Journal of Applied Physiology Add DOIs to your references at manuscript stage!
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

J Appl Physiol 90: 2330-2340, 2001;
8750-7587/01 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (6)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by BuSha, B. F.
Right arrow Articles by Leiter, J. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by BuSha, B. F.
Right arrow Articles by Leiter, J. C.
Vol. 90, Issue 6, 2330-2340, June 2001

Identification of respiratory vagal feedback in awake normal subjects using pseudorandom unloading

Brett F. BuSha1, Brooke G. Judd2, Harold L. Manning1,2, Peggy M. Simon1,2, Brian C. Searle1, J. Andrew Daubenspeck1, and J. C. Leiter1,2

Departments of 1 Physiology and 2 Medicine, Dartmouth Medical School, Lebanon, New Hampshire 03756


    ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

Evidence of the Hering-Breuer reflex has been found in humans during anesthesia and sleep but not during wakefulness. Cortical influences, present during wakefulness, may mask the effects of this reflex in awake humans. We hypothesized that, if lung volume were increased in awake subjects unaware of the stimulus, vagal feedback would modulate breathing on a breath-to-breath basis. To test this hypothesis, we employed proportional assist ventilation in a pseudorandom sequence to unload the respiratory system above and below the perceptual threshold in 17 normal subjects. Tidal volume, integrated respiratory muscle pressure per breath, and inspiratory time were recorded. Both sub- and suprathreshold stimulation evoked a significant increase in tidal volume and inspiratory flow rate, but a significant decrease in inspiratory time was present only during the application of a subthreshold stimulus. We conclude that vagal feedback modulates respiratory timing on a breath-by-breath basis in awake humans, as long as there is no awareness of the stimulus.

proportional-assist ventilation; system identification; vagus; control of respiration


    INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

HERING AND BREUER FIRST DESCRIBED the reflex effects of volume feedback on respiratory timing in 1868 (3). They found that passive lung inflation beyond normal tidal volumes (VT) either shortened inspiratory time (TI) or prolonged expiratory time (TE), depending on the timing of the inflation stimulus within the respiratory cycle. Conversely, preventing inflation or expiration prolonged TI and TE, respectively. These reflex effects were abolished by sectioning the vagi. The vast majority of studies of the Hering-Breuer reflex have been in anesthetized or decerebrate animals, but the Hering-Breuer reflex has also been studied in humans. Electrical stimulation of the vagi in anesthetized humans produced reflex shortening of TI, and bilateral blockade of the vagi abolished this reflex (8). However, unilateral electrical stimulation of the vagi in awake humans elicited a sense of respiratory discomfort but no reflex changes in breathing (1). During anesthesia, airway occlusion at end-expiration in adult humans increased TI during the following breath (18). However, electrical stimulation and airway occlusion are potent stimuli and not typical of eupneic breathing. It has been difficult to demonstrate that subtle changes in VT actually modulate respiratory timing in awake subjects. There are two methodological problems. If the change in VT is small, it is difficult to measure the small changes in respiratory timing that may ensue. However, if the stimulus is greater, conscious perception of the stimulus and behavioral responses to the stimulus may modify and obscure any reflex responses. To avoid anesthesia and the confounding influence of conscious control of respiratory responses, investigators have examined the Hering-Breuer reflex in humans during sleep. Hamilton and co-workers (9) investigated the effect of passive lung inflation during wakefulness and during stable non-rapid eye movement (NREM) sleep in laryngectomized patients. During sleep, the Hering-Breuer reflex was elicited only when inflation volumes of 0.54-2.10 liters were applied at end-inspiration to augment spontaneous respiratory efforts and thereby prolong TE. Reflex prolongation of TE was not elicited in most subjects until volumes >1.40 liters were used, indicating that there may be a threshold necessary to elicit the reflex during sleep. No change in respiratory timing was apparent in waking subjects during similar maneuvers. Iber and colleagues (10) found that, during NREM sleep, a mean VT of ~1.5 liters delivered at the end of inspiration prolonged TE in normal subjects, but VT values as large as 2.5 liters had no effect on respiratory timing in patients who had undergone lung transplantation. Hence, the Hering-Breuer reflex can be elicited in sleeping humans, but a large inspiratory volume threshold must be exceeded before reflex effects on TE can be seen.

During proportional-assist ventilation (PAV), breathing is initiated by the subject and is assisted in proportion to each individual's effort. Unloading of the inspiratory muscles occurs when the assist augments the subject's instantaneous effort by a predetermined percentage of that individual's respiratory resistance (Rrs) and elastance (Ers; Refs. 21, 22). Steady-state unloading of the respiratory muscles with PAV during wakefulness did not significantly reduce neural output when chemical stimuli were controlled (7). Similar results were found when PAV was applied during rapid eye movement and NREM sleep (14). In contrast to previous studies of the Hering-Breuer reflex during sleep, Meza et al. (14) found no evidence of modulation of breathing by lung stretch receptor feedback, and they concluded that CO2 levels in the blood control respiration during unloading. However, VT was not augmented beyond 30% of the control values during these sleep studies, and the increase in VT associated with unloading may not have exceeded the threshold of the Hering-Breuer reflex during sleep. Nonetheless, these studies confirm past studies of waking subjects in that there was little evidence of an active Hering-Breuer reflex during wakefulness at normal VT values, and even during sleep VT values that would be well above the perceptual threshold during wakefulness are necessary to elicit the reflex.

In the present study, we reinvestigated the role of vagal feedback in breath-by-breath control of inspiratory timing during wakefulness. We hypothesized that, if lung volume were increased in awake subjects unaware of the stimulus, vagal feedback would modulate breathing on a breath-to-breath basis. Our study differs from past studies in that we systematically increased VT and inspiratory flow (VT/TI) below and above the threshold of sensation. PAV allows augmentation of respiratory effort while the subject breathes spontaneously, and we increased VT and VT/TI in normal awake humans on a breath-by-breath basis after a pseudorandom binary sequence (PRBS) of sub- or suprathreshold PAV. The respiratory response to the PRBS was modeled by using a linear system estimation procedure based on the prediction error method. These methods permit an analysis of subtle changes in inspiratory variables resulting from small increases in VT and VT/TI that could not be detected by using classic techniques such as single-breath or steady-state changes in volume. In this way, we were able to assess the combined effect of volume- and flow-related feedback on respiratory timing during wakefulness in the absence of cortical influences.


    METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

Subjects

We recruited 17 normal adult volunteers without prior or current respiratory disease. One subject had a history of smoking. None of the subjects was aware of the objectives of the study, and none of the subjects had significant knowledge of respiratory psychophysics. The local Institutional Review Board approved the study, and informed consent was obtained from all subjects before participation in the study.

Instrumentation and Setup

Each subject sat recumbent in a dental chair throughout the study. Each subject wore a nose clip and breathed though a mouthpiece attached to a PAV (University of Manitoba, Winnipeg, MB) and listened to white noise so that he or she was not distracted or given clues about the activity of the ventilator during the experiment. The dead space of the breathing circuit was ~0.1 liter. We used a prototype PAV machine, the Winnipeg Proportional Assist Ventilator (loaned to us by M. Younes). We recorded analog signals from the ventilator proportional to airway pressure and flow. We supplied an external, analog command signal that caused the piston within the PAV to generate an inflation pressure proportional to the command signal. The assist function of the PAV machine was not used; rather, we used software written in LabVIEW (National Instruments, Austin, TX) to calculate the assist level and command the ventilator to give the appropriate assist pressure. The assist pressure was calculated by using the instantaneous flow and volume and each subject's Rrs and Ers as described by Younes (21). It was necessary to command the prototype PAV machine with the computer to control the level of PAV on a breath-by-breath basis.

For analysis, airway pressure measured at the mouth (Pm) and inspiratory airflow were recorded directly from an analog output supplied by the ventilator, and flow was integrated with a zero-crossing reset to calculate VT. End-tidal CO2 partial pressure (PETCO2) was measured at the mouth with a CAPSTAR-100 carbon dioxide analyzer (CWE, Ardmore, PA). During the estimation of respiratory mechanics, airflow at the mouth was occluded with a two-way shutoff valve (9340 series inflatable balloon controlled with an 8230 series automatic controller, Hans Rudolph, Kansas City, MO). Programs written with LabVIEW software controlled the ventilator and balloon valves and displayed and recorded the data.

Experimental Protocols

Each experiment consisted of three parts. First, we estimated the passive mechanical characteristics of the respiratory system of each subject (Rrs and Ers). Next, we determined the threshold of detection of respiratory assistance in each subject. Finally, we gave PAV on a breath-by-breath basis by using a PRBS.

Estimation of respiratory mechanics. We used a modification of the interrupter method to estimate Ers and Rrs (5, 20). We obtained passive measurements of respiratory mechanics in our subjects by controlled mechanical hyperventilation. After a few minutes of spontaneous breathing to acclimate to the ventilator circuit, the ventilator was switched to the controlled mandatory ventilation mode. Respiratory rate, VT, and VT/TI rate were adjusted until the subject was comfortable and passive ventilation was achieved. Once comfortable ventilator settings were achieved, the subject was allowed to breathe for 10-15 min to acclimate to positive-pressure ventilation. The subjects were invariably hypocapnic during this period.

Airway pressure at the mouth, flow rate, and VT were monitored continuously to ensure passive ventilation. We defined passive ventilation by the absence of variability in the respiratory waveforms and the absence of negative deflection of Pm during inspiration. When the subject was relaxed, the airway was occluded every five to eight breaths for 400-600 ms during late inspiration after delivery of 60-80% of the subject's VT. The occlusions were judged satisfactory when no subject effort was visible during the breath, when the flow and pressure waves were similar to previous breaths, and when the plateau during the occlusion was flat. Approximately 20 satisfactory occlusions were recorded from each subject.

Respiratory mechanics were calculated from VT/TI, VT, and Pm recorded during occluded breaths. Static Ers was calculated from the difference between the plateau pressure and the end-expiratory pressure divided by the VT of that breath. Rrs was calculated from the difference between the Pm measured before the occlusion and the subsequent plateau pressure divided by the flow immediately preceding the occlusion.

Determination of 50% threshold of sensation. We used a forced choice protocol to determine the threshold of perception of assisted ventilation (11). In this protocol, equal levels of elastic and resistive assistance (e.g., 5% elastic and 5% resistive assistance) were applied on individual breaths. The levels of assistance varied from imperceptible to well above the perceptual threshold (0-25% assistance). A forced-choice selector box (two buttons, one "yes" and the other "no") was placed on the subject's lap, and a red light-emitting diode (LED) was placed in front of the subject. The subject was required to decide whether the ventilator helped or did not help on the breath that followed immediately after the LED was lit; the subject indicated his or her choice by pressing the appropriate button on the box. Forced-choice decisions were recorded on-line. The ventilator randomly applied five levels of PAV ranging from 0 to 25% with three unassisted breaths between each assisted breath. The 20-breath sequence of assisted breaths was repeated 11 times in a randomized-block design. A computer program controlled the ventilator, illuminated the LED at the desired time, recorded the subject's choice, and displayed and recorded VT/TI, VT, and Pm.

The probability of detection of ventilator assistance was calculated at each level of assistance by dividing the number of times the subject detected assistance by the total number of times assistance was applied. The first 20-breath sequence was discarded from the analysis to remove any effect of learning the procedure. We defined the perceptual threshold as the level of PAV at which assistance was detected >50% of the time. To determine the perceptual threshold of PAV in each subject, the probability of detection at each assistance level was transformed into logits [units of probability in which the probability of detection at each level of assistance, P, is transformed by using the following equation: a logit = ln(P/1 - P)]. For each subject, the logits were plotted as a function of percent assistance, and linear regression was used to fit the data (Fig. 1). The x-axis crossing in this logit analysis is equal to the 50% threshold of sensation (TOS50%). We chose the TOS50% as the greatest stimulus that was still below the perceptible threshold.


View larger version (10K):
[in this window]
[in a new window]
  		Fig. 1.   The probability (P) of detecting proportional-assist ventilation (PAV) was plotted as a function of the level of PAV. The probabilities were converted to logits, and regression was performed to determine the best linear fit of the logits on the assist levels. The x-axis crossing (solid line at x = 0) represents the 50% detection threshold.

PRBS assist protocols. Each breathing trial was composed of three sequences applied in the following order: 70 unassisted control breaths, a 127-PRBS of assisted and unassisted breaths, and another 70 unassisted control breaths (267 sequential breaths were measured). The unassisted control breaths recorded before and after the pseudorandom sequence enhance the ability of the system identification procedure to model the low-frequency response characteristics of the system that occur during the initiation and the conclusion of the PRBS. A minimum of 10 min of rest separated the two assist protocols that each subject performed during an experiment.

In protocol 1, subjects were tested twice with separate 267-breath sequences: one sequence at the individual's previously described TOS50% and another sequence at twice TOS50%. The order of the application of these test sequences was determined by a randomized block design. In protocol 2, subjects were tested twice with separate 267-breath sequences at the subject's TOS50%.

Determination of First Breath Response of Respiratory Variables

Respiratory variables (VT, TI, and TE), mean VT/TI, instantaneous respiratory muscle pressure output (Pmus), inspiratory respiratory muscle output integrated per breath (int Pmus), and the pressure applied to the mouth by the PAV were calculated on-line by using programs written in LabVIEW and MATLAB (Math Works, South Natick, MA). The PETCO2 was recorded continuously, and breath-by-breath values were calculated for each subject. By use of the continuous measurements of flow, VT, and Pm and the values of Ers and Rrs determined previously in each subject, Pmus was calculated only during inspiration for each breath according to the equation (23)
Pmus<IT>=</IT>Ers<IT>×</IT>V<SC>t</SC><IT>+</IT>Rrs<IT>×</IT>flow<IT>−</IT>Pm.
We used a general system identification technique known as the prediction-error method to calculate the response to a single assisted breath for each respiratory variable for each PRBS (13). This method assumes that the output of the system (e.g., VT) is linearly related to the input function (e.g., application of PAV) and to random noise. Individual data sets were transformed to remove the mean of the signal. A linear difference equation was constructed to model the data [specifically the ARX variant, where AR refers to the autoregressive component ay(t) and Y refers to the exogenous input bu(t)] as follows
y(t)=−<IT>a<SUB>1</SUB>y</IT>(<IT>t−1</IT>)<IT>−…−a</IT><SUB><IT>n<SUB>a</SUB></IT></SUB><IT>y</IT>(<IT>t−n<SUB>a</SUB></IT>)<IT>+b<SUB>1</SUB>u</IT>(<IT>t−1</IT>)

<IT>+…+b</IT><SUB><IT>n<SUB>b</SUB></IT></SUB><IT>u</IT>(<IT>t−n<SUB>b</SUB></IT>)<IT>+e</IT>(<IT>t</IT>)
where t was the sample number, u(t) was the system input (in our case the PRBS), and e(t) was the error term. The parameters a1...n and b1...n were estimated by using a least squares algorithm. To find the best fitting model for the data, models of different order and with different values of coefficients were constructed. The fit of the model was determined by the difference between model estimates and real output data (i.e., the prediction errors or residuals). Because the data used to determine the model structure and to check the fit of the model were the same, as model order was increased the prediction errors decreased. To account for this automatic decrease in prediction errors, selection of the optimal order of the model and the actual coefficients was made by using Akaike's information theoretic criterion (AIC). The AIC was calculated for the multiple models as follows
AIC<IT>≈</IT>log [(<IT>1+2n/N</IT>)<IT>×V</IT>]
where n is the total number of estimated parameters; N is the length of the data record; and V is the loss function (the sum of the squared differences between the model and the actual data). The value of AIC increases whenever the model uses a greater number of coefficients or the fit of the data to the model is worse. Therefore, the optimal model has the smallest order that still provides a reasonable fit to the data. After constructing multiple models from the respiratory data recorded during the PRBS, we used the coefficients producing the model with the lowest AIC value to define the impulse-response characteristics of the first through fifth breaths.

To ascertain the ability of the linear model to accurately describe the respiratory response to a single breath of PAV (which is potentially nonlinear), a cross-correlation function was calculated between the input data and the prediction errors. A correlation between the input and the prediction errors would indicate that information in the residuals is related to the input, and thus the model did not fully describe the response (13). All system identification procedures were analyzed on a computer using MATLAB and the MATLAB System Identification Toolbox (Math Works).

Model Estimation of Respiratory System Responses

We implemented the model of respiratory mechanics developed by Younes et al. (19, 23, 24) to estimate the response of the respiratory system to the pseudorandom sequence of PAV in the absence of neural reflexes; the model response depends solely on the mechanical characteristics of the system. The model converts a preselected respiratory neural output into the appropriate mechanical output of the respiratory system. We used a simplified neural drive in which the phrenic neural output was modeled as a triangle with an inspiratory ramp and short postinspiratory decline in neural activity and no inspiratory or expiratory neural activity during expiratory time. The neural output was used to determine the muscle force and pressure output of the diaphragm after the initial conditions were chosen. Iterative values for flow and volume were calculated and summed over time. The output of the model consisted of Pmus, VT, VT/TI, and respiratory timing for a sequence of breaths. We adapted the model to run the same sequence of control breaths (70), a PRBS of PAV as applied to our human subjects, and a final sequence of unassisted breaths (70). Once the model produced a string of control and assisted breaths, we used the same analytic tools described above to estimate the impulse response of the model. The model was implemented in SIMULINK (Mathworks), and we used an explicit Runge-Kutta method (ODE45 in SIMULINK) to estimate the solutions to the differential equations in the model with a maximum step size of 0.001 s.

Statistics

Data were expressed as means ± SD. Individual respiratory variables were compared across trials with paired t-tests or with unpaired t-tests when appropriate. A one-sample t-test was used to determine whether the change in a respiratory variable was significantly different from zero. Changes in PETCO2 between the control periods and the PRBS were analyzed with a one-way repeated-measures ANOVA, and pair-wise comparisons were made by using the Bonferroni adjustment. A P value <= 0.05 was considered significant.


    RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

Acquisition software malfunctioned during one experiment, and the data from this subject were not included in the analysis. A second subject was unable to relax effectively during the measurement of respiratory system mechanics, and this subject was not tested further. The remaining 15 subjects completed all of the protocols. Subject characteristics, Ers, Rrs, and TOS50% of PAV are shown in Table 1. The mean value for Ers for all subjects was 12.0 ± 5.2 cmH2O/l, and the mean Rrs was 5.4 ± 2.1 cmH2O · l-1 · s. The mean value for TOS50% was 12.7 ± 4.7% assist.

                              
View this table:
[in this window]
[in a new window]
  		Table 1.   Subject characteristics

There were two technical problems calculating firstbreath responses. First, we could not estimate Pmus in some subjects. The piston in our prototype ventilator was underdamped (M. Younes, personal communication), and it oscillated at very low VT/TI rates. The oscillation prevented accurate assessment of Pmus and prevented accurate calculation of Pmus and int Pmus during periods of low VT/TI. This problem occurred in two subjects: in one subject, we obtained no Pmus estimates under either test condition, and in the other, we obtained accurate estimates of Pmus only at twice TOS50%. All other measurements were made in these subjects. The second problem was that some subjects had augmented breaths (sighs). These breaths were much greater in amplitude than the rest of the breaths and would have disproportionately affected the average breath size. Inspiratory sighs were defined as a VT greater than the mean VT plus three times the standard deviation, and the corresponding values of TI, VT/TI, Pmus, and int Pmus for the sigh were labeled. If the sigh occurred before or after the PRBS, all of the data for the breath were omitted from analysis. If a sigh occurred during the PRBS, individual values were replaced with the mean value calculated only from the PRBS. Breaths were not removed from the PRBS to avoid altering the randomness of the assisted breaths. Approximately 50% of the PRBSs contained one to three sighs.

Responses to Respiratory System Unloading with PAV

Brief segments of the PRBS from two subjects during assist at the TOS50% level are shown in Figure 2. Most subjects had imperceptible changes in VT, VT/TI, TI, and Pmus during the PRBS as shown in Fig. 2A. However, two subjects had much larger increases in VT than the other subjects during PAV at both TOS50% and twice TOS50%; >10% of the breaths within the PRBS were twice the normal VT (shown in Fig. 2B). When VT or flow is large, the absolute magnitude of the assist is large, and Pm increases and may exceed the perceptual threshold even though the percent level of assist is low. Large breaths defeated the purpose of PAV at the TOS50% level because subjects could reliably detect a stimulus that should have been below the threshold of sensation. Therefore, data from these two subjects were excluded from analysis.


View larger version (16K):
[in this window]
[in a new window]
  		Fig. 2.   Flow, tidal volume (VT), calculated inspiratory respiratory muscle pressure output (Pmus), and the level of PAV during a 5-breath segment of the pseudorandom binary sequence (PRBS) have been plotted for 2 subjects (A and B) as a function of time. The respiratory system elastance (Ers) was 9.4 cmH2O/l, the respiratory system resistance (Rrs) was 7.4 cmH2O · l-1 · s, and the assist was 18.5% in subject A. The results from subject A were typical of most subjects; the level of PAV was small and resulted in single-breath changes in flow, VT, and Pmus that were imperceptible to the subject. In contrast, 2 subjects augmented respiratory effort during the application of PAV, and flow, VT, and Pmus increased well into the perceptible range, as shown for 1 of these subjects (B). The Ers was 8.7 cmH2O/l, the Rrs was 5.6 cmH2O · l-1 · s, and the assist was 16.5% in subject B.

The effects of unloading on VT, VT/TI, TI, and int Pmus with subthreshold PAV from a single subject are shown in Fig. 3. The impulse responses were examined over five breaths: the first assisted breath and the four following breaths. In this subject, there were significant first-breath effects of the unloading. The VT of the assisted breath exceeded the unassisted VT by more than the 95% confidence interval and was below the 95% confidence interval on the second, unassisted breath. Mean VT/TI was also elevated on the assisted breath, fell below the control level on the second breath, and returned to baseline on subsequent breaths. The changes in VT and VT/TI are the expected mechanical consequences of PAV. TI and int Pmus were reduced on the first breath in this subject as well. As shown in Fig. 3, there were significant second-breath effects in some subjects, but these were inconsistent when the entire group was examined. When present, second-breath effects were always compensatory: for example, an increase in VT on the first breath was followed by a decrease in VT on the second breath as shown in Fig. 3.


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3.   Estimated impulse responses of VT, inspiratory flow rate (VT/TI), inspiratory time (TI,) and integrated respiratory muscle pressure per breath (int Pmus; iPmus) in 1 subject after 1 breath of subthreshold PAV were plotted as a function of breath number. The impulse response shows the change in each variable as if the stimulus had been applied on the first breath only. The data shown are means ± 95% confidence interval (the confidence intervals were derived from the estimated model standard errors). There was a significant increase in VT and VT/TI and a significant decrease in TI and int Pmus on the assisted breath. This subject also demonstrated compensatory second-breath effects; VT and VT/TI fell below the control level. The second-breath changes in TI and int Pmus were small, and all variables returned to control levels by the third breath in the estimated impulse response.

The average first-breath responses for all subjects for subthreshold and suprathreshold unloading are shown in Tables 2 and 3, respectively. We obtained data from 12 subjects during subthreshold unloading. Of the 15 subjects completing protocol 1, two subjects, shown in Fig. 2, were excluded because of aberrant responses, and one subject was excluded from the subthreshold unloading because we could not obtain an estimate of Pmus secondary to underdamping of the ventilator. Two of the five subjects who participated in protocol 2 (subthreshold unloading only) also participated in protocol 1, and their mean responses from protocols 1 and 2 are presented in Table 2. The average Pm applied during TOS50% was 0.83 ± 0.40 cmH2O. This was significantly less than the mean Pm applied during twice the TOS50% PAV, which was 2.00 ± 0.96 cmH2O (P = 0.002). The average response to subthreshold PAV was a significant increase in VT (0.040 ± 0.039 liters, P = 0.004) and in VT/TI (0.044 ± 0.035 l/s, P = 0.008). Subthreshold PAV also resulted in a significant reduction in TI (-0.040 ± 0.044 s, P = 0.006). Integrated Pmus was reduced, but the reduction was not significant (-0.164 ± 0.295 cmH2O/breath, P = 0.092). There was no significant change in PETCO2 (P = 0.39) during subthreshold unloading. During suprathreshold unloading, VT, VT/TI, and Pm were increased significantly. Although the increase in Pm during suprathreshold unloading was slightly more than double the subthreshold Pm, the distribution of VT and VT/TI differed slightly compared with subthreshold unloading: the increase in VT was proportionately greater and the increase in VT/TI proportionately less during suprathreshold unloading compared with subthreshold unloading. Despite greater unloading during suprathreshold PAV, the reduction in TI was less. The reduction in int Pmus was greater than the reduction during subthreshold unloading but still not significantly different from zero. The TI response was much more variable during suprathreshold unloading. The coefficient of variation for the group was 344% during suprathreshold unloading compared with 110% during subthreshold unloading. The PETCO2 fell significantly by ~1.9 ± 1.6 Torr (P < 0.005) during suprathreshold unloading.

                              
View this table:
[in this window]
[in a new window]
  		Table 2.   Summary of first-breath responses to subthreshold unloading


                              
View this table:
[in this window]
[in a new window]
  		Table 3.   Summary of first-breath responses to suprathreshold unloading

Consistency of PRBS Output

To determine whether the PRBS of PAV provided consistent results across trials, five subjects were tested twice at TOS50%. The changes in the respiratory variables (VT, VT/TI, TI, and int Pmus) were similar between the two identical protocols (P > 0.30) and similar to the results shown in Fig. 2.

Model Estimation of Respiratory Responses

We estimated the effect of the PRBS of PAV on the respiratory system by using a computer model of the respiratory system (19, 23, 24). We used the mean values of Ers and Rrs from our subjects, the average level of subthreshold assist, and the mean TI from all subjects during the control period to set the initial conditions of the model (Table 4). We estimated the vital capacity from the height, age, and sex of each subject (15). We chose a level of neural drive sufficient to generate a VT of 0.5 liters. We held neural TI and neural drive constant during the simulation; hence, the results of the model reflect the effect of PAV on a respiratory system with no reflex responses to the unloading. We used the model to generate a sequence of control breaths, a sequence of breaths in which PAV was applied by using the PRBS, and a final period of control breaths. The impulse response of the model was estimated with the same system identification method used to characterize the impulse response of the actual data from each subject. The predicted changes from control values for the first through fifth breath impulse responses derived from the simulation are shown in Fig. 4 (dashed lines), and the actual average changes ±95% confidence intervals of our subjects to subthreshold unloading are plotted along with the predicted model responses. The VT and VT/TI responses of the passive system were very close to the actual values we measured. In contrast to the significant shortening of TI observed in our subjects, there was no change in TI or TE estimated from the model. Integrated Pmus declined slightly, but the decline estimated from the model was approximately one-half the size of the average decline in Pmus we measured. The decline in Pmus occurred in the passive model as a result of the increase in VT during PAV and the associated change in the length-tension relationship of the diaphragm, so that even at constant neural drive, the muscle force developed was less throughout each assisted breath.

                              
View this table:
[in this window]
[in a new window]
  		Table 4.   Respiratory model parameters



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 4.   Actual (solid line, mean ± 95% confidence intervals) and predicted impulse responses (dashed lines) of VT, VT/TI, TI, and iPmus to 1 breath of assist have been plotted as a function of breath number. Predicted responses were derived from a model of the respiratory system that reflects only the mechanical characteristics of the system; i.e., no reflex responses are incorporated into the model.


    DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

We investigated the role of the Hering-Breuer reflex in awake human subjects. We applied a PRBS of PAV at levels below and above the perceptual threshold to "overinflate" the respiratory system and stimulate the Hering-Breuer reflex. We modeled the PRBS data with linear difference equations to obtain impulse-response profiles of VT, VT/TI, TI, and int Pmus. The respiratory responses to PAV unloading may be nonlinear and thus may be difficult to characterize with a linear model. A cross-correlation function was calculated between the model residuals and the input data to determine whether the respiratory response to PAV could be accurately characterized with the linear model used in this analysis. There was no correlation between the prediction errors and the input data. Thus the prediction errors are independent of the input, and the linear model adequately describes these respiratory responses to PAV. VT and VT/TI were the possible stimuli of the Hering-Breuer reflex, and TI and int Pmus were the possible response variables. We limited the possibility of cortical influences on respiratory responses to our stimuli by applying subthreshold levels of PAV, and we limited the possibility of humoral effects by augmenting mechanical respiratory output intermittently and by small amounts. We found that VT and VT/TI significantly increased in response to a PRBS of PAV during the application of subthreshold and suprathreshold stimuli. We expected, as evidence of the Hering-Breuer reflex, TI shortening and/or a reduction in int Pmus during assisted breaths (4, 12, 18). There was a significant decrease in TI during unloading of the respiratory system below the perceptual threshold, and although int Pmus was reduced by PAV, the reduction did not achieve statistical significance for the group as a whole. When the stimuli were greater and potentially perceptible, there were no significant changes in TI and int Pmus. These data support the hypothesis that vagal feedback modulates the pattern of breathing on a breath-by-breath basis during wakefulness. Furthermore, the failure of stimuli within the perceptual range to modulate breathing suggests that cortical responses to loading and unloading may have masked the activity of vagally mediated reflexes in previous studies of waking subjects.

Pseudorandom Application of Small Stimuli

During wakefulness in humans, perturbations to the respiratory system result in a response that can be separated into three components: reflex, humoral, or conscious responses. This investigation concentrated on the reflex responses, and thus it was necessary to limit humoral and conscious stimuli. Application of PAV in a PRBS permitted us to apply the methods of system identification and model the impulse responses of selected respiratory variables by using linear difference equations. We were able to apply multiple stimuli over a relatively short period of time and calculate a single-breath response. The stimuli were subtle, below the threshold of sensation. The randomness of the application of the stimuli also helped limit perception of the stimuli. Finally, application of small stimuli on a pseudorandom sequence limited changes in humoral factors that might alter the reflex ventilatory response being tested. During application of subthreshold PAV, the reduction in TI cannot be attributed to humoral factors; PETCO2, for example, did not change during the subthreshold PRBS of PAV.

The actual change in TI that we measured during subthreshold unloading was small. Before concluding that modulation of TI by the Hering-Breuer reflex is unimportant, it is worth noting that the stimulus-response curve of the reflex described by Clark and von Euler (4) was very steep when VT was increased. Even for large increases in VT, only small changes in TI resulted. Pulmonary stretch-receptor feedback prolongs TI when changes in VT are retarded more effectively than it shortens TI when changes in VT are accelerated. The small changes in TI that we detected do, therefore, provide strong evidence of vagally mediated breath-to-breath modulation of respiratory activity. Vagal feedback also inhibits phrenic nerve activity within each breath in animals (2, 12), and we thought we might see a fall in int Pmus as a result of unloading with PAV. This expectation of a change in int Pmus may have been unrealistic. Inhibition of phrenic activity has a significant volume threshold in animal experiments, and, as a result, there is little modulation of the profile of phrenic activity within each breath until a volume threshold is reached at which time phrenic activity is abruptly terminated. In contrast, hypoglossal and laryngeal nerve activities have a low volume threshold, and the profile of hypoglossal and laryngeal activity is modulated within each breath before the termination of inspiration is reached (12). Thus the reflex change in int Pmus that we were hoping to detect was only the change arising from the early termination of phrenic activity, which was very small. Despite that, the change in int Pmus was in the direction we expected. However, some of the change in int Pmus that we observed can be accounted for on the basis of the effect of PAV on the length-tension relationship of the inspiratory muscles. The actual change in int Pmus was slightly but not significantly greater than the change predicted solely on the basis of the mechanical characteristics of the respiratory system. Therefore, we conclude that subthreshold unloading with PAV did elicit a Hering-Breuer reflex and that this was manifested by shortening of TI. We found no evidence of a reflex reduction in int Pmus, which may be attributed to the very small reflex change that we were trying to detect.

Although the directions of the changes in TI and int Pmus were similar during both sub- and suprathreshold unloading, the responses were less consistent among subjects during suprathreshold unloading and therefore failed to reach statistical significance. We attribute this variability, which may mask Hering-Breuer reflex effects, to conscious perception of and responses to the respiratory stimulus. There is an additional factor confounding interpretation of TI and int Pmus during suprathreshold unloading. The subjects were hypocapnic during the PRBS with suprathreshold PAV compared with the initial and final sequences of control breaths used to define the impulse response. The hypocapnia may have reduced respiratory drive and slowed respiration, thereby prolonging TI and TE, reducing peak phrenic activity, and reducing int Pmus. Subjects were aware of the suprathreshold assistance: some of the subjects spontaneously mentioned it; this never happened during subthreshold unloading. We believe that conscious perception contributed to the variable TI and int Pmus responses, but it is also possible that the changes in TI and int Pmus resulted from hypocapnia during the suprathreshold unloading.

The results obtained during subthreshold unloading with PAV are consistent with other investigations of the Hering-Breuer reflex in humans. During anesthesia, Polachek et al. (18) elicited the Hering-Breuer reflex by occluding the airway at end-inspiration and by augmenting VT, and the Hering-Breuer reflex was active within the range of a spontaneous VT during anesthesia. By augmenting VT and achieving inflation volumes >1 liter, Hamilton and co-workers (9) demonstrated significant vagal influence on breathing during sleep in laryngectomized subjects. The authors were unable to demonstrate a similar reflex effect during wakefulness, even with greater inflation volumes. These studies, coupled with our results from subthreshold unloading of the respiratory system in awake subjects, demonstrate the ability of vagal reflexes to modify breathing on a breath-by-breath basis but also suggest that the effects of the Hering-Breuer reflex become apparent only after the removal of cortical influences by anesthesia or sleep or during the application of subthreshold stimuli. However, sleep and anesthesia may blunt vagally mediated reflexes (14, 16), and relatively large volume changes were required to modify respiratory timing in anesthetized and sleeping subjects. Thus experiments during sleep and anesthesia may not accurately reflect the capacity of the respiratory system to respond to small changes in volume and flow that are well below the thresholds of stimulation described in sleeping and anesthetized subjects.

Aberrant Responses

We excluded two subjects from our analysis. During the application of PAV, both subjects augmented respiratory effort (increased TI and increased effort) to an extraordinary degree (one subject's aberrant response is presented in Fig. 2). The response was so striking and so frequent that we wondered about other reflex mechanisms that might increase VT. Head's paradoxical reflex is a vagally mediated response to small changes in VT that results in a large augmented VT. The reflex is well described in neonates, but not, so far as we know, seen in adults. However, the response of these two subjects was reminiscent of Head's paradoxical reflex. We have no explanation why these two subjects responded so consistently to PAV in this way when the majority of subjects had more subtle responses in the opposite direction (shorter TI and a trend toward reduced effort).

Limitations of the Method

Application of PAV resulted in small but significant increases in VT and VT/TI. In the anesthetized dog, increased airflow administered during a single breath results in a vagally mediated increase in the moving time-averaged phrenic neurogram and shortening of TI (17). Georgopoulos et al. (6) investigated a similar reflex in adult humans. Subjects were trained to breathe on a volume-cycled ventilator in the assist/control mode in which all breaths were subject triggered. As VT/TI increased, TI, TE, and Pmus all decreased. These data suggest that the reflex decrease in TI found in our data could be the result of an increase in VT/TI rather than the increased VT. Our experiments were not designed to distinguish between volume and flow stimuli, and we conclude that small, imperceptible increases in VT and/or VT/TI significantly modify breathing on a breath-by-breath basis. On the basis of past work by others and the pattern of the response, we believe that these responses are vagally mediated, but we cannot exclude nonvagal reflexes that may originate from the upper airway, for example.

To calculate Pmus, it was necessary to determine the resting Rrs and Ers for each subject. Although the range of our resistance and elastance measurements was large (see Table 1), the mean values were similar to other published results (14). The interrupter technique of measuring respiratory mechanics was developed for use in passively ventilated patients. Despite our best efforts, our subjects may have been incompletely relaxed during measurements of respiratory mechanics. Any errors estimating Ers and Rrs would increase the variability of the results, and our estimates of Ers and Rrs are more variable than measurements made by others (14). Furthermore, errors in the measurement of Rrs and Ers would cause errors in the calculation of Pmus and int Pmus. Such an error would change the magnitude of Pmus and int Pmus but would not change the direction of the response. This may explain why the largest value for the coefficient of variation occurred for int Pmus and may also explain why, at low levels of assist, there was a significant reduction in TI but not in int Pmus.

The main finding of this study was a change in TI, and we relied on measures of timing derived from mechanical events. We recorded the activity of the diaphragm with electromyogram surface electrodes placed on the anterior axillary line over the sixth and seventh intercostal spaces on the right side of the chest in five subjects. We found that the signal-to-noise ratio was satisfactory in the data from two subjects. We found no change in the relationship between neural events measured by the surface electrodes and the mechanical events measured by TI, whether breaths were assisted or not (data not shown). Hence, as best we can tell, mechanical events faithfully represented neural events during control and PAV conditions. Therefore, the reduction in TI during subthreshold unloading, although derived from measurements of mechanical events, consistently reflected neural events.

In summary, it has been difficult to demonstrate that the Hering-Breuer reflex is active in awake adult humans by using large-volume step changes as the stimulus. However, we found that subtle changes in VT and flow elicited by using a PRBS of PAV modified respiratory timing in a way consistent with the Hering-Breuer reflex in awake normal humans. The stimuli were small and below the perceptual threshold. The responses were also small and apparent only when the subject was not aware of the stimulus. We believe the results of this study demonstrate that vagally mediated reflexes modulate the timing of respiration on a breath-to-breath basis in awake humans.


    ACKNOWLEDGEMENTS

We thank Dr. Magdy Younes for lending a prototype Winnipeg Proportional Assist Ventilator to us.


    FOOTNOTES

This work was support by National Heart, Lung, and Blood Institute Grants HL-07449 and HL-29068 (to J. A. Daubenspeck).

Address for reprint requests and other correspondence: B. F. BuSha, Dept. of Physiology, Borwell Bldg., Dartmouth Medical School, 1 Medical Center Drive, Lebanon, NH 03756 (E-mail: brett.bu.sha{at}dartmouth.edu).

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 11 September 2000; accepted in final form 11 January 2001.


    REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

1.   Banzett, RB, Guz A, Paydarfar D, Shea SA, Schachter SC, and Lansing RW. Cardiorespiratory variables and sensation during stimulation of the left vagus in patients with epilepsy. Epilepsy Res 35: 1-11, 1999[ISI][Medline].

2.   Bartlett, D, Jr, and St. John WM. Influence of lung volume on phrenic, hypoglossal and mylohyoid nerve activities. Respir Physiol 73: 97-110, 1988[ISI][Medline].

3.   Breuer, J, and Hering E. Self-steering of respiration through the nervus vagus. In: Breathing: Hering-Breuer Centenary Symposium, edited by Porter R.. London: Churchill, 1970, p. 359-394.

4.   Clark, FJ, and von Euler C. On the regulation of depth and rate of breathing. J Physiol (Lond) 222: 267-295, 1972[Abstract/Free Full Text].

5.   Gay, PC, Rodarte JR, Tayyab M, and Hubmayr RD. Evaluation of bronchodilator responsiveness in mechanically ventilated patients. Am Rev Respir Dis 136: 880-885, 1987[ISI][Medline].

6.   Georgopoulos, D, Mitrouska I, Bshouty Z, Webster K, Anthonisen NR, and Younes M. Effects of breathing route, temperature and volume of inspired gas, and airway anesthesia on the respiratory output to varying inspiratory flow. Am J Respir Crit Care Med 153: 168-175, 1996[Abstract].

7.   Georgopoulos, D, Mitrouska I, Webster K, Bshouty Z, and Younes M. Effects of inspiratory muscle unloading on the response of respiratory motor output to CO2. Am J Respir Crit Care Med 155: 2000-2009, 1997[Abstract].

8.   Guz, A, Noble IM, Trenchard D, Cochrane HL, and Makey AR. Studies on the vagus nerves of man: their role in respiratory and circulatory control. Clin Sci (Colch) 27: 293-304, 1964.

9.   Hamilton, RD, Winning AJ, Horner RL, and Guz A. The effect of lung inflation on breathing in man during wakefulness and sleep. Respir Physiol 73: 145-154, 1988[ISI][Medline].

10.   Iber, C, Simon P, Skatrud JB, Mahowald MW, and Dempsey JA. The Breuer-Hering reflex in humans. Am J Respir Crit Care Med 152: 217-224, 1995[Abstract].

11.   Killian, KJ, Campbell EJM, and Howell JBL The effect of increased ventilation on resistive load detection. Am Rev Respir Dis 120: 1233-1238, 1979[Medline].

12.   Kuna, ST. Inhibition of inspiratory upper airway motoneuron activity by phasic volume feedback. J Appl Physiol 60: 1373-1379, 1986[Abstract/Free Full Text].

13.   Ljung, L. System Identification: Theory for the User. Upper Saddle River, NJ: Prentice-Hall, 1999, p. 609.

14.   Meza, S, Giannouli E, and Younes M. Control of breathing during sleep assessed by proportional assist ventilation. J Appl Physiol 84: 3-12, 1998[Abstract/Free Full Text].

15.   Morris, JF. Spirometry in the evaluation of pulmonary function. West J Med 125: 110-118, 1976[ISI][Medline].

16.   Nishino, T, Anderson JW, and Sant'Ambrogio G. Responses of tracheobronchial receptors to halothane, enflurane, and isoflurane in anesthetized dogs. Respir Physiol 95: 281-294, 1994[ISI][Medline].

17.   Pack, AI, DeLaney RG, and Fishman AP. Augmentation of phrenic neural activity by increased rates of lung inflation. J Appl Physiol 50: 149-161, 1981[Abstract/Free Full Text].

18.   Polachek, J, Strong R, Arens J, Davies C, Metcalf I, and Younes M. Phasic vagal influence on inspiratory motor output in anesthetized human subjects. J Appl Physiol 49: 609-619, 1980[Abstract/Free Full Text].

19.   Riddle, W, and Younes M. A model for the relation between respiratory neural and mechanical outputs. II. Methods. J Appl Physiol 51: 979-989, 1981[Abstract/Free Full Text].

20.   Rossi, A, Gottfried SB, Higgs BD, Zocchi L, Grassino A, and Milic-Emili J. Respiratory mechanics in mechanically ventilated patients with respiratory failure. J Appl Physiol 58: 1849-1858, 1985[Abstract/Free Full Text].

21.   Younes, M. Proportional assistance ventilation, a new approach to ventilatory support. Am Rev Respir Dis 145: 114-120, 1992[ISI][Medline].

22.   Younes, M, Puddy A, Roberts D, Light RB, Quesada A, Taylor K, Oppenheimer L, and Cramp H. Proportional assist ventilation. Results of an initial clinical trial. Am Rev Respir Dis 145: 121-129, 1992[ISI][Medline].

23.   Younes, M, and Riddle W. A model for the relation between respiratory neural and mechanical outputs. I. Theory. J Appl Physiol 51: 963-978, 1981[Abstract/Free Full Text].

24.   Younes, M, Riddle W, and Polachek J. A model for the relation between respiratory neural and mechanical outputs. III. Validation. J Appl Physiol 51: 990-1001, 1981[Abstract/Free Full Text].

J APPL PHYSIOL 90(6):2330-2340
8750-7587/01 $5.00 Copyright © 2001 the American Physiological Society


This article has been cited by other articles:


Home page
 J. Appl. Physiol.Home page
L. Kubin, G. F. Alheid, E. J. Zuperku, and D. R. McCrimmon
Central pathways of pulmonary and lower airway vagal afferents
J Appl Physiol, August 1, 2006; 101(2): 618 - 627.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Respir. Crit. Care Med.Home page
A. J. Rice, H. C. Nakayama, H. C. Haverkamp, D. F. Pegelow, J. B. Skatrud, and J. A. Dempsey
Controlled versus Assisted Mechanical Ventilation Effects on Respiratory Motor Output in Sleeping Humans
Am. J. Respir. Crit. Care Med., July 1, 2003; 168(1): 92 - 101.
[Abstract] [Full Text] [PDF]

Home page
 J. Appl. Physiol.Home page
B. F. BuSha, M. H. Stella, H. L. Manning, and J. C. Leiter
Termination of inspiration by phase-dependent respiratory vagal feedback in awake normal humans
J Appl Physiol, September 1, 2002; 93(3): 903 - 910.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (6)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by BuSha, B. F.
Right arrow Articles by Leiter, J. C.
Right arrow Search for Related Content
PubMed
Right arrow