INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Recent advances have altered the approach to mechanical ventilation in the settings of acute respiratory distress syndrome (ARDS) and severe asthma. The current focus centers on decreasing pressure-induced pulmonary injury by lowering tidal volume (VT) and allowing systemic arterial CO₂ tension (Pa_CO2) to increase to levels previously thought to be unacceptable (permissive hypercapnia) (1). These changes in ventilatory approach have been combined with the traditional practice of applying positive end-expiratory pressure (PEEP) to increase end-expiratory lung volume (EEVL) in ARDS. The relationships between increased levels of Pa_CO2, alterations in lung volumes and the activity of pulmonary receptors with regard to the control of breathing pattern could be better defined.

Following the observations of Clark and von Euler (6), many studies of the control of breathing have concentrated on the relationships between tidal volume and the corresponding duration of inspiration and expiration. These studies were performed under a variety of conditions in which ventilation was increased. It is now generally accepted that phasic activity from slowly adapting pulmonary stretch receptors is the principal volume-related feedback influencing breathing pattern (7, 8). Increasing pulmonary stretch receptor activity during inspiration is thought to act centrally to terminate inspiration. Several studies have also suggested that EEVL may influence expiratory time independently of inspiratory time (9).

Green and Kaufman (12) investigated in dogs the reflex changes in breathing elicited by graded reductions in EEVL, and the vagal nerves responsible for those reflex changes. Prior studies of pulmonary afferent activity had indicated that reduction of EEVL stimulates rapidly adapting receptors (13- 16), reduces activity in slowly adapting receptors (13), and has little or no effect on pulmonary or bronchial C-fibers (17). In their series of dogs, Green and Kaufman found that with a 50% reduction in end-expiratory transpulmonary pressure (EEPtp) from an EEVL approximating functional residual capacity ("FRC"), VT, and breathing frequency (f) immediately increased significantly. VT returned to, or below, control values after 90 s. The increase in f was due entirely to a reduction in expiratory time (TE). Vagotomy completely abolished these reflex changes. In the same study, they recorded action potentials from all known categories of pulmonary vagal afferents and demonstrated that the changes in breathing pattern induced by a 50% reduction in EEPtp were due solely to a withdrawal of slowly adapting stretch receptor activity. Slowly adapting receptors were the only receptors whose level of activity significantly changed at a 50% reduction in EEPtp, with rapidly adapting receptor activity insignificantly altered until EEPtp was further reduced well below the 50% of control level.

In light of the identification by Green and Kaufman of the pulmonary receptors activated at different EEPtp levels below "FRC," we wanted to examine how changes in breathing pattern induced by progressive hypercapnia were modified by three different levels of EEVL: "FRC," 0.5 "FRC," and 1.5 "FRC." Investigation of the responses to these differing lung volumes may help clarify the role of pulmonary afferent receptors during permissive hypercapnia used in the treatment of ARDS and asthma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General

We anesthetized eight mongrel dogs weighing 31.4 ± 4.7 kg with -chloralose 60 to 80 mg/kg intravenously (21). The trachea was cannulated low in the neck, and the lungs were initially ventilated by a Harvard pump (Harvard Apparatus Co., South Natick, MA) at a tidal volume of 15 ml/kg (22). All animals were then paralyzed intravenously with succinylcholine chloride 0.5 mg/kg administered at appropriate intervals. Anesthesia was maintained intravenously with supplemental doses of -chloralose 6 to 8 mg/kg administered every 60 min or as needed (21). Arterial blood gases were measured at frequent intervals with a Mark-II blood-gas analyzer (Radiometer America, Westlake, OH). If the acid-base balance was abnormal, appropriate corrections were made. Arterial blood pressure and heart rate were measured continuously via a catheter advanced into the aorta through the femoral artery, and tracheal pressure was monitored from a sidearm in the tracheal cannula; pressures were recorded by Statham transducers (Statham Instruments, Oxnard, CA). The C5 root of the left phrenic nerve was isolated from surrounding tissues. The chest was then opened widely in the midsternal line to eliminate possible reflexes from chest wall afferent receptors. Inspiratory and expiratory airflow rates were monitored with a Hans Rudolph pneumotachograph connected to the tracheal cannula (Hans Rudolph, Kansas City, MO). This allowed the following respiratory variables to be measured for each breath: tidal volume (VT), inspiratory time (TI), and TE. Inspiratory time was that period of the respiratory cycle that began with the onset of inspiratory flow and ended with the onset of expiratory flow (peak tracheal pressure). Similarly, expiratory time was that period of the respiratory cycle that began with expiratory flow and ended with the onset of inspiratory flow. Respiratory frequency (f) was calculated as 60/(TI + TE) and minute ventilation (I) was calculated as VT · f, measuring inspiratory tidal volume. All electronic signals were recorded on a Gould ES2000 multichannel recorder (Gould, Inc., Oxnard, CA).

Ventilation Methods

To measure breathing pattern in the open-chest dog, we used an electronically controlled positive-pressure ventilator driven by inspiratory phrenic neural activity (23). This ventilator was a modified Bird Mark 14 (Bird Corp., Palm Springs, CA). The left fifth cervical phrenic nerve was desheathed and its action potentials were recorded using standard techniques. A window discriminator was used to distinguish inspiratory activity from expiratory noise. The inspiratory activity was then rectified and passed through a resistance-capacitance network with a time constant of 100 ms, approximating a moving average with a 100 ms window. Flow was delivered by advancing an air valve within the Bird ventilator with a servomotor controlled by the output of a differential amplifier. The phrenic neural and tracheal pressure signals were processed by the differential amplifier to yield an inspiratory flow proportional to the difference between the phrenic moving average and tracheal pressure. Thus the VT (flow integrated over a given inspiration) was proportional to phrenic activity. We adjusted the ventilator gain and the level of tracheal pressure feedback at the beginning of each experiment to create a VT of approximately 15 ml/ kg. Once the desired ventilation was obtained, no further adjustments were made to the ventilator. While inspiration was active, expiration was passive. During expiration a valve, which was closed during inspiration, was opened, allowing the lungs to recoil against a PEEP created by directing the expired volume through low resistance Inspiron tubing (three-quarter-inch ID) placed under water. Under these conditions, volume feedback from the lungs could regulate VT, TI, and TE. We therefore considered the ventilator cycles as "breaths" for the purpose of analyzing the effects of changes in EEVL upon the breathing pattern during progressive hypercapnia.

Experimental Protocol

We measured breathing pattern (I, VT, f, TI, and TE) as a function of systemic arterial PCO₂, first at an EEVL that approximated "FRC," then after reduction in EEVL to a volume that was approximately 0.5 "FRC," and finally at an EEVL that was approximately 1.5 "FRC." To approximate "FRC" in open-chest dogs we carefully measured transpulmonary pressure (PL) in paralyzed dogs at the moment the chest was opened (by occluding the tracheal cannula and measuring the tracheal pressure). Subsequently, to reestablish "FRC" in the now open chest, we set PEEP equal to the PL obtained at the moment the chest was opened. This was done after hyperinflation to 25 cm H₂O for several seconds to provide a constant-volume history by reexpanding any atelectatic lung segments. To obtain 0.5 "FRC," we set PEEP, after hyperinflation, at approximately one-half the value of the PL at "FRC." Similarly, 1.5 "FRC" was obtained by setting the PEEP at approximately 1.5 times the value of the PL at "FRC."

The animal was hyperventilated, using a manual override switch on the servo-ventilator, to lower Pa_CO2 and was allowed to remain apneic until the Pa_CO2 increased to a level at which ventilation spontaneously began. Blood gas measurements were recorded at 15-s intervals during this period. To further increase Pa_CO2, we bled 100% CO₂ into the inspiratory line leading from the ventilator to the lungs. At minute intervals, the flow rate of CO₂ was increased to produce a progressive hypercapnia rising over a period of approximately 5 to 8 min. This protocol caused a rate of rise of CO₂ similar to that observed in dogs during rebreathing from a 6-L anesthesia bag (24). Just before each increase in CO₂ flow, Pa_CO2 and breathing pattern were measured, and the data were plotted to construct a CO₂-response curve. Blood gas levels were then allowed to return to baseline levels as the dog stabilized over the next 10 to 15 min. This protocol was repeated at "FRC," 0.5 "FRC," and 1.5 "FRC."

Statistics

Statistical analysis was performed in the following manner. The measured data points for all variables fell within three areas of distribution over the range of Pa_CO2: (1) below the apneic threshold all measured values were zero; (2) immediately above the apneic threshold there was a sharp increase in all variables as Pa_CO2 rose approximately 2 to 5 mm Hg; (3) as Pa_CO2 rose further there was a distinct change in the slope of each variable as it continued to rise in a linear fashion. This latter linear segment was the portion of data used for statistical analysis (Figure 3). Analysis was performed using a multiple regression equation that encompassed the slope and intercept coefficients for all three lung volumes. We then applied a linear statistical model based on the analysis of variance (25), which allowed us to separate the coefficients and determine the statistical significance of changes in slope and intercept between the regressed relationships of the three end-expiratory lung volumes. The slope of each individual relationship was tested for significant difference from zero using a t test for paired variants. Analysis of variance was used to compare the difference in the apneic thresholds (Pa_CO2 levels) seen between the three end-expiratory lung volumes. All values are stated as the mean ± SD. Statistical significance was accepted at a value of p < 0.05. 

View larger version (18K): [in this window] [in a new window]   Figure 3.   CO2-response curves for each element of breathing pattern obtained at three different end-expiratory lung volumes. Data from a representative dog. Solid lines are linear regression curves. Data points not included in each regression fall on the sharp vertical rise that occurs immediately above the apneic threshold, or fall below the apneic threshold. For clarity, regression curves are not shown for the inspiratory time-PaCO2 relationship.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Changes in EEVL at Any Given Pa_CO2 Level

The mean values of measured respiratory variables and arterial blood gases during control conditions (open-chest preparation without any alteration of EEVL or inspired CO₂ levels) are listed in Table 1. The effects of changes in EEVL on breathing pattern are shown in Figure 1. Although this figure was obtained from one dog at a Pa_CO2 of approximately 60 mm Hg, it is qualitatively similar to that obtained at all levels of Pa_CO2 in all eight dogs. Decreasing EEVL from "FRC" to 0.5 "FRC" increased both f (p < 0.01) and VT, although the increase in VT was not statistically significant. The increase in f was due to a reduction in TE (p < 0.01) with little change in TI. This increase in f was responsible for a corresponding increase in I (p < 0.01). Increasing EEVL from "FRC" to 1.5 "FRC" decreased f through a very prolonged TE (p < 0.01). Inspiratory time did not change. Although VT decreased slightly, the decrease was not statistically significant. Minute ventilation also decreased significantly (p < 0.01). Thus, the overall effect of reducing EEVL from "FRC" to 0.5 "FRC" at any level of Pa_CO2 was to induce a rapid and slightly deeper breathing pattern, whereas the opposite was true when EEVL was increased from "FRC" to 1.5 "FRC."

View larger version (20K): [in this window] [in a new window]   Figure 1.   Changes in breathing patterns at three different levels of end-expiratory lung volume. PaCO2 = 60 mm Hg. From top to bottom: AP = action potentials of phrenic nerve; PT = tracheal pressure; VT = tidal volume.

CO₂-Response Curves at Varying EEVL

CO₂-response curves obtained from representative dogs are presented in Figures 2 and 3. These curves are consistent with those obtained from all dogs. Pa_CO2 was slowly increased from low levels (ventilation = 0) to produce a progressive hypercapnia. We defined the apneic threshold as the Pa_CO2 level obtained within 15 s prior to the initial onset of ventilation. The apneic thresholds for 0.5 "FRC," "FRC," and 1.5 "FRC" were 37.6 ± 5.6, 42.4 ± 10.1, and 47.1 ± 11.1 mm Hg, respectively. There were statistically significant differences between the thresholds for "FRC" and 1.5 "FRC," and between 0.5 "FRC" and 1.5 "FRC" (p < 0.05). Although the apneic thresholds for 0.5 "FRC" and "FRC" were quite different, and we believe physiologically significant, the change was not statistically significant (p = 0.108). The effects of changes in EEVL on the apneic threshold of the I-Pa_CO2 relationship of one dog are illustrated in Figure 2.

View larger version (13K): [in this window] [in a new window]   Figure 2.   CO2-response curves at three different levels of end-expiratory lung volume. I = minute ventilation; PaCO2 = systemic arterial CO2 tension. Data from a representative dog. Arrows indicate apneic thresholds. Note reduction in apneic threshold as end-expiratory lung volume is decreased from 1.5 "FRC."

As CO₂ increased beyond the apneic threshold, ventilation increased until Pa_CO2 reached levels where all parameters of breathing pattern (I, VT, f, and TE), except TI, changed in a relatively linear manner. Regressions were obtained from data over these upper linear regions (solid lines in Figure 3). Average regression coefficients obtained from the statistical model are presented in Table 2. The slopes of all the response curves, except TI, were found to be significantly different from zero (p < 0.01) at any EEVL. As a consequence, as Pa_CO2 was increased, VT, f, and I increased, whereas TE decreased and TI remained unchanged.

How changes in EEVL caused shifts in all types of CO₂- response curves are shown in Figure 3. This figure was selected because it closely matches the average dog. On average, as EEVL changed, there were no significant differences found between the slopes of any relationship (Table 2). Therefore, the slopes were considered equal, and a shifting curve was defined as a significant change in the intercept of the regressed relationship. Changes in EEVL had no statistically significant effect on the position of the VT or TI response curves. As EEVL was decreased from "FRC" to 0.5 "FRC," upward shifts in the response curves were seen with I and f, and a downward shift occurred in the TE response curve. As EEVL was increased from "FRC" to 1.5 "FRC," there was a downward shift in the frequency-response curve, caused entirely by an upward shift in the TE-response curve. The response curve for I showed an apparent downward shift as EEVL was increased to 1.5 "FRC." However, this was due to a visible, but not statistically significant (p < 0.10), decrease in the slope (the intercept shifted in an upward direction, p < 0.01).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During the past several years, permissive hypercapnia has gained widespread acceptance as an effective ventilatory method with decreased risk of barotrauma in cases of ARDS and severe asthma requiring mechanical ventilation (1, 3, 5, 26). This involves controlled hypoventilation by lowering VT and increasing f, with increased Pa_CO2 levels (4). In ARDS, FRC and pulmonary compliance are typically reduced to one-third or less of normal (27). In asthma and chronic obstructive pulmonary disease (COPD), FRC is often significantly increased. The results of our study help clarify the relationships between alterations in lung volumes (such as those seen in ARDS and patients with asthma in which permissive hypercapnia may be used), increased levels of Pa_CO2, and the respiratory reflexes initiated by pulmonary receptors.

Our data demonstrated that, as EEVL is decreased from 1.5 times normal "FRC" to "FRC" and then to 0.5 "FRC," physiologically significant shifts occur in the CO₂-response curve for I. These shifts were due to a significant increase in f secondary to a marked decrease in TE as EEVL was lowered. Neither VT nor TI showed a significant change as EEVL decreased. Given the findings of Green and Kaufman (12) that only slowly adapting stretch receptors (SARs) showed significant changes in activity level with lung volume decreases to 0.5 "FRC," our results suggest that the shift in the CO₂-response curve is due to a partial withdrawal of the inhibitory influence of SARs on respiration that is seen at higher lung volumes. The decrease in SAR activity with decreased EEVL causes a marked increase in the level of ventilation, but it does not modify the relative response to progressive hypercapnia.

At any one of the three levels of EEVL investigated, progressive hypercapnia produced a predictable and previously described response curve of breathing pattern (28, 29). Tidal volume, f, and hence I increased with progressive hypercapnia, whereas TE decreased. Inspiratory time did not change as we altered the level of EEVL, and it also did not vary with progressive hypercapnia. Therefore, in our experiments, TE changed independently of TI, lending support to the idea that end-expiratory activity of slowly adapting pulmonary stretch receptors influences TE without significant changes of TE (9- 11). Additionally, there was a qualitative, but statistically insignificant, tendency for VT to increase as EEVL was decreased from one level to another. This can perhaps be accounted for by the observation of Green and Kaufman (12) that the initial effect of decreasing EEVL on VT was a significant increase, which, over a 90-s period, returned to but remained slightly above the control level.

We recognize that the method of plugging the trachea at end-expiration and measuring the PL does not measure FRC exactly. An increase or decrease by one-half of the PL will change end-expiratory reserve volume by approximately one-half, but will not alter the residual volume. Thus, this increase or decrease results in a less than one-half change in the relative "FRC." Additionally, a paralyzed animal has a decreased chest wall recoil component of FRC, and hence would exhibit an intrinsically lower EEVL. However, for the purpose of our study, we needed only an approximation of "FRC" to make the relative changes in the observed EEVL. Therefore, the levels of 0.5 "FRC," "FRC," and 1.5 "FRC" were considered as relative values for comparison.

Several past studies have examined the effects of lung volume changes on breathing pattern, both during hypercapnic and hypoxic conditions. These studies have been performed in both humans and anesthetized animals (30). In 1973, Cherniack and colleagues (33) investigated changes in phrenic nerve activity in dogs during progressive and steady states of hypoxia and hypercapnia. Studies were performed at levels of lung inflation similar to those in our current study. Whereas the study of Cherniack and colleagues was performed on closed-chest dogs in which both chest wall and lung afferent receptors would be expected to be active, we felt an open-chest preparation would allow for increased isolation of the pulmonary afferent reflexes we desired to investigate. Prior studies have suggested that mechanical restriction may influence VT in decreased EEVL induced by pneumothorax (34), and some of the changes in breathing pattern seen with decreased EEVL may come from afferent input from the chest wall (35). An open-chest study, keeping the chest wall widely opened away from the lungs, would eliminate most of the relative influence of chest-wall proprioceptors with EEVL changes.

Our findings are generally consistent with those of Cherniack and colleagues (33) regarding the effects of lung volume changes on respiratory drive during progressive hypercapnia. The response to progressive hypercapnia was equivalent to the classic expected results of increased I, increased f, and decreased TE. One observation of our study was at variance with the findings of Cherniack and colleagues, and it may be explained by the differences in the two methods. We found that the shifts in the CO₂-response curves with differing EEVL were essentially parallel, showing no significant slope changes at the different levels. The prior study, on the other hand, found a decrease in the slope of the hyperinflation response curve to 67% of the normal "FRC" curve slope, and an increase in slope of the deflation response curve to 137% of the control value. Although the slope values in our study did not change significantly, there was a trend of decreasing slope as "FRC" was increased (see Table 2). This difference in findings could be attributable to an effect from chest-wall afferent receptors on breathing pattern at different levels of EEVL. Although a closed-chest animal may be considered more physiologically "natural," it does not allow for separation of pulmonary afferent receptor activity from effects of afferent receptors in the chest wall. Additionally, the servo-ventilator used in our study offers an advantage, in that dogs have been shown to "train" ventilatory response to a constant-volume ventilator.

Increased pulmonary arterial PCO₂ has been shown to attenuate the activity of slowly adapting pulmonary stretch receptors (36). Because slowly adapting receptors are the apparent afferent nerves affected when EEVL is lowered within the range of 1.5 "FRC" to 0.5 "FRC," attenuation by increased PCO₂ may have altered the absolute results of the breathing pattern response to changes in EEVL. However, in our study any attenuating effect from CO₂ on slowly adapting receptors was approximately equal at all three levels of EEVL. Therefore, any decrease in receptor activity that occurred as EEVL was lowered was in addition to the decrease that would have resulted from increased Pa_CO2 alone.

The effect of altering EEVL on the apneic threshold was consistent with somewhat similar reports from prior studies. Earlier work by Younes and colleagues (37) indicated that static lung volume directly affects the duration of apnea. Our study quantified the apneic threshold in the conditions imposed by the experiment, which had not been previously investigated. As EEVL was increased, the apneic threshold increased, which is explained by increased activity of slowly adapting pulmonary stretch receptors. This was very apparent at an EEVL of 1.5 "FRC," when receptor activity would have been greatest. The apneic threshold at this EEVL was increased to a Pa_CO2 level well above that at which central drive should normally initiate respiration when EEVL is at control levels. We suggest that slowly adapting receptor activity plays an important role in determining the apneic threshold, with an ability to override central drive at levels of EEVL elevated above "FRC."

We recognize that it is often difficult to extrapolate findings in a study performed on healthy animals to disease states seen in human patients. However, the questions remain regarding the interactions and roles of pulmonary afferents in disease states where permissive hypercapnia may be utilized. Our findings may help clarify the role of pulmonary afferent receptors and breathing patterns in the setting of permissive hypercapnia and its use in the disease processes of ARDS and severe asthma. In patients with ARDS, FRC and pulmonary compliance are reduced, and I is lower because of a reduced physiologic lung volume from large areas of pulmonary edema. In this setting, SAR activity would be lowered, and hence may lead to a relative increase in I, a rapid deeper respiratory pattern, and a slightly intensified respiratory response to hypercapnia. The addition of PEEP in the ventilation of these patients would then increase FRC, with some possible attenuation of these responses. On the other hand, a higher level of FRC such as that seen in patients with severe asthma, would lead to increased SAR activity, and therefore, a possible increase in the apneic threshold and some blunting of the respiratory response to hypercapnia. Additionally, in a recent study by Mansoor and colleagues (38), the results pointed toward myelinated vagal afferents playing a role in the rapid, shallow breathing pattern seen in pulmonary fibrosis. The data from a rat model of pulmonary fibrosis suggested that the activity of SARs is decreased at any given lung inflation pressure, based upon a blunted Hering-Breuer inflation reflex.

In summary, the findings of our study are: (1) a significant parallel upward shift in the CO₂-response curves of I and f as EEVL is decreased from 1.5 "FRC" to "FRC" and then to 0.5 "FRC," and (2) a decrease in the apneic threshold as EEVL is decreased. Both phenomena may be attributable to partial withdrawal of slowly adapting pulmonary stretch receptor activity.