INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The carbonic anhydrase inhibitor acetazolamide is used in patients with pulmonary disease and sleep-disordered breathing syndromes and in the prevention and treatment of acute mountain sickness. Improvement of blood gas values in patients with chronic obstructive pulmonary disease (COPD) is known to occur especially in cases with a metabolic alkalosis related to the use of steroids and diuretics (1). The beneficial effect of acetazolamide in these patients is probably primarily due to an increase in ventilatory drive secondary to a metabolic acidosis induced by effective inhibition of renal carbonic anhydrase (7, 8). The enzyme, however, is also present in the central nervous system (glial cells, and possibly also neurons), in red cells, peripheral chemoreceptors, muscle cells, and in the endothelium of capillaries supplying brain, muscles, and lungs (5, 9, 10).

Because of its physical-chemical properties, acetazolamide does not easily cross the blood-brain barrier (9, 11). However, penetration of acetazolamide into erythrocytes and other peripheral tissues may result in effective inhibition of local carbonic anhydrase, even if the drug is administered in low doses. Effective local inhibition (i.e., > 99% enzyme inhibition) can already be achieved at low concentrations of acetazolamide (< 5 mg/kg) (9, 12). Small variations in dose, particularly in the lower range, may thus give rise to very complicated pharmacodynamics of the drug and to a large variation in respiratory effects depending on the degree of local enzyme inhibition in peripheral tissues. For example, a low intravenous dose of acetazolamide (4 mg/kg, i.e., a dose not causing effective inhibition of red cell carbonic anhydrase) in the cat causes a reduction in the CO₂ sensitivities of both the peripheral and central chemoreflex loops by 30% (13). In the same animal preparation, a large dose (50 mg/kg) even results in a total loss of the ability of the ventilation to respond adequately to changes in arterial oxygen tension (14, 15). The effect of low-dose acetazolamide on the ventilatory CO₂ sensitivity in humans is less clear. Several investigators have studied the effect of clinical doses of the drug on the ventilatory CO₂ response curve. Although it is a common finding that acetazolamide increases resting ventilation, its reported effects on the CO₂ response slope vary from no change (6, 16) or an increase (4, 17, 18) after chronic application, to a decrease of the CO₂ sensitivity during hypoxia after acute administration (18). Differences in dose regimens and methodology to determine the slope of the CO₂ response curve (e.g., steady state methods versus rebreathing) may account for these variable study outcomes.

For patients with COPD but also for sojourners at high altitude, adequate functioning of the peripheral chemoreceptors is important in determining the chemical drive to breathe. It is of particular interest therefore to document the effect of clinical doses of acetazolamide on the carbon dioxide sensitivity of the peripheral chemoreceptors. A useful means to study the influence of drugs on the ventilatory CO₂ response curve is to apply square-wave changes in end-tidal PCO₂ (while keeping the end-tidal PO₂ constant) and to analyze the ventilatory responses with a two-compartment model comprising a fast and slow component (19, 20). These two components are represented by the peripheral and central chemoreflex loops, each of which is characterized by a CO₂ sensitivity and a single offset, respectively. This technique enables one to separate the effects of drugs on the peripheral and central chemoreflex loops from each other. In this study we applied this technique to examine the effects of a clinical dose of acetazolamide (i.e., 250 mg by way of mouth, every 8 h for 3 d) on both the peripheral and central chemoreflex loops in healthy volunteers.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects and Test Study

Twelve healthy, nonsmoking, subjects, 4 women and 8 men, age 21 to 34 yr, were recruited to participate in the protocol approved by the Leiden University Committee on Medical Ethics, after giving their informed consent. All subjects performed a series of test carbon dioxide studies to familiarize them with the apparatus and the experimental procedure and to test their suitability to perform respiratory studies. This procedure is similar to that of some of our previous studies (e.g., 21). After these test studies, 11 subjects were asked to participate in the placebo-controlled acetazolamide study. One male subject felt uncomfortable during the test studies and refrained from further participation. Two other subjects (two males) did not return for unknown reasons. Subjects were asked to refrain from stimulants and depressants (for example: alcohol, coffee, tea, chocolate) from at least 72 h prior to the studies.

Study Design

The acetazolamide study was placebo-controlled with a double-blind, randomized design. All nine subjects (4 women, 5 men) were studied on two sessions (a drug session and a placebo session), at least 2 wk apart. Before one session, the subjects took 250 mg acetazolamide at 8-h intervals for 72 h (3 d). Before the other session, the subjects took 250 mg placebo (cellulose) at 8-h intervals for 72 h. Immediately after the 3-d drug or placebo intake, resting steady-state values for end-tidal PCO₂ (PET_CO2) and PO₂ (PET_O2) and ventilation (I) were determined, after which 2 to 4 ventilatory carbon dioxide responses were obtained. Subsequently, after returning to the resting steady-state situation, blood was drawn from the capillary bed of a hyperemic finger for determination of the acid-base status of our subjects. This was performed as follows. The samples were tonometered with 4% and 8% CO₂, respectively, and the pH values were read (Astrup equilibration technique; BMS Mk2 Blood Micro System, Gas Mixing Apparatus GMA 1 and Acid-Base Analyzer pHM72; Radiometer, Copenhagen, Denmark). This enabled us to determine the subjects' buffer (Astrup) lines, which were plotted in a Siggaard-Andersen curve nomogram. Because resting end-tidal PCO₂ values were determined at each session, we were able to read the subjects' arterial pH from the nomogram (assuming no essential difference between end-tidal and arterial PCO₂). Subsequently, resting arterial pH and PCO₂ values were used to calculate the subjects' arterial bicarbonate concentrations and base excess values using the Siggaard-Andersen alignment nomogram. To study the ventilatory response to carbon dioxide, we used the dynamic end-tidal forcing technique. With this technique, we are able to force the end-tidal gas tensions to follow a prescribed pattern in time by manipulating the inspired gas concentrations independently of the ventilatory responses (19). In this study we performed steps in end-tidal carbon dioxide tension (PET_CO2) against a background of constant normoxia (end-tidal oxygen tension [PET_O2] = 110 mm Hg). The PET_CO2 pattern was as follows. After a 10- to 15-min period of steady-state ventilation (I), without any inspired carbon dioxide, the PET_CO2 was raised above resting values by 2 to 3 mm Hg. After another 15 min, the PET_CO2 was increased by 7.5 to 13 mm Hg in a step-wise fashion and kept constant for 8 min. Subsequently, the PET_CO2 was returned to its original value and maintained for another 8 min. The maximal number of studies obtained was four per session.

Apparatus

Subjects were seated in a comfortable position and breathed through a face mask (Vital Signs, Totowa, NJ). The inspired and expired gas flows were measured with a pneumotachograph connected to a differential pressure transducer (Hewlett-Packard, Andover, MA) and electronically integrated to yield a volume signal. The volume signal was calibrated with a motor-driven piston pump. The subjects received a gas mixture with a flow of 45 L/min from a gas mixing system consisting of three mass flow controllers (Bronkhorst High Tec, Veenendaal, The Netherlands) through which the flow of O₂, CO₂, and N₂ could be set individually at a desired level. A PDP mini computer (Digital Equipment Co., Maynard, IA) provided control signals to the mass flow controllers so that the composition of the inspired gas mixture could be adjusted to obtain the desired PET_CO2 and PET_O2. The O₂ and CO₂ concentrations of inspired and expired gas were measured with a Datex gas monitor (Multicap, Helsinki, Finland). The gas monitor was calibrated with gas mixtures of known concentrations. A monitor (Satellite Plus; Datex) continuously measured the arterial hemoglobin-oxygen saturation from pulse oximetry (Sp_O2) with a finger probe and the electrocardiogram (ECG). The following variables were stored on disc on a breath-to-breath basis for further analysis: PET_CO2, PET_O2, Sp_O2, I, tidal volume (VT), and respiratory rate (f).

Data Analysis

The steady-state relation of I to PET_CO2 at constant PET_O2 in humans is described by:

(1)

where G_p = the carbon dioxide sensitivity of the peripheral chemoreflex loop, G_c = the carbon dioxide sensitivity of the central chemoreflex loop, and B = the apneic threshold or extrapolated PET_CO2 at zero I. The sum of G_p and G_c is the total carbon dioxide sensitivity or Gtot.

For the analysis of the dynamic response of ventilation to a stepwise change in PET_CO2 we used a two-compartment model (19, 20):

(2)

(3)

where _p and _c = the time constants of the peripheral and central chemoreflex loops, respectively, _p(t) and _c(t) = the outputs of the peripheral and central chemoreflex loops, respectively, PET_CO2[t  T_p] = the stimulus to the peripheral chemoreflex loop delayed by the peripheral transport delay time (T_p), and PET_CO2[t  T_c] = the stimulus to the central chemoreflex loop delayed by the central transport delay time (T_c).

To model _c of the ventilatory on transient (i.e, the central time constant of the ventilatory response to a step increase in PET_CO2 = _on) to be different from the ventilatory off-transient (i.e., the central time constant of the ventilatory response to a step decrease in PET_CO2 = _off), _c is written as:

(4)

where x = 1 when PET_CO2 is high and x = 0 when PET_CO2 is low.

It is our experience that some subjects show a small (i.e., < 50 ml · min²) positive or negative drift in ventilation. We therefore included a drift term (C · t) in our model. The total ventilatory response [I(t)] is made up of the contributions of the central and peripheral chemoreflex loops, C · t, and a measurement white noise term [W(t)]:

(5)

The parameters of the model were estimated by fitting the model to the breath-to-breath data with a least-squares method. To obtain optimal time delays a "grid search" was applied, and all combinations of T_p and T_c, with increments of 1 s and with T_p  T_c, were tried until a minimum in the residual sum of squares was obtained. The minimum time delay was chosen, arbitrarily, to be 1 s, the _p was constrained to be at least 0.3 s.

In order to obtain information on resting I, VT, f, PET_CO2, PET_O2, and Sp_O2, steady-state data points were obtained when inspired CO₂ was zero. The data points were averaged over 10 breaths.

Statistical Analysis

To detect the significance of difference between the treatments (acetazolamide versus placebo), a two-way analysis of variance was performed on parameters B, G_p, G_c, Gtot, _P, _on, _off, T_c, and T_p using a fixed model. A Student paired t test was performed on resting I, resting PET_CO2, resting PET_O2, resting Sp_O2, base excess, pH, calculated HCO₃, and slopes of buffer lines. Probability levels < 0.05 were considered significant. All values are mean ± SD, unless otherwise stated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Resting Conditions

The effects of acetazolamide on resting ventilation, end-tidal gas pressures, and acid-base status are shown in Figure 1 and summarized in Table 1. The means (± SD) of the slopes of the Astrup buffer lines in nine subjects were 1.74 ± 0.18 after placebo and 1.97 ± 0.31 after acetazolamide (p > 0.13). In all subjects these Astrup lines were displaced to lower pH values (within the physiological pH range), indicating metabolic acidosis. Base excess decreased from 0.08 ± 1.2 to 7.48 ± 2.07 mmol · L¹. Ventilation increased after intake of acetazolamide resulting in a decrease in end-tidal PCO₂ and a rise in PO₂ (Table 1). In Figure 2 the relationship between the resting end-tidal PCO₂ and arterial bicarbonate concentration is shown for all nine subjects. The mean PET_CO2/HCO₃ for all subjects was 0.90 ± 0.39 mm Hg · mmol¹ · L¹.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Changes in base excess, resting end-tidal PCO2, and resting ventilation after acetazolamide intake. *Significantly different from placebo.

View larger version (14K): [in this window] [in a new window]   Figure 2.   Relationship between resting end-tidal PCO2 and arterial bicarbonate concentration in nine subjects. Each subject is represented by a separate symbol. Placebo condition and situation after acetazolamide intake are represented by closed and open symbols, respectively. The mean (± SD) slope of the relationship in the nine subjects was 0.90 ± 0.39 mm Hg · mmol1 · L1.

Dynamic Ventilatory Response to Step Changes in End-tidal PCO₂

Altogether, 50 dynamic end-tidal forcing (DEF) runs were analyzed in nine subjects. An example of DEF experiments in one subject after placebo and acetazolamide is shown in Figure 3. The results of the parameter estimations in all subjects are summarized in Table 2. The data show that the only significant change after acetazolamide intake consisted of a decrease in the apneic threshold (x-intercept of the ventilatory CO₂ response curve) from 33.2 ± 3.9 to 25.9 ± 7.1 mm Hg. The time constants, delays, and CO₂ sensitivities of both the peripheral and central chemoreflex loops were not altered by acetazolamide. These results imply that the effect of acetazolamide was to shift the ventilatory CO₂ response curve to lower PCO₂ values, without changing the slope.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Examples of individual DEF runs in one subject after placebo and acetazolamide intake, respectively. Upper traces are end-tidal PCO2. Breath-by-breath ventilatory data are represented by small open circles. The solid line through these actual ventilatory data is the model output. c and p are the calculated contributions of the central and peripheral chemoreflex loops, respectively. CO2 sensitivities of both chemoreflex loops are similar after placebo and acetazolamide. In both the placebo and acetazolamide condition, 2 to 3 DEF runs were performed in all subjects.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main result of this study is that chronic acetazolamide administration (250 mg every 8 h, during 3 d) did not change the carbon dioxide sensitivities of the peripheral and central chemoreflex loops in healthy subjects, but caused a parallel shift of their CO₂ response curve to the left. Mean resting PET_CO2 in our subjects decreased by approximately 7 mm Hg.

An important finding was the unchanged carbon dioxide sensitivity of the peripheral chemoreflex loop after acetazolamide. Normal activity of the peripheral chemoreflex loop is also indicated by the fact that the observed decrease in PCO₂ in our subjects can be entirely attributed to the ventilatory compensation of the metabolic acidosis, for which the peripheral chemoreceptors are of primary importance (22). Analysis of the acid-base status of our subjects showed a parallel shift of the logPCO₂-pH buffer line to the left, and a decrease in mean base excess of 7.40 mmol · L¹, indicating metabolic acidosis. Normal subjects without chest wall and lung impairments have a linear chronic ventilatory response to metabolic acid- base derangements over a range of arterial bicarbonate concentration of 20 mmol · L¹, with a mean slope of 0.9 mm Hg · mmol¹ . L¹ (22). Our results fit very well in this scenario, because we found a mean PET_CO2/HCO₃ of 0.90 ± 0.39 mm Hg · mmol¹ · L¹. Effective inhibition of carbonic anhydrase in the peripheral chemoreceptors of the cat reduces carotid body output and steady-state CO₂ sensitivity in vivo (23, 24) and abolishes the hypoxic ventilatory response (15). In the in vitro cat carotid body baseline activity and the speed of response appear to be affected (25). Our finding of an unchanged peripheral time constant and CO₂ sensitivity thus suggests that the dose in which acetazolamide was given, was insufficient to cause effective (i.e., > 99%) inhibition of carotid body carbonic anhydrase. Higher doses, however, or different dose regimens or routes of administration resulting in different pharmacokinetics may lead to (temporal) carotid body inhibition. This, for example, is illustrated by the observation that acute intravenous infusion of 500 mg in normal subjects inhibits the CO₂-O₂ interaction and also the ventilatory response to hypoxia (18), a finding similar to the inhibiting effects of low doses in the cat (13). For clinical (and high-altitude) applications, it is recommendable therefore not to use higher doses than necessary to block renal carbonic anhydrase (8).

The speed and magnitude of the CO₂ response of the central chemoreflex loop remained also intact during acetazolamide. An important factor determining the slope of the CO₂ response curve is cerebral blood flow regulation (26). It has been reported that clinical doses of acetazolamide do not result in measurable changes in baseline cerebral blood flow (27). Whether this is also true for the cerebral blood flow response to CO₂ remains to be verified experimentally. Central chemoreceptors may contain carbonic anhydrase (28, 29). In clinical doses, however, acetazolamide will not penetrate into the brain in sufficient amounts to cause effective inhibition of central nervous system (CNS) carbonic anhydrase (9, 11, 12). Selective inhibition of the CNS enzyme results in an increase in central CO₂ sensitivity (10, 30).

The unchanged dynamics of the ventilatory response to step changes in PCO₂ suggests that the kinetics of the CO₂ (de)hydration reaction in the blood was not substantially altered by acetazolamide. Slower (or much slower) pH dynamics would have slowed or even masked the fast peripheral component in the ventilatory response. Large doses of inhibitor (resulting in effective erythrocytic enzyme inhibition) will considerably slow CO₂/H₂CO₃ equilibration in the blood (9). This would lead then to slower ventilatory dynamics after a step change in end-tidal PCO₂ (31) not because of blocking CNS carbonic anhydrase, but rather because of slower stimulus dynamics secondary to slow kinetics of the CO₂ (de)hydration reaction.

One point concerning the usually applied methods to assess the effect of metabolic acidosis on ventilatory CO₂ sensitivity merits further discussion. Application of (modified forms of) the Read rebreathing method is not preferable, because during acidosis this technique yields a carbon dioxide sensitivity which exceeds steady-state sensitivity considerably (26, 32). The reason is that, in contrast to the steady-state method, the CO₂ response slope becomes larger when the magnitude of the initial step increase in end-tidal CO₂ increases (26, 33). Because, at least in normal subjects, the initial end-tidal PCO₂ after acetazolamide will be lower than in the control situation, an increase in response slope may then reflect a change in the initial experimental condition rather than an effect of the agent on ventilatory CO₂ sensitivity. Therefore, our results are difficult to compare with those obtained with rebreathing methods (17, 34). Finally, because many patients with COPD have arterial-to-end-tidal PCO₂ differences, it is preferable in clinical studies to use the arterial PCO₂ as the independent variable for the CO₂ response curve, rather than the end-tidal PCO₂. First, acetazolamide may alter this difference by blocking lung carbonic anhydrase (35) or by causing sufficient partial inhibition of the erythrocytic enzyme, particularly at higher doses. Second, arterial-to-end-tidal PCO₂ differences will change with changing levels of ventilation.

Use of Acetazolamide in Patients with COPD

Acetazolamide has been shown to be an efficient pharmacological tool to lower plasma bicarbonate concentration and arterial PCO₂ in COPD patients with severe metabolic alkalosis and hypoventilation-induced hypercapnia (1, 6). For these patients it is essential to be able to lower their PCO₂ (and increase their Pa_O2) with minimal extra muscular effort. A simple approach of the ventilatory system consisting of a controlled system (represented graphically by the metabolic hyperbola for CO₂ in the left hand panel of Figure 4 and a controller (visualized by the CO₂ response curve) can be helpful to envisage how acetazolamide may achieve this in some of these patients.

View larger version (9K): [in this window] [in a new window]   Figure 4.   (Left-hand panel ) Metabolic hyperbola and CO2 response curves before and after acetazolamide intake in a hypercapnic patient (hypothetical). Points A and B are the operating points before and after acetazolamide, respectively. The agent causes a leftward shift of the CO2 response curve, similar to that in healthy individuals. At point B, after acetazolamide intake, arterial PCO2 is substantially lower with only a minor increase in ventilation. (Right-hand panel ) Metabolic hyperbola and CO2 response curves before and after acetazolamide in healthy subjects. Operating points A and B are resting conditions in nine healthy subjects after placebo and acetazolamide intake, respectively, in nine healthy subjects. The CO2 response curves shown are based on data summarized in Table 1. A decrease in PCO2 now results from a substantial larger rise in ventilation than in the left-hand panel.

We consider a closed loop situation, i.e., a resting condition of an air-breathing patient with an arterial PCO₂ of 60 mm Hg (the operating point A in Figure 4, left hand panel ). Note the relatively high value of the y-asymptote of the metabolic hyperbola in the left hand panel of Figure 4 owing to abnormally high dead space ventilation, and the relatively flat CO₂ response curve, which is a common observation in COPD patients. If acetazolamide in this patient has a similar effect on base excess as we found in the subjects of our study, the CO₂ response curve would shift to lower values of PCO₂, without a change in slope. Initially this yields the new operating point B, implying a decrease in Pa_CO2 of approximately 7 mm Hg with hardly any measurable change in ventilation. Thus, acetazolamide would not act as a ventilatory stimulant but rather as an effective acidifying agent.

The actual situation would be somewhat more complicated because, owing to the initial hypoventilation, the patient was not only hypercapnic but also hypoxic. Because the metabolic hyperbola for oxygen is relatively flat in the hypoxic range (as is the case for the metabolic hyperbola for CO₂ in the hypercapnic range), the minor increase in ventilation will be sufficient to cause a considerable rise in arterial PO₂ (unless this is prevented by the presence of many lung regions with extremely low ventilation-perfusion ratios). This would have a difficult-to-predict effect on the CO₂ response curve: on the one hand, due to CO₂-O₂ interaction in the carotid bodies provided they function properly in the patientthe curve would become less steep and move somewhat rightward, and on the other hand the depression of ventilation caused by chronic hypoxia (the so-called hypoxic ventilatory depression) would be diminished, causing a shift of the I-CO₂ response curve to the left.

Whatever the exact mechanisms leading to the final (minor) change in ventilation, the essential point is that if the operating point of patients is within the relatively flat portion of the metabolic hyperbola (for CO₂ and O₂), treatment with acetazolamide will be more efficient than when this point lies on the steep portion of the curve. A decrease in Pa_CO2 of approximately 20 mm Hg in severe hypercapnic patients with COPD after acetazolamide as reported by Miller and Berns (2) may serve as an illustrative example in this context. A possible substantial improvement of blood gas values with acetazolamide in hypercapnic patients with COPD might thus be predicted by the effect of mild voluntary hyperventilation on the arterial blood gases, rather than by the outcome of FEV₁ scores. This is supported by data of Vos and coworkers (4), who showed that in patients with COPD of whom 89% were able to lower their Pa_CO2 upon voluntary hyperventilation, acetazolamide caused an improvement of blood gases without a significant rise in ventilation. Skatrud and Dempsey (6) reported data on "noncorrectors" to acetazolamide who were unable to lower their Pa_CO2 upon hyperventilation. Normal renal function would also be a prerequisite for an improvement of blood gas values by acetazolamide, and the ventilatory control system should be able to detect (and respond to) changes in base excess.

If the operating point of a subject lies on a less flat portion of the metabolic hyperbola, acetazolamide would cause a larger increase in ventilation, because the operating point is expected to move into an even steeper portion of the curve. This is illustrated by the present findings in healthy volunteers, who showed a decrease in mean PET_CO2 of approximately 7 mm Hg upon a rise in mean ventilation of approximately 2 L · min¹ (see Figure 4, right panel ). For some categories of patients with COPD similar, or larger, increases in ventilation might demand too much muscular effort and discomfort.

In summary, in our view an essential improvement of blood gas values with acetazolamide in hypoxic (and hypercapnic) COPD patients with severe metabolic alkalosis may be expected in those patients with normal renal and carotid body functions who are able to substantially lower their Pa_CO2 upon mild volitional hyperventilation. An effect on their Pa_O2 will depend on the occurrence of lung regions with very low ventilation-perfusion ratios. Possible inhibiting effects of acetazolamide on the peripheral and central chemoreflex loops may be prevented by using no higher doses than necessary to block renal carbonic anhydrase. In healthy volunteers we found that 250 mg each 8 h (for 3 d) did not impair the CO₂ sensitivities of both the peripheral and central chemoreflex loops.