INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It has recently been shown that when awake normal subjects are connected to a ventilator in the assist mode (volume-cycled or pressure-support) and ventilator output is increased to produce larger tidal volumes (VT) than those during spontaneous breathing, spontaneous respiratory rate does not decrease to the same extent as the increase in VT (1). As a result, hypocapnia, often severe, develops while spontaneous breathing continues. Persistence of rhythmic breathing efforts despite hypocapnia has provided an opportunity to re-examine CO₂ response in the hypocapnic range, an issue that currently remains controversial (2, 3, 6). It had been thought previously, based largely on anecdotal data (see References 2 and 6 for review), that CO₂ response in hypocapnia is discontinuous, with an essentially flat segment up to a certain PCO₂ and a linear segment beyond that point. The point of intersection of these two segments thus defined a discrete CO₂ threshold (6).

Two recent studies have examined CO₂ response, beginning from fairly severe hypocapnia induced during assisted mechanical ventilation (2, 3). In one study, in which assisted mechanical ventilation in the volume-cycled (A/C) mode was used, it was found that CO₂ responsiveness extended well into hypocapnia (i.e., PCO₂ between 20 and 30) and, although the slope of the response during hypocapnia was relatively lower, there was no discontinuity; the slope gradually increased as PCO₂ increased (2). In the other study, in which pressure-support ventilation (PSV) was used to produce the initial hypocapnia, the investigators concluded that at a pressure-support level of 10 cm H₂O the output of the respiratory centers appears to be insensitive to CO₂ as long as PCO₂ falls below normocapnic values (3).

The reason for the substantial differences in the results of these two studies is not clear. Although it is conceivable that the CO₂ response differs in the PSV mode from that in the A/C mode, several technical differences between the two studies could have accounted for the discrepant results. Chief among these is the use of respiratory muscle pressure (Pmus) as the output variable in the study of Patrick and colleagues (2) and VT as the main output variable in the study of Scheid and coworkers (3). There are practical and theoretical reasons for VT to be relatively insensitive to changes in Pmus in the PSV mode. Thus, in most PSV delivery systems, including the one used by Scheid and coworkers (3), the pressure assist decreases as flow increases (7). This tends to attenuate the spirometric response (i.e., flow and VT) to increased subject effort. Furthermore, even with a fixed pressure delivery, small increases in Pmus may fail to increase VT because of peculiar interactions between subject and ventilator in the PSV mode (7). Accordingly, in the present study we re-evaluated the response to CO₂ in the PSV mode, but, in addition to the usual spirometric variables (flow, VT, respiratory rate), we used Pmus as a respiratory output variable.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sixteen normal adults were studied (seven men and nine women). They were medical students, physicians, and technicians. Their average age was 31 yr (range, 24 to 41 yr). None had any known respiratory illness, and all but one were naive to the purpose of the study. All subjects were accustomed to breathing with inspiratory pressure assist. The study was approved by the institutional ethics committee, and informed consent was obtained from all subjects.

The subjects were studied while awake and seated in a comfortable chair. We told them that "you will breathe through a ventilator, which will assist your breathing." We asked them to breathe "as you want," without giving any other instructions. However, they were informed that during the experiments "you may feel shortness of breath and may breathe harder." If this feeling was very uncomfortable and they wanted to terminate the experiment, they were asked to raise their hands. We encouraged them to read during the experiment, and one of us followed them closely in order to ensure wakefulness, which was judged on behavioral criteria.

The subjects, wearing noseclips, breathed through mouthpieces connected by conventional tubing to a prototype ventilator (Winnipeg ventilator; Respiratory Investigation Unit, Winnipeg, Canada) able to ventilate the subjects in both volume-cycled and pressure-support modes. The design and operation of this ventilator has been described previously (8). Briefly, the gas delivery system consists of a freely moving piston reciprocating within a chamber. Electronics control a motor that moves the piston toward the subject. When the piston is activated it creates a pressure in the piston chamber, and the forward movement of the piston produces air flow. The pressure in the piston chamber is measured by a pump pressure transducer. The difference in pressure between the chamber pressure and proximal airway pressure creates a trigger signal. This signal in turn switches on a three-way solenoid valve connecting the pump pressure to the exhalation valve line and causes the exhalation valve to close. When the criterion for cycling off is met (see below) the trigger signal goes off, causing the exhalation valve to open. This allows passive deflation through the exhalation valve while the piston returns to the starting position, taking gas from the ventilator input as it moves backward and resetting for the next inspiration. The motor activates and applies force to the piston according to different command signals regulated by the operating mode. With volume-control the piston is activated by time, and a preset flow-time profile is the command signal that applies pressure in the piston chamber. With pressure-support and the piston at the starting position, as the subject breathes in, the piston moves freely into the chamber, providing an initial flow and volume. When the velocity of the piston movement (inspiratory flow) reaches a preset threshold value (0.05 L/s) the motor is activated and applies force to the piston, such that the pressure in the chamber increases to the preset level at a fast rate (attack time, 0.1 s). The pressure remains at that level until inspiratory flow decreases to 0.1 L/s (cycling off criterion).

An external demand blended gas system was attached to the ventilator input opening. The oxygen inlet of the blender was connected to a tank containing 10% CO₂ (balance O₂), and the air inlet was connected to a 100% O₂ tank. The fraction of inspired oxygen (FI_O2) concentration dial was then used to adjust inspired concentration of CO₂ (FI_CO2) between a minimum value of zero (setting at 21%) and a maximum value of 10% corresponding to tank CO₂ concentration (setting at 100%).

Airflow was measured using a pneumotachograph (Fleisch no. 3; Laussane, Switzerland) placed between the mouthpiece and the Y ventilator connector. The flow signal was electronically integrated to provide inspiratory and expiratory volume. Airway pressure was measured from a side port near the mouthpiece (Micro-Switch, 140PC; Ontario, Canada). CO₂ concentration was also monitored near the mouth using a mass spectrometer (Perkin-Elmer, Oakbrook, IL) or an infrared sensor (Datex, Helsinki, Finland). The total sample rate for all channels was 100 Hz (Codas; Dataq Instruments, Akron, OH). All signals were stored on a computer disk for later analysis.

Initially, the subjects breathed spontaneously through the ventilator (mouthpiece and noseclips), which was set on CPAP mode with zero positive end-expiratory pressure. When ventilation (I) and end-tidal PCO₂ (PET_CO2) stabilized, VT, breathing frequency (f), and PET_CO2 were recorded for 2 min (baseline). Respiratory system mechanics were then measured using the technique of airway occlusion at end-inspiration (9, 10). Briefly, the ventilator was set to volume-controlled mode delivering a comfortable VT with a square-wave flow-time profile. Mandatory breathing frequency was adjusted to lower the subject's PET_CO2 and inhibit intrinsic respiratory muscle activity. The cessation of inspiratory muscle activity was inferred from the absence of the initial negative deflection of airway pressure, stabilization of mouth pressure waveform, constancy of peak positive end-inspiratory pressure from breath to breath, and exponential decline in expiratory flow (11). When the above criteria were met the airways were occluded at the end of inspiration for 2 to 3 s. Airway pressure (Paw) exhibited an immediate drop from a peak pressure (Ppeak) to a lower value, followed by a gradual decline to a plateau pressure (Pp). Maneuvers with airway pressure waveform distortion caused by respiratory muscle activation were excluded from the analysis. Assuming that end-expiratory lung volume was at passive FRC, respiratory system elastance (Ers) was computed according to the formula:

(1)

The resistance of the respiratory system (Rrs) was obtained by dividing the difference between Ppeak and Pp by the preceding constant inspiratory flow.

After determination of mechanics the subjects breathed spontaneously (CPAP mode) until ventilation and PET_CO2 returned to baseline levels. The ventilator was then switched to the PSV mode with the pressure set to the highest comfortable level for each subject. Initially, the FI_CO2 was set to zero (Figure 1). When PET_CO2 was stabilized to the new value, usually after 5 to 10 min, FI_CO2 was increased in small steps, each step lasting 4 min. The study continued until the subject indicated discomfort and wished to terminate the experiment. In four subjects, in order to assess changes in end-expiratory lung volume, inspiratory capacity maneuvers were performed at the end of each step.

View larger version (50K): [in this window] [in a new window]   Figure 1.   End tidal PCO2 (PETCO2), airway pressure (Paw), airflow, and volume (inspiration positive, integrator resets at both zero flow crossings) in one representative subject during spontaneous breathing (CPAP mode) and during pressure-support (PS) ventilation with two different concentrations of inspired CO2 (FICO2).

In each subject variables reflecting different aspects of respiratory motor output were obtained by averaging breaths occurring during the last 30 s preceding each change in FI_CO2. The following variables were computed: (1) f. (2) VT. (3) peak inspiratory flow (peak). (4) mechanical inspiratory time (TIm). This was measured as the interval from the onset of inspiratory flow to the onset of expiratory flow. (5) Paw, airflow, and volume at different fractions of the inflation time. These were derived from computer-averaged data, and the values were provided at 5% intervals of TIm for the relevant breaths. (6) Muscle pressure (Pmus) at different PET_CO2 levels. Pmus was calculated using the equation of motion, as proposed by Mead and Agostoni (12) and modified for mechanically ventilated subjects by Younes (13). Briefly, in subjects ventilated with pressure-support, the total pressure applied to the respiratory system (Ptotal) is the sum of the ventilator pressure (Paw) and the pressure generated by the respiratory muscles (Pmus). Assuming that inertia is negligible, Ptotal is dissipated to overcome resistive and elastic properties of the respiratory system. The former is a function of flow, whereas the latter is a function of volume above passive FRC. Thus:

(2)

and

(3)

where Ptotal(t), Pmus(t), and Paw(t) are the relevant pressures at time t from the onset of inspiratory flow, and V(t) and (t) are, respectively, instantaneous volume above passive FRC and instantaneous inspiratory flow.

The response of any variable to CO₂ was assessed from the change in the value of this variable between different PET_CO2.

Although calculated Pmus was available at 5% intervals of TIm, we will present only the data from 20, 50, and 80% of TIm (Pmus₂₀, Pmus₅₀, and Pmus₈₀). These represent Pmus at early, middle, and late points of the subject's inspiratory effort.

Data were analyzed by paired t test and analysis of variance for repeated measurements (ANOVA), followed by paired t test with the Bonferoni correction for multiple comparisons if the F value was significant; p < 0.05 was considered statistically significant. Values are expressed as mean ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Rrs and Ers, measured during controlled mechanical ventilation, averaged 4.3 ± 0.4 cm H₂O/L/s and 13.3 ± 0.8 cm H₂O/L, respectively.

Pressure-support was set to a mean value of 10.1 ± 0.6 cm H₂O (range, 8 to 17). At zero FI_CO2, this pressure was associated with a mean VT of 1.16 ± 0.1 L, significantly higher than spontaneous VT (0.85 ± 0.1 L). Breathing frequency did not differ significantly from that observed during spontaneous breathing, 15.98 ± 0.9 versus 15.63 ± 1.11 breaths/min, respectively. As a result, the subjects hyperventilated with pressure-support and developed hypocapnia, which in some cases was quite severe (Figure 2). At zero FI_CO2, PET_CO2 was 23.5 ± 1.2 mm Hg, considerably lower than PET_CO2 during spontaneous breathing (PET_CO2sb, 35.5 ± 1.1 mm Hg). The final PET_CO2 achieved was 49.2 ± 1.4 mm Hg. The range examined, therefore, covered an average range of 12 mm Hg below to 14 mm Hg above eupneic PET_CO2 (Figure 2).

View larger version (18K): [in this window] [in a new window]   Figure 2.   CO2 range examined in each subject. Horizontal line in each bar represents end-tidal PCO2 during spontaneous breathing (eupnea).

The different ranges of CO₂ examined in different subjects (Figure 2) made it difficult to provide average responses of all subjects over the same PET_CO2 range. Furthermore, we observed that the response to CO₂ was not linear and could not be described adequately in all subjects using a mathematical formula (i.e., exponential). Thus, to maximize the use of the data, in each subject PET_CO2 of the initial step (lowest PET_CO2) and of the step in which PET_CO2 was closest to but below PET_CO2sb were recorded and the range covered by these two steps was calculated. Then, the step in which PET_CO2 was closest to midpoint of the above range was identified. Furthermore, the final step (i.e., highest PET_CO2) and the step in which PET_CO2 was similar to or slightly higher than PET_CO2sb were recorded and the step in which PET_CO2 was closest to midpoint of this range was identified. Data during these six steps were analyzed. Therefore, in each subject the response of various variables was measured across three PET_CO2 values below and three above PET_CO2sb. These values averaged 23.5 ± 1.2, 28.2  ± 1.0, 33.3 ± 1.2, 37.0 ± 1.1, 42.9 ± 1.2, and 49.2 ± 1.4 mm Hg.

In four subjects in whom IC maneuvers were performed at the end of each step, end-expiratory lung volume did not change from hypocapnia to normocapnia (mean change, 0.003  ± 0.03 L; range, 0.07 to 0.09 L). Above PET_CO2sb end-expiratory lung volume decreased slightly with increasing PET_CO2 (mean decrease at the highest PET_CO2, 0.15 ± 0.05 L).

Airway pressures at the point of peak inspiratory flow (Pawpeak) and at Paw₂₀, Paw₅₀, and Paw₈₀ of inflation time are shown in Table 1. At high levels of PET_CO2, Pawpeak, Paw₂₀, and Paw₅₀ decreased with increasing PET_CO2, indicating that the function of the ventilator was influenced by the subject's effort. Paw₈₀ remained relatively stable.

Mean values of peak, TIm, Pmus, VT, f, and I at the six levels of PET_CO2 are shown in Table 2 and in Figures 3 and 4, whereas individual responses of Pmus (Pmus₂₀ and Pmus₅₀) against PET_CO2 of the different steps are shown in Figures 5 and 6. TIm remained relatively constant at all PET_CO2 levels (Table 2), and f remained stable in hypocapnia and increased significantly only at the highest values of PET_CO2 (Figure 3). VT, I, peak, and Pmus increased with increasing PET_CO2. The response was detectable even below PET_CO2sb.

View larger version (15K): [in this window] [in a new window]   Figure 3.   Average tidal volume and breathing frequency at six levels of end-tidal PCO2 (PETCO2). Bars are SE. *Significant difference from values at PETCO2 of 23.5 mm Hg. #Significant difference from values at PETCO2 of 28.2 mm Hg (ANOVA).

View larger version (10K): [in this window] [in a new window]   Figure 4.   Average ventilation at six levels of PETCO2. Bars are SE. See Figure 3 for abbreviations, symbols, and significance values.

View larger version (28K): [in this window] [in a new window]   Figure 5.   Individual response of Pmus at 20% (solid lines) and at 50% (dashed lines) of inflation time to CO2 in Subjects 1 to 8. Vertical dotted lines indicate PETCO2 during spontaneous breathing (PETCO2sb).

View larger version (30K): [in this window] [in a new window]   Figure 6.   Individual response of Pmus at 20% (solid lines) and at 50% (dashed lines) of inflation time to CO2 in Subjects 9 to 16. Vertical dotted lines indicate PETCO2 during spontaneous breathing (PETCO2sb).

Twelve of the 16 subjects had data at two or more levels of PET_CO2 below 30 mm Hg and below PET_CO2sb. In these subjects respiratory variables at the lowest PET_CO2 (21.63 ± 1.0) were compared with those observed at PET_CO2, which was closest to but below 30 mm Hg (28.32 ± 0.4). There was a significant increase in VT, I, and Pmus (Pmus₂₀, Pmus₅₀, and Pmus₈₀) in this PET_CO2 range (Table 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critique of Methods

The response to CO₂ was examined in normal conscious subjects during positive pressure breathing. When conscious subjects perform experiments such as these, it is possible that the results may be influenced by behavioral responses. However, we tried to limit the influence of these responses on the results. We studied subjects who were accustomed to positive pressure by previous experiments, and all of them breathed through the experimental setup on several occasions before the trial. Furthermore, 15 of the 16 subjects were not aware of the purpose of the study and did not know what changes in ventilatory parameters were expected. Also, the design of the study permitted us to keep constant the distracting stimuli (i.e., mouthpiece, noseclips, pressure-support level), and FI_CO2 was the only changing variable. Finally, the response to CO₂ was measured at the end of the 4-min period after a change in FI_CO2, an interval that likely was sufficient to achieve steady state. However, although all these reasons make us doubt that the response we observed was voluntarily influenced by these trained subjects, behavioral responses to the experimental setup (i.e., mouthpiece, noseclips) and to positive pressure breathing can not be entirely excluded.

Respiratory motor output, in addition to directly measured respiratory variables, was assessed by muscle pressure. Muscle pressure was calculated by the equation of motion (12), modified for mechanically ventilated subjects (13) and based on previously measured respiratory system mechanics. This technique of Pmus measurement has been validated in patients during pressure-support ventilation; a fairly good correlation existed between work of breathing derived from the Pmus estimate with the patient's work of breathing measured by esophageal pressure (14, 15). Thus, it appears that Pmus is a good index of respiratory motor output, even in patients with abnormal lung mechanics (14, 15).

At all PET_CO2 levels, volume change during inflation was related to passive FRC, assuming, thus, that end-expiratory lung volume did not change as a function of PET_CO2. This was most probably true in the hypocapnic range, as indicated by the results of four subjects in whom changes in end-expiratory lung volume were measured directly using IC maneuvers. On the other hand, above eupneic levels, end-expiratory lung volume decreased with increasing PET_CO2, in accordance with the existing literature (16, 17). During hypercapnia expiratory muscle contraction forces end-expiratory lung volume below the equilibrium point of the respiratory system, and thus passive inspiration is accomplished by the elastic recoil pressure of the system at end-expiration. In this case Pmus, as we measured it, represented the total pressure generated by contraction of inspiratory and expiratory muscles. Although we do not know to what extent expiratory muscle activity contributes to Pmus, this is unlikely to influence our conclusion. The aim of the study was to examine the response of respiratory motor output (inspiratory and expiratory muscle activity) to CO₂. To the extent that expiratory muscle contraction shares part of total muscle work and contributes to ventilation, calculation of Pmus, as we did, is justified for the purpose of this study.

From the values of Pmus reported, Pmus₂₀ and Pmus₅₀ are the most accurate and representative of the response to CO₂ for two reasons. First, Pmus₂₀ and Pmus₅₀ were measured, contrary to Pmus₈₀, closest to the onset of neural inspiration and thus these measurements are likely to occur during the rising phase of Pmus in all subjects. On the other hand, with PSV there is no automatic linkage between end of neural inspiration and end of machine inspiratory time; inflation can outlast neural inspiratory time (7). Therefore, Pmus₈₀, which was measured at the end of inflation, may be on the declining phase of Pmus in some subjects. This is supported by the fact that in several subjects Pmus₈₀ was smaller than Pmus₅₀. Second, Pmus was calculated assuming constant values of Ers. At low levels of PET_CO2, the inflation volumes at which Pmus was measured were relatively low, and therefore it is reasonable to assume constant values of Ers throughout the inflation cycle. This may not be the case during hypercapnia, where inspiratory volumes were high and approached total lung capacity. In these volumes Ers of the respiratory system decreases (18), and therefore Pmus at end-inspiration (i.e., Pmus₈₀) might be underestimated in some subjects. Indeed, at the highest PET_CO2, Pmus₈₀ were calculated at inflation volume of approximately 1.9 L (range, 1.12 to 2.83). On the other hand, Pmus₂₀ and Pmus₅₀ were calculated at lower volumes where the Ers most likely is constant. For these reasons individual response of Pmus₈₀ to CO₂ was not reported.

The intensity of respiratory effort was evaluated by peak inspiratory flow, VT, and Pmus. Although all these indices increased progressively with increasing PET_CO2, the increase in Pmus was more evident. Compared with values at the first PET_CO2 level, Pmus₂₀, Pmus₅₀, and Pmus₈₀ increased by 11, 16, and 20%, respectively, at the second PET_CO2 level and by 75, 46, and 47% at the third. The corresponding values of VT and peak were 7 and 1%, at the second level, and 19 and 6% at the third, respectively. Similar findings were observed in 12 subjects who had data at two or more levels of PET_CO2 below 30 mm Hg (Table 3). This observation, however, was not unexpected. There are factors other than the intensity of respiratory effort that may influence considerably peak and VT. First, it has been shown, on theoretical grounds, that with pressure support VT, particularly at low levels of inspiratory effort, is not very sensitive to changes in muscle pressure (7). Second, peak and VT depend not only on the intensity of a subject's effort but on pressure generated by the ventilator (Paw). Theoretically, with pressure-support the time course of Paw should be independent on the intensity of respiratory effort. This was not the case in our study. We observed that Paw, particularly early in inspiration, decreased with increasing PET_CO2, deviating from the target level, sometimes by a substantial amount. This phenomenon has been reported previously and is a feature of ventilator function (7). This technical feature inevitably reduced the ventilatory consequences of a given patient effort. It follows that VT and peak underestimated the intensity of respiratory effort with increasing CO₂ stimulus. On the other hand, Pmus was calculated taking into account Paw (see METHODS) and, therefore, was a much better index of a subject's effort.

The PET_CO2 values used to generate an average of 28.2 and 33.3 mm Hg (second and third PET_CO2 level) included five subjects (Subjects 3, 5, 6, 7, and 8) when these values were in excess of 31 and 35 mm Hg, respectively. Therefore, one could argue that the progressive increase we observed in various variables in the hypocapnic range was due to these subjects. Data in 12 subjects who had at least two steps below PET_CO2 of 30 mm Hg and below PET_CO2sb do not support such an argument. In these subjects VT, I, and Pmus increased significantly when PET_CO2 increased from an average of 21.6 to 28.3 mm Hg, indicating that the response observed was not due to the method of analysis we used.

The subjects were hyperoxic throughout the experiment, with the FI_O2 ranging between 100 and 94%. It has been previously shown that in normal subjects hyperoxia results in hyperventilation (19, 20). This hyperoxic hyperventilation is attributed to a decrease in Haldane effect, increase of the excitatory state of respiratory neurons in the brain that are depressed by normoxia or cerebral vasoconstriction (19, 20). Although we did not measure PET_CO2 during normoxic conditions, PET_CO2 data during hyperoxic spontaneous breathing (CPAP) indicate that some of the subjects hyperventilated (Figure 2), in agreement with the existing literature. Breathing through the mouthpiece may also contribute to hyperventilation (21). However, this excitatory effect of hyperoxia and mouthpiece breathing is unlikely to influence our data. The response to CO₂ was measured for at least 10 to 15 min of hyperoxic breathing, a time interval that probably permitted the excitatory effect of hyperoxia and mouthpiece breathing to be fully expressed during ventilation (19).

Ventilatory Response to CO₂

The results of the present study demonstrated that in normal conscious subjects ventilated on the PSV mode, the ventilatory response to CO₂ is detectable even when PCO₂ is well below eupneic levels and gradually increases as a function of CO₂ stimulus. Over the range of PET_CO2 examined, a threshold below which the ventilatory response to CO₂ does not exist could not be identified. The response to CO₂ was expressed mainly by the intensity of respiratory effort, as indicated by the progressive increase of VT, peak, and Pmus as PET_CO2 increased. On the other hand, breathing frequency was relatively insensitive to CO₂ over a wide range of PCO₂. The response of breathing frequency appeared at values of PCO₂ well above eupneic levels.

The present results during PSV confirm recent findings by Patrick and colleagues (2) during volume-cycled ventilation and indicate that there is no fundamental difference in response to CO₂ between these two modes. That the respiratory rate response was essentially flat over the PCO₂ range of 23 to 45 mm Hg whether tidal volume was allowed to increase (current study) or not (2), indicates that tidal-volume-related influence (22, 23) on respiratory rate is minimal in awake humans in this volume range.

The demonstration of a significant response to CO₂ during hypocapnia associated with PSV contrasts with the findings of Scheid and coworkers (3). These investigators studied ventilatory sensitivity to CO₂ in eight normal awake humans during two levels of pressure-support using a device that had similar functional characteristics to our ventilator. At each pressure support level subjects inhaled four different concentrations of CO₂. They reported that at the level of pressure-support comparable to ours (10 cm H₂O), respiratory drive, as assessed by VT and P_0.1, was insensitive to CO₂ as long as PET_CO2 fell below normocapnic values. Although direct comparison of our study with that of Scheid and coworkers (3) is difficult because of the different study designs, it is possible that the different conclusions reached by these studies could be related to at least three reasons. First, we calculated Pmus during different PET_CO2 levels. This takes into account the decrease in applied pressure associated with greater inspiratory effort and the discrepancy between Pmus and ventilatory output observed in PSV (7). In fact, as it was stated earlier, if we examine VT response, the changes in the hypocapnic range will be less impressive than in Pmus (Tables 2 and 3 and Figures 3 and 4). Similar to our pressure-support system, the system of Scheid and coworkers (3) was sensitive to subject effort, as indicated by the experimental records shown (see Paw waveform in Figure 2 of Reference 3). Second, in our study, respiratory motor output was measured, particularly below eucapnia, at several PET_CO2 levels. This improves the definition of CO₂ response. In the study of Scheid and coworkers (3) at each level of pressure-support only two measurements separated the lowest and highest PET_CO2. Furthermore, at 10 cm H₂O only one measurement below 30 mm Hg PET_CO2 was obtained. Third, Scheid and coworkers examined only eight subjects and, given the variability and the small response we observed in the hypocapnic range, this number might be too small to detect statistically significant differences. This argument is supported by their results; P_0.1 increased by 66% from hypocapnic to normocapnic range, in line with our conclusion.

We studied normal humans during wakefulness. It is clear from several studies that during sleep and under anesthesia the maintenance of respiratory rhythm is critically dependent on chemical feedback; when PCO₂ is reduced by only a few mm Hg, apnea will occur (24). In the face of mechanical ventilatory support, there are compensatory changes in breathing such that PCO₂ is forced to hover around the CO₂ set point (26). Indeed, periodic breathing has been observed during NREM sleep in subjects mechanically ventilated with pressure-support (4). Thus it appears that the sleep/awake stage is a critical factor for ventilatory response to CO₂ in the hypocapnic range. However, the difference in ventilatory response to CO₂ between sleep and wakefulness does not necessarily mean that respiratory centers respond differently to CO₂ (see Reference 2 for discussion).

In the present study the response to CO₂ was studied during one level of pressure support. Thus, it is not known to what extent different assist levels may alter the response. Nevertheless, recently, using the rebreathing method, we studied the response of neuromuscular output to CO₂ in normal subjects with and without unloading the respiratory system (27). At a similar PCO₂ in peripheral and central chemoreceptors, neuromuscular output, expressed by transdiaphragmatic pressure, remained virtually unchanged by an approximately 50 to 60% reduction of the normal mechanical load; the neuromuscular output was tightly linked to CO₂ and not to load reduction. Furthermore, data from patients during constant flow synchronized intermittent mandatory ventilation (SIMV) have shown that for a given level of assist, inspiratory effort did not differ between spontaneous and mandatory breaths, indicating that inspiratory output is preprogrammed, probably by chemical feedback, and is relatively insensitive to breath-by-breath changes in load reduction seen during SIMV (28, 29). All these studies indicate that the response of respiratory centers to CO₂ may be independent of assist level. However, further studies are needed to clarify this issue.

Clinical Implications

The results of our study may be important for the management of mechanically ventilated patients. The observations that during wakefulness (1) breathing frequency is relatively insensitive to CO₂ over a wide range of PCO₂ and (2) both below and above the eucapnic level, the intensity of respiratory effort increases with increasing PCO₂, have at least three implications for the mechanically ventilated patient. First, mechanical ventilation with PSV greatly compromises the ability of chemical feedback to prevent respiratory alkalosis, an important cause of various arrhythmia (30) and weaning failure (31). This is because with PSV, in the absence of active termination of inspiration, the VT has a minimum value that depends on PS level, mechanical properties of respiratory system and cycling-off criteria (7). It follows that breathing frequency plays a key role in defending respiratory alkalosis. To the extent that breathing frequency is insensitive to CO₂, mechanically ventilated awake subjects receiving PS may easily develop respiratory alkalosis because of an inappropriate high level of assist, reduced metabolic rate, or improvement in gas exchange properties of the lung. Second, pressure-support, contrary to assist-volume, permits, to some extent, the respiratory system to compensate for changes in PCO₂. This compensation is, however, partial because of the minimum VT delivered and the limited ability of respiratory effort to modulate VT, particularly in patients with abnormal mechanical properties of respiratory system (7). Third, respiratory rate during pressure-support is an insensitive index of evolving hypercapnia. All these issues should be considered in the management of patients ventilated on PSV.

Finally, our finding that respiratory motor output is sensitive to CO₂ even during hypocapnia might have physiologic implications in those situations where Pa_CO2 falls as a result of various stimuli (i.e., ascent to altitude, metabolic acidosis) (6). It is possible that ventilation may be determined by the combined action of the stimuli and the magnitude of the PCO₂ drop. However, the type of this interaction (multiplicative, additive, subadditive) is unknown and remains to be elucidated by further studies.

·

·