© 2003 American Thoracic Society Hypoxic Respiratory Response during Acute Stable HypocapniaDepartment of Respiratory Medicine, University of Manitoba, Winnipeg, Manitoba, Canada Correspondence and requests for reprints should be addressed to Stephen Corne, RS 318-810 Sherbrook Street, Respiratory Hospital, Winnipeg, MB, Canada R3A 1R8. E-mail: scorne{at}hsc.mb.ca
The hypoxic ventilatory response during hypocapnia has been studied with divergent results. We used volume-cycled ventilation in spontaneously breathing normal subjects to study their hypoxic ventilatory response under conditions of stable hypocapnia. Subjects were studied at three different levels of end-tidal (partial) carbon dioxide pressure (PETCO2), eucapnia and 6 and 12 mm Hg below eucapnia (mild and moderate hypocapnia, respectively). The response to hypoxia was assessed by changes in muscle pressure output (Pmus) and respiratory rate. Compared with the Pmus response at eucapnia (0.53 ± 0.59 cm H2O/percentage oxygen saturation [% O2sat]), the response at mild hypocapnia was attenuated (0.26 ± 0.33 cm H2O/% O2sat), whereas the response at moderate hypocapnia was negligible (0.003 ± 0.09 cm H2O/% O2sat). Similar reductions were seen with the respiratory rate (eucapnia, 0.17 ± 0.2 breaths/minute/% O2sat; mild hypocapnia, 0.11 ± 0.11 breaths/minute/% O2sat; moderate hypocapnia, 0.01 ± 0.06 breaths/minute/% O2sat). The Pmus and respiratory rate responses at the three levels of PETCO2 were significantly different (p < 0.05, analysis of variance). The responses at moderate hypocapnia were not significantly different from zero. We conclude that when apnea occurs under conditions in which central PCO2 is well below the CO2 setpoint, subjects are at risk of developing dangerous hypoxemia due to absence of a hypoxic ventilatory response.
Key Words: control of breathing • rebreathing • hyperventilation • diving Acute stable reduction in PaCO2 is not uncommon under physiologic and clinical conditions. Subjects may hyperventilate for several minutes before diving to increase underwater time. Patients placed on mechanical ventilation may be exposed to ventilator settings that reduce PaCO2 well below their pre-existing spontaneous levels. In addition, patients with chronic hypercapnia secondary to severe upper airway obstruction (e.g., severe obstructive apnea) may develop an acute decrease in PaCO2 after relief of the obstruction with a tracheostomy. During acute hypocapnia, the respiratory drive is influenced minimally (1) or not at all (2–5) by changes in PaCO2. With diving, apnea is produced voluntarily, whereas in the previously noted clinical situations, apnea may develop spontaneously, particularly during sleep, where in the absence of hypoxemia, rhythmic breathing is entirely dependent on the CO2 drive (6, 7). Evolution of a hypoxic drive during such apneas may, under some conditions, be critical for survival. For example, increasing hypoxic drive would force the diver to resurface or may be necessary for the patient to resume spontaneous breathing in the event of a ventilator malfunction (e.g., disconnect) or when a central apnea develops after sleep in recently tracheostomized previously hypercapnic patients. The ventilatory response to hypoxia is augmented when PCO2 is increased above eucapnia (8–10). Although it is clear from previous human studies that the hypoxic response decreases during hypocapnia, there are uncertainties about the magnitude of this depression and, in particular, whether there is a PCO2 level below which hypoxic response is lost and, if so, what is this level. Thus, some studies demonstrated persistence of a vigorous response at end-tidal (partial) carbon dioxide pressure (PETCO2) up to the mid 20s (e.g., 8), whereas others reported that hypoxic response disappeared at PETCO2 levels only a few mm Hg lower than eupneic PCO2 (11). In previous human studies, the extent of central hypocapnia, at the time of the hypocapnic hypoxic challenge, was uncertain and likely varied considerably depending on the experimental approach used, thereby possibly explaining the divergent results (see DISCUSSION). Patrick and colleagues (1) recently described an approach that results in acute stable hypocapnia of adjustable magnitude. The subject is connected to a volume-cycled ventilator in the strictly assist mode (all ventilator cycles are subject triggered). As the VT setting is increased, rhythmic respiratory efforts continue at nearly the same rate. Acute stable hypocapnia with maintained respiratory efforts results. The magnitude of hypocapnia is directly related to the set VT . Under these conditions, inspired gas concentrations can be manipulated to effect surreptitiously desired changes in alveolar gas tensions. Respiratory responses to these changes are assessed from changes in respiratory rate (RR) and respiratory muscle pressure output (Pmus). In this study, we used this approach to study hypoxic response at constant levels of reduced PCO2, where the change in central PCO2 can be estimated with reasonable confidence. We felt that the results obtained under conditions of stable central PCO2 should provide additional insights into the nature of interaction between CO2 and O2 drives and may help reconcile the differences between results obtained using other approaches in which central PCO2 was not constant.
We studied eight normal nonsmoking subjects, four males and four females, ranging in age from 22 to 39 years. The protocol was approved by the Human Ethics Committee of the University of Manitoba, and informed written consent was obtained from all subjects. We used volume-cycled ventilation to create acute stable hypocapnia, as previously described (1). The experimental setup (Figure 1) is similar to that described previously for mechanical unloading during CO2 rebreathing (12). The ventilator was set in the assist control mode, with a backup rate of two, such that all breaths were subject triggered. A t-piece with two unidirectional valves connected subjects to an inspiratory and expiratory circuit. A three-way valve allowed us to control the proportion of inspiratory gas that passed through a CO2 absorber, thereby permitting adjustment of the concentration of CO2 in the inspired gas (FICO2) in the range of zero (all gas directed through the absorber) to an upper range determined by the CO2 concentration of the rebreathing bag. A pneumotach measured inspiratory VT and flow. Airway pressure and PETCO2 were monitored with a pressure transducer and mass spectrometer, respectively. Oxygen saturation (O2sat) was monitored with a finger oximeter. All measurements were recorded at 125 Hz using data acquisition software (Windaq; Dataq Instruments, Akron, OH).
Eupneic PETCO2 was established for each subject during spontaneous ventilation through a mouthpiece and the pneumotach (apparatus disconnected). Subjects were then attached to the circuit, and the tidal volume was adjusted for comfort. At this time, oxygen was added to the expiratory line to maintain normoxia. VT was then progressively increased with the goal of lowering PETCO2 to 12 mm Hg below the baseline level. Inspiratory flow was increased in parallel to preserve the mechanical inspiratory time of the ventilator. FICO2 for this trial was zero. Once a stable target PETCO2 was achieved, the oxygen supply to the bag was shut off to create progressive hypoxia. The trial was terminated when O2sat fell to 80%. The same procedure was repeated twice, but at two different levels of PETCO2, eucapnia and 6 mm Hg below eucapnia. In these two trials, VT was initially set at the level used in the first trial (  PCO2  -12 mm Hg), and PETCO2 was adjusted to the target level by increasing FICO2. Particularly in the eucapnic trials, several subjects requested a higher volume than what they were satisfied with in the hypocapnic trials. The process of adjusting ventilator settings and FICO2 to attain both comfort and the desired PCO2 consumed 10–15 minutes. In the last 5–10 minutes of these, manipulations of FICO2 and ventilator settings consisted of a minor fine-tuning so that PETCO2 changed little. This period was used to obtain baseline measurements.
After the three trials, measurements of respiratory system elastance and resistance were made using the end-inspiratory hold technique (13). The respiratory response to hypoxia was assessed by measuring both changes in RR and Pmus. The latter was necessary because the ventilatory response was in part determined by the ventilator settings for tidal volume, not by the subject. Pmus was determined using the equation of motion: Pmus = (elastance x volume) + (resistance x flow) - airway pressure (Paw) (14, 15). VT, RR,
The mean respiratory system elastance and resistance were 14.1 ± 2.8 cm H2O/L and 5.4 ± 1.0 cm H2O/L/second, respectively. The response in one subject is illustrated in Figure 2 . The time necessary to lower O2sat to 80% in this example, between 4 and 5 minutes, was typical of the trials in the eight subjects. Note that at the lowest level of PETCO2 (Figure 2, top panel), there is no discernable increase in Pmus in response to hypoxia.
Table 1 shows the ventilatory and Pmus data at baseline preceding the three hypoxic trials. Baseline PETCO2 was 42.8 ± 3.3, 36.8 ± 3.8, and 30.7 ± 2.6 mm Hg, corresponding to eucapnia, mild hypocapnia, and moderate hypocapnia, respectively. VT was slightly, but significantly, higher during eucapnia than at lower PETCO2 levels. This is because some subjects requested a higher volume when CO2 was titrated up from the initial hypocapnic level to eucapnia. As a result,  E was also higher during baseline in the eucapnic trials. There were no significant differences among the baseline values of RR or Pmus. Baseline O2sat (not shown in Table 1) was 98.3 ± 1.2, 97.8 ± 1.0, and 97.9 ± 1.1 mm Hg during baseline at eucapnia and mild and moderate hypocapnia, respectively (no significant difference).
Table 1 shows the corresponding end-hypoxic values. The values are those obtained at the lowest O2 saturation common to all three runs in the same subject. The lowest common O2sat was 82.6 ± 0.9%. Under eucapnia, Pmus increased significantly (81% on average). The increase in RR was relatively smaller (15% on average) and less consistent. E increased, in part because of the higher RR and in part because airway pressure did not increase as much during the inflation phase on account of the increase in Pmus. Thus, less of the air delivered by ventilator was used to compress the gas in the box (Figure 1), and more went to the subject. The tendency for PETCO2 to decrease as a result of the increase in E was not always perfectly offset by increasing FICO2 so that PETCO2 decreased in some subjects in the course of hypoxia. The decrease in PETCO2 was, on average, not significant.
The changes, from baseline, with hypoxia during mild hypocapnia were qualitatively similar, although less pronounced (Table 1). Changes in
There were no significant changes from baseline in VT, RR, Figures 3 and 4 show the individual regression slopes of Pmus and RR versus O2sat. It can be seen that at moderate levels of hypocapnia, the response is minimal (not significantly different from zero by t-test), both in terms of Pmus (0.003 cm H2O per percentage change in oxygen saturation, ± 0.09) and RR (0.01 breaths/min per percentage change in oxygen saturation, ± 0.06). The response is stronger at mild levels of hypocapnia in terms of both Pmus (0.26 ± 0.33) and RR (0.11 ± 0.11) but is still attenuated as compared with eucapnia (0.53 ± 0.59 and 0.17 ± 0.20 for Pmus and RR, respectively). Both the Pmus and RR responses at the three levels of PETCO2 showed significant differences by analysis of variance (p < 0.05 for both Pmus and RR). Tukey's test for multiple comparisons demonstrated that the response at moderate hypocapnia was significantly different from that at eucapnia for both Pmus (p < 0.05) and RR (p < 0.005), whereas the response at mild hypocapnia was not significantly different from the other responses for either Pmus or RR.
We have demonstrated that the hypoxic response, as measured by RR and Pmus, is attenuated at mild levels of steady hypocapnia and becomes negligible at moderate levels of steady hypocapnia in normal subjects.
Previous Approaches to Study Hypoxic Response in Hypocapnia Notwithstanding its shortcomings, this approach (poikilocapnic hypoxia) provided an important piece of information, namely the minimum range of hypocapnia over which a hypoxic response continues to exist. Thus, if a PETCO2 of 25 mm Hg is reached in a subject during a hypoxic test, one can reasonably conclude that the subject displayed a hypoxic response, strong enough to nearly double ventilation, at least down to this level. The lowest reported PETCO2 levels during such hypoxic tests varied considerably. Nielsen and Smith (8) reported two subjects in whom PETCO2 decreased to 20 mm Hg through the action of hypoxia, implying a continued vigorous hypoxic response even at this very reduced PETCO2. In a study by Hall (19), a PETCO2 of 26 mm Hg was reached in only 1 of 14 subjects. A PETCO2 of 28 mm Hg was reached in 10 subjects, and a PETCO2 of 30 mm Hg was reached in all subjects. In the study by Moore and colleagues (16), the lowest PETCO2 reached at a PETCO2 of 40 mm Hg ranged from 24.1 to 37.0 mm Hg (32.3 ± 3.0 mm Hg). The second approach was introduced by Roberts and colleagues (20). Stable hypocapnia of different degrees was produced using a mechanical ventilator in the controlled mechanical ventilation (CMV) mode. The response to a standard, brief hypoxic challenge (O2sat decreasing to 85%) was estimated from changes in Paw and diaphragm electromyogram. They found that when PETCO2 was reduced by 7.5 mm Hg below eucapnia, there was no neuromuscular response to the hypoxic challenge. Between -7.5 and 2.5 mm Hg relative to eucapnia, the response was graded. Although this approach is somewhat similar to ours, it differs in one fundamental respect, namely that they used controlled, as opposed to subject-triggered, ventilation. The ventilator was set to result in apnea before all hypoxic challenges. The mechanism of central apnea during this CMV protocol (initially introduced by Prechter and colleagues [3]) is not clear (21, 22). What is clear, however, is that once apnea is produced in this fashion (i.e., CMV), the chemical drive required to reinitiate spontaneous efforts is considerably higher than what exists during spontaneous breathing (this is referred to as "control system inertia " [4]). For example, when PETCO2 is gradually increased during the apnea, to reinitiate efforts, the level required (so called recruitment threshold) is, on average, 6–8 mmHg (with a range of up to 12 mm Hg) higher than eucapnic PETCO2 (4, 5, 23). Such a difference between recruitment threshold PCO2 and spontaneous PETCO2 cannot be interpreted as evidence of lack of response to CO2 in the PCO2 range between spontaneous PETCO2 and recruitment threshold PCO2. For the same reason, failure to initiate spontaneous efforts with a hypoxic challenge delivered during the CMV-induced apnea (20) need not reflect lack of hypoxic response. It could reflect that the increase in O2 drive was not enough to offset the inhibitory influence that caused the apnea in the first place.
The third approach was introduced by Rapanos and Duffin (11). Stable hypocapnia (PETCO2 In summary, previous approaches have produced divergent results ranging from loss of hypoxic response when PETCO2 decreases by 1–2 mm Hg below eucapnia (11) to persistence of a strong response when PETCO2 is in the low to mid 20s. Our approach obviates the shortcomings of previous studies. First, stable hypocapnia of different magnitudes can be induced regardless of whether a hypoxic response is present. Second, the same stable PETCO2 is maintained before and during the hypoxic challenge. Because of the very different equilibration (with blood) dynamics of peripheral and central chemoreceptors (2), there are always uncertainties about the status of central PCO2 when peripheral PCO2 is rapidly changing. These uncertainties do not apply to our approach. Third, hypoxia can be induced surreptitiously, obviating behavioral responses at the time where critical measurements are made. Fourth, spontaneous rhythmic respiratory efforts are present throughout. Accordingly, any change in chemical drive should result in a measurable change in respiratory output. The issue of "control system inertia," for reinitiation of efforts after the onset of apnea (4), is thus avoided. This study is the first to demonstrate convincingly that hypoxic response disappears below a threshold stable PETCO2. In all subjects, including those who displayed a very vigorous response to hypoxia at eucapnia, no response was detectable when PETCO2 was reduced by an average of 11 mm Hg. Because our study involved only eight subjects, we cannot exclude the possibility that an occasional individual may retain a hypoxic response when stable PETCO2 is reduced by more than 11 mm Hg.
Possible Role of Neuromechanical Inhibition
Interaction between Responses to CO2 and Hypoxia Although this study did not directly address the issue of interaction, its results, particularly when combined with results of other approaches, provide support for the existence of a central mechanism that contributes to the interaction in humans. In particular, the total disappearance of hypoxic response during moderate hypocapnia provides an important clue. Thus, in this study, respiratory motor activity was present throughout. If an excitatory input, associated with hypoxia, was received by the respiratory motor centers in the brainstem, an increase in respiratory activity should have resulted. The lack of such increase during moderate hypocapnia, therefore, suggests that no excitation was received by the respiratory motor centers at this level of PCO2. This lack of input cannot be because peripheral chemoreceptors do not respond to hypoxia at the PCO2 levels used in this study: First, peripheral chemoreceptors in animals retain a substantial sensitivity to PO2 when PCO2 is 30 mm Hg or even much less (28, 29). Second, in human studies using the poikilocapnic hypoxia approach (e.g., 8, 16, 19), a substantial ventilatory response to hypoxia, enough to lower PCO2 in the mid 20s, was present even though PCO2 was lower than in our study. Thus, even in humans, the peripheral chemoreceptors are not silenced when peripheral PCO2 is in the 20s. It seems reasonable, therefore, to conclude that central PCO2 determines whether peripheral chemoreceptor input is conveyed to the respiratory motor centers. This may result if central PCO2 controlled the gain of intermediate neural pathways that process peripheral chemoreceptor input before its arrival at the motor centers. Below a threshold central PCO2, the gain is zero. Alternatively, as proposed by Duffin and colleagues (30), central and peripheral chemoreceptor activities are summed, and a threshold total amount is required before chemoreceptor activity of either source can influence respiratory motor output. The current results cannot distinguish between these two possibilities. The conclusion that central PCO2 controls the traffic between peripheral chemoreceptors and respiratory motor centers may help reconcile the divergent results obtained in previous studies. As indicated earlier, the range of threshold stable PETCO2 suggested by previous studies is wide, extending from near eucapnia (11) to less than 20 mm Hg (8). These divergent results can be explained by the relatively slow equilibration (with blood) dynamics of central PCO2. Thus, in experiments in which PETCO2 rises during the hypoxic challenge (11), the increase in central PCO2 will lag, and instantaneous PETCO2 will thus overestimate central PCO2, leading to the false conclusion that the hypoxic response disappears with minimal hypocapnia. Conversely, when PETCO2 progressively falls during a fast hypoxic challenge (16), instantaneous PETCO2 underestimates central PCO2, leading to the conclusion that the hypoxic response may survive severe hypocapnia.
Clinical Implications
The analysis in the online supplement shows that a normal subject (functional residual capacity [FRC] = 3.0 L; hemoglobin = 15 gram percent; blood volume = 5.0 L; respiratory quotient [RQ] = 0.8) with a PaCO2 threshold of 30 mm Hg who hyperventilates to a steady PaCO2 of 15 mm Hg (i.e., 15 lower than PaCO2 threshold) before breathholding will sustain a decrease in PAO2 to Whereas hyperventilation before diving has been known to cause death, the postulated mechanism relied heavily on the fact that PO2 drops rapidly during resurfacing from depth due to decompression (33). By reducing PaCO2, hyperventilation before diving reduces the PO2 at which the diver feels a strong urge to breathe. There is then a very rapid further decrease in PO2 as the diver resurfaces from depth, due to decompression. It is thus postulated that consciousness is lost during ascent (33). The elucidation of a central PCO2 threshold before any hypoxic drive can be sensed introduces another explanation; the diver who hyperventilates before diving may never even have an urge to breathe before losing consciousness. It would also explain some instances of death in shallow pools, where ascent-related decompression is not a factor. The same sequence of events may, theoretically, occur in the clinical setting. This study was motivated by a patient who had moderate chronic hypercapnia, an extremely crowded pharynx and extremely severe obstructive sleep apnea (OSA) on polysomnography. The continuous positive airway pressure required to eliminate OSA (28 cm H2O) could not be tolerated. After a tracheostomy, the patient was initially placed on supplemental O2 (routine postoperative orders). The nurses noted that the patient stopped breathing when she was falling asleep. During a repeat polysomnography on room air (2-days postoperative), the patient developed a protracted central apnea whenever she fell asleep and made no breathing efforts until she awoke spontaneously or was awakened by the technologists on account of extremely low O2sat (less than 35%). The occurrence of central apneas during sleep after tracheostomy for severe OSA is well documented (34, 35). We speculate that by acutely relieving the nocturnal hypercapnia, daytime PaCO2 decreases to a new stable level that is lower than before with consequent loss of hypoxic and CO2 drives. Breathing during wakefulness is maintained by wakefulness stimuli (36). With loss of these stimuli, during sleep, breathing stops until arousal occurs or chemical drive rises enough to initiate breathing. Although we do not know whether deaths have actually occurred as a result, our findings clearly show that a sequence of events that may result in life-threatening hypoxemia can occur under these conditions. Oxygen supplementation in the few days after surgery would, accordingly, be prudent, and the occurrence of protracted central apneas during sleep should be excluded before discharge or discontinuation of O2. This sequence of events, leading to dangerous hypoxemia, would not apply to other forms of central sleep apnea (e.g., Cheyne-Stokes respiration), as the potential for central PCO2 to decrease well below apneic threshold does not exist; breathing stops as soon as central PCO2 decreases below threshold, and it is only a matter of several seconds before it rises again to reinitiate breathing. Nonetheless, our findings suggest that during the period of central apnea the deteriorating PAO2 likely does not contribute much to the reinitiation of breathing. In this sense, a loss or marked attenuation of hypoxic responses during the hypocapnic phase of the cycle may result in somewhat greater hypoxemia than what would otherwise occur. In summary, we have shown that regardless of their strength at eucapnia, hypoxic responses are lost when PaCO2 is reduced, in a stable manner, by 5–10 mm Hg relative to steady-state eucapnia. This may provide explanation for deaths in shallow pools and, also, suggests that patients in whom PCO2 is acutely reduced through tracheostomy, or mechanical ventilation, are at increased risk of developing dangerous hypoxemia during sleep or after ventilator malfunction.
Supported by the Canadian Institutes of Health Research. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form March 12, 2002; accepted in final form January 15, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
