INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

INSPIRATORY FLOW () varies over a very wide range in health and disease. During exercise, for example, may increase from a resting level of 0.5 l/s to values in excess of 6 l/s. In mechanically ventilated patients, is an important independent variable that is often adjusted, over a wide range, to accomplish a variety of clinical and physiological objectives. The effect of on respiratory motor output in humans is not well documented. Such information would be relevant to the understanding of mechanisms of spontaneous hyperpnea in health and disease and of the consequences of changes in ventilator on respiratory muscle energetics in ventilator-dependent patients.

Based on experiments in anesthetized animals, in which can be readily manipulated while other relevant variables are controlled, may influence respiratory motor output in one of three ways.

1) With changes in inspiratory duration [neural inspiratory time (TI_n)] consequent to the Hering-Breuer (H-B) volume (V)-related inspiratory inhibitory reflex (6, 16), an increase in () would result in an earlier attainment of the V threshold for inspiratory termination and a shorter TI_n. Because inspiratory muscle activity rises in a ramplike fashion, a deliberate , with a consequent reduction in TI_n, should result, with all else being the same (i.e., no change in rate of rise of inspiratory activity), in lower peak activity and vice versa (6, 31, 33).

2) With changes in inspiratory activity before inspiratory termination, whether the rate of inflation affects inspiratory activity before inspiratory termination is controversial. In several animal studies, the rate of change in inspiratory activity during inspiration was found to be unaffected by (6, 31, 33). Other studies, however, demonstrated an increase in this rate of rise when was increased (4, 8, 9, 18, 27). It is very likely that these differences in response to are related to the depth of anesthesia (9, 27). This suggests that this -related excitation may be particularly prominent in consciousness.

3) Changes in , and, consequently, instantaneous V, could affect inspiratory muscle pressure output, independent of muscle activity, via strictly mechanical effects (intrinsic properties of respiratory muscles). Thus at a given activity respiratory muscles generate less pressure in the presence of higher [via the force-velocity relation (1, 13, 28)] and V [via the force-length relation (10, 14, 28)], and vice versa.

There is no information in humans regarding the possible excitatory effect of (see 2 above). A small effect of on TI_n (1 above) was found in several human studies (20, 22, 29, 32). However, the range of examined was very small [essentially between zero (occlusion) and resting (~0.3-0.5 l/s)], and the subjects were anesthetized (29) or asleep (20, 22, 32). Our laboratory has recently demonstrated a marked effect of inspiratory on TI_n in normal, awake subjects over the range of 0.8-2.5 l/s (11). Interestingly, TI_n reduction was not related to earlier attainment of a V threshold. In fact, V at inspiratory termination was significantly lower when was increased. The operation of the intrinsic properties of respiratory muscles has been well demonstrated in humans (1, 13, 14, 28). It is, however, difficult to infer the quantitative impact of these responses during spontaneous breathing in view of the nature of the methods used in these studies (see DISCUSSION).

In the present study, we describe the response of diaphragmatic electrical [electromyogram (EMG)] activity (EMGdi) and pressure output [transdiaphragmatic pressure (Pdi)] to brief (~0.2 s) increases in inspiratory in awake humans. From this information, we extract the magnitude of the intrinsic properties and document the occurrence of substantial -related inspiratory excitation in awake humans. The response of TI_n to such brief increases in was also examined in an effort to define the mechanism of reduction in TI_n observed earlier (11) with sustained increases in .

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Eight normal subjects were studied: four men and four women. Subject age ranged from 27 to 38 yr. No subject had any clinical evidence of respiratory disease. Three subjects were aware of the general nature of the study (to study the effect of inspiratory ) but not the specific protocol or expected results. One subject (S. Corne) was aware of the specific protocol.

A gastroesophageal catheter was placed in all subjects. The catheter was fluid-filled and measured gastric and esophageal pressures from two ports separated by a distance of 20 cm. The catheter also recorded the EMGdi from two silver electrodes placed between the gastric and esophageal pressure ports. The catheter was inserted through the nose and advanced until negative pressure deflections were observed in both the gastric and esophageal pressure waveforms during inspiration. The catheter was then slowly advanced until positive deflections became visible in the gastric waveform, confirming position in the stomach. Minor adjustments were then made in the catheter position to optimize the EMGdi waveform and minimize the amplitude of the cardiac artifact in the Pdi tracing. The distance between the proximal pressure port and the EMG electrodes was 8.5 and 13.5 cm (average 11 cm), respectively, for the proximal and distal EMG electrodes. This ensured an appropriate location of the proximal port in the lower one-third of the esophagus when the EMG electrode was at the level of the diaphragm (and hence providing the optimal EMG signal). The Pdi was derived by subtracting esophageal pressure from gastric pressure. The raw EMG signal was filtered by a band-pass filter (30-1,600 Hz).

Subjects were seated and connected to a mechanical ventilator (Winnipeg Ventilator) through a mouthpiece. The ventilator was set to allow spontaneous respiration through the ventilator circuit. Nose clips were applied. Airway pressure (Paw) was measured from a side port near the mouthpiece with a differential pressure transducer. Inspiratory and expiratory were measured with a heated pneumotachograph (Hans-Rudolph 3700, Kansas City, MO) connected to the mouthpiece, and V was derived from the electronic integration of . End-tidal carbon dioxide was monitored from a side port near the mouthpiece by using a mass spectrometer (MGA-1100 Medical Gas Analyzer, Perkin-Elmer, Pomona, CA). All waveforms were continuously recorded on two personal computers with the use of a data-acquisition program (WINDAQ, DATAQ Instruments, Akron, OH) for later analysis. EMGdi and were sampled at 500 Hz on one computer. All other waveforms, including the identical signal, were sampled at 125 Hz on a second computer. The signal, common to both recordings, was used to align the EMG signal to other recorded signals in the second computer.

An external pulse-generating box was connected to the pressure control mechanism of the ventilator. This permitted the delivery of positive Paw pulses of varying amplitudes (0-23 cmH₂O). The duration of the pulse was 0.2 s.

The ventilator was triggered, and the threshold for triggering was set at the lowest level that did not result in autotriggering. This typically resulted in a trigger sensitivity of ~0.1 l/s. The pulse apparatus sensitivity was set at a level that caused triggering of the pulse to occur at 0.1 l/s as well. Subjects breathed room air, and a period of ~20 min was allowed for them to adjust to the ventilator.

Subsequently, brief pulses of positive pressure were delivered near the onset of inspiration. Pulses were delivered for one breath, and then ~30-60 s were allowed to elapse before the next pulse was delivered. The duration between pulses was varied so that the subjects could not anticipate when a pulse was about to be delivered. Pulses of at least three different voltages, small [increase () in Paw, 5-10 cmH₂O], medium (Paw, 10-15 cmH₂O), and large (Paw, 15-20 cmH₂O), were given to all subjects, with the exception of subject 1, to whom only medium and large pulses were given. Eight to fifteen pulses of a given voltage were applied, and then the voltage was changed. The order of applying the different voltages was randomized. For the purposes of describing the data, we shall refer to a sequence of pulses (8-15 pulses) of the same external voltage in a given subject as a "trial."

The effect of pulses on peak inspiratory , inspiratory at the termination of TI_n (@TI_n), inspiratory tidal V at the termination of TI_n (V@TI_n), Paw, Pdi, EMGdi, TI_n, neural expiratory time (TE_n), and duration of respiratory cycle (Ttot) were analyzed. Control breaths, usually the breath immediately preceding the pulse breath, were analyzed as well for comparison.

For each pulse trial, we derived averaged waveforms for pulse and control breaths for each of the relevant parameters, , V, Pdi, Paw, and EMGdi. This was done by converting each individual waveform into a series of numerical values. These values were then averaged for the 8-15 breaths analyzed and subsequently reconverted back to waveforms and graphed against time.

Quantitative analysis of the EMGdi waveform necessitated removal of the electrocardiogram (ECG) artifact. In all subjects but one, this required removal of only the QRS artifact. The QRS artifact, typically 100 ms in duration, was first removed from the signal. The EMG signal was then rectified. Subsequently, the deleted 100-ms segment corresponding to the ECG artifact was replaced with a straight line that began at an amplitude equal to the average of values obtained for 40 ms before the QRS artifact and ended in a value equal to the average of values obtained 40 ms after the QRS artifact. In one subject (subject 2), the T-wave artifact was of sufficient magnitude that it required removal as well. The T-wave artifact was also ~100 ms in duration and was removed with a process identical to that described for the QRS artifact. The EMG signal was then averaged (100-ms moving average). An example of the raw EMG and the processed EMG with artifact removed is shown in Fig. 1. For the purposes of averaging EMGdi values among trials, we used normalized data, whereby the average peak control EMGdi in each trial was assigned a value of 100 (%).

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Example of a control breath (left) and a pulse breath (right) in 1 subject illustrating the processing of diaphragmatic electromyogram (EMGdi). Note the increase in inspiratory flow caused by the positive pressure pulse, as well as the increased rate of rise of EMGdi and the shortening of neural inspiratory time in the pulse breath. Paw, airway pressure.

For the purposes of analysis, the onset of TI_n in an individual breath was defined as the initial negative deflection in the Paw tracings and a rapid change in the direction of from expiratory (or zero) to inspiratory. Where possible, the Pdi and EMGdi tracings were also inspected to help define the onset of TI_n. However, cardiac artifact often made it difficult to utilize Pdi alone for the determination of the onset of TI_n. The very gradual increase in EMGdi, typically observed at the onset of a breath, together with ECG artifact in the EMGdi waveform, often made it difficult to determine precisely where inspiration began simply from inspection of the EMGdi waveform. The end of TI_n, and by definition the onset of TE_n, was defined in an individual breath as the onset of a rapid decline from the peak Pdi.

Values for TI_n, TE_n, Ttot, peak EMGdi, peak Pdi, @TI_n, and V@TI_n were determined on individual breaths, and mean values were then determined for pulse and control breaths in each trial. The 23 trials were then sorted into three groups for analysis based on the , that is, the achieved in the pulse breath over that of the control breath: small (n = 7), medium (n = 8), and large (n = 8). Mean values for each of the three groups were determined, and comparisons were made between pulse and control breaths utilizing a paired t-test.

After the onset of the pulse, there was a period (40-200 ms) during which EMGdi of the pulse breath (EMGdi_pls) was not appreciably different from control EMGdi (EMGdi_con) during the same interval, but Pdi was lower (see, e.g., Figs. 2, 3, and 5). We attributed the reduction in Pdi over this interval to the increased V and acting via the force-length and force-velocity relationships of the diaphragm (see RESULTS and DISCUSSION). We attempted to quantitate this relationship by calculating the difference in Pdi, , and V between pulse and control breaths at multiple data points during this isoactivity interval and creating a linear regression equation
where A is equal to the force-length relationship (in cmH₂O/l) and B is equal to the force-velocity relationship (in cmH₂O · l¹ · s) of the diaphragm.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Example of EMGdi, transdiaphragmatic pressure (Pdi), flow, and volume (V) in 1 subject in the medium-pulse trial. Each waveform represents an averaging of 15 breaths. Solid line, control breaths; dotted line, pulse breaths. Thin vertical lines (1st and 2nd) indicate the onset and termination, respectively, of the increased flow caused by the positive pressure pulse. Dashed vertical lines (1st and 2nd) indicate end of neural inspiration in control and pulse breaths, respectively. Down arrow in the EMGdi tracing represents the onset of neural excitation. V@TIn (pls) and V@TIn (con), volume at termination of inspiration in pulse and control breaths, respectively. (The end of neural inspiratory time and V@TIn are shown here for illustrative purposes; however, the quantitative data for these variables were derived from individual breaths and not averaged waveforms.)

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Control (solid lines) and pulse EMGdi (dotted lines) plotted against time in subjects 1-8. Each waveform represents averaging of 8-15 breaths. Vertical lines (1st and 2nd) mark the onset and termination, respectively, of the increased flow in the pulse breaths. Note excitation in the pulse breaths.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As expected, the application of positive pressure pulses resulted in increased inspiratory during the duration of the pulse. As outlined in METHODS, trials were sorted into three groups based on the peak achieved: small (mean , 0.51 l/s), medium (mean , 1.11 l/s), and large (mean , 1.65 l/s). Figure 2 shows responses of EMGdi, Pdi, , and V to a pulse application in a representative subject. These responses can be described in terms of three major effects.

Effects of Increased Inspiratory on Respiratory Timing

There was no significant difference in V@TI_n between pulse and control breaths in the small and medium trials. In the large trials, V@TI_n was significantly larger in the pulse breaths (Table 1). For the group of 23 trials, V@TI_n was larger in the control breaths in 11 trials and larger in the pulse breaths in 12 trials. In the large group, there was no significant correlation between the extent of increase in V@TI_n and the extent of decrease in TI_n (r = 0.32).

Excitatory Effects of on Respiratory Motor Output

We measured the amplitude of the EMGdi_pls signal at the cessation of TI_n and compared it with the EMGdi_con at an identical time point in inspiration, with this time point being referred to as EMGdi_isotime. EMGdi_pls was almost invariably greater than EMGdi_con in the eight subjects (See EMGdi_isotime, Table 1). EMGdi_pls was 131 (P = 0.02), 142 (P = 0.055), and 155% (P = 0.007) of EMGdi_con in the small, medium, and large trials, respectively. We also compared the mean values for EMGdi_pls and EMGdi_con at 0.1-s intervals from the onset of the . Results for the large trials are shown in Fig. 4. The increase in EMGdi_pls was statistically significant at 0.3 and 0.4 s.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Average (means ± SE) EMGdi pulse (dotted line) and EMGdi control (solid line) of all large flow increase trials. Time refers to the onset of flow increase in the pulse breath relative to control breath. * P < 0.05.

We defined the onset of excitation as the point at which the rate of rise of the EMGdi_pls clearly deviated from that of EMGdi_con (arrow in EMGdi tracing, Fig. 2). This point could be identified in 21 of 23 trials. The latency of the EMGdi response to a pulse was measured as the interval between the increase in inspiratory above control and the point at which excitation of EMGdi occurred. The mean latency for the 21 trials was 126 ± 42 (SD) ms. The latencies for the small, medium, and large trials were not statistically significantly different.

Inhibitory Effects of Due to Force-Velocity and Force-Length Relationship

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Control Pdi (solid lines) and pulse Pdi (dotted lines) plotted against time in subjects 1-8. Each waveform represents averaging of 8-15 breaths. Vertical lines (1st and 2nd) mark the onset and termination, respectively, of the increased flow in the pulse breaths. Note initial negative deflection in the pulse Pdi waveform, reflecting the force-velocity and force-length relation of the diaphragm, as well as a subsequent increase in rate of rise in the Pdi, particularly in subjects 1, 2, 7, and 8, reflecting diaphragmatic excitation.

In fact, to the extent that any changes in the EMGdi_pls tracing relative to EMGdi_con were visible during this segment of early inspiration, they involved an excitation of the motor output of the diaphragm ~40-200 ms after the onset of the pulse, as described above. This concave deflection, observed in the pulse breath Pdi (Pdi_pls) relative to the control breath Pdi (Pdi_con), therefore, appears to represent a change due to the force-velocity and force-length relationship of the diaphragm. The magnitude of this depression effect was quantified as described earlier in METHODS. Some trials were not suitable for analysis. The two major reasons for excluding trials were 1) very early onset of EMGdi excitation, such that the interval in which EMGdi_pls and EMGdi_con were visually comparable was too short to permit sufficient data points for analysis, and 2) excessively large cardiac artifact in the Pdi tracings, such that the signal-to-noise ratio was unacceptably low. Twelve trials (each trial consisting of the averaging of 8-15 pulse and control breaths) in six subjects were deemed suitable for analysis. The mean values for the force-length and force-velocity-relationship were 11.2 ± 2.5 and 0.2 ± 0.6 (SD) cmH₂O · l¹ · s, respectively. An example of the results of analysis in one trial is shown in Fig. 6. The intercept (11.2 in this case) represents the change in Pdi per unit change in V, whereas the slope (0.8 in this case) reflects the change in Pdi per unit change in . Excellent correlation coefficients were obtained in all cases, with a mean r of 0.95 ± 0.05 (SD).

View larger version (9K): [in this window] [in a new window]   Fig. 6.   Mechanical relationship between diaphragmatic pressure output and flow (force-velocity) and volume (force-length) in 1 subject. Results shown here represent averaging of 15 breaths. Note the importance of the force-length relation (intercept = 11.2 cmH2O/l) relative to the force-velocity relation (slope = 0.8 cmH2O · l1 · s). P, , and V represent differences in pressure, flow, and volume, respectively, between control and pulse breaths at different points following the increase in .

The effect of inspiratory on peak Pdi was minimal when it was looked at for the groups as a whole (Table 1). Mean peak Pdi decreased from 12.4 to 11.4, from 11.9 to 10.2, and from 12.1 to 11.8 cmH₂O in the small, medium, and large trials, respectively. None of these differences was statistically significant. However, there was a great deal of intersubject variability in the response of peak Pdi to . In those with a large decrease in TI_n and a weak excitatory response to , peak Pdi decreased at higher (subjects 5 and 6, Fig. 5). Conversely, in those with only a small decrease in TI_n and a stronger excitatory response to (subjects 1, 2, 7, and 8), peak Pdi increased at the higher . We also analyzed the difference in Pdi_pls and Pdi_con at 100-ms intervals from the onset of . The results for the large group are shown in Fig. 7. There were statistically significant decreases in Pdi_pls at 0.1 and 0.2 s, reflecting the force-length and force-velocity effect. There was also a subsequent increase in Pdi_pls over Pdi_con from 0.4 to 0.7 s, reflecting the excitatory effects of . This failed to reach statistical significance, because of the large variability in responses (SE bars shown).

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Average (means ± SE) Pdi pulse (dotted line) vs. average Pdi control (solid line) of all large flow increase trials. Time refers to the time from onset of flow increase in pulse breath relative to control breath. * P < 0.05.

Figure 8 demonstrates the change in Paw, , V, Pdi, and EMGdi between pulse and control breaths in the small, medium, and large groups. The time period analyzed begins with the first data point at which pulse increased over that of control and extends to the end of TI_n for the individual trial with the shortest TI_n in that group. This limited the period of analysis to ~0.2 s for the small group and to ~0.4 s for the medium and large groups. By definition, there was a graded increase in and V from the small to the large groups associated with a graded increase in Paw. There was also a graded increase in the negative Pdi deflection, reflecting the force-velocity and force-length relation of the diaphragm. Conversely, there was a graded increase in EMGdi from the small to the large group. The EMGdi was significantly different at 0.2 s among the three groups using ANOVA (P < 0.05).

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Average (means ± SE) differences () between control and pulse breaths in various variables in the small (dotted lines), medium (dashed lines), and large (solid lines) flow increase groups. Note that the response in each variable is graded. EMGdi values were normalized in reference to the mean peak control EMGdi in each trial, which was assigned a value of 100. * Significant difference between responses in the three groups by ANOVA (P < 0.05). Changes in Paw, flow, and volume were significant (by design) at all time points (asterisks not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have examined the effect of changes in inspiratory on respiratory motor output in normal subjects. These effects can be grouped into three categories: 1) effects on respiratory timing, 2) effects on excitation of diaphragmatic motor output, and 3) the force-length and force-velocity relationship of the of the diaphragm.

Respiratory Timing

With respect to TI_n shortening, the delayed effect may be due to the higher V accrued during the period of increased (e.g., Figs. 2 and 8). Because such excess V is retained past the period of the transient, it could cause earlier termination of TI_n via the traditional V-related H-B reflex. Although this may have contributed to some extent in some cases, it is very unlikely that this mechanism provides the entire explanation. According to the H-B reflex, the V threshold for inspiratory termination declines progressively with inspiratory time (6, 16). It would thus require a higher V to terminate inspiration sooner. In the small and medium groups, V@TI_n was not higher (Table 1). Although V@TI_n was higher with the large pulses, the difference was too small to account for the TI_n shortening. Thus, with the large pulses, TI_n was reduced by 25% (1.17 vs. 1.58 s, Table 1), whereas V@TI_n increased by ~25% (0.92 vs. 0.74 liter, Table 1). Even in animals in which this reflex is very strong [e.g., pentobarbital-anesthetized cats (6, 16)], V@TI_n increases much more for equivalent reductions in TI_n. Furthermore, as mentioned, there was no correlation between the extent of increase in V@TI_n and the extent of decrease in TI_n. These observations suggest that the delayed effects were primarily related to neural processing.

Information from animal studies regarding delayed central effects of inspiratory-terminating inputs is highly contradictory. In pentobarbital-anesthetized cats, removal of V (36) or inspiratory inhibitory vagal stimulus (34) before the end of inspiration results in paradoxical delayed effects: TI_n would have been lengthened relative to its duration had the stimulus not been introduced at all. By contrast, in chloralose-anesthetized dogs, Cross et al. (8) found that TI_n continued to be shortened when lung inflation was withdrawn before inspiratory termination. These observations clearly indicate that inspiratory inputs can produce central effects that outlast the stimulus but that the directions of these effects can be diametrically opposite (concordant with, or paradoxical to, the primary effect of the stimulus) in different species and/or under different experimental conditions (type or depth of anesthesia). The present findings indicate that, in awake, normal humans, the delayed effects are concordant with the primary effect of the stimulus; there is memory for the inspiratory-terminating influence of increased . Whether the same response will hold under other conditions, for example, during sleep or anesthesia, or in the presence of disease remains to be determined.

Earlier studies in anesthetized animals demonstrated a linkage between TI_n and TE_n (6, 16, 35). When TI_n is shortened by an inspiratory-terminating input, the following TE_n is also shortened. The reduction in TE_n observed in the present study can thus be explained on the basis of this central linkage. In our previous study, TE_n was not consistently shortened when increased and TI_n decreased (11). In this latter study, however, inflation usually continued past the end of TI_n, representing further inflation into expiration. Because inflation during expiration lengthens TE_n via the expiratory-prolonging component of the H-B reflex (23-25), extension of inflation into expiration would obscure the reduction in TE_n that might otherwise have occurred as a result of TI_n shortening [for a more detailed discussion, please see Fernandez et al. (11)]. By limiting the inflation to the inspiratory phase in the present study, the TI_n-TE_n linkage was demonstrated.

Excitation of Diaphragmatic Activity

To our knowledge, the present study is the first to deliberately look for and demonstrate -related inspiratory excitation in humans. Accordingly, it is necessary to address some technical issues.

Technical considerations. The EMGdi signal is subject to being artifactually increased and will appear larger if it is measured at a larger V (12). Therefore, one may question whether the increase in EMGdi we observed was due to the fact that V rose more quickly in the pulsed breath. We do not feel that the increase in EMGdi was an artifact for a number of reasons. First, there was a finite delay before the excitation was evident. Second, the increase in V observed in the pulsed breaths, which was typically only a few hundred milliliters (e.g., Fig. 8), would be insufficient to account for the large increases observed in EMGdi, based on the relation between V and EMGdi reported by Gandevia and McKenzie (12). Third, the Pdi response was biphasic, initially showing a decreased rate of rise, followed by an increased rate of rise (Fig. 8). The secondary increase in the rate of rise of Pdi began at a time when V was continuing to rise at a faster rate (Fig. 8) and despite the negative effect this would have on Pdi secondary to the diaphragm's intrinsic properties. An increase in the rate of rise of Pdi under such conditions can only result from the increased rate of rise in EMGdi. In four subjects, absolute Pdi_pls ultimately exceeded Pdi_con, despite V being higher (Fig. 5). This can only occur if EMGdi increased enough to more than offset the depressant effect of the larger V via the force-length relation. Finally, some studies in animals (4, 8, 9, 18, 27) have documented a similar excitation while recording was done directly from the phrenic nerve, a measurement that is not subject to this artifact.

View larger version (62K): [in this window] [in a new window]   Fig. 9.   Example, in 1 subject, of applying consecutive pulses (last 2 breaths). Note the presence of the increased rate of rise in the Pdi tracing in the consecutively pulsed breaths (arrows), demonstrating the absence of a refractory period for the excitation response.

Mechanism of reflex, -related inspiratory excitation. Animal studies on the effect of inflation rate on inspiratory activity before inspiratory termination produced very conflicting results. In early work, it was found that the pattern or intensity of inspiratory activity was not affected by until shortly before termination (6, 31, 33, 36). This led to the concept that V (and ) affects inspiratory activity in an all-or-none fashion (6, 31). In several later experiments, however, a vagally mediated, -related increase in inspiratory activity was demonstrated before termination of the inspiratory phase (4, 8, 9, 27). The earlier lack of such effect was likely related to the depth of anesthesia because, in at least two studies (9, 27), the excitatory response was eliminated by additional anesthetic doses. It is very likely that the response observed here is the same as the one described in the above-cited animal studies and is, therefore, vagal in origin. -sensitive upper airway receptors are not likely to be responsible: McBride and Whitelaw (24) found that increased through the upper airway (without lung inflation) in awake humans inhibits inspiratory motor output. Accordingly, any excitation we observed represents the net effect of excitation produced by lower receptors, less any inhibition produced by upper airway receptors. A possible contribution from muscle receptors cannot be entirely discounted, although it is unlikely to be significant. Newsom Davis and Sears (26) monitored external intercostal activity during strong, sustained voluntary contractions against a closed airway (Mueller maneuver). On sudden release of occlusion, there was a short-latency (22-25 ms) inhibition of activity, which they attributed to unloading of the respiratory muscle spindles, followed by facilitation with longer latency (50-60 ms), which was attributed to unloading of the inhibitory tendon organs. The excitation we observed could, therefore, theoretically be due to reduction in the tension of diaphragmatic tendon organs produced by unloading during the pulse. Pdi at the time of pulse application was, however, of the order of 1-4 cmH₂O (Fig. 5), a mere 1-3% of maximum Pdi. It is highly improbable that tendon organ inhibition at these very low tensions is such that its elimination (by increased and unloading) caused a >40% increase in activity (EMGdi_isotime, Table 1).

Physiological significance. The -related excitatory response may be a mechanism that serves to counteract the negative consequences of the obligatory intrinsic properties of respiratory muscles. Without such an excitatory mechanism, the ventilatory response to a given respiratory stimulus would be attenuated, because as increases, the efficiency of the respiratory muscles as pressure generators decreases (secondary to the force-length and force-velocity relation). Regardless of whether this response was developed for this specific purpose, its net effect is in the direction of compensating for reduced respiratory muscle efficiency with increased ventilatory demand. Our results permit an analysis of the balance between the two opposite -related responses. To the extent that respiratory drive is not altered during the brief period of the pulse, a perfect compensatory response would result in Pdi remaining constant as is artificially increased; the reduction in Pdi produced by the obligate mechanical properties of muscles would be exactly offset by an increase in muscle activation. Figure 8 shows that, on average, the two mechanisms canceled each other out at ~0.35 s after the onset of (Pdi returning to zero, Fig. 8). Before this time, compensation was inadequate, resulting in a decrease in Pdi. This shortfall is, to a large extent, related to the fact that the operation of the intrinsic properties of muscles is instantaneous, whereas neural delays preclude an instantaneous excitatory response. Given the abrupt transition in this study, such an initial shortfall is unavoidable. It may be expected, however, that, with less abrupt changes in , the discrepancy would be less pronounced.

Intrinsic Properties of the Diaphragm

In the present study, we took advantage of the fact that EMGdi was not affected for a period of time (latency of the excitatory response), whereas and V changed, to calculate the mechanical effects of and V on Pdi. We believe that this approach has several advantages over previously used techniques and that the results should, therefore, be closer to the situation during spontaneous breathing. First, activation of the diaphragm was spontaneous. Therefore, there is no reason to believe that the distribution of activity within the diaphragm was anything but normal. Second, the distribution of activity between diaphragm and other muscles, which affects thoracoabdominal configuration and hence pressure output (14), was normal (i.e., not constrained by protocol). Third, although we did not monitor thoracoabdominal configuration, expansion during the imposed pulses must have closely followed the relaxed thoracoabdominal configuration.¹ Although some deviation from passive configuration may occur during large, spontaneous increases in ventilation (15), chest expansion usually follows the resting configuration at resting and moderately increased levels of ventilation (13). Thus the changes in Pdi when the chest expands at different rates along its relaxed configuration should provide a closer approximation of what happens during spontaneous increases in ventilation.

Our results indicate that the effect of per se on pressure output (the force-velocity relation) is negligible (0.2 ± 0.6 cmH₂O · l¹ · s) over the range studied (0.65-2.4 l/s, Table 1). This is in agreement with findings in spontaneously breathing dogs in which EMGdi and diaphragm velocity of shortening were recorded by using implanted sensors (25). The independent effect of V via the force-length relation and configuration factors was, however, large. Its magnitude (11.2 ± 2.5 cmH₂O/l) is comparable to the passive elastance in normal subjects (2). When uncompensated for, via the excitatory response, for example, V would have the same effect on ventilatory responses as would the doubling of the passive elastance. Given these findings, it is of interest to note that the time course of the excitatory response follows much more closely the time course of V (Fig. 8). This suggests that the excitatory response is well suited to offset the unavoidable effects of intrinsic muscle properties.