| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Departments of *Experimental Therapeutics,
Neurosurgery, and
Veterinary Sciences and
||Division of Pharmacy, The University of Texas M. D. Anderson Cancer Center, Houston, Texas; and
Medtronic, Inc., Minneapolis, Minnesota
Address correspondence and reprint requests to Mary J. Johansen, PharmD, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Unit 601, Houston, TX 77054. Address e-mail to mjohanse{at}mdanderson.org
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
IMPLICATIONS: We investigated the toxicity of preservative-free intrathecal midazolam delivered continuously via implanted infusion systems in sheep and pigs. Doses of 5–15 mg/d were well tolerated. The lack of neurotoxicity observed suggests that intrathecal midazolam may be an alternative for the treatment of intractable pain that is unresponsive to opioids.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Evidence also suggests that some pain states, such as neuropathic pain, can be relatively resistant to opioids (4). Approximately 30% of cancer-related pain is neuropathic in origin, secondary to tumor involvement in nervous tissue or treatment-related tissue damage. Toxicity and the development of tolerance also remain a concern with the chronic use of intrathecal opioids. Aside from the more common manageable side effects, long-term intrathecal morphine infusion has been associated with the development of dose-related inflammatory lesions along the catheter tract (5).
Alternative drugs for intrathecal administration that act at spinal adrenergic or 
-aminobutyric acid (GABA) receptors are now under investigation. GABA receptors play a role in antinociception and the mechanisms that cause allodynia and hyperesthesia (6). Intrathecal delivery of the benzodiazepine GABAA receptor agonist midazolam has produced significant relief of chronic lower back pain and spasticity (7–9), and its antinociceptive properties have been well described in animal models (10–12). Combined intrathecal therapy with midazolam and morphine has also produced significant analgesia without side effects in cancer patients resistant to large oral doses of morphine (up to 400 mg/d) (9).
Numerous clinical reports (7–9,13) and animal studies (11,14) have demonstrated the safety of intrathecal midazolam administration; sedation and somnolence are the most commonly reported side effects. Motor dysfunction has occurred only at larger doses in animals (15). More recent studies have reported neurotoxicity in rats (16) and rabbits (17,18) after bolus administration. These studies have focused attention on the toxic potential of intrathecally administered midazolam, but they raise several questions regarding the validity of results obtained in smaller animal models which may not be directly comparable to use in humans. Larger drug concentrations achieved by bolus administration into the smaller intrathecal space in the rat and rabbit models and the possibility of trauma induced by direct injection remain significant concerns. In addition, the nature of the midazolam preparations used with regard to diluent and the presence of preservative is not always reported clearly. To more accurately evaluate the neurotoxic potential of intrathecal midazolam, we performed a toxicity study of preservative-free midazolam administered by continuous intrathecal infusion in the larger, more relevant sheep model. Additional studies were performed in pigs to further evaluate possible differences in tolerance between species.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Sheep (juvenile Rambouillet cross, 47–82 kg; n = 15) were instrumented with intrathecal catheters and infusion pumps (SynchromedTM programmable pumps; Medtronic, Inc., Minneapolis, MN) under aseptic conditions. Cefazolin 1 g IV and glycopyrrolate 0.4 mg IV were administered before anesthetic induction. Anesthesia was induced with diazepam 0.2 mg/kg and ketamine 6.0 mg/kg by IV bolus, and animals were intubated with a cuffed Murphy endotracheal tube (8- to 10-mm inner diameter). Anesthesia was maintained with halothane or isoflurane at an inspired concentration of 2%–3% in oxygen (Ohio ventilator). Distention of the rumen and attendant ventilatory depression were avoided by oral rumen cannulation with a large-bore stomach tube. Body temperature was maintained at 100°F–104°F by using a circulating water pad. IV fluids (0.9% NaCl) were administered throughout the procedure, and vital signs were monitored with an electrocardiogram temperature respirator monitor (Vet/Ox Plus).
After preparation of a sterile surgical field, a midline incision was made at the level of the L6 to S2 vertebrae, and the muscle fasciae were exposed. A 16-guage Tuohy needle was inserted into the intravertebral space at L7-S1. The needle was slowly advanced until the dura was punctured, as evidenced by the free flow of cerebrospinal fluid (CSF) from the needle hub. A silicone elastomer intraspinal catheter (4F [18 gauge]; inner diameter, 0.6 mm; outer diameter, 1.2 mm) was threaded into the Tuohy needle and advanced 10 cm cephalad to the subarachnoid space (L5). The catheter was secured to the muscle fascia with 2-0 silk sutures.
A pocket was then fashioned in the left paralumbar fossa. The pump was inserted and anchored to the muscle at 3 locations in 90°–120° intervals with 2-0 silk or 0 Braunamid suture. The catheter was then tunneled to the pocket and connected to the pump. The pump was filled with sterile saline and programmed via a telemetry wand to deliver 1 mL/d.
The incisions were flushed with saline/gentamicin and a local anesthetic. Wounds were closed in layers with Vicryl suture. Analgesics (butorphanol tartrate 5 mg IM or morphine up to 10 mg per dose) were administered before anesthetic emergence, again in the evening and the following morning, and as needed thereafter. Postoperative antibiotics consisted of cefazolin 1 g IM twice daily for 2 days, followed by penicillin A benzathine/penicillin G procaine 5 mL subcutaneously once daily for an additional 3 days. Each animal was observed for evidence of neurological deficit for 7 days postsurgery.
One sheep was initially administered midazolam 3 mg/d by continuous infusion for 43 days and was evaluated off-study for toxicity. After no observations of toxicity, experimental sheep were then administered midazolam 5 or 15 mg/d or saline for 43 days (Table 1). One additional sheep received 10 mg/d. Pigs were randomized to receive either midazolam 15 mg/d or saline.
|
Midazolam powder was obtained from Diosynth BV, The Netherlands (purity not less than 98.5%; total impurities <1%). The aqueous solubility was found to be pH dependent, with greater solubility at lower pH. Midazolam dissolved in 0.9% sodium chloride had a maximum solubility of approximately 2.4 mg/mL (pH 4.2). To maximize the solubility, 1.0, 1.67, and 2.5 mg/mL solutions were prepared by initially dissolving in 1 N hydrochloric acid and then diluting with 0.9% sodium chloride injection (USP) to the desired concentration. The pH was then adjusted to 3.85 with 1 N sodium hydroxide to minimize the potential of corrosion of the infusion pump.
Because of lower solubility of larger concentrations at the desired pH, 5.0 mg/mL solutions were initially dissolved in 1 N hydrochloric acid and then diluted with 0.45% sodium chloride injection (USP) to avoid precipitation. The pH was then adjusted to 3.6 with 1 N sodium hydroxide. A pH of 3.6 was considered comparable to 3.85 while ensuring adequate solubility of the larger concentration. Osmolalities measured in representative samples of the 2.5 and 5.0 mg/mL formulations were 316 and 200 mOsm/kg, respectively. Midazolam solutions contained no preservatives or other additives. All solutions were sterile-filtered with a 0.22-µm filter. The control article was 0.9% sodium chloride injection (Baxter, Deerfield, IL). The labeled pH range of this normal saline is 4.5 to 7.0, with an osmolality of 308 mOsm/kg. The pH measured in a representative lot used in this study was 5.0.
Behavior and motor function were evaluated in sheep for a 15-min period daily and were recorded with a four-grade scale as follows: Grade 0, animal standing, eating, and ruminating with normal respiratory and heart rates and is able to rise and ambulate without difficulty; Grade 1, shuffling of either rear leg or slight limp, slight distortion of normal spinal axis; Grade 2, loss of righting reflex (inability to extend through the pastern in the normal locomotion typical of this species) in one of the rear legs, able to stand without assistance but with some difficulty, decreased interest in eating, ruminating, and environment; and Grade 3, inability to maintain standing posture, attempts by technician to help animal stand are unsuccessful.
Indirect arterial blood pressure, heart rate, temperature, and weight were documented on Day 0 (before drug infusion), Day 1, and Day 3 and weekly thereafter (with the exception of the initial sheep [Sheep 473] that received 3 mg/d). Arterial blood pressure was documented by using an instrument equipped with a pediatric cuff (Colin Press-mate, BP8800; Colin Electronics Co., Japan) placed over the cephalic vessels.
Venous blood samples (7–10 mL) were drawn from sheep at baseline (before surgery) and on Day 1 (postdrug initiation), Day 15, and Day 43 from the jugular vein for complete blood count, electrolytes, and serum chemistry. CSF samples (1–2 mL) for cytology and chemistry were obtained at the time of catheter placement and at termination on Day 43 via dural puncture with a 21-gauge needle at the L7-S1 level.
Gross and microscopic evaluation of tissues was performed on all animals by a board-certified veterinary pathologist (WBB). Initially, a dorsal laminectomy was performed from S1 to C1, the drug pump was carefully dissected out, and the spinal cord was removed with the intrathecal catheter intact. The appearance of the tissue near the catheter, the position and path the catheter traveled, and the length of the catheter in the subarachnoid space were documented. The spinal cord was fixed in 10% formalin for a minimum of 1 wk. After fixation, representative cord sections (
19) were taken from the catheter tip, from an area 5 cm cranial to the catheter tip, and from the midcatheter area. All tissues were processed routinely for light microscopic evaluation, paraffin-embedded, and sectioned at 4–6 µm and stained with hematoxylin and eosin. Spinal cord sections were evaluated for the presence or absence of (a) inflammation or other reactions (e.g., fibroplasia/fibrosis) around the catheter; (b) inflammation in the epidural space, meninges, and spinal cord parenchyma; and (c) compression or other damage to the cord parenchyma. Inflammation was classified as acute, chronic, or both. The degree of inflammation or spinal cord compression was graded as mild (+), moderate (++), or severe (+++).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
|
|
The observed lesions occurred in both saline- and midazolam-treated animals in both species (Figs. 1 and 2) to similar degrees, with no other obvious differences between control and treated animals. Figure 1 shows mild infiltration of inflammatory cells surrounding the catheter site at the S1 (Fig. 1A) and L4 (Fig. 1B) levels in a sheep that received 15 mg/d. Figure 1A represents the most reactive response observed in sheep. Figure 2 illustrates mild spinal cord compression (Fig. 2A) and axonal changes in the white matter (Fig. 2B) in a saline-treated sheep. This is consistent with pressure/mechanical injury secondary to the presence of the catheter in the vertebral canal. Nonnervous tissues (lungs, trachea, heart, liver, gallbladder, gastrointestinal tract, lymph nodes, spleen, kidneys, urinary bladder, pituitary, and adrenal and thyroid glands) had no significant lesions.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Midazolam is a water-soluble imidazobenzodiazepine that differs structurally from other benzodiazepines by the presence of an imidazole ring. In its closed form at physiological pH, this ring imparts increased lipid solubility, facilitating tissue penetration. This characteristic has contributed to its being an extensively studied benzodiazepine for spinal administration. As a single drug administered by intrathecal bolus in humans, midazolam doses up to 2 mg/d have effectively treated chronic nonmalignant back pain (7) and pain of somatic origin (13). In rodent, dog, and sheep models, intrathecal midazolam alone has shown effects of sensory blockade and antinociception (10,12,14) and significant increases of mechanical pain thresholds (14).
The antinociceptive effects of benzodiazepines are thought to be mediated primarily via spinal cord benzodiazepine-aminobutyric acid receptors (6,19); however, multiple pathways may be involved in their proanalgesic action. Synergistic analgesic effects have been described with intrathecal midazolam combined with clonidine (15), local and general anesthetics (20), and opioids (9,11). Additional studies demonstrate antinociceptive synergism between midazolam and glutamate receptor antagonists that is suggestive of a functional coupling of benzodiazepine-aminobutyric acid receptors with opioid and glutamate receptors in acute nociceptive mechanisms in the spinal cord (21).
The relative lack of toxicity associated with intrathecal midazolam in clinical reports and initial animal studies enhanced interest in its development for intraspinal use. However, three recent reports of neurotoxicity in rats and rabbits have necessitated further evaluation of its neurotoxic potential (16–18). In rats, evidence of neuronal death and cellular abnormalities was reported after 20 daily 100-µg subarachnoid bolus injections (20 µL each) of Dormicum (Roche, Switzerland), a commercially available preservative-free midazolam solution (16). In addition to the toxicity that may result from multiple bolus injections, others have suggested that the commercial preparation used in that study was hypotonic, and this may implicate hypotonicity in the observed tissue damage (22). Hypotonicity has been shown to produce permanent nerve injury in isolated nerve preparations (23) and has been implicated in the neurotoxicity of spinal sufentanil in sheep (24).
In the first of 2 rabbit studies demonstrating neurotoxicity, histologic abnormalities were observed in spinal tissue harvested 8 days after a single percutaneous intrathecal injection of 0.1% midazolam (0.1 mg/kg in 300 µL) (17). In this study, several possible sources of artifact have also been suggested, including the diffuse nature of the histopathologic abnormalities that uncharacteristically extended from the cervical to lumbar sections, the presence of significant and persistent diastolic hypotension in the treatment group, and the delayed postmortem tissue fixation (22).
More recently, a second rabbit study by Erdine et al. (18) demonstrated histologic lesions after five daily bolus injections of both preservative-containing and preservative-free 0.1% midazolam preparations (0.3 mL per injection followed by 0.3 mL of 0.9% saline). The study of Erdine et al. also has a number of limitations that raise concerns about the validity of the results. The rabbit infusion system used a 20-gauge polyethylene catheter introduced into the intrathecal space from the bottom of the spinal cord and directed upward. These catheters were only slightly smaller than that used in our sheep and pig model (18 gauge) and were much less pliable than silicone elastomer catheters. Silicone elastomer catheters are by far the most common material used in human spinal catheters. In addition, the nature of the midazolam test substance used in the Erdine et al. study remains unclear with regard to whether the Dormicum preparation contained preservative and how the preservative-free preparation was formulated.
In this study, preservative-free midazolam preparations were prepared within a similar pH range of 3.6–3.85. This pH range was selected to maximize solubility and minimize the possibility of corrosive effects on pump components. We did not adjust the pH of the commercial sodium chloride injection preparation used in control animals. Although the labeled pH range for this preparation is somewhat large (4.5–7.0), the pH measured in a representative lot used in this study was 5.0. Because we did not observe deficits in behavior or motor function and because no drug-related lesions were observed in this study, the difference in pH between midazolam and normal saline control solutions was not considered a significant factor.
Concentrations of midazolam delivered in this study ranged from 1.0 to 5.0 mg/mL to allow a consistent infusion rate of approximately 3 mL/d (125 µL/h) in most animals. Osmolality of concentrations up to 2.5 mg/mL was comparable to that of normal saline. Preparation of a 5.0 mg/mL solution necessitated a reduced amount of sodium chloride (0.45%) to avoid midazolam precipitation, which was slightly hypotonic.
It is also important to note that in the Erdine et al. study, as well as in all published animal studies, midazolam was administered by intrathecal bolus, and pressure from direct injection into the intrathecal space and the resulting larger drug concentrations and injection volumes may have played a role in the toxic effects of the intraspinally administered drug. The subarachnoid space of a rabbit or rat is much smaller than in humans, and there is therefore much less of a dilution effect compared with humans receiving microliter volumes of an intrathecal drug into a large CSF pool. Finally, the Erdine et al. study began bolus intrathecal injections one day after catheter placement, thus leaving little time for healing after catheter placement. The authors did not state how many rabbits were excluded from the study because of weakness and/or paralysis after placement of the 20-gauge catheters. In this study, a full week was allowed postsurgical manipulations before midazolam was infused.
The sheep and pigs implanted with infusion systems provided several advantages over the smaller animal models that used indwelling intrathecal catheters and provided further evidence of the safety of preservative-free midazolam preparations given by continuous intrathecal infusion. Sheep have been used to study the intrathecal infusion of test substances because they more closely resemble humans with regard to spinal cord length, CSF volume, CSF production rate, and body weight (25). The total doses administered, therefore, more closely approximate those likely required for analgesia in humans, and infusion systems that are used in the clinical setting could be incorporated into the model. In addition, infusion systems could be placed surgically in an identical manner to the human permanent spinal infusion pumps, with percutaneous needle placement caudally, passage of the catheter through the needle, and then upward threading of the needle into the intrathecal space.
The single histologic finding in this study was characterized as mild to moderate infiltration of inflammatory cells surrounding the catheter tract; this was present in all animals. Because it was observed equally in both saline- and midazolam-treated animals, the most likely cause relates to the presence of the chronic indwelling catheters and not to midazolam. Fibrosis and lymphocytic infiltrations secondary to chronic spinal cannulation have been documented in various animal models (5). In this study, evaluation of sections stained with hematoxylin and eosin was sufficient to identify evidence of drug-induced toxicity. In the absence of such evidence on qualitative evaluation by light microscopy, additional stains and further quantitative analyses were not performed.
Although not the purpose of this study, there was some suggestion of an analgesic effect from midazolam by nociceptive testing performed in sheep by using a mechanical stimulus device as previously described (14). The analgesic efficacy of intrathecal midazolam will require further study, and data will be presented in a future article.
In summary, continuous intrathecal infusion of preservative-free midazolam at doses of 5–15 mg/d appear to be well tolerated. Ours is the first study to critically evaluate the neurotoxic potential of this drug administered by continuous infusion—the method of administration most likely to be used for the treatment of intractable pain in humans. The lack of neurotoxicity observed in this study in two large animal species provides further evidence of the safety of long-term intrathecal therapy with this drug, suggesting that midazolam may be a valuable alternative to intrathecal morphine for severe intractable pain or pain syndromes unresponsive to opioids.
|
Acknowledgments |
|---|
The authors thank Kyle Voelker, BS (large animal section supervisor), Troy King, ALAT (Animal Technician I), Don Hudson, BS (animal technologist), and Gerald Costello, BS, HT (ASCP) (senior research histology technician, The University of Texas M. D. Anderson Cancer Center, Department of Veterinary Sciences, Bastrop, TX) for expert technical assistance in performing this study. They also thank Dr. Keith Hildebrand (Medtronic Neurological, Inc., Minneapolis, MN) for his scientific expertise in performing this study.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
