Midazolam causes analgesia at the spinal level, possibly via γ-aminobutyric acid (GABA) receptors . Midazolam does not cause histopathological changes when given intrathecally in rats , and has been used both intrathecally  and epidurally  for pain relief in humans without adverse effects.
This does not mean that perispinal midazolam is harmless. Malinovsky and colleagues  showed microscopic changes in the blood-brain barrier of rabbits. Local anaesthetics, especially chloroprocaine, can be neurotoxic, possibly because of their low pH , and midazolam solution is acidified.
Clouding of a solution indicates the formation of particles which might be neurotoxic, or some other change in its characteristics. Assuming that clouding could indicate neurotoxicity, we examined the effect on pH and transparency of adding midazolam to human cerebrospinal fluid.
Materials and methods
The protocol was approved by the ethics committee of The University of Tokyo, Tokyo, Japan, and informed consent was obtained from all patients.
The transparency and pH of the test drugs were checked by taking six samples from each drug preparation: up to 10 mg midazolam (Dormicum™, Yamanouchi Pharmaceutical Co. Ltd, Tokyo, Japan; 10 mg 2 mL−1) was added to 10 mL of saline solution (Otsuka Pharmaceutical Co. Ltd, Tokyo, Japan) and 6 mL of 0.25% bupivacaine (Marcain™, Astra Japan Co. Ltd, Osaka, Japan). The tested combinations of the agents, and their doses and concentrations, were chosen because these are used in epidural injection for post-operative pain relief [7,8] in human subjects. The preparation of bupivacaine contains preservatives, but is the only preparation available in Japan.
Transparency was checked by the naked eye. The solutions were placed in glass test-tubes and were checked against a black background. Without knowing the contents of the test-tubes, a single observer with good near vision compared the test solutions with the original drugs and with the cerebrospinal fluid . The pH was measured by pH Strip™ (Merck Co. Ltd, Darmstadt, Germany), which has a detectable range of 0-14 and an interval of 0.2-0.3.
A sample of 1.5 mL of cerebrospinal fluid was drawn from each patient. This volume was chosen because it is the same as the normal intrathecal dose of local anaesthetic. The sample was poured into a clear glass test-tube. Four groups of tests were done. In the first group(group M), midazolam was added in increments of 0.1 mg to a total of 10 mg; in the second group (MS), 5 mg of midazolam in 10 mL of saline was added in increments of 0.02 mL to a total of 2 mL; in the third group (B), 0.25% of bupivacaine was added in increments of 0.04 mL to a total of 3 mL; and in the fourth group (MB), 5 mg of midazolam in 6 mL of 0.25% bupivacaine was added in increments of 0.04 mL to a total of 3 mL. The transparency and pH were checked after each increment in all groups.
Statistical analysis of the changes in pH was by analysis of variance for repeated measures followed by Fisher's protected least significant difference; P<0.05 was considered statistically significant.
We studied samples from 24 patients (10 male and 14 female, aged between 30 and 65 years, weighing between 40 and 72 kg) who were scheduled for elective operations suitable for spinal anaesthesia. There were six samples in each of the tested groups.
The addition of midazolam decreased the pH of the saline and 0.25% bupivacaine solutions (Fig. 1). No clouding was observed.
Adding 3 mg of midazolam to 1.5 mL of cerebrospinal fluid (group M) decreased the pH to below 7.0, but even after adding 10 mg of midazolam, the pH was still above 6.0 (Fig. 2). Clouding of the cerebrospinal fluid was first seen in some of the samples at 0.7 mg midazolam and all the samples were clouded after the addition of more than 1 mg midazolam (Fig. 2).
Adding the solution of midazolam (group MS) did not lower the pH to below 7.0 (Fig. 3) and there was no clouding.
Adding 0.25% of bupivacaine(group B) lowered the pH to below 7.0 after 1.9 mL had been added. Clouding was first seen in some of the samples with 1.1 mL of bupivacaine, and all the samples were clouded at the 1.4 mL level (Fig. 4). Adding the mixture of midazolam and bupivacaine (group MB) lowered the pH to below 7.0 after 1.3 mL; clouding first occurred at 0.6 mL and all the samples were clouded after the addition of 1.1 mL(Fig. 4).
Solutions with a low pH are thought to be neurotoxic . Bupivacaine has a pH of 4 and is not neurotoxic intrathecally , but the reported pH was of the bupivacaine solution, not the pH when added to cerebrospinal fluid.
We have shown in this study that the pH of a sample of 1.5 mL of cerebrospinal fluid was decreased below 7.0 by more than 3 mg midazolam, more than 1.9 mL 0.25% bupivacaine or more than 1.3 mL of midazolam in 0.25% bupivacaine solution. Clouding was seen after adding more than 0.7 mg of midazolam, more than 1.1 mL of bupivacaine or more than 0.6 mL of the combined solution. Midazolam in saline had little effect on pH and there was no clouding.
If drugs injected into the cerebrospinal fluid dilute rapidly, we can extrapolate to what would happen to the whole cerebrospinal fluid volume, which is about 70-80 mL based on the standard textbooks published in Japan. This gives a factor of 46.7 (70 divided by 1.5), and means that more than 32.6 mg of midazolam, more than 51.3 mL of 0.25% bupivacaine or more than 28 mL of the combined solution would be needed to cause clouding of the cerebrospinal fluid, and that more than 140 mg of midazolam, more than 88.7 mL of 0.25% bupivacaine or more than 60.7 mL of the combined solution would be needed to lower the pH below 7.0. Even 93 mL of midazolam in saline would not have any important effect.
Intrathecally, 2 mg of midazolam is effective in relieving chronic low back pain without any side effects . The effective dose of epidural midazolam for post-operative analgesia is 0.05 mg kg−1 in 10 mL of saline  or in 6 mL of 0.25% bupivacaine . A dose of 1.25-3.75 mg in 3 mL of 5% dextrose is effective for the relief of spasticity. Clouding and lowering pH to below 7.0 needed doses far in excess of these. We are presuming that neurotoxicity will not occur unless the pH is less than 7, but we have also ignored the secretion of fresh cerebrospinal fluid (0.35 mL min−1), which is an extra diluent of any injected drug.
In experimental studies, injection of midazolam intrathecally in rats  and epidurally in rabbits , midazolam caused no histopathological changes in the spinal cord. In only one study (Malinovsky et al.) , necrosis, haemorrhage and other histopathological changes occurred in three out of the nine spinal cords of rabbits that had received single-shot intrathecal injections of midazolam [1-4].
There have been no studies of the effects of the speed of injection of midazolam, either epidurally or intrathecally; at the moment of injection, the drug will be in its highest concentration and the possibility of neurotoxicity may be greater. However, from our study, and presuming that clouding and changes in pH are indications of possible neurotoxicity, we suggest that perispinally injected midazolam is unlikely to be neurotoxic.
1 Edwards M, Serrao JM, Gent JP, Goodchild CS, Chir B. On the mechanism by which midazolam
causes spinally mediated analgesia. Anesthesiology
2 Auroy P, Schoeffler P, Maillot C, Haberer JP, Woda A. Tolerance intrathecale du midazolam
: etude histologie. Ann Fr Anesth Reanimi
81-82. (In French.)
3Goodchild CS, Noble J. The effects of intrathecal midazolam
on sympathetic nervous system reflexes in man: a pilot study. Br J Clin Pharmacol
4 Nishiyama T. The post-operative analgesic action of midazolam
following epidural administration. Eur J Anesthesiol
5 Malinovsky JM, Cozian A, Lepage JY, Mussini JM, Pinaud M, Souron R. Ketamine and midazolam
neurotoxicity in the rabbit. Anesthesiology
6 Covino BG, Marx GF, Finster M, Zsigmond EK. Prolonged sensory/motor deficits following inadvertent spinal anesthesia. Anesth Analg
7 Nishiyama T, Hirasaki A, Odaka Y, Konishi H, Seto K, Goto I. Epidural midazolam
with saline: optimal dose for postoperative pain. Jpn J Anesthesiol
49-54. (In Japanese with English abstract.)
8 Nishiyama T, Hirasaki A, Odaka Y, Mikane T, Kobayashi O, Seto K. Epidural midazolam
with bupivacaine: optimal dose for postoperative pain. Jpn J Anesthesiol
1113-1118.(In Japanese with English abstract.)
9 General Rules. In: The Interpretation of the Japanese Pharmacopoeia,
12th Edn. Tokyo: Hirokawa Publishing Co. Inc., 1991: A44. (In Japanese.)
10Ready LB, Plumer MH, Haschke RH, Austin E, Sumi M. Neurotoxicity of intrathecal local anesthetics in rabbits. Anesthesiology
11 Serrao JM, Marks RL, Morley SJ, Goodchild CS. Intrathecal midazolam
for the treatment of chronic mechanical low back pain: a controlled comparison with epidural steroid in a pilot study. Pain
12 Dahm LS, Beric A, Dimitrijevic MR, Wall PD. Direct spinal effect of a benzodiazepine (midazolam
) on spasticity in man. Stereotact Funct Neurosurg
13 Cutler RWP, Page L, Galicich J, Watters GTV. Formation and absorption of cerebrospinal fluid
in man. Brain
14 Madsen JB, Jensen FM, Crawford ME, Toftdahl DB. Catheterization of the epidural space in the rabbit. Neuropathological effects of epidural meptazinal and midazolam
1990; Suppl. 5: