The ventilatory response to hypoxia is composed of the stimulatory activity from peripheral chemoreceptors and a depressant effect from within the central nervous system. Morphine induces respiratory depression by affecting the peripheral and central carbon dioxide chemoreflex loops. There are only few reports on its effect on the hypoxic response. Thus the authors assessed the effect of morphine on the isocapnic ventilatory response to hypoxia in eight cats anesthetized with alpha-chloralose-urethan and on the ventilatory carbon dioxide sensitivities of the central and peripheral chemoreflex loops.
The steady-state ventilatory responses to six levels of end-tidal oxygen tension (PO2) ranging from 375 to 45 mmHg were measured at constant end-tidal carbon dioxide tension (P[ET]CO2, 41 mmHg) before and after intravenous administration of morphine hydrochloride (0.15 mg/kg). Each oxygen response was fitted to an exponential function characterized by the hypoxic sensitivity and a shape parameter. The hypercapnic ventilatory responses, determined before and after administration of morphine hydrochloride, were separated into a slow central and a fast peripheral component characterized by a carbon dioxide sensitivity and a single offset B (apneic threshold).
At constant P(ET)CO2, morphine decreased ventilation during hyperoxia from 1,260 +/- 140 ml/min to 530 +/- 110 ml/ min (P < 0.01). The hypoxic sensitivity and shape parameter did not differ from control. The ventilatory response to carbon dioxide was displaced to higher P(ET)CO2 levels, and the apneic threshold increased by 6 mmHg (P < 0.01). The central and peripheral carbon dioxide sensitivities decreased by about 30% (P < 0.01). Their ratio (peripheral carbon dioxide sensitivity:central carbon dioxide sensitivity) did not differ for the treatments (control = 0.165 +/- 0.105; morphine = 0.161 +/- 0.084).
Morphine depresses ventilation at hyperoxia but does not depress the steady-state increase in ventilation due to hypoxia. The authors speculate that morphine reduces the central depressant effect of hypoxia and the peripheral carbon dioxide sensitivity at hyperoxia.
It is well known that morphine and all other opioids influence ventilatory control: Arterial carbon dioxide tension increases and the ventilatory response to inspired carbon dioxide is reduced or shifted to higher end-tidal carbon dioxide tension (PETCO2) values. [1–5] The scarce data reported on the hypoxic response indicate a reduction by opioids in awake humans. [3–5] The steady-state ventilatory response to hypoxia is the result of stimulation by the peripheral chemoreceptors of the carotid and aortic bodies and a central (i.e., residing within the central nervous system) depressant effect. [6–9] This depressant effect in anesthetized cats and probably also in anesthetized humans is related to the washout of carbon dioxide at the site of the central chemoreceptors due to the increase of brain blood flow and to central inhibitory neurotransmitters. [8,9] The magnitude of the stimulatory and depressant components or an effect a drug has on either component cannot be determined in human or “intact” animal studies. In anesthetized animals, elaborate invasive studies (such as those using artificial perfusion of the brain stem or peripheral chemoreceptor denervation) should be performed to obtain information on these opposing ventilatory components.
In the current study, we investigated the effects of intravenous morphine on the steady-state ventilatory responses to step-wise changes in end-tidal oxygen concentration (PETO2) against the background of isocapnia in anesthetized cats. We also determined the ventilatory response to carbon dioxide and its central and peripheral components using the dynamic end-tidal forcing (DEF) technique).  The combination of parameters (e.g., carbon dioxide sensitivities of the peripheral and central chemoreflex loop, apneic PETCO2threshold, steady-state hypoxic ventilatory sensitivity, hyperoxic ventilation) obtained in one group of intact cats will allow us to infer the influence of morphine on the two hypoxia-related ventilatory components without performing invasive procedures.
Materials and Methods
Experiments were performed on eight anesthetized female adult cats (body weight 2.6–3.8 kg). The use of the animals was approved by the Ethical Committee for Animal Experiments of the University of Leiden. Anesthesia was induced with 15 mg/kg ketamine hydrochloride given intramuscularly, followed by halothane inhalation. The right femoral vein was cannulated; 25 mg/kg alpha-chloralose and 125 mg/kg urethan were slowly administered intravenously, and the volatile anesthetic was withdrawn. About 1 h later, an infusion of an alpha-chloralose-urethan solution was started at a rate of 1 or 2 mg [center dot] kg sup -1 [center dot] h sup -1 alpha-chloralose and 5–10 mg [center dot] kg sup -1 [center dot] h sup -1 urethan. This regimen leads to conditions in which the level of anesthesia is sufficient to suppress the pain-withdrawal reflex but light enough to preserve the corneal reflex. Comparison of our studies with those of others in awake cats shows that this anesthetic regimen has little effect on the ventilatory response to elevated PETCO2compared with the awake state and does not yield systematic changes over time (more than 6 h). [10–13]
The oxygen and carbon dioxide responses were studied with the DEF method before and after the administration of morphine. Because this method has been described previously, we restrict ourselves to a brief description.  In the DEF technique, the end-tidal oxygen and carbon dioxide concentrations are forced to follow a specific pattern in time (in the current study, step increases and decreases in PETO2and P sub ET CO2) independent of the changes in ventilation. This is performed by computer-driven adjustment of the inspired carbon dioxide and oxygen concentrations. The ventilatory response after a prescribed change in PETCO2or PETO2was assessed on a breath-by-breath basis.
To measure inspiratory and expiratory flow, the trachea was cannulated at midcervical level and connected via a Fleisch (Lausanne, Switzerland) number 0 flow transducer head to a T-piece of which one arm received a continuous gas flow of 5 l/min. With the aid of three computer-steered mass flow controllers (High-Tec, Veenedaal, The Netherlands), a prescribed composition of the inspirate from pure oxygen, carbon dioxide, and nitrogen could be obtained. The respiratory fractions of oxygen and carbon dioxide were measured continuously with a fast-responding zirconium oxide cell Jaeger, O2-test, Wurzburg, Germany) and an infrared analyzer (Gould Godart MK-2 capnograph, Bilthoven, The Netherlands). Temperature was controlled to within 1 degree Celsius in each cat and ranged among cats from 36.1 to 38.2 degrees Celsius.
All signals were recorded on polygraphs and processed by a PDP 11/23 minicomputer (sample frequency, 100 Hz). Tidal volume, breathing frequency, ventilation (V with dotI), and end-tidal carbon dioxide and oxygen were determined using the minicomputer and stored on a breath-by-breath basis. For monitoring purposes during the experiment, averages for 20 breaths of ventilation, end-tidal gas tensions, and arterial pressure were calculated, displayed on a monitor, and stored on computer disk.
In each cat, the hypercapnic and hypoxic ventilatory responses were measured before and after the intravenous administration of 0.15 mg/kg morphine hydrochloride. Morphine studies were performed from about 30 min after the administration when ventilation had stabilized.
Carbon Dioxide Studies. Hypercapnic experiments consisted of P sub ET CO2challenges of approximately 7–10 mmHg during normoxia (P sub ET O2controlled at 110 mmHg). After a period of steady-state ventilation during which PETCO2was slightly increased above resting values, PETCO2was increased in a stepwise manner and kept constant for about 7 min. Thereafter, the PETCO2was decreased to its original value and kept constant for another 7 min. In each cat, two or three control and two or three morphine studies were obtained. In the eight cats, we obtained 22 control and 21 morphine responses.
Oxygen Studies. In the hypoxic experiments, the PETO2was forced according to the following pattern: 1) 300 mmHg, 2) 52 mmHg, 3) 113 mmHg, 4) 60 mmHg, 5) 42 mmHg, 6) 188 mmHg, and back to 300 mmHg. Each oxygen level was maintained for about 6 min. To be able to perform the hypoxic control and morphine studies at identical PETCO2levels, the end-tidal carbon dioxide concentration was fixed at approximately 41 mmHg, which is about 6 mmHg higher than the normal value in cats. Two control and two morphine studies were performed in each cat.
Oxygen Studies. The hypoxic response of each cat was evaluated by taking mean values of ventilation and PETO2of the last 20 breaths of each oxygen level. Using a least-squares method, ventilation was fitted to PETO2according to the exponential function :Equation 1in which G is the overall hypoxic sensitivity, D is a shape parameter, and A is the ventilation during hyperoxia. Especially going from hyperoxia to hypoxia and when ventilation is depressed by morphine, several breaths are required before the target PETO2was approached. Therefore, we only present results of the dynamics of the response for steps out of hypoxia, which can be performed more quickly. The ensemble average of the transitions from hypoxia to normoxia and to hyperoxia were determined by indexing on the time of the step change in P sub ET O2and with linear interpolation between breaths at 3-s intervals.
Carbon Dioxide Studies. The steady-state relation of ventilation to PETCO2at constant PETO2in the cat is linear down to the PETCO2axis and well described by Berkenbosch et al.  and by DeGoede et al. Equation 2.
The parameters Scand Spare the central and peripheral ventilatory carbon dioxide sensitivities, and the offset B represents the apneic threshold or extrapolated PETCO2of the steady-state ventilatory response to carbon dioxide at zero ventilation.
To estimate Sc, Sp, and B, we used the dynamic response of the ventilation and a two-compartment model as described previously. [1,12] We estimated all the parameters of the model simultaneously by fitting the data of each carbon dioxide study with a least-squares method.
To compare the control and morphine group, we performed a paired t test on the parameters of the oxygen response. An analysis of variance with a two-way layout using a fixed model was performed on the estimated parameters of all the individual carbon dioxide studies. The level of significance was set at 0.05.
(Figure 1) shows the recording of a step transition from hyperoxia to hypoxia and back to normoxia. It took several breaths before the target hypoxic PETO2was reached. The ventilatory response shows nearly no overshoot. The transition from hypoxia to normoxia occurs within two or three breaths. Nevertheless, the ventilatory response is relatively slow and shows a slight undershoot. Not all cats respond with a clear over- or undershoot in the response. In this sample of cats, we found a manifest over- or undershoot in two cats. In Figure 2, the ensemble average of the responses from hypoxia to normoxia and to hyperoxia is shown. It shows that although the step out of hypoxia occurs in about 6 s, the ventilatory response is relatively slow. The control responses show a slight undershoot, which was no longer visible after administration of morphine.
(Figure 3) shows a representative example of the steady-state ventilatory response curves to oxygen of one cat. It illustrates the general finding that after morphine administration, hyperoxic ventilation is significantly depressed. However, the increase in ventilation by hypoxia is about the same, so that the response is approximately parallel and displaced to lower ventilation levels. This is also illustrated by Figure 4, which shows the difference in ventilation (Delta VI) between control and morphine experiments at each oxygen level of all cats. Analysis of variance revealed that Delta VIwas not significantly different at each PETO2level (P = 0.65). The mean steady-state ventilation and its components tidal volume and breathing frequency are shown in Table 1. In Figure 5, scatter diagrams of the hypoxic sensitivity (A), the shape parameter (B), and the ventilation at hyperoxia (C) are shown. The mean of the estimated parameters of the fits to Equation 1are summarized in Table 2. Only the ventilation at hyperoxia (parameter A) decreased significantly from 1,260 +/- 140 ml/min to 530 +/- 110 ml/min.
Carbon Dioxide Studies
Twenty-two control studies and 21 morphine studies were obtained. Administration of morphine shifted the carbon dioxide response curve to higher end-tidal carbon dioxide values and decreased the slope (i.e., total ventilatory carbon dioxide sensitivity [STOT]). The value of parameter B increased significantly after morphine by approximately 6 mmHg (Table 3). The central (Sc) and peripheral (S sub p) carbon dioxide sensitivities decreased by about 30%, causing the ratio of Spto Scnot to differ between treatments. Mean values of the parameters are collected in Table 3.
Our study in anesthetized cats showed that although morphine depressed hyperoxic ventilation significantly by about 60%, the increase in ventilation due to isocapnic hypoxia was not diminished. This is reflected by the finding that the hypoxic sensitivity and the shape parameter are not significantly changed (Table 1). Comparative data from the literature are scarce. Weil et al.  and Santiago et al.  investigated the effects of morphine in awake humans. They consistently found a diminished response to hypoxia. Using a rebreathing technique, Weil et al.  observed a reduction of the hypoxic ventilatory sensitivity (defined by the shape parameter H of the hyperbolic function VE= H [center dot][PETO2- 32] sup -1) by more than 50% 1 h after subcutaneous administration of 7.5 mg morphine (rebreathing was performed for 8–10 min). Santiago et al.  used a steady-state technique and determined the hypoxic ventilatory sensitivity after 5 min of hypoxia.  They found that the hypoxic sensitivity (defined as Delta VI/Delta [arterial oxygen saturation]) was 30% of control after a 0.2 mg/kg dose of morphine given intramuscularly. Gross et al.  studied the influence of another micro-agonist, alfentanyl, on hypoxic ventilatory sensitivity (defined as Delta VI/Delta [arterial oxygen saturation]; oxygen saturation was decreased to about 70% for 2 or 3 min). They observed a reduction of 20% of the hypoxic sensitivity during alfentanyl administration. As indicated in these studies, different procedures to measure the ventilatory response to hypoxia were used. The results of these studies are difficult to interpret and to compare with our data for other reasons, too. In the studies of Weil et al.  and Santiago et al.,  the end-tidal or arterial PCO2in the morphine studies was higher compared with control studies (more than 2–5 mmHg). Further, in all three studies, steady-state ventilation may not have been reached after 5–10 min of hypoxia. Recent studies have shown that in awake humans it takes 15–20 min before the hypoxic depressant effect on ventilation is fully developed and steady-state ventilation is reached.  In the anesthetized cat, steady-state ventilation after induction of isocapnic hypoxia is attained after about 6 min.  The evident contrast in central nervous arousal states (awake vs. anesthetized) and species difference may hinder a comparison of our data with those of Weil et al.,  Santiago et al.,  and Gross et al. 
In the awake human and the anesthetized cat model, the isocapnic hypoxic ventilatory response is the result of two opposing effects: Hypoxia stimulates ventilation by an action on the peripheral chemoreceptors and depresses ventilation via an effect within the central nervous system. [6–8] The depressant effect of hypoxia in the anesthetized cat is related to an increased wash-out of acid metabolites from the brain compartment due to an increase in brain blood flow [8,19,20] and probably also to the release or synthesis of inhibiting neurotransmitters such as gamma-aminobutyric acid.  However, at moderate levels of hypoxia, as also applied in this study, we could not reduce the central depressant effect of hypoxia by blocking the gamma-aminobutyric acid receptor with bicuculline.  Furthermore, other researchers report that gamma-aminobutyric acid concentrations only increase at severe hypoxic levels. 
Because it is reasonable to assume that the development of central depression of ventilation starts immediately on exposure of the brain stem to hypoxia, the magnitude of the ventilatory output from the peripheral chemoreflex loop will a fortiori be underestimated in studies of intact bodies. A recent study by Berkenbosch et al.  using the technique of artificial brain stem perfusion showed in anesthetized cats that the ventilatory response of the peripheral chemoreceptors to a stepwise change in PETO2contains a fast and a slow component.  In humans, with the termination of a 1-min hypoxic episode, ventilation also shows a fast and a slow component.  The existence of this slow component in the anesthetized cat explains the variability that is commonly observed in the time course of the overall hypoxic ventilatory response.  Depending on the magnitude and time constant, the slow component of the peripheral chemoreceptor response can mask the central depressant effect of hypoxia, preventing a biphasic appearance of the response. This was the case in six of the cats in this study. This indicates that separation of the measured hypoxic ventilatory responses into peripheral and central components is impossible from studies in intact animals. Because the overall hypoxic sensitivity (parameter G in Equation 1) estimated in our study is the result of the two opposing effects of hypoxia (peripheral stimulating and central depressant), it provides little or no information on if and how morphine affected either of these.
To increase our understanding of how morphine influenced the hypoxic ventilatory response, we used the results obtained in carbon dioxide studies. Our carbon dioxide studies showed, corresponding to an earlier study in anesthetized cats,  that the main depressant effect of morphine on ventilation was by an effect on the apneic threshold. Due to this increase in apneic threshold, we had to perform the control oxygen studies at a relative hypercapnic PETCO2. The peripheral and central carbon dioxide sensitivities were depressed by an equal percentage (approximately 30%). As a consequence, the ratio of these sensitivities (S sub p/Sc) did not differ in the treatments (Table 2). Previously we reasoned that the change in apneic threshold we observed was not due to an effect of morphine on the chemoreceptors themselves, although we cannot entirely exclude such an effect.  The finding that the peripheral and central carbon dioxide sensitivities are depressed to the same extent supports the idea that the depressant effect of morphine is mainly on the neuronal structures common to the peripheral and central chemoreflex pathway; that is, the respiratory centers within the brain stem rather than the peripheral and central chemoreceptors per se. Because signals from the peripheral chemoreceptors due to hypoxia and hypercapnia are transported to the brainstem via identical pathways, we would expect, based on our carbon dioxide studies, that the response of the peripheral chemoreflex loop to hypoxia also will be depressed by morphine. We cannot exclude the possibility that the peripheral carbon dioxide sensitivity is depressed somewhat more than the peripheral hypoxic sensitivity by morphine. This is explained by an effect of morphine on peripheral carbon dioxide sensitivity at hyperoxia, which is an appreciable part of the total peripheral normoxic carbon dioxide sensitivity in the anesthetized cat. 
Our remarkable observation that the overall hypoxic sensitivity is not diminished suggests that the counteracting central depressant effect of hypoxia was lessened by morphine. This may be due to a diminished synthesis or release of inhibiting neuromodulators. However, we believe that in anesthetized cats the depressant effect of moderate hypoxia on ventilation is due to an increase in brain blood flow. [8,19] Our results then suggest that morphine reduces brain blood flow reactivity to hypoxia at constant PETCO2. Morphine decreases resting brain blood flow coupled to a reduction of brain metabolism.  To the best of our knowledge, no data are available on the influence of morphine on brain blood flow reactivity to isocapnic hypoxia.
Do our data apply to humans, and if so, what is their clinical importance? There are now several examples of striking similarities between feline and human ventilatory responses to isocapnic hypoxia in the awake and anesthetized states (for example, see  and ). Not only do the responses appear similar in shape, but the underlying mechanisms (peripheral stimulation and central depression) probably have identical origins. It is interesting to note that the central nervous system arousal states (i.e., awake vs. anesthetized or sedated) are more important in causing differences in the response shape and mechanisms than are species differences. 
Application of our results to anesthetized (or sedated) persons breathing spontaneously then suggests that intravenous morphine reduces hyperopic, normoxic, and hypoxic ventilation. The ability of the respiratory control system to increase ventilation in response to imposed hypercapnic loads is reduced but not abolished. Evidently, patients will remain (or resume) breathing only if their resting PETCO2is above the apneic threshold, which in our study was increased by about 25%. The ventilatory response to hypoxia would be altered by morphine in the sense that steady-state hypoxic sensitivity will remain identical to control, but the level of ventilation is diminished at all PO2levels.
Obviously, several caveats should be stated before extrapolating our results to humans. First, all agents used for general anesthesia or sedation affect ventilatory control, and an interaction between these drugs and opioids also may occur. [24,25] Although the general anesthetic used in our preparation (alpha-chloralose-urethan) does not seem to affect resting PETCO2and the slope of the carbon dioxide response curve compared with the awake cat, [10,11] an interaction of this particular anesthetic regimen (or the central nervous system arousal state it induces) and morphine is not unlikely. Second, during and after surgery, additional respiratory drives (noxious stimulation, shivering, verbal, and tactile stimulation by recovery room personnel) play an important role in the balance between respiratory depression and stimulation.  Third, we successfully opened the feedback loop with the dynamic end-tidal forcing technique. In patients, the feedback loop remains intact, and variations in PETCO2occur due to morphine alone or due to ventilatory responses. If we take this into account, our conclusions about the effect of morphine on ventilatory control as well as the mechanisms involved are valid for anesthetized cats and probably also for humans. However, extrapolation of our results to the perioperative setting should be made with caution because several factors that we have not studied should be considered.