Ventilatory Response to Hypoxia in Humans: Influences of Subanesthetic Desflurane

After approval by the Leiden University Committee on Medical Ethics, 14 healthy, aged 24-30 yr subjects (5 women) without a history of smoking, drug, or alcohol abuse were selected for the study. None had participated in other respiratory studies. Before the studies, subjects participated in a training session to become familiarized with the protocol and apparatus. The magnitude of the hypoxic response was not measured in these training sessions. The subjects were advised not to eat or drink for the 6 h before the study. This procedure was identical to that used in our earlier studies. [2-6].

During the study, the subjects were in a semirecumbent position. An oronasal mask was fitted before the experiment started. The airway gas flow was measured with a pneumotachograph (Fleisch, Lausanne, Switzerland) connected to a differential pressure transducer (model 270, Hewlett Packard, Andovar, MA) and electronically integrated to yield a volume signal. This signal was calibrated with a motor-driven piston pump (stroke volume 11 at a frequency of 20 strokes/min). The pneumotachograph was connected to a T piece. One arm of the T piece received a gas mixture, with a flow of 45 l/min from a gas mixing system that consisted of four mass flow controllers (F201-F203, Bronkhorst High Tec, Veenendaal, The Netherlands) with which the flow of oxygen, carbon dioxide, nitrogen, and desflurane in nitrogen could be set individually at a desired level. Flows were calibrated with flow resistance standards (Godart, Bilthoven, The Netherlands). A Programmable Digital Processor 11/23 microcomputer (Digital Equipment Corporation, Maynard, IA) provided control signals to the mass flow controllers, so that the composition of the inspiratory gas mixture could be adjusted to force the end-tidal oxygen tension (P^ETO²) to follow a specific pattern in time while the end-tidal carbon dioxide tension (P^ETCO²) is kept constant. Part of the nitrogen (5 l/min) passed through the desflurane vaporizer (Tec 6, Ohmeda, West Yorkshire, United Kingdom). During the initial part of the study, the vaporizer was kept in the "off" position.

The oxygen and carbon dioxide concentrations of the inspired and expired gases were measured with a gas monitor (Datex Multicap, Helsinki, Finland) by paramagnetic and infrared analysis, respectively. The gas monitor was calibrated with gas mixtures of known concentrations delivered by a gas-mixing pump (Wosthoff, Bochum, Germany). The desflurane concentration was measured at the mouth with a Datex monitor (Capnomac Ultima, Helsinki, Finland). A pulse oximeter (Datex Satellite Plus, Helsinki, Finland) continuously measured the arterial hemoglobin-oxygen saturation via a finger probe (S^PO²). Throughout the study, the electrocardiograph was monitored. Inspiratory minute ventilation (V^I), tidal volume (V^T), respiratory rate (RR), S^PO², P^ETCO², and P^ETO²were stored on a breath-to-breath basis.

The ventilatory responses to hypoxia at the background of normocapnia (Normocapnic Studies) and the responses at a background of hypercapnia (Hypercapnic Studies) were determined on different morning sessions, at least 1 week apart. Four hypoxic responses were obtained before and during desflurane administration at the following target end-tidal oxygen fractions: 0.10 (P^ETO²approximately 75 mmHg), 0.07 (P^ETO²approximately 53 mmHg), 0.058 (P^ETO²approximately 44 mmHg), and 0.050 (P^ETO²approximately 38 mmHg; Figure 1).

Each session started with a 30-min relaxation period. Thereafter, individual resting P^ETCO²was determined at the end of 10 min of steady-state V^Iwith no inspired carbon dioxide. Subsequently, the P^ETCO²was increased by 2 mmHg in the normocapnic studies and 8 mmHg in the hypercapnic studies. This level was maintained throughout the control and desflurane studies.

Hypoxia was induced with a "dynamic end-tidal forcing" system [9]: steps from normoxia (P^ETO²approximately 110 mmHg) into hypoxia (target P^ETO²level obtained within 4-8 breaths) were applied. Hypoxia was maintained for 3 min, after which normoxia was reintroduced. Between hypoxic challenges there was a 10- to 15-min interlude. For each study (control-normocapnia; control hypercapnia; desflurane-normocapnia; desflurane-hypercapnia), the subjects were allocated randomly to 1 of 24 possible study sequences (Figure 1).

The control studies always preceded the desflurane hypoxic studies. The end-tidal fraction of desflurane was brought to 0.72% (approximately 0.1 MAC) [7]within 1 min by means of an "overpressure" technique. Thereafter a 10- to 15-min equilibration period preceded the hypoxic challenges.

Hypoxic sensitivity--The average V with dot^Iof the last 30 s of each hypoxic challenge was calculated. In addition, normoxic ventilation was determined before the hypoxic challenges longer than 30 s. This yielded 5 V with dot^I-S^PO²data points.

We expressed V with dot^Ias a linear function of arterial oxygen-hemoglobin desaturation: Equation 1where S is the hypoxic sensitivity and V with dot^I(0) the ventilation at normoxia (S^PO²= 98%). The parameters were determined by linear regression of V with dot^Ion (100 - S^PO²). Similar procedures were performed for respiratory rate and tidal volume.

One of the investigators (AD, ES, or MvdE) continuously observed the subjects. During the study, soft music was played, but there was no visual stimulation. Experiments were performed in a normal lighted room that was sealed off from the data acquisition- and end-tidal steering-apparatus.

At the start and end of the studies, we recorded the central nervous system (CNS) arousal state of the subjects by applying a five-point scale:

0 = normal alertness;

1 = drowsy, open eyes;

2 = closed eyes, opened in response to verbal command;

3 = closed eyes, opened in response to touch; and

4 = closed eyes, unarousable.

Data were included for analysis if the subjects were in scale 0 for control and scales 0, 1, or 2 for the desflurane studies. In addition, data were discarded when a subject displayed discomfort during the study.

To detect a significant difference (in the normocapnic and hypercapnic studies) between the control and desflurane studies, a Student's t test was performed on the S, V with dot^I(0), Delta V^T/Delta S^PO², Delta RR/Delta S^PO², respiratory rate at normoxia (RR(O)), and tidal volume at normoxia (V^T(0)). Probability levels < 0.05 were considered significant. All values reported are mean+/-SE.

All fourteen subjects completed the normocapnic studies without problems or discomfort. Two subjects did not attend the hypercapnic session for unknown reasons; three others failed to complete the hypercapnic studies because of nausea and vomiting (two subjects) or discomfort (one subject). There were no signs of airway irritation, such as coughing, from desflurane during the studies.

End-tidal carbon dioxide tension averaged 43+/-0.8 mmHg in the control and desflurane studies. The mean end-tidal desflurane concentration was 0.71+/-0.01% in the desflurane studies. In the control studies, the CNS arousal state of all subjects was 0 (normal alertness), and in the desflurane studies, 4 subjects were at level 1 and 10 were at level 2.

The V with dot^I-S^PO²relation over the S^PO²range tested by us (98 to approximately 70%) was linear (Figure 2). The hypoxic sensitivities obtained from the linear regression of V with dot^Ion (100 - S^PO²) did not differ significantly between control and desflurane studies: 0.45+/-0.07 l *symbol* min sup- 1 *symbol* % sup -1 (control) versus 0.43+/-0.09 l *symbol* min sup -1 *symbol* % sup -1 (desflurane). Inspiratory minute ventilation averaged 11.8+/-0.6 l *symbol* min sup -1 in the control studies and 10.8+/-0.8 l *symbol* min sup -1 in the desflurane studies (not significant). In Figure 3, the hypoxic sensitivities of all subjects are collected in a scatter diagram. The three subjects that had an increase of their hypoxic sensitivity with desflurane had an CNS arousal score of 2. The values of the estimated parameters are listed in Table 1.

The hypoxic experiments were performed at a mean P^ETCO²background of 49+/-1 mmHg, approximately 8 mmHg above resting P sub ET CO²and 6 mmHg above the P^ETCO²level of the normocapnic studies. The end-tidal desflurane concentration averaged 0.72 +/-0.01% in the desflurane studies. In the control studies, the CNS arousal state of all subjects was 0 (normal alertness)l; in the desflurane studies, 4 subjects were at level 1, and 5 subjects were at level 2.

The slopes of the linear regression of V with dot^Ion (100 - S^PO²) decreased from 0.74+/-0.09 l *symbol* min sup -1 *symbol* % sup -1 (control) to 0.53+/-0.06 1 *symbol* min sup -1 *symbol* % sup -1 (desflurane; P = 0.04). In Figure 3, the individual hypoxic sensitivities are plotted. Inspiratory minute ventilation did not differ significantly between treatments: 20.7+/-1.7 1 *symbol* min sup -1 and 20.9+/-1.6 l *symbol* min sup -1 in the control and desflurane studies, respectively. See Table 1for results of all estimates.

In the current study, we observed that subanesthetic desflurane in a group of 14 healthy volunteers did not influence significantly the ventilatory response to isocapnic hypoxia over the range of 110 to 38 mmHg (S^PO²98 to approximately 70%) when experiments were performed against a background of normocapnia. This indicates that desflurane at 0.1 MAC lacks the well described and pronounced effect on the carotid bodies that the other inhalational anesthetic agents (halothane, isoflurane, and enflurane) possess. [2-6,10]Still, the carotid bodies did not remain unaffected by their exposure to 0.1 MAC desflurane, because the ventilatory response to the combination of hypoxia and hypercapnia (asphyxia) was reduced by 30%.

Control of Exhaled Gas Concentrations. To study the ventilatory response to acute hypoxia (i.e., hypoxia < 5 min), we used the computer-driven dynamic end-tidal forcing technique. [9]This allowed us to perform steps in P^ETO²while maintaining a constant P^ETCO²by manipulation of the inspired gas concentrations independently of the ventilatory response. The P^ETCO²was set at 2 mmHg above the resting values in the normocapnic studies and at 8 mmHg above resting in the hypercapnic studies. The reason for the increase in the normocapnic studies is twofold. First, a small increase in inspired carbon dioxide concentration is necessary to control the end-tidal carbon dioxide concentrations adequately. The precision of end-tidal P^CO2control was 0.6 mmHg; that is, the standard deviation of the breath-to-breath P^ETCO²of single periods [A or B] was 0.6 mmHg or less. A similar precision of control was obtained for P^ETO²in periods A and B. Second, this procedure permits the control and drug baseline P^ETCO²'s to be similar, despite differences in level of baseline V with dot^I. Differences in P^ETCO²between control and drug studies make the interpretation of the results difficult, and sometimes impossible, because the additive or synergistic effects of separate stimuli (for example pH, arterial PCO²and PO², or circulating catecholamines) on the measured response differ between treatments and may offset a specific drug effect. Other noncomputer-driven systems also are able to control end-tidal gas concentrations adequately to achieve near steps in P^ETO²at constant P^ETCO²(for example, see [11]).

Mask Versus Mouthpiece. The subjects were attached to the pneumotachograph through an oronasal mask. We prefer a mask above a mouthpiece-noseclip arrangement for two reasons. It allows normal movement of mouth and lips and is less disruptive to normal breathing. The fit of the mask was loose to prevent excessive pressure on the face. However, the combination of mask and pneumotachograph increased the dead space to approximately 200 ml. This, together with the increased P^ETCO²level, explains the higher baseline V with dot^Eof the normocapnic studies compared with studies that do not use a mask and/or inspired carbon dioxide. [1-7].

Corrections for Changes in Inspired Oxygen Concentrations. Because of the changes in oxygen concentration of the inspired air, the viscosity of the inspired gas changes and, consequently, so does the flow measured through the pneumotachograph. As in previous studies, we performed an online correction for these changes in oxygen viscosity. [2-6].

Step Hypoxic Test. We performed 3-min hypoxic steps in this study. It is our experience that peak V with dot^Iis reached in the first 3 min of hypoxia, when the drop in P^ETO²is achieved within 4-6 breaths and the control of P^ETCO²is strict. We tested four levels of hypoxia for two reasons: (1) It allows the study of the hypoxic response to a broad hypoxic range (S^PO²down to approximately 70%); and (2) It reduces the intra-subject response variability. In previous studies, we determined the hypoxic response at a single P^ETO²level (approximately 44 mmHg). [2-6]Analysis of the single step test at P^ETO²approximately 44 mmHg in the current study showed, however, that similar conclusions are drawn from a single test in comparison to the multiple step test. In addition, because a single step test reduces study duration, this test may be preferable above multiple step tests to study the effects of drugs on the hypoxic response.

The obtained ventilatory step responses are the sum of different factors that determine V with dot^Ion exposure to hypoxia: (1) the hypoxic drive from the carotid bodies; [12](2) the depressive effects of "central" hypoxia [12]; and (3) the hypoxic effect on the ventral medullary blood flow that determines the carbon dioxide tension at the site of the central chemo-receptors. [4,6,13]After 3 min of hypoxia specially, factors 1 and 3 are thought to be involved in determining the hypoxic response. Our measured responses, therefore, include some central (depressant) effect due to the increased brain blood flow (BBF) that may not have reached a steady state within 3 min. [13].

The complete recovery from 20 min of hypoxia requires as long as 1 h of room air breathing or 7 min of breathing 100% O². [14]When breathing room air after sustained hypoxia, a subsequent hypoxic response will be smaller in magnitude compared with the first one. This is most probably caused by non-BBF-related hypoxic depression. [14,15]A 3-min hypoxic episode generates hypoxic depression, [16]most of which is due to an increase in ventral medullary blood flow. However, we are unable to exclude an effect of non-BBF-related hypoxic depression on a subsequent hypoxic response in our study. A bias due to this effect was not observed. We relate this to the weak potency of non-BBF-related hypoxic depression generated during 3 min of hypoxia, the random order of the hypoxic levels, and the intra-subject response variability.

Analysis of V with dot I-O2 Data. The relation between V with dot^Iand P^ETO²obtained is nonlinear and were described using either hyperbolic or exponential functions, whereas linear functions were found empirically to describe the V with dot^I-S^PO²relation. [17-19]There is no physiologic reason to select one over the other. However, with a linear response, the sensitivity S should in principle be independent of the level of hypoxia at which it is determined, simplifying the method. In addition, when the number of data points is limited, the linear relation between V with dot^Iand S^PO², with only two parameters to estimate, is preferred. Using exponential, hyperbolic, or other nonlinear functions (see below) increases the number of parameters to estimate. With limited data, this reduces the reliability of estimation.

We observed that the isocapnic semi-steady-state relation between V with dot^Iand P^ETO²was curvilinear (Figure 2). It is of interest to discuss to some extent the linearity of the V with dot^I-S^PO²relation. To the best of our knowledge, there are no studies in which researchers investigated the linearity of the V with dot^I-S^PO²relation when V with dot^Idata points are obtained under (semi)-steady-state conditions. In some of our subjects, we observed a convex relation between V with dot^Iand S^PO²(e.g., subject A of Figure 2). Although not designed for this purpose, we tested the V with dot^I-S^PO²relation for nonlinearity. The data of individual responses were fitted to a second-order polynomial function of the form: Equation 2where H = [100 - S^PO²], epsilon is a constant, beta is a first order parameter, and gamma is a second order parameter. If the data are adequately described by a linear model, the value of gamma is not different from zero. [20]Analysis of all 46 hypoxic responses revealed that nonlinearity occurred in only 6 responses (Figure 4). Although this seems to indicate that a linear function adequately describes the (semi)-steady-state V with dot^I-S^PO²relation (in the S sub P O²range from 98% to approximately 70%), further studies, using multiple responses in individual subjects on one day--to increase the reliability of the estimates--are necessary to clarify this matter. The finding that some of the V with dot^I-S^PO²responses are nonlinear does not affect our main conclusions.

Desflurane Administration. As in previous studies, the drug studies were performed at a target concentration of 0.1 MAC (in the age group 18-30 yr, the MAC of desflurane is 7.2%). [7]Hypoxic studies began after a 10- to 15-min equilibration time at the target end-tidal desflurane concentration. Although the time constant for desflurane brain uptake at a constant end-tidal concentration is not reported, data in rabbits using in vivo nuclear magnetic resonance indicate that desflurane brain uptake is 1.7 times faster than isoflurane brain uptake. [21]The time constant for brain uptake at constant isoflurane concentration is 4 min. [22]Therefore, we estimated the time constant for desflurane brain uptake at approximately 2-2.5 min. The 10-min equilibration period, therefore, constitutes 4 to 5 times the estimated time constant. A similar equilibration period was selected in our previous studies on the effects of subanesthetic isoflurane on ventilatory control. [5,6].

Sequence of Studies. The order of experiments was fixed to control before desflurane studies. We did not want to perform control and drug studies on separate days, because day-to-day variability of the hypoxic responses is greater than within-day variability. [23]A randomized crossover design on one day was not considered, because it leads to long sessions and subject discomfort, a possible influence of a drug experiment on a subsequent control experiment, and an increased run-to-run variability. Because of the structured nature of the study design, we were unable to exclude an order or time effect on the measured responses. However, such an effect favorably outweighs the influences of a randomized crossover design on single or separate days.

The individual desflurane hypoxic sensitivities as fraction of the control responses ranged between 0.02 and 2.0 (mean 0.91+/-0.14; Figure 3). The coefficient of variation was 57%, indicating the "large" intersubject variability. The difference in the outcome of the normocapnic studies compared with our previous findings with halothane and isoflurane is surprising, and may not be related to methodologic differences, because in this investigation, we used identical standards, as were used in previous studies regarding subject selection and preparation, randomization of drug and control experiments, the step hypoxic test, the magnitude of the inspired carbon dioxide concentration during the studies, the CNS arousal state of the subjects, the MAC-fraction studied, and the equilibration time for desflurane brain uptake (see Methods section and [2,3,5,6,24]).

We cannot exclude the fact that our data underestimated the effects of desflurane somewhat (type II error). Eliminating the four outliers from the normocapnic data pool (the subjects with closed symbols in Figure 3: S^DES/S^CON= 0.02, 1.6, 1.6, and 2.0) yielded a significant 13% reduction of the hypoxic response. This reduction is still less than observed with isoflurane and halothane in our laboratory (50% or more reduction of the hypoxic response at 0.1 MAC; Figure 5). [2,6].

Three of the subjects had a normocapnic control response < 0.2 l *symbol* min sup -1 *symbol* % sup -1 (Figure 3). The average control normocapnic hypoxic sensitivity (0.45 l *symbol* min sup -1 *symbol* % sup -1) is low in comparison to one of our earlier studies on subanesthetic halothane (S = 0.9 l *symbol* min sup -1 *symbol* % sup -1). [2]It is possible that a group with a low hypoxic response makes a type II error more likely. For this reason, some investigators select subjects with high hypoxic responses. [11]However, the control responses of the current population volunteers are of similar magnitude in comparison to those of two groups of volunteers in studies on subanesthetic isoflurane (S = 0.54 and 0.44 l *symbol* min sup -1 *symbol* % sup -1). [5,6]In these studies, 0.1 MAC isoflurane caused a reduction of approximately 50% of the hypoxic responses.

Taking into account the above, a different effect of desflurane, compared to halothane and isoflurane, should be taken into consideration (for example, on the carotid bodies or on the CNS).

The ventilatory changes due to a sudden decrease in P^ETO sub 2 (step hypoxic test) are caused by an increase in tidal volume and respiratory rate. Halothane and isoflurane at 0.1 MAC reduce these increases of both ventilatory components. [2,5,6]Desflurane caused a reduction in Delta V^T/Delta S^pO²similar to that observed with isoflurane and halothane (Table 1). This indicates that low-dose desflurane had a depressant effect on the carotid bodies with respect to tidal volume. Conversely, the somewhat higher levels of RR in periods A and B compared with control counteracted the depression of tidal volume such that their product was not significantly different in control and desflurane studies. The sustained or even increased RR in normoxia and hypoxia may be related to stimulation of specific airway, lung, or systemic (for example, at the carotid bodies or within the CNS) receptors. [25]At high inspired concentrations, stimulation of these receptors causes transient increases in sympathetic activity. [8]The integrity of the hypoxic V^Iresponse with low-dose desflurane may be related to the stimulation of these receptors that trigger a central mechanism that counteracts depression of the peripheral chemoreceptors in terms of V^I(for instance, by affecting rythmogenesis or increasing sympathetic activity).

During hypercapnia, desflurane inhalation reduced the hypoxic sensitivities of 8 of the 9 subjects (individual desflurane hypoxic sensitivities as fraction of the control responses ranged between 0.42 and 1.0; mean 0.77+/-0.08; coefficient of variation 34%; Figure 3). Desflurane reduced the carbon dioxide-induced augmentation of the hypoxic ventilatory drive: increase of S from normocapnia to hypercapnia = 65% and 23% in the studies before and during desflurane inhalation, respectively. From our data, we were able to estimate the ventilatory carbon dioxide sensitivities in normoxia and hypoxia. Remember, however, that normocapnic and hypercapnic studies were performed on separate days. Performing experiments on different days increases the run-to-run variability. [23,26]The ratio of the carbon dioxide sensitivity obtained during desflurane inhalation to the control carbon dioxide sensitivity decreased from 1.1 to 0.87 and 0.78 at an S^pO²of 97.5%, 80%, and 70%, respectively. Our findings suggest that desflurane affects the carbon dioxide-induced augmentation of the hypoxic response (which is mediated by the carotid bodies) [9]more than either the hypoxic or hypercapnic responses. This may be clinically of interest because the combination of hypercapnia and hypoxia may occur more frequently in perioperative patients than does hypoxia or hypercapnia alone.

The discrepant findings in the normocapnic and hypercapnic studies suggest that the translation of the oxygen-carbon dioxide interaction into V^Iat the carotid bodies involves a separate mechanism from the hypoxic response at normocapnia or mild hypercapnia. To the best of our knowledge, there are no studies to support this suggestion. Because we do not conceive the hypoxic response at high inspired carbon dioxide concentrations to involve a fundamentally different mechanism from the hypoxic response at low or no inspired carbon dioxide concentrations, we relate the inability to find a significant effect of 0.1 MAC desflurane in the normocapnic studies to a type II error. Clearly, the number of volunteers was too small to detect a small effect of subanesthetic desflurane on the hypoxic response. Inflation of the hypoxic response by carbon dioxide unearthed the depressant influence of desflurane.

In conclusion, we report that, in contrast with the results on subanesthetic halothane and isoflurane, subanesthetic desflurane has only a small effect on the normocapnic ventilatory response to hypoxia in healthy volunteers. In common with the other inhalational anesthetic agents, subanesthetic desflurane decreases the ventilatory response to asphyxia.

The authors thank Dr. John W. Severinghaus for suggestions concerning the design of the study and for critical reading of the manuscript.