Analgesic Interaction between Intrathecal Midazolam and Glutamate Receptor Antagonists on Thermal-induced Pain in Rats 

The protocol was approved by the Research and Education Institute of Harbor-UCLA Medical Center. Sprague-Dawley rats (280–300 g; B. K, Universal, Fremont, CA) were implanted with chronic lumbar intrathecal catheters under halothane (2%) anesthesia according to the method described by Yaksh and Rudy. [13] Briefly, an 8.5-cm polyethylene catheter (PE-10; Clay Adams, Parsippany, NJ) was advanced caudally through an incision in the atlantooccipital membrane, to the thoracolumbar level of the spinal cord. The external part of the catheter was tunneled subcutaneously to exit on the top of the skull and plugged with a 28-gauge stainless steel wire. Only rats with normal motor function and behavior 5 days after surgery were used.

Midazolam (a benzodiazepine-GABA^Areceptor agonist; Sigma, St. Louis, MO) 1, 3, 10, 30, and 100 [micro sign]g, and AP-5 (2-amino-5-phosphonovaleic acid, an NMDA receptor antagonist; Sigma) 1, 3, 10, and 30 [micro sign]g were dissolved in saline 10 [micro sign]l. YM872 ([2,3-Dioxo-7-(1H-imidazol-1-yl)-6-nitro-1,2,3,4-tetrahydro-1-quinoxalinyl] acetic acid), an AMPA receptor antagonist (Yamanouchi Pharmaceutical, Tsukuba, Japan) 10 mg was dissolved in 0.97 ml distilled water with 30 [micro sign]l 1 N NaOH to adjust pH to 7.3–7.5. Solutions of 0.3 (0.86), 1 (2.86), 3 (8.59), 10 (28.63), and 30 (85.89)[micro sign]g (nM) per 10 [micro sign]l were made using normal saline. After intrathecal drug injection, the catheter was flushed with a subsequent injection of 10 [micro sign]l of normal saline to clear the dead space of the catheter (7 +/- 0.4 [micro sign]l, mean +/- SE). Microinjector syringes were used for all injections. In each dose group, eight randomly selected rats were used. Normal saline 10 [micro sign]l was injected in the control group.

Each rat was placed in a clear plastic cylindrical cage with its tail extended through a slot provided in the rear of the tube. Noxious stimulation was provided by a beam of high-intensity light (Tail-flick Analgesia Meter 0570–001L, Columbus Instruments International, Columbus, OH) focused on the tail 2–3 cm proximal to the end. The response time was measured and defined as the interval between the onset of the thermal stimulation and the abrupt flick of the tail. The cutoff time in the absence of a response was set to 14 s to prevent tissue injury.

The general behavior (including agitation and allodynia), motor function, pinna reflex, and corneal reflex were examined. Their presence or absence was recorded. Agitation was judged to be present when the rat spontaneously vocalized or became restless. The presence of allodynia was examined by looking for agitation (escape or vocalization) evoked by lightly stroking the flank with a pencil. The stimulus was sufficient to move hair but not dent the skin. Motor function was evaluated by the placing/stepping reflex and by the righting reflex. The former was evoked by drawing the dorsum of either hind paw across the edge of the table. The latter was assessed by placing the rat horizontally with its back on the table, which normally gives rise to an immediate, coordinated turning of the body back to an upright position. Flaccidity was judged as a muscle weakness. Pinna and corneal reflexes were examined with a paper string.

The first series of experiments was performed to determine the dose dependency and time course of the analgesic actions of intrathecally administered midazolam, AP-5, and YM872 on acute thermal nociception. The tail flick test, behavioral test, and motor function test were performed before and 5, 10, 15, 30, 60, 90, 120 min after drug injection and at 1-h intervals until the response time returned to baseline (maximum 360 min).

To investigate the interaction between midazolam and AP-5 or YM872, and isobolographic analysis was used. [14] The method is based on comparisons of dose ratios that are determined to be equieffective. First, the respective 50% effective dose (ED⁵⁰) values are determined from the dose-response curves of the agents alone. Subsequently, a dose-response curve is obtained by coadministration of the two drugs in a constant dose ratio based on the ED⁵⁰values of the single agents. From the dose-response curve of the combined drugs, the ED⁵⁰value of the mixture was calculated.

Data were expressed as mean +/- standard error (SEM). Tail flick response latency was converted to percentage maximum possible effect (%MPE) according to the formula:%MPE =[(postdrug latency - baseline latency)/(cutoff time - baseline latency)] x 100. ED⁵⁰was calculated by a computer program, which was created in the laboratory of University of California, San Diego, according to Tallarida and Murray, [15], as the dose that produces a value of 50% MPE.

To describe the magnitude of interaction between the agents, a total fractional dose value was calculated as follows:[(ED⁵⁰dose of drug 1 in combination)/(ED⁵⁰value for drug 1 alone)]+[(ED⁵⁰dose of drug 2 in combination)/(ED⁵⁰value for drug 2 alone)]. The values were normalized by assigning the ED⁵⁰values of the agents given alone a value of 1. Values near 1 indicate an additive interaction, values greater than 1 imply an antagonistic interaction; values less than 1 indicate a synergistic interaction. To compare the theoretic additive point with experimentally derived ED⁵⁰, isobolographic analysis [14] was used.

Differences between doses were analyzed with two-way analysis of variance followed by the Newman-Keuls test. Student t test was used to compare the calculated ED50 values with the theoretic additive values. A P value less than 0.05 was considered statistically significant.

The baseline latency (before drug injection) in the tail flick test was 3.0 +/- 0.2 s (mean +/- SE). Intrathecal administration of midazolam, AP-5, and YM872 resulted in dose-dependent increases in the tail flick latency (Figure 1). The ED⁵⁰values were 1.57 +/- 0.34 [micro sign]g, 5.54 +/- 0.19 [micro sign]g, and 1.0 +/- 0.22 [micro sign]g with midazolam, AP-5, and YM872, respectively.

Coadministration of midazolam and AP-5 intrathecally shifted a dose response curve to the left (Figure 2) and showed a significant increase in the thermal escape latency compared with the agents alone by an isobolographic analysis (Figure 3). The experimentally obtained ED⁵⁰of the combination of midazolam and AP-5 was midazolam 0.21 +/- 0.18 [micro sign]g and AP-5 0.75 +/- 0.18 [micro sign]g. These doses were significantly lower than the theoretic additive doses (midazolam 0.79 +/- 0.38 [micro sign]g and AP-5 2.77 +/- 0.24 [micro sign]g). The total fractional dose value of the combination was calculated to be 0.27 +/- 0.11, which indicates a synergistic interaction.

Coadministration of midazolam and YM872 intrathecally shifted a dose-response curve to the left (Figure 2) and showed a significant increase in the thermal escape latency compared to the agents alone (Figure 4). The experimentally obtained ED⁵⁰of the combination of midazolam and YM872 was midazolam 0.15 +/- 0.09 [micro sign]g and YM872 0.24 +/- 0.09 [micro sign]g. These doses were significantly lower than the theoretic additive doses (midazolam 0.79 +/- 0.38 [micro sign]g and YM872 0.5 +/- 0.22 [micro sign]g). The total fractional dose value of the combination was calculated to be 0.34 +/- 0.12, which indicates a synergistic interaction.

Midazolam >or= to 3 [micro sign]g (each one rat in 3, 10, 30, and 100 [micro sign]g) and AP-5 >or= to 10 [micro sign]g (one in 10 [micro sign]g and two in 30 [micro sign]g) induced agitation and allodynia. Motor disturbances (tested by the placing/stepping reflex and by the righting reflex) occurred with midazolam >or= to 30 [micro sign]g (two rats in 30 [micro sign]g and six in 100 [micro sign]g), AP-5 >or= to 10 [micro sign]g (two in 10 [micro sign]g and three in 30 [micro sign]g) or YM872 >or= to 10 [micro sign]g (three in 10 [micro sign]g and four in 30 [micro sign]g). Flaccidity was seen in the rats with midazolam >or= to 30 [micro sign]g (one in 30 [micro sign]g and two in 100 [micro sign]g), or YM872 >or= to 10 [micro sign]g (two in 10 [micro sign]g and six in 30 [micro sign]g). AP-5 30 [micro sign]g induced loss of pinna reflex (two rats). In contrast, combination of midazolam and AP-5 induced no observable side effects. Allodynia was seen with midazolam 0.2 [micro sign]g plus YM872 0.125 [micro sign]g (one rat), and loss of righting reflex occurred with midazolam 0.8 [micro sign]g plus YM872 0.5 [micro sign]g (one rat). The combinations of midazolam and AP-5 or YM872 displayed fewer side effects than the equieffective doses of the individual agents (Table 1). No rats showed paralysis in this study.

We found that intrathecally administered midazolam (benzodiazepine-GABA^Areceptor agonist), AP-5 (NMDA receptor antagonist) and YM872 (AMPA receptor antagonist) produced dose- dependent increases in tail flick latency. Midazolam showed synergistic analgesic effects with both AP-5 and YM872.

In the dorsal horn of the spinal cord, GABA receptors mediate presynaptic inhibition on the primary afferent terminals. [16] At these endings, GABA produces a mild depolarization of the primary afferents and thereby reduces the release of the excitatory transmitter onto the second-order neurons in the spinal cord. [17] Binding sites for benzodiazepine are in lamina II of the dorsal horn. [18] Radioligand binding assays and electrophysiologic studies showed the linkage of the benzodiazepine sites to the GABA^Areceptor complex in the spinal cord. [19,20] Enhancement of presynaptic inhibition might be a possible mechanism for the action of midazolam, because benzodiazepines are known to increase GABA transmission via their specific binding site colocated with the GABA^Areceptor. [21] Benzodiazepines were reported not to block the transmission of sensory impulses through nerve fibers. [22]

Yanez et al. [23] reported that intrathecally administered midazolam 20–60 [micro sign]g produced dose-dependent antinociception on thermally induced pain and larger doses (60–100 [micro sign]) induced motor dysfunction. These published data are similar to those of our current study, in which midazolam 1–100 [micro sign]g induced dose-dependent analgesia and doses higher than 30 [micro sign]g induced motor dysfunction. In the study of Bahar et al. [6], 75 [micro sign]g intrathecal midazolam induced sleep. However, 100 [micro sign]g did not induce sleep in our study. We did not apply higher doses of midazolam because of the limitation of its solubility in saline and because the 30 and 100 [micro sign]g doses already induced motor dysfunction.

NMDA receptors are involved in the wind-up phenomena of deep dorsal-horn cells evoked by C-fiber activation. [24] NMDA receptor antagonists are therefore the most efficacious against the continuously stimulated state of nociception, induced for example by formalin. [8] Generally, they are inefficacious on acute nociception, [11] although some studies [25,26] have shown analgesic effects on acute thermal stimuli. In the present study, AP-5 (NMDA receptor antagonist) produced dose-dependent analgesic effects on acute thermal stimulus, although the ED⁵⁰value was relatively high. In a previous study, AP-5 had only weak analgesic effects at the maximum usable dose in the hot-plate test. [27] Considering these results together, NMDA antagonists might have some analgesic effects on acute nociception depending on the experimental settings.

AMPA receptors are found throughout all superficial laminae of the dorsal horn pre- and postsynaptically. [28,29] These receptors are thought to mediate the acute excitation from primary afferent fibers to dorsal horn neurons evoked by high intensity stimuli. Intrathecal application of AMPA receptor antagonists produces dose-dependent antinociception on acute pain in animal models. [27,30] The results of the present study are consistent with these previous studies. [22,30]

No single agent of these classes (benzodiazepines, NMDA, or AMPA receptor antagonists) administered alone is effective enough to block nociception without any adverse effects. One reason is that pain is not mediated by a single receptor or a single neurotransmitter. The other is that the receptors and neurotransmitters mediating pain are also connected to other neuronal networks in the central nervous system that may induce adverse effects. Thus, combination of agents acting through different mechanisms may be one of the best ways to arrive at better analgesic methods.

The present study showed a significant synergistic antinociception between midazolam, a benzodiazepine-GABA^Areceptor agonist, and AP-5, an NMDA receptor antagonist, or YM872, a new AMPA receptor antagonist, on acute thermal stimulation. We used only the tail flick test. To confirm the results of the present study, further investigation using other methods is necessary. Aanonsen et al. [12] reported that GABA^Areceptor agonist inhibited behavioral effects of NMDA, quisqualic acid, and kainic acid. Only in the presence of NMDA, did GABA^Areceptor agonist have antinociceptive effect in the tail flick test. GABA produces a mild depolarization of the primary afferents and thereby reduces the release of the excitatory transmitter onto the second-order neurons in the spinal cord. [17] The GABA^Areceptor might have some functional coupling with glutamate receptor.

With regard to the side effects, no paralysis was seen in this study. Therefore, we considered tail flick latency not to be affected by motor dysfunction. Midazolam plus AP-5 and midazolam plus YM872 decreased behavioral changes and motor dysfunction and enhanced analgesic effects. These combinations could enhance the therapeutic efficacy of the acute pain treatment and safety. However, one of the important concerns in applying the results to clinical pain management is toxicity of the agents. There are still some controversies surrounding the neurotoxicity of intrathecal midazolam. [31,32] Current formulations of NMDA receptor antagonists are also neurotoxic. [33] AMPA receptor antagonists have poor water solubility and nephrotoxicity. [34] YM872 is a new AMPA receptor antagonist, which is much more water-soluble than the other formulations of AMPA receptor antagonists. [35] YM872 had no neurotoxicity in cat, [36] rat and monkey brains in toxicologic studies (unpublished data). However, there are no studies investigating the toxicity of YM872 on the spinal cord. Therefore, further studies of their toxicity and of new compounds should be performed before applying the results to humans.

In conclusion, intrathecal coadministration of midazolam (a benzodiazepine-GABA^Areceptor agonist) with AP-5 (an NMDA receptor antagonist) or midazolam with YM872 (an AMPA receptor antagonist) produced significant synergistic analgesia with decreased side effects on acute thermal nociception measured by tail flick test. These results suggest a functional coupling of benzodiazepine-GABA^Areceptors with NMDA and AMPA receptors in acute nociception in the spinal cord.

The authors thank Dr. Ang Ji, Dr. Young-moon Cho, and Nguyen B. Nguyen for their assistance.