Epilrpsia.
39(3):29&294, I998
Lippincott-Raven Puhlishers, Philadelphia
0 International League Against Epilepsy
Buccal Absorption of Midazolam: Pharmacokinetics and
EEG Pharmacodynamics
Rod C. Scott, *Frank M. C. Besag, ?Stewart G. Boyd, $David Berry, and Brian G. R. Neville
Instilute of Child Health (UCLMS), London; *St Piers Lingfield; ?Great Ormond Street Hospital for Children, London; and
$Medical Toxicology Unit,London, U.K.
Summary: Purpose: To determine whether buccallsublingual
administration of midazolam (MDL) would lead to detectable
venous concentrations and EEG changes in 10 healthy volun-
teers.
Methods: The study consisted of an open-label and a double-
blind phases. Subjects held 10 mg MDL in 2 ml peppermint-
flavored fluid or peppermint-flavored placebo in their mouth
for 5 min and then spat it out. Cardiorespiratory and EEG
monitoring was performed in all subjects.
Results: Venous MDL concentrations measured on 10 occa-
sions from 5 to 600 min after administration showed a rapid
Convulsive status epilepticus (CSE) is the most com-
mon neurological medical emergency and continues to
be associated with significant morbidity and mortality
(1). If prompt prehospital treatment is given, fewer an-
tiepileptic drugs (AEDs) (2,3) are required in the emer-
gency department and seizures tend to be shorter. Diaz-
epam (DZP) is often administered rectally for acute pre-
hospital treatment of seizures but its use may be socially
embarrassing and undesirable. Buccal/sublingual admin-
istration of medications is likely to be more acceptable to
patients and caregivers than rectal administration. The
liquid preparation of midazolam (MDL) was chosen be-
cause it is more rapidly absorbed than the solid prepara-
tion, analogous to the situation with rectal DZP. The
mouth and the rectum have similar surface areas (200
cm') and are richly supplied with blood and lymphatic
vessels; the absorbed drug bypasses the liver, avoiding
the problem of first-pass metabolism (4).
We wished to determine whether buccal/sublingual
MDL would be rapidly absorbed and whether a cerebral
effect would be detectable within a few minutes in
healthy volunteers. Confirmation of these factors was
considered a prerequisite to a clinical trial of treatment of
frequent or prolonged seizures.
Accepted November 6 , 1997.
Address correspondence and reprint requests to Dr. R. C. Scott at
The Wolfson Centie, Mecklenburgh Square, London WClN 2AP, U.K.
290
increase for the first 20-30 min. However, changes in the 8- to
30-Hz frequencies identified by spectral analysis of the EEG
showed changes in S5-10 min in test but not in control sub-
jects-more rapid than were expected from the venous absorp-
tion data. There were no significant adverse effects.
Conclusions: Our data provide direct evidence of the speed
of cerebral effect of a drug. Our results suggest that the buccal/
sublingual route of administration should be tested in emer-
gency treatment of seizures as an alternative to the rectal route,
over which it has clear practical advantages. Key Words: Mid-
azolam-Buccallsublingual-Status epilepticus.
METHODS
Ten healthy adult volunteers ( 5 men and S women
aged 23-47 years) were enrolled in the study. An open-
label phase was followed 6-8 weeks later by a double-
blind phase. Before being accepted into the trial, each
volunteer was confirmed to have normal results of clini-
cal examination, complete blood count, creatinine, elec-
trolytes, and electrocardiogram. The volunteers had no
restrictions on eating or drinking on the morning of the
investigations. To each volunteer was attached a multi-
channel recorder that monitored ECG, thoracic and ab-
dominal respiratory excursion, nasal aifflow, mean arte-
rial blood pressure, and 10 channels of EEG. Silver-
silver chloride EEG electrodes were placed at selected
positions according to the International 10-20 system of
electrode placement (F3, F4, F7,F8, TS, T6, C3, C4, P3,
P4, and with reference between Cz and Pz). A venous
cannula was inserted in a peripheral vein to facilitate
regular blood sampling. The volunteer then lay on a bed,
awake, with arousal maintained and eyes open, in a quiet
room for the first 30 min of the testing period and was
then allowed to sleep.
Because MDL has a bitter taste, it was flavored with
peppermint essence before being administered. The vol-
unteers held 2 ml of the intravenous preparation of MDL
( 5 mg/ml) in their mouth for 5 min and then spat out the
remainder. In the first arm of the study, venous blood
 BUCCAL ABSORPTION OF MDL
samples (5 ml) were drawn before MDL administration
and then at 5, 10, 15, 20, 30, 45, 60, 120, 300, and 600
rnin after administration. The blood was placed in
lithium heparin tubes and spun down, and the serum was
stored at -8°C until analyzed by high-performance liquid
chromatography. MDL was analyzed by a method s i m -
lar to that described by Lehmann and Boulieu (3,with a
minor modification in that methyl t-butyl ether was used
in a liquid-liquid procedure to extract the drug from
plasma. The assay has a lower limit of quantification of
3 pg/L and a coefficient of variance within and between
batches at the 2 0 - p g L level of 6-9%.
The EEG data were filtered (100-Hz low pass, 0.4-Hz
high pass, and 50-Hz notch), digitized at 250 samples per
second, and stored on optical disk. Spectral analysis was
made off-line on the EEG data collected referentially
from the central channels bilaterally. Thirty-second ep-
ochs from 5 min before to 20 min after administration
were analyzed, allowing determination of the relative
power of the EEG at each frequency band ( 1 4 , 4-8,
8-1 2, 12-1 6, and 16-30 Hz) during each epoch. The raw
data were assessed for artifact, especially from electro-
myography and for sleep spindles. Use of central and
midline reference electrodes minimized muscle artefact.
In the second arm of the study, the same volunteers
were randomly assigned to receive either 2 ml pepper-
mint-flavored water or 10 mg (2 ml) peppermint-
flavored MDL in a double-blind study design. We moni-
tored the same parameters as in the initial study.
RESULTS
There was a rapid increase in venous MDL concen-
trations. In 2 of the 10 subjects, the concentration rose to
a level of 5 pg/L in c 5 min; in 7 subjects, this level was
reached by I0 min; in only 1 volunteer did it take I5 min
to reach this concentration. The venous levels continued
2.5 1
FIG. 1. Box and whisker plot of
venous midazolam concentration
versus time confirms rapid in-
crease in serum concentration for
s-
E
1'5-
1.0-
the first 15-20 min after adminis-
tration. Median (horizontal line),
interquartile range (box), and
complete range (whiskers). Outli-
ers (>2 SD from the mean) = as +.
0 5
Time (mins)
291
to rise between 20 and 60 rnin after administration, but
the increase in venous concentration in this time did not
reach statistical significance [one-way analysis of vari-
ance (ANOVA) from 20 to 60 min: p = 0.09 121. A mean
maximum concentration of 32.73 -C 6.4 pg/L (C,,, k 2
SD) was achieved in 48 + 28 rnin (T,,, f 2 SD). Serum
levels then decreased and were either very low or unde-
tectable by 10 h (Fig. 1). The mean elimination half-life
of MDL was 202 f I13 min (t% 2 SD). *
For statistical analysis, we used SPSS for Windows
(version 6) on a PC, combining the EEG data from the
right and left sides. All data were logarithmically trans-
formed before analysis to equalize variances. One-way
ANOVA was performed to determine whether changes
occurred in the power spectrum over time. In the first
arm of the study, buccal/sublingual administration of
MDL was associated with a reduction in 8- to 12-Hz
activity for the first 5 min, followed by a progressive
increase in the relative power of the EEG in the 8- to 12-,
12- to 16-, and 16- to 30-Hz frequency bands (one-way
ANOVA; p < 0.001 for all frequency bands). Statistically
significant changes in the relative power were evident in
c 1 0 min. The increase in the 8- to 12-, 12- to 16-, and
16- to 30-Hz frequency bands correlated with the plasma
MDL concentration (Spearman correlation coefficients:
8-12 Hz, r = 0.29 and p = 0.015; 12-16 Hz, r = 0.44
and p < 0.001; 16-30 Hz, r = 0.25 and p = 0.037). The
4- to 8-Hz band was not correlated with venous MDL
concentration ( r = 0.03 and p = 0.8180).
Similar results were obtained in the double-blind
phase of the study, although the initial reduction in rela-
tive power in the 8- to 12-Hz band was not clearly iden-
tified (Fig. 3). There were significant increases over time
in the relative power of the EEG in the 8- to 30-Hz
frequency bands in the volunteers who received MDL
(one-way ANOVA for all 8- to 30-Hz frequency bands,
10 15 20 30 45 60 120 300 600
Epilepsia, Vol. 39, No. 3, 1998
2 92 R. C. SCOTT ET AL.

30 i

251
20 ' " ) : 0 .
FIG. 2. Venous midazolam
(MDL) concentration versus rela-
tive power. Increasing MDL con-
centration in serum is associated
with an increase in the relative
power of the faster EEG frequen-
cies: 16-30 Hz (dashed line, x),
12-16 Hz (solid line, solid circle),
8-12 Hz (dotted/dashed line,
' 16-3OHz
__ open diamond).
. 12-16Hz
, 8-12Hz
-5 0 5 10 15 20 25 30 35 40 45

Midazolam Concentration (mcg/L)

p < 0.001). There were no significant changes over time
in the EEG of the volunteers who received peppermint-
flavored water (one-way ANOVA 8-1 2 Hz, p = 0.695 1 ;
12-16 Hz, p = 0.9862; 16-30 Hz, p = 0.7365). TO
determine whether the effects of time on the relative
power of the 4- to 30-Hz frequency band were dependent
on the presence of MDL, we used two-way ANOVA.
There was no significant interaction between time and
drug in the relative power of the 4- to 8-Hz frequency
band (p = 0.6510). The interaction between drug and
time on the relative power of the 8- to 30-Hz frequency
bands was highly significant (p < 0.001).
No clinically significant cardiorespiratory adverse
events occurred. The lowest recorded oxygen saturation
was 83%; it lasted 17 s before returning to normal. Seven
of the 10 volunteers slept, with onset of sleep occurring
between 30 and 45 min after drug administration.
DISCUSSION
Our data demonstrate changes in the EEG within 5-10
rnin of buccal/sublingual administration of MDL. That
MDL was undetectable in venous blood at 5 min in 8 of
TABLE 1. Individual pharmacokinetic parameters in I0
volunteers after buccal/suhlingual administration
of midazolam
Volunteer c,,, (Pgk) T,,, (min)
1 86.7
2 18.9
3 37.2
4 83.3
5 29.0
6 24.3
I 2.5.1
8 21.8
9 21.6
10 42.6
the 10 volunteers may not mean that MDL is slowly
absorbed from buccal/sublingual mucosa. Rapid clear-
ance from the arterial circulation into brain and fat may
result in initial venous blood levels that are lower than
arterial levels. Arterial sampling theoretically would pro-
vide better data but was considered too invasive for this
volunteer study (6). No data relate concentrations of ve-
nous MDL to anticonvulsant effect, but initial EEG
changes occurring in S5-10 min lend support to the
hypothesis that MDL had bound to y-aminobutyric acid,
(GABA,) receptors when the venous blood level had
reached 5 kg/L. Intravenous MDL, through GABA, re-
ceptor agonism, has been shown to be very effective in
the treatment of CSE (7). Power spectrum and voltage
changes in the EEG, similar to those observed in the
present study, have been used as evidence of pharmaco-
dynamic effect of intravenous MDL (8,9). Therefore,
processed EEG signals may be used to assess cerebral
absorption in circumstances in which venous samples
may be misleading and acquiring arterial samples is not
feasible.
Some MDL may have been swallowed, accounting for
the persistent high level of MDL recorded in plasma
from 20 to 60 rnin after administration, but because the
increase during this time did not reach statistical signifi-
cance, most of the MDL probably was absorbed through
the buccal mucosa.
The major EEG changes occurred in the 12- to 16-Hz
tin b i n s ) frequency band. We confined spectral analysis of the
54 201 volunteers' EEGs to the waking state to avoid contami-
41 214
31 200 nation of the 12- to 16-Hz band by sleep spindles. All
59 269 volunteers were alert for the first 20 min of monitoring;
41 213 therefore, the acute changes we identified were probably
41 210
47 I89 a direct result of drug absorption into the CNS. In the
48 203 double-blind phase, there were significant differences
54 203 between the volunteers who received MDL and those
54 306
who received water, with marked increases occurring in

Epilepsia. Vol. 39, No. 3. 1998

 BUCCAL ABSORPTION OF MDL 293

=Log 16-30 Hz

Log 12-16 Hz
FIG. 3. Changes in the relative
power of the EEG with time after
buccal/sublingual water (A) or
buccallsublingual midazolam
(MDL) administration (B) showing
marked increases in the relative
power of the faster EEG frequen-
cies in subjects who received
MDL. The reference line repre-
sents the time of administration of
drug or placebo.

the 8- to 30-Hz bands when active drug was administered
(Fig. 3 ) .
The volunteers had no difficulty holding MDL in the
mouth long enough to demonstrate unequivocal cerebral
effect by EEG. If MDL is swallowed once a seizure has
ceased, the drug will be further absorbed from the gut.
This may have therapeutic benefit by maintaining anti-
convulsant concentrations of MDL for a longer time,
potentially reducing the chance of recurrent seizures. A
child in status epilepticus or having frequent serial sei-
zures might have difficulty holding MDL in the mouth
long enough for it to be absorbed. Subsequent testing in
children having frequent serial seizures showed that buc-
cal/sublingual MDL is effective treatment (10); the drug
was squirted around the mouth and under the tongue in
each child. These preliminary data indicate that buccal/
sublingual MDL administration can be easily achieved in
practice.
Our study, by demonstrating rapid effects on the brain
when MDL is administered by the buccal/sublingual
route, supports the view that such administration of MDL
may offer an alternative to rectal administration of DZP
in the emergency treatment of seizures. A clinical study
to compare the use of buccal/sublingual MDL with that
of rectal DZP is currently in progress.
Acknowledgment: We thank Dr. Robert Surtees for valu-
able help and advice on the manuscript preparation.
REFERENCES
1. Shorvon S. Status epilepticus: its clinica1,fearure.s and treatment in
adults and children. Cambridge: Cambridge University Press,
Cambridge, 1994.
2. Alldredge BK, Wall DB, Ferriero DM. Effect of prehospital treat-
ment on the outcome of status epilepticus in children. Pediatr
Neurol 1995;12:213-6.
3. Lombroso CT. Intermittent home treatment of status and clusters
of seizures. E p i l e p i n 1989;3O(suppl 2):s I 1 4 .
4. De Boer AG, De Leede LG, Breimer DD. Drug absorption by
sublingual and rectal routes. Br J Anaesth 1984;56:69-82.
5. Lehmann B, Boulieu R. Determination of midazolam and its un-
conjugated I -hydroxy metabolite in human plasma by high-
performance liquid chromatography. J Chromatogr [B] Biomed
Appl 1995;674:13842.
6. Donati F, Varin F, Ducharme J, Gill SS, Theoret Y, Bevan DR.

Epilepsia, Vol. 39, N o . 3, I998

294 R. C. SCOTT ET AL.
Pharmacokinetics and pharmacodynamics of atracurium obtained
with arterial and venous blood samples. Clin Pharmacot Ther
I99 I ;49:5 15-22.
7. Rivera R, Segnini M, Baltodano A, Perez V. Midazolam in the
treatment of status epilepticus in children. Crir Care Med 1993;
2 1 :99 1 4 .
X. Buhrer M, Maitre PO, Hung 0, Stanski DR. Electroencephalo-
graphic effects of benzodiazepines. I. Choosing an electroencepha-
lographic parameter to measure the effect of midazolam on
Epilepsia, Voi. 39, No. 3, 1998
the central nervous system. Clin Pharmncol Ther 1990;48:544
54.
9. Aeschbach D, Dijk DJ. Trachsel L, Brunner DP, Borbely AA.
Dynamics of slow-wave activity and spindle frequency activity in
the human sleep EEG: effect of midazolam and zopiclone. Neuro-
psychopharmacology 1994;1 1123744.
10. Scott RC, Neville BGR, Boyd SG, Besag FMC. Nasal rather than
rectal benzodiazepines in the management of acute childhood sei-
zures [Letter in reply]. Deu Med Child NeuroE 1997;39:137-8.