Vous êtes sur la page 1sur 7

European Journal of Anaesthesiology 2008; 25: 403–409

r 2007 Copyright European Society of Anaesthesiology

doi: 10.1017/S0265021507003079

Original Article

The efficacy and neurotoxicity of dexmedetomidine

administered via the epidural route

S. Konakci*, T. Adanir*, G. Yilmaz*, T. Rezankoy

Ataturk Training and Research Hospital, Departments of *Anaesthesiology, y Pathology, Izmir, Turkey

Background: a2-Adrenoceptor agonists administered into the intrathecal and epidural space have been found to
be effective in the treatment of chronic pain. Moreover, it was shown that they increase the analgesic effects of
local anaesthetics and provide sedation, anxiolysis and haemodynamic stability. Dexmedetomidine, a potent
and highly selective a2-adrenoceptor agonist, is in current clinical use, particularly in the intensive care unit.
Our aim was to investigate whether dexmedetomidine produced motor and sensory blockade and neurotoxic
effects when administrated via the epidural catheter in rabbits. Methods: Twenty-one New Zealand white
rabbits were included in the study. Animals were randomized into three groups. In Group L: lidocaine (2%),
in Group LD: lidocaine (2%) 1 dexmedetomidine (5 mg) and in Group D: dexmedetomidine (10 mg) were
administered by epidural catheter. Motor and sensory blockade were evaluated. After the evaluation of block,
the animals were euthanized and their spinal cords removed for neuropathological evaluations. Results: Motor
and sensory blockade were lower in Group D than in Group L and Group LD (P , 0.01). Although there were
no differences between the groups for ischaemia of the medulla spinalis, evidence of demyelinization of the
oligodendrocytes in the white matter in Group D was significantly higher than in Group L (P 5 0.035).
Conclusions: We observed that dexmedetomidine does not have motor and sensory effects, but it may have a
harmful effect on the myelin sheath when administered via the epidural route.



Introduction a2-adrenoceptor agonists predominantly result from

supraspinal actions [4]. An a2-adrenergic compound
In order to provide ideal analgesia via the epidural
that diffuses poorly through the blood–brain barrier
route, the antinociceptive side-effects should be
and within the central nervous system might provide
minimal. Because there is no such ideal local
effective locally mediated antinociception with minor
anaesthetic, many drugs (e.g. opioids, benzodiaze-
side-effects after intrathecal or epidural administra-
pines, magnesium, ketamine, clonidine, baclofen,
tion. a2-Adrenoceptor agonists have been used in
somatostatin, neostigmine and steroids) have been
association with local anaesthetics to increase the
used as adjuvants. a2-Adrenoceptor agonists have
duration of spinal anaesthesia. Intrathecal or epidural
antinociceptive effects [1–3]. This antinociception
administration of clonidine prolongs motor blockade
is in part caused by action on spinal a2-adreno-
induced by local anaesthetics. Since the affinity of
ceptors. In contrast, the sedative side-effects of
dexmedetomidine to a2-adrenoceptors is eight times
greater than clonidine, it is expected that dexmede-
Correspondence to: Tayfun Adanir, P.B. 12 Hatay, 35370 Izmir, Turkey. E-mail:
tadanir@tnn.net; Tel: 190 232 2444444 380; Fax: 190 232 2434848
tomidine could be advantageous in regional anaes-
Accepted for publication 16 October 2007 EJA 4302
thesia. Establishing the safety of neuroaxial drugs is
First published online 19 December 2007 very important before these drugs are given to
404 S. Konakci et al

humans. Despite extensive use in animal experi- Five hours after the insertion of the epidural
mentation, dexmedetomidine has not been tested catheter, in order to verify that the catheter was
clinically by the epidural or intrathecal route, or within the epidural space, 1 mL of 1% lidocaine was
evaluated for neurotoxicity. administered via the epidural catheter and flushed
The purpose of this prospective, randomized, with 0.2 mL of 0.9% NaCl. If any motor or sensory
double-blind study was to investigate whether blockade was observed within 5 min, it was concluded
dexmedetomidine has motor and sensory blockade that the catheter was in the subdural space, not in
and side-effects when it was administered via an the epidural space. It was verified that motor or
epidural catheter in rabbits. sensory blockade was observed after 20 min of drug
The ear lobe veins of the animals were cannulated
Methods with 24-G catheters and 0.9% NaCl was admi-
This study was approved by the Institutional Animal nistered (5 mL kg21) 24 h after the placement of
Investigation Ethics Committee of Izmir Ataturk the epidural catheter. The animals were sedated
Training and Research Hospital. Twenty-eight New with intravenous (i.v.) midazolam (0.2 mg kg21)
Zealand White Rabbits weighing 1650–2100 g were (Dormicum, Roche, Fontenay, France). Arterial
planned to be included in the study; however, seven cannulation of the other ear was performed after
rabbits were excluded for failure of placement of sedation. Invasive blood pressure (BP) and heart rate
epidural catheter at the beginning of the study. (HR) monitoring (KMA 275; Petas Corporation,
Twenty-one rabbits were studied. All procedures were kara, Turkey) were performed through this catheter.
performed between 4:00p.m. and 12:00p.m. Each We waited for the sedation effect to resolve. On the
animal was used for only one set of studies to elim- second day, we performed the main drug study.
inate possible interactions between different doses and Animals were assigned to three groups with seven
routes of drugs. The animals were kept in the animal animals in each group chosen by computer-gener-
research laboratory for 10 days and fed a standard diet ated random numbers.
before the procedures started.
Group L: Lidocaine 2% (Aritmal Biosel amp,
Before the placement of the epidural catheter,
Beykoz, Turkey) 0.8 mL and NaCl 0.9% 0.2 mL
ketamine 50 mg kg21 (Ketalar-Eczacibasi Warner
were administered through the epidural catheter
Lambert 39780, Luleburgaz, Turkey) was admi-
and the catheters were flushed with NaCl 0.9%
nistered intramuscularly. The hair on the tails
0.2 mL.
and the waist regions of the animals were shaved
Group LD: Lidocaine 2% 0.8 mL and dexmedeto-
after they were prone-positioned. The tail regions
midine (5 mg) 0.2 mL (Precedex flacon, Abbott
were scrubbed with 10% povidone iodine. A skin
Laboratory, North Chicago, IL, USA) were adminis-
incision was performed 1 cm distal to the anus,
tered through the epidural catheter and the
following sterile coverage and the subcutaneous
catheters were flushed with NaCl 0.9% 0.2 mL.
injection of 1% lidocaine. After skin incision,
Group D: Dexmedetomidine (10 mg) 0.8 mL and
connective tissue and paraspinal muscles were dis-
NaCl 0.9% 0.2 mL were administered through the
sected and the sacral hiatus was opened. An 18-G
epidural catheter, and the catheters were flushed
epidural catheter (Minipack SIMS Portex Ltd,
with NaCl 0.9% 0.2 mL.
Hyde, Kent, UK) was inserted to the epidural space
through the sacral canal up to 4–5 cm cranially. We did not create another group using only
Following catheter placement into the sacral canal, saline for control since lidocaine and saline injec-
we tested whether the dura matter was punctured or tions through the catheter would inevitably verify
not by aspiration. Then the free edge of the catheter that the catheter was within the epidural space.
was placed in a subcutaneous tunnel using a Tuohy Therefore this group would be identical to the
needle, it was shortened and connected to a screw lidocaine group.
connector for subsequent use. Then the catheter was Mean arterial pressures (MAP) and HR of the
fixed to the skin and the incision was sutured. animals were recorded during the study. Arterial
Neurological injury associated with the use of the blood samples were obtained at 0, 5, 15, 30 and
epidural catheter was evaluated after recovery from 60 min to evaluate pH, PaO2, PaCO2 and SaO2.
the ketamine. Paraplegia and the absent flexion Motor function was assessed according to the cri-
reflex of both the lower extremities following teria of Drummond and Moore [5] by an investi-
painful stimuli on the toes of the animals were gator blind to the treatment (0: free motion without
accepted as positive signs of neurological injury. limitation at lower extremities; 1: asymmetry and
Any animals exhibiting signs of a neurological or limitation in providing body support and walking
motor deficit were eliminated from the study. at lower extremities; 2: inability to provide body

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409
Dexmedetomidine neurotoxicity 405

support by lower extremities; and 3: paralysis of Statistical analysis

both the lower extremities). Statistical analyses were carried out using SPSS for
A painful stimulus was applied by nipping the tail, Windows version 11.0. Data were expressed as
lower extremities, upper and lower abdomen, upper mean 6 SD. As three of the groups had n , 30 and
extremities and ear using a surgical clamp by a were independent of each other and variables had
blinded investigator. If there was no response to the been measured at equal intervals, statistical analyses
surgical clamp, the response to surgical painful were performed with non-parametric tests. HR,
stimulus was evaluated by an incision on the abdomen. MAP, motor block, painful stimuli, intestinal loops
After the surgical stimulus, if there was no response if exposed, pH, PaO2, PaCO2 and SaO2 were com-
to the painful stimulus the peritoneum was incised pared with the Kruskal–Wallis test according to
and intestinal loops were exposed similar to a sur- time in each group. U-test with Bonferroni’s cor-
gical operation. Sedation was assessed by placing the rection and Wilcoxon signed rank sum test were
rabbit on its back – if it spontaneously recovered to used to assess differences found between the three
the prone position, there was no sedation. groups. Comparison of the neuropathological eva-
After the evaluation of sensory blockade, the luations between the three groups was made using
animals were euthanized using thiopental Fisher’s exact x2 test. The tolerance of intestinal
(120 mg kg21, i.v.) and then the perfusion fixation loops that were exposed to evaluate the response to
method was applied with 10% formaldehyde. The painful stimuli was also compared with Fisher’s
spinal cords of the animals were removed en bloc exact x2 test. P , 0.05 was considered as the
from their spinal columns and fixed in 4% para- threshold of significance.
formaldehyde solution. After fixation, the sections
of spinal cords were embedded in paraffin. In total,
five segments (at the catheter tip level and two
segments up and down) were examined in each MAP, HR and blood gas data did not change in any
rabbit and two sections of each segment were of the groups (Fig. 1). Although epidural admin-
examined. The blocks were cut at a thickness of istration of lidocaine and lidocaine combined with
5 mm and stained with haematoxylin–eosin to dexmedetomidine produced clear motor block,
examine the ischaemic injury and immunestaining dexmedetomidine alone did not produce the same
with antibodies to myelin basic protein (MBP) effect. There were significant differences at 2, 3, 4,
(Clone RB-1460 R7; Neomarker Inc., Fremont, 5, 15 and 30 min between Group D and the other
CA, USA) [6] for evaluation of the myelin damage. two groups (P , 0.01). The motor block that was
Immunohistochemical staining was performed seen in Group LD was similar to that in Group L
using the Strept–Avidin–Biotin Peroxidase method. (P . 0.05) (Fig. 2).
Neuropathologic assessment was provided by a Epidural administration of dexmedetomidine
pathologist, blinded to the study group. Ischaemic (10 mg dose) did not produce antinociception to the
neurons were identified by cytoplasmic eosinophilia painful stimulus that was produced by surgical
with loss of Nissl substance and pyknotic nuclei. To clamp. In the evaluation of painful stimuli that
evaluate the amount of myelin loss in white matter on were applied on the tail, lower extremities and
immunostained sections for MBP, the scoring system upper and lower abdomen, there were significant
of Petersson and colleagues [7] was modified. The differences at 2, 3, 4, 5, 15, 30 and 45 min between
whole mounted coronal sections were scored by a Group D and the other two groups (P , 0.01)
pathologist, who was not aware of each group. The (Fig. 3). No differences were observed between
pathologist scored the most severely injured area of Group L and Group LD in the evaluation of the
each section using the following scoring system: response to the surgical stimulus produced by an
incision on the abdomen. For ethical reasons, the
0 – no myelin loss;
stimulus could not be applied because of possible
1 – myelin loss between 1% and 25% (minimal
intolerance of the animals in Group D. In the eva-
myelin loss);
luation of the intestinal loop tolerance that was
2 – myelin loss between 26% and 50% (moderate
exposed to evaluate response to painful stimuli,
myelin loss);
there were no significant differences between Group
3 – myelin loss between 51% and 100% (severe
LD and Group L. We did not find any differences
myelin loss).
between groups for painful stimuli that were
The percentage of myelin loss was calculated as applied on the upper extremities and the ear by
the number of unstained oligodendrocytes for MBP pinprick test.
in the total oligodendrocytes count of white matter There were no statistically significant differences
for each section. between the three groups for the degree of ischaemic

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409
406 S. Konakci et al

Figure 1.
Mean arterial pressure (MAP, mmHg) and heart rate (HR, beats min21).

Dexmedetomidine is a highly selective, short-acting
central a2-adrenoceptor agonist. Adrenergic agonists
acting at the a2-adrenoceptors produce antinocicep-
tion. These responses have been well documented in
animals and in human clinical studies [8]. It has been
demonstrated that dexmedetomidine produces anti-
nociception in systemic [9,10], intrathecal [11–14],
epidural administration [15,16] or into the locus
coeruleus administration [17].
In general, dexmedetomidine is used as an anti-
nociceptive agent, and it has been investigated with
regard to its antinociceptive effect or prevention of
postoperative pain. Its intraoperative use has been
under investigation recently. Memis and colleagues
[18] have shown that the addition of dexmedetomi-
dine to the local anaesthetic solution during i.v.
regional anaesthesia improved the quality of anaes-
Figure 2. thesia and perioperative analgesia. Another study by
Motor block. There were significant differences between Group D Esmaoglu and colleagues [19] has demonstrated that
and the other two groups (P , 0.01). the addition of dexmedetomidine to local anaesthetic
solution in i.v. regional anaesthesia improved the
injury. Mild, diffuse and parenchymatous swell- quality of anaesthesia and decreased analgesic require-
ing of the white matter was observed in all cases ments, but had no effect on the onset and regression
(Fig. 4). The oligodendrocytes in the white matter times of sensory and motor blocks.
showed evidence of moderate or severe demyelin- Since epidural administration of dexmedetomi-
ization in three cases in Group D and in two cases dine is approximately five times more effective than
in Group LD. The difference in the degree of systemic administration in producing an anti-
demyelinization of the myelin sheaths between nociception effect [15], we chose the epidural route.
Group D and Group L was statistically significant Our results were similar to that of the study of
(P 50.035) (Figs 5 and 6, respectively). In addition Esmaoglu and colleagues [19], although the admi-
to these results, we observed deep sedation in nistration route of dexmedetomidine was different
Group D and Group LD. in our study. We were not able to show motor

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409
Dexmedetomidine neurotoxicity 407

Figure 3.
Evaluation of painful stimuli applied on the tail, lower extremities, and upper and lower abdomen. There were significant differences at 2, 3,
4, 5, 15, 30 and 45 min between Group D and the other two groups (P , 0.01).

Figure 4. Figure 5.
Mild diffuse parenchymatous swelling of the white matter (left Myelin loss of the oligodendrocytes in the white matter; Group D
side; H&E 3220). (myelin basic protein 3440).

blockade and antinociception from the epidural also showed that dexmedetomidine produced a dose-
administration of dexmedetomidine (10 mg dose) in dependent (1, 2, 4, 10 and 20 mg kg21 epidurally)
response to twisting with a surgical clamp. We are antinociceptive effect. Although we observed adequate
aware that the dexmedetomidine doses (5 and sedation with these drug doses, even with these doses,
10 mg) that we used might not be enough to produce myelin sheath injury occurred. The onset and regres-
adequate block. We did not create dose-response sion of motor block in Group L were similar to Group
curves for rabbits and there is no previous study for LD. We did not observe any synergistic effect between
evaluating a dose-response curve in rabbits. However, dexmedetomidine and lidocaine.
similar drug doses had been used in rat studies [15]. To the best of our knowledge, there has been no
Asano and colleagues [15] examined the dose- study in the English medical literature about the
dependent (0.5, 1, 5 and 10 mg epidurally) antinoci- effect of the epidural administration of dexmede-
ceptive dexmedetomidine effect in rats. They showed tomidine on neural injury. In contrast, dexmedeto-
that dexmedetomidine produced dose-dependent midine provides neuroprotection in models of brain
antinociceptive effect. Walker and colleagues [16] ischaemia in animals [20,21]. A neuroprotective

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409
408 S. Konakci et al

rather than a specific effect of dexmedetomidine.

However, all of the rabbits removed from the study
were the first seven animals (learning curve) and no
animal was excluded from the study for failure of
placement of the epidural catheter afterwards. It is
conceivable that the presence of an epidural catheter
alone could have produced the injury we found. A
control group that received only NaCl would have
allowed us to separate catheter injury from drug-
induced injury. However, previous studies on the
safety of epidural catheters found no neurotoxic
effect of these catheters. Canduz and colleagues [24]
showed that there was no histopathological sign of
neurotoxicity in the control group, which had only
an epidural catheter without receiving any drug. In
Figure 6.
Group L cytoplasmic staining with myelin basic protein antibody
addition, Lim and colleagues [25] epidurally
in oligodendrocytes (3 oil immersion field). administered hyaluronic acid to rabbits and the
subsequent electron microscopic findings showed
no neurotoxicity in the control group.
effect of dexmedetomidine is mediated by the In summary, i.v. dexmedetomidine may be an
activation of the a2-adrenergic receptor subtype appropriate and safe adjuvant agent for local or general
[22]. As dexmedetomidine has neuroprotective anaesthesia. However, it caused significant neural
effects, it could be expected that dexmedetomidine injury when it was applied via the epidural route in
by epidural or intrathecal administration would not rabbits. In order to administer dexmedetomidine via
be harmful. However, this study has shown that the epidural route as a safe adjuvant agent, further
dexmedetomidine (10 mg) produced the moderate studies using lower doses of dexmedetomidine and
or severe demyelinization of myelin sheaths in advanced pathologic investigations are required.
the white matter, when it was administered via the
epidural route. This effect could be related to Acknowledgements
vasoconstriction of the medullary spinal vessels [23]
and pH of dexmedetomidine. Precedex (dexmede- The authors are grateful to Mehmet Haciyanli for
tomidine) is supplied as a clear, colourless, isotonic his careful English review of the manuscript.
solution with a pH of 4.5–7.0, and this pH may
cause demyelinization. Each 1 mL of Precedex con- References
tains 118 mg of dexmedetomidine HCl (equivalent
to 100 mg dexmedetomidine base) and 9 mg of 1. Yaksh TL. Pharmacology of spinal adrenergic systems
NaCl in water. The solution is preservative-free and which modulate spinal nociceptive processing. Pharmacol
contains no additives or chemical stabilizers. Thus, Biochem Behav 1985; 22: 845–858.
2. Pertovaara A. Antinociception induced by alpha-2
we do not think that demyelination could be related adrenoceptor agonists, with special emphasis on dexme-
to additives or chemical stabilizers. detomidine studies. Prog Neurobiol 1993; 40: 691–709.
Furthermore, dexmedetomidine did not have an 3. Kingery WS, Davies MF, Maze M. Molecular mechanisms
antinociception effect against surgical stimulus or for the analgesic properties of alpha-2 adrenergic agonists.
twisted stimulus by surgical clamp. We observed Prog Pain Res Manage 1997; 9: 275–304.
that dexmedetomidine (5 mg) with local anaesthetic 4. Maze M, Tranquili W. Alpha-2 adrenoceptor agonists:
solution in epidural anaesthesia had sedative and defining the role in clinical anesthesia. Anesthesiology 1991;
haemodynamic stabilization effects. In spite of 74: 581–605.
administering dexmedetomidine, which causes a 5. Drummond JC, Moore SS. The influence of dextrose
dose-dependent decrease in BP and HR, we did not administration on neurological outcome after temporary
observe any haemodynamic side-effects such as spinal cord ischemia in the rabbit. Anesthesiology 1989; 70:
hypotension or bradycardia requiring intervention.
6. Chanimov M, Cohen ML, Grinspun Y et al. Neurotoxicity
The reason for this might be the use of low-dose after spinal anaesthesia induced by serial intrathecal
dexmedetomidine. injections of magnesium sulphate. An experimental study
A significant number of rabbits (seven rabbits) in a rat model. Anaesthesia 1997; 52(3): 223–228.
were removed from the study due to the failure of 7. Petersson KH, Pinar H, Stopa EG et al. White matter
placement of the epidural catheter. One may spec- injury after cerebral ischemia in ovine fetuses. Pediatr Res
ulate that demyelinization may be a result of trauma 2002; 51(6): 768–776.

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409
Dexmedetomidine neurotoxicity 409

8. Hayashi Y, Maze M. Alpha 2 adrenoceptor agonists and dexmedetomidine on inflammatory hyperalgesia in rat
anesthesia. Br J Anaesth 1993; 71: 108–118. pups. Anesthesiology 2005; 102: 1226–1234.
9. Otsuguro K, Yasutake S, Ohta T, Ito S. Effects of opioid 17. Guo TZ, Jiang JY, Buttermann AE, Maze M. Dexmede-
receptor and alpha 2-adrenoceptor agonists on slow ventral tomidine injection into the locus ceruleus produces
root potentials and on capsaicin and formalin tests in neonatal antinociception. Anesthesiology 1996; 84: 873–881.
rats. Brain Res Dev Brain Res 2005; 158(1–2): 50–58. 18. Memis D, Turan A, Karamanlioglu B, Pamukcu Z, Kurt I.
10. Wahlander S, Frumento RJ, Wagener G et al. A Adding dexmedetomidine to lidocaine for intravenous
prospective, double-blind, randomized, placebo-controlled regional anesthesia. Anesth Analg 2004; 98: 835–840.
study of dexmedetomidine as an adjunct to epidural 19. Esmaoglu A, Mizrak A, Akin A, Turk Y, Boyaci A.
analgesia after thoracic surgery. J Cardiothorac Vasc Anesth Addition of dexmedetomidine to lidocaine for intra-
2005; 19: 630–635. venous regional anesthesia. Eur J Anaesthesiol 2005; 22:
11. Antti P, Hong W. Attenuation of ascending nociceptive 447–451.
signals to the rostroventromedial medulla induced by a 20. Jolkkonen J, Puurunen K, Koistinaho J et al. Neuropro-
novel alpha 2-adrenoceptor agonist, MPV-2426, following tection by the alpha 2-adrenoceptor agonist, dexmedeto-
intrathecal application in neuropathic rats. Anesthesiology midine, in rat focal cerebral ischemia. Eur J Pharmacol
2000; 92: 1082–2002. 1999; 372: 31–36.
12. Xu M, Kontinent V K, Kalso E. Effects of radolmidine, 21. Vincent L, Jean M, Hugo L, Jean-Marie D, Philippe E,
a novel alpha 2-adrenergic agonist compared with Pierre G. Effects of a2-adrenoceptor agonists on perinatal
dexmedetomidine in different pain models in the rat. excitotoxic brain injury: comparison of clonidine and
Anesthesiology 2000; 93: 473–481. dexmedetomidine. Anesthesiology 2002; 96: 134–141.
13. Gabriella J, Gyöngyi H, Walter K et al. The effects of 22. Ma D, Hossain M, Rajakumaraswamy N et al. Dexmede-
ketamine and its enantiomers on the morphine- or tomidine produces its neuroprotective effect via a2-
dexmedetomidine-induced antinociception in rats. adrenoceptor subtype. Eur J Pharmacol 2004; 502: 87–97.
Anesthesiology 2000; 93: 231–241. 23. Iida H, Ohata H, Iida M, Watanabe Y, Dohi S. Direct
14. Calasans-Maia JA, Zapata-Sudo G, Sudo RT. Dexmedeto- effects of alpha 1- and alpha 2-adrenergic agonists on
midine prolongs spinal anaesthesia induced by levobupi- spinal and cerebral pial vessels in dogs. Anesthesiology
vacaine 0.5% in guinea-pigs. J Pharm Pharmacol 2005; 57: 2000; 92: 1488–1489.
1415–1420. 24. Canduz B, Aktug H, Mavioglu O et al. Epidural
15. Asano T, Dohi S, Ohta S, Shimonaka H, Iida H. lornoxicam administration-innocent. J Clin Neurosci
Antinociception by epidural and systemic alpha 2007; 14: 968–974.
2-adrenoceptor agonists and their binding affinity in rat 25. Lim YJ, Sim WS, Kim YC, Lee SC, Choi YL. The
spinal cord and brain. Anesth Analg 2000; 90: 400–407. neurotoxicity of epidural hyaluronic acid in rabbits: a
16. Walker SM, Howard RF, Keay KA, Fitzgerald M. light and electron microscopic examination. Anesth Analg
Developmental age influences the effect of epidural 2003; 97: 1716–1720.

r 2007 Copyright European Society of Anaesthesiology, European Journal of Anaesthesiology 25: 403–409