartate (NMDA)-mediated muscle relaxant action that it is possible in both neuropathic and in- of flupirtine in rats. Neuroreport 1994;5:1981–4. flammatory pain models to cause such significant 9 Schwarz M, Schmitt T, Pergande G, Block F. antinociception with flupirtine administered in N-methyl-D-aspartate and a2-adrenergic mecha- combination with morphine as to reverse hyperal- nisms are involved in the depressant action of flu- gesia, such that animals with these pain states are pirtine on spinal reflexes in rats. Eur J Pharmacol 1995;276:247–55. rendered normal with respect to pain sensitivity. 10 Osborne NN, Pergande G, Block F, Schwarz M. This strongly suggests that flupirtine may be used Immunohistochemical evidence for flupirtine acting effectively in the clinic as an adjunct to opioid as an antagonist on the N-methyl-D-aspartate and analgesics especially in pain states such as inflam- homocysteic acid- induced release of GABA in the matory and neuropathic pain. This is important rabbit retina. Brain Res 1994;667:291–4. given that these pain states are reported to be 11 Kornhuber J, Bleich S, Wiltfang J, Maler M, opioid resistant to the extent that only partial anal- Parsons CG. Flupirtine shows functional NMDA gesia can be achieved with opioid drugs unless one receptor antagonism by enhancing Mg2+ block administers excessive doses that cause side effects via activation of voltage independent potassium such as sedation [25,31]. The coadministration of channels. Rapid communication. J Neural Transm flupirtine with opioid offers improved pain control 1999;106(9,10):857–67. 12 Zimmermann M. Ethical guidelines for investiga- in inflammatory and neuropathic pain with doses tions of experimental pain in conscious animals. and combinations that are not accompanied by Pain 1983;16(2):109–10. sedation. We conclude that this multimodal 13 Cartmell SM, Gelgor L, Mitchell D. A revised approach of using flupirtine, a potassium channel rotarod procedure for measuring the effect of anti- opener, in combination with an opioid such as nociceptive drugs on motor function in the rat. J morphine, is worthy of clinical evaluation. Pharmacol Methods 1991;26:149–59. 14 Courteix C, Eschalier A, Lavarenne J. Streptozocin- Financial Disclosure Statement induced diabetic rats: Behavioural evidence for a model of chronic pain. Pain 1993;53:81–8. Professor Goodchild and Dr. Ian Cooke are offic- 15 Courteix C, Bardin M, Chantelauze C, Lavarenne J, ers of CNSBio Pty Ltd, which holds rights to Eschalier A. Study of the sensitivity of the diabetes- commercialize new intellectual property arising induced pain model in rats to a range of analgesics. from the research described in this article. Pain 1994;57(2):153–60. 938 Goodchild et al.

16 Randall LO, Selitto JJ. A method for measurement relieves neuropathic pain: A randomized controlled of analgesic activity of inﬂamed tissue. Archiv Int trial in painful diabetic neuropathy. Pain 2003; Pharmacdynamie 1957;111:409. 105(1–2):71–8. 17 Max MB. Towards physiologically based treatment 25 Portenoy RK, Foley KM, Inturrisi CE. The nature of patients with neuropathic pain. Pain 1990;42: of opioid responsiveness and its implications for 131–3. neuropathic pain: New hypotheses derived from 18 Max MB. U.S. government disseminates acute pain studies of opioid infusions [see comments]. Pain treatment guidelines: Will they make a difference? 1990;43(3):273–86. Pain 1992;50:3–4. 26 Kumbhat S, Meyer N, Schurr MJ. Complex 19 Sindrup SH, Gram LF, Brosen K, Eshoj O, regional pain syndrome as a complication of a

Mogensen EF. The selective serotonin reuptake chemical burn to the foot. J Burn Care Rehabil Downloaded from https://academic.oup.com/painmedicine/article/9/7/928/1861898 by guest on 01 October 2021 inhibitor paroxetine is effective in the treatment of 2004;25(2):189–91. diabetic neuropathy symptoms. Pain 1990;42(2): 27 Pavelka K. Symptomatic treatment of osteoar- 135–44. thritis : paracetamol or NSAIDs? Int J Clin Pract 20 Arner S, Meyerson BA. Lack of analgesic effect of Suppl 2004;144:5–12. opioids on neuropathic and idiopathic forms of pain 28 Portenoy RK. Adjuvant analgesic agents. Hematol [see comments]. Pain 1988;33(1):11–23. Oncol Clin North Am 1996;10:103–19. 21 Avidan A, Devor M. 1995 intravenous lidocaine 29 Shang AB, Gan TJ. Optimising postoperative pain infusion for chronic pain therapy. Isr J Med Sci management in the ambulatory patient. Drugs 1996;32:331–4. 2007;63(9):855–67. 22 Davis BL, Manzanares J, Lookingland KJ, Moore 30 Passmore GM, Selyanko AA, Mistry M, et al. KE, Clemens LG. Noradrenergic innervation to the KCNQ/M currents in sensory neurons: Signiﬁ- VMN or MPN is not necessary for lordosis. Phar- cance for pain therapy. J Neurosci 2003;23(18): macol Biochem Behav 1991;39:737–42. 7227–36. 23 Galer BS. Neuropathic pain of peripheral origin: 31 Portenoy RK, Foley KM, Inturrisi CE. Genuine Advances in pharmacologic treatment. Neurology resistance to opioids–fact or ﬁction? Response. Pain 1995;45(suppl 9):S17–25. 1991;47:119–21. 24 Watson CP, Moulin D, Watt-Watson J, Gordon A, Eisenhoffer J. Controlled-release oxycodone