Full text of DOI:10.1007/BF03019871

1216
REPORTS OF INVESTIGATION
Flumazenil abolishes
midazolam-induced
increase in the work of
nasal breathing
Yasuko Kawauchi MD,*
Tsutomu Oshima MD PhD,†
Yuhji Saitoh MD PhD,*‡
Hidenori Toyooka MD PhD §
Purpose: To evaluate the effects of midazolam sedation followed by flumazenil antagonism on the work of nasal
breathing in normal humans.
Methods: We measured minute ventilation through the nasal route, respiratory frequency, nasal resistance (Rn)
and the work of nasal breathing under three conditions: awake, during midazolam sedation, and after flumazenil
antagonism in eight healthy human subjects. A custom-made, partitioned face mask enabled nasal and oral airflow
to be measured separately. To calculate Rn and the work of nasal breathing, nasal mask and oropharyngeal pres-
sure was also measured.
Results: Total resistive work spent on the upstream segment of the nasal route per minute (Wn) (J·min–1) was
greater during midazolam sedation (3.6 ± 2.9) than while awake (1.6 ± 0.9) and after flumazenil antagonism (1.7
± 0.6), respectively (mean ± SD) (P < 0.05). Total resistive work spent on the upstream segment of nasal
breathing (Wn/V nE) (J·L–1) increased from 0.31 ± 0.14 to 0.75 ± 0.61 after midazolam administration (P < 0.05)
and decreased to 0.31 ± 0.10 after flumazenil. Following midazolam administration, a strong correlation was
observed between changes in Wn/VnE and changes in Rn r = 0.852, P < 0.0001), whereas there was no corre-
lation between changes in Wn and changes in Rn r = 0.159, P = 0.279).
Conclusion: The work of breathing spent on the upstream segment of the nasal route increases during mida-
zolam sedation and returns to baseline after flumazenil antagonism.
Objectif : Évaluer les effets sédatifs du midazolam dont l’administration est suivie de celle d’un antagoniste, le
flumazénil, sur le travail ventilatoire nasal chez des humains normaux.
Méthode : Nous avons mesuré la ventilation-minute par voie nasale, la fréquence respiratoire, la résistance
nasale (Rn) et le travail ventilatoire nasal sous trois conditions : à l’état d’éveil, pendant une sédation avec du mida-
zolam et après l’action antagoniste du flumazénil, chez huit sujets en santé. L’utilisation d’un masque cloisonné,
fabriqué sur mesure, a permis la mesure séparée de l’écoulement d’air nasal et oral. On a aussi mesuré la pres-
sion du masque nasal et de l’oropharynx pour calculer la Rn et le travail ventilatoire nasal.
Résultats : Le travail de résistance total effectué, par minute, sur le segment en amont de la voie nasale (W n)
(J·min–1) est plus important pendant la sédation au midazolam (3,6 ± 2,9) qu’à l’état d’éveil (1,6 ± 0,9) et
qu’après le flumazénil (1,7 ± 0,6), respectivement (moyenne ± écart type) (P < 0,05). Le travail de résistance
–1
total en amont de la respiration nasale (Wn/VnE) (J·L ) augmente de 0,31 ± 0,14 à 0,75 ± 0,61 après l’adminis-
tration de midazolam (P < 0,05) et diminue à 0,31 ± 0,10 après celle du flumazénil. Avec le midazolam, une
forte corrélation est observée entre les changements de Wn/VnE et ceux de la Rn (r = 0,852, P < 0,0001), tan-
dis qu’il n’y a pas de corrélation entre les changements de Wn et ceux de la Rn (r = 0,159, P = 0,279).
Conclusion : Le travail ventilatoire effectué sur le segment en amont de la voie nasale augmente pendant la
sédation au midazolam et reprend les valeurs de base sous l’action antagoniste du flumazénil.
From the Department of Anesthesiology and Critical Care Medicine, School of Medicine, Tokyo Medical and Dental University, Tokyo,*
Gifu University School of Medicine, Gifu,† Fukushima Medical University, Fukushima,‡ and University of Tsukuba, Ibaraki,§ Japan.
Address correspondence to: Tsutomu Oshima MD, Department of Anesthesiology and Critical Care Medicine, Gifu University School of
Medicine, 40 Tsukasamachi, Gifu-City, Gifu 500-8705, Japan. Phone: 81-58-267-2295; Fax: 81-58-267-2961; E-mail: oshimat@cc.gifu-u.ac.jp
This study was supported in part by Grant-in-Aid 12671460 from the Ministry of Education, Science, and Culture, Tokyo, Japan.
Accepted for publication September 6, 2000.
CAN J ANESTH 2000 / 47: 12 / pp 1216–1219

Kawauchi et al.: NASAL BREATHING WORK
1217
H E velopharynx, that portion of the
for 10 sec. The dead space for nasal and mouth cham-
bers was 90 and 120 ml, respectively. Nasal airflow,
nasal mask pressure (Pnm), and Pop were monitored
with a C-P pulmonary monitor (model C-P 100,
BICORE monitoring system, USA) connected to the
breathing circuit. Oral airflow was measured using a
Fleisch No. 2 pneumotachograph.
After inserting a 22-Gauge catheter in the left
cephalic vein, we started an intravenous infusion of
acetated Ringer’s solution at 100 ml·hr– 1. Oxygen satu-
ration, ECG, and noninvasive blood pressure measure-
ments were recorded. After determining baseline
respiratory measurements, we began midazolam
administration: 0.01 mg·kg–1 by rapid (< 1 sec) intra-
venous injection, administered every minute, until sub-
jects reached a consciousness level where they exhibited
spontaneous eye closure and responded only to prod-
ding or shaking. This approach to assessing conscious-
ness level has been previously described.3,6,7 We made
respiratory measurements three minutes after the last
dose of midazolam. We then gave 0.5 mg flumazenil iv
over five minutes, and made respiratory measurements
three minutes after flumazenil administration.
Nasal resistance (R) was calculated by dividing the
transnasal pressure (nPnm-Pop) by inspiratory flow
when Pop was least during inspiration. At this time,
upper airway dimensions are minimal.8 Resistive work
was measured for six consecutive breaths at each peri-
od studied. Total resistive work spent on the upstream
segment of the nasal route per minute (Wn) was
obtained by multiplying work per breath by the corre-
sponding respiratory frequency. Total resistive work
spent on the upstream segment per litre of ventilation
through the nasal route (Wn/VnE) was obtained by
dividing Wn by minute ventilation through the nasal
nasopharynx bounded by the soft palate, is
the most common site of narrowing during
1
T
sedation, which increases upper airway
resistance.2–3 However, a change from nasal to oral
breathing occurs if the velopharynx is occluded.4
Therefore, the final result of sedation on the work of
breathing spent on the nasal route is difficult to predict
and has not been previously reported. Furthermore, no
data are available regarding the effects of flumazenil on
midazolam-induced increase in the work of breathing.5
The present study was performed to elucidate the
effects of midazolam sedation followed by flumazenil
antagonism on the work of breathing through the
nasal route in normal humans.
Methods
The study protocol was approved by the Ethics
Committee of Tokyo Medical and Dental University.
Five male and three female healthy volunteers, aged
27.5 3.9) yr (mean SD), participated in this study.
The height and weight were 168.0 7.0 cm and 60.2
8.3 kg, respectively. As demonstrated by low BMI
– 2
(21.3 1.8 kg·m , mean SD), none was obese. After
obtaining written informed consent from each subject,
we conducted a history and physical examination.
Exclusion criteria were clinical evidence of nasal, pul-
monary, cardiac, or CNS disease, obesity, use of cen-
trally acting medications within the previous week, a
history of smoking, sleep disordered breathing, recent
upper respiratory infections, drug or alcohol abuse.
Subjects refrained from oral intake for eight hours, and
from alcohol and caffeine for 24 hr before the study.
Each subject had one nasal passage anesthetized
with 0.05 ml lidocaine 8% spray. No nasal deconges-
tants were used. Oropharyngeal pressure (Pop) was
measured using a balloon catheter (Millar, o.d. 1.7
mm) introduced through the anesthetized nostril.
The tip of the catheter was located just caudal to the
inferior margin of the soft palate and rostral to the
tongue in the oropharynx. The position was verified
by direct visual examination of the posterior orophar-
ynx. No migration of the oropharyngeal balloon
catheter was observed throughout the study.
route. The latter was obtained by multiplying the V
T
and respiratory frequency (f) measured on the same
breaths used to compute the work of breathing. For
statistical analysis, Scheffe’s multiple comparison test
after analysis of variance was used to determine if dif-
ferences existed among the five parameters: V , f, R,
nE
n
Wn, and Wn/VnE under the three conditions. A signif-
icance level of 0.05 was used.
An airtight custom-made, partitioned face mask
(Senkousha, Japan) was fitted while the subject rested
in the supine position. A hard rubber septum separat-
ed the mask into nasal and oral chambers, each with its
own port (i.d. 9 mm, length 40 mm) and a side-arm
outlet (i.d. 3.5 mm). The rubber septum was posi-
tioned above the upper lip. Each mask chamber was
tested separately for air leaks by pressurizing to 10
cmH2O and ensuring that the pressure was constant
Results
In all the subjects sedated with midazolam, inspirato-
ry airflow limitation was sometimes observed. After
midazolam administration, four subjects continued
breathing though the nasal route and the other four
exhibited oronasal breathing. As shown in the Table,
Rn was greater during midazolam sedation than while
awake and after flumazenil antagonism (P < 0.05).
Furthermore, Wn, and Wn/ V were greater during
nE

1218
CANADIAN JOURNAL OF ANESTHESIA
decrease in nasal airflow could contribute to the lack
of correlation between changes in Wn and changes in
Rn while V exhibited no change (Table).
TABLE Changes in respiratory measurements during wakeful-
ness, midazolam-induced sedation, and after antagonism with
flumazenil. Mean SD values for minute ventilation through the
nasal route (V ), nasal resistance (R), total resistive work spent
on the upstream segment of the nasal route per minute (Wn), and
total resistive work spent on the upstream segment per litre of
Patients nwE ith obstructive sleep apnea have obstruc-
tion at various sites of the upper airway during
sleep.10,11 Even normal subjects may have narrowing at
different sites during sedation.12 If the oropharynx
and/or hypopharynx are narrowed during midazolam
sedation, the resistive work spent on the downstream
segment should increase, thereby resulting in a decrease
n E
n
ventilation through the nasal route (Wn/V ). Median for respira-
n E
tory frequency (f). (A) Before midazolam administration. (B) After
midazolam administration. (C) After flumazenil administration. *P
< 0.05 compared with A, C. See text for details.
VnE
f
Rn
Wn
Wn/V
(J·L )
n E
– 1
– 1
– 1
– 1
(L·min ) (breaths·min ) (cmH2O·L– 1 s) (J·min )
in Wn and Wn/ V . Therefore, narrowing at different
nE
sites of the upper airway during midazolam sedation
may contribute to the discrepancy between changes in
Wn and changes in Rn. However, further studies are
needed to demonstrate this hypothetical phenomenon.
In the previous studies conducted by Montravers et
al.,2,5 the subject was allowed to breathe exclusively
through the nasal route by sealing the lips with tape.
In the present study, the subject was allowed to
breathe without limitation of the breathing route.
However, our partitioned face mask, which enables
nasal and oral airflow to be measured separately, exerts
a considerable pressure on the orofacial region.3,4,7 We
cannot exclude the possibility that our results may be
affected under these highly artificial circumstances. A
limitation of the present study is the small number of
subjects showing each type of response to midazolam
sedation followed by flumazenil antagonism.
Confirmation of these observations in a large number
of subjects is necessary.
A
B
C
5.3 2.0 1 3
5.0 2.4 1 7
5.4 1.2 1 5
2.8 4.0
33.6 47.9* 3.6 2.9* 0.75 0.61*
2.7 2.5 1.7 0.6 0.31 0.10
1.6 0.9 0.31 0.14
(n = 8)
midazolam sedation than while awake and after
flumazenil antagonism, respectively (P < 0.05). After
midazolam administration, a strong correlation was
observed between changes in Wn/ VnE and changes in
Rn r = 0.852, P < 0.0001). However, there was no
correlation between changes in Wn and changes in R
r = 0.159, P = 0.279).
n
Discussion
The Wn and Wn/VnE were greater during midazolam
sedation than while awake and after flumazenil antago-
nism, respectively (P < 0.05). The present study is the
first to demonstrate that the work of breathing spent on
the upstream segment of the nasal route increases dur-
ing midazolam sedation, and is reversed by flumazenil
We conclude that the work of breathing spent on
the upstream segment of the nasal route increases dur-
ing midazolam sedation and returns to baseline after
flumazenil antagonism.
antagonism. The changes in R are consistent with our
n
previous study.3 Since the tip of the oropharyngeal
catheter was located just below the inferior margin of
the soft palate, anatomical structures associated with R,
Wn and Wn/ VnE included the nasal passage rostral tno
the inferior margin of the soft palate.
References
1 Mathru M, Esch O, Lang J, et al. Magnetic resonance
imaging of the upper airway. Effects of propofol anes-
thesia and nasal continuous positive airway pressure in
humans. Anesthesiology 1996; 84: 273–9.
2 Montravers P, Dureuil B, Desmonts JM. Effects of i.v.
midazolam on upper airway resistance. Br J Anaesth
1992; 68: 27–31.
During midazolam sedation, inspiratory airflow
limitation sometimes occurs.7 This leads to remarkably
non-linear pressure-airflow relationships within the
upper airway.9 Consequently, measurement of Wn and
Wn/ V as well as R may be crucial to reveal the
nE
n
midazolam-induced upper airway behaviour.
3 Oshima T, Masaki Y, Toyooka H. Flumazenil antago-
nizes midazolam-induced airway narrowing during
nasal breathing in humans. Br J Anaesth 1999; 82:
698–702.
4 Nishino T, Sugiyama A, Tanaka A, Ishikawa T. Effects
of nasal anaesthesia on shift of breathing route in
adults. Lancet 1992; 339: 1497–500.
A strong correlation between changes in Wn/ V
nE
and changes in R following midazolam administra-
tion is not surprisinng because these quantities depend
upon the same variables for their derivation. However,
there was no correlation between changes in Wn and
changes in R. During midazolam sedation, four sub-
n
jects exhibited oronasal breathing. The shift from
nasal to oral breathing should allow nasal airflow to
decrease. Following midazolam administration, a
5 Montravers P, Dureuil B, Molliex S, Demonts JM.
Effects of intravenous midazolam on the work of
breathing. Anesth Analg 1994; 79: 558–62.

Kawauchi et al.: NASAL BREATHING WORK
1219
6 Gross JB, Weller RS, Conard PF. Flumazenil antagonism
of midazolam-induced ventilatory depression.
Anesthesiology 1991; 75: 179–85.
7 Kawauchi Y, Oshima T, Suzuki S, Saitoh Y, Toyooka H.
Advancement of the mandible facilitates nasal breath-
ing in human subjects sedated with intravenous mida-
zolam. Can J Anesth 2000; 47: 215–9.
8 Henke KG, Dempsey JA, Kowitz JM, Skatrud JB. Effects
of sleep-induced increases in upper airway resistance on
ventilation. J Appl Physiol 1990; 69: 617–24.
9 Hudgel DW, Hendricks C, Hamilton HB. Characteristics
of the upper airway pressure-flow relationship during
sleep. J Appl Physiol 1988; 64: 1930–5.
10 Morrison DL, Launois SH, Isono S, Feroah TR,
Whitelaw WA, Remmers JE. Pharyngeal narrowing and
closing pressures in patients with obstructive sleep
apnea. Am Rev Respir Dis 1993; 148: 606–11.
11 Hudgel DW. Variable site of airway narrowing among
obstructive sleep apnea patients. J Appl Physiol 1986;
61: 1403–9.
12 Hudgel DW, Hendricks C. Palate and hypopharynx-
sites of inspiratory narrowing of the upper airway dur-
ing sleep. Am Rev Respir Dis 1988; 138: 1542–7.