American Journal of Respiratory and Critical Care Medicine

In six dogs studied in nonrapid eye movement (NREM) sleep, we found that the frequency, volume, and timing of application of mechanical ventilator breaths had marked and sustained inhibitory effects on diaphragm electromyogram (EMGdi). Single ventilator breaths of tidal volume (Vt) 75–200% of control caused apnea (up to three times eupneic expiratory time [Te]) when applied during the initial 25–65% of expiratory time. When continuous controlled mechanical ventilation (CMV) was applied with ventilator frequency increased as little as 1 cycle/min > eupnea and PaCO2 and Vt maintained at near eupneic control levels, EMGdi was silenced and triangularis sterni EMG (EMGts) became tonic within 2 to 5 ventilator cycles. On cessation of normocapnic CMV, apnea ensued with Te ranging from 1.2 to five times eupneic Te. The spontaneous Vt and EMGdi determined immediately after these prolonged apneas were also markedly reduced in amplitude. The larger the Vt applied during the isocapnic CMV (120–200% of eupnea) and the longer the duration of the CMV (3–90 s), the longer the duration of the postventilator apnea. Significant postventilator apneas and postapneic hypoventilation also occurred even when end-tidal CO2 pressure (PET CO2 ) was raised 3–5 mm Hg > eupnea (and 7–10 mm Hg > normal apneic threshold) throughout CMV trials at raised frequency and Vt. Our findings demonstrate that the increased frequency of CMV was critical to the elimination of inspiratory motor output and the onset of tonic expiratory muscle activity; furthermore, once EMGdi was silenced, the tidal volume and duration of the passive mechanical ventilation determined the magnitude of the short-term inhibition of inspiratory motor output after cessation of CMV.

When humans and dogs during sleep were subjected to prolonged mechanical ventilation in the control mode at high ventilator frequency and tidal volume, respiratory motor output was eliminated even when normocapnia was maintained (1, 2). Furthermore, apnea persisted in the postventilator period. It has not been determined which ventilatory parameters were primarily responsible for producing this nonchemical elimination of respiratory motor output and its aftereffects.

Studies using electrical stimulation of sensory nerve inputs in anesthetized animals have identified the importance of three parameters, namely, phase, intensity, and duration of stimulus as important influences on resetting the respiratory rhythm (3). Increasing tidal volume (Vt) (or positive pressure), by itself, using assist control mechanical ventilation (ACMV), caused a time-dependent decrease in the amplitude of respiratory motor output in sleeping humans with maintained normocapnia, which persisted after cessation of mechanical ventilation (4, 5). However, expiratory time was unchanged or only slightly prolonged during normocapnic ACMV at increased Vt and was immediately restored to eupneic control values on cessation of ACMV (4). Similarly, in sleeping dogs (6) or humans (7) increasing ventilator volume alone reduced the amplitude of the diaphragm electromyogram (EMGdi) during normocapnic or hypercapnic breathing; but spontaneous breathing frequency remained unchanged. The phase of the breath during which sensory input is applied is also important, as superior laryngeal, vagal, or intercostal nerve electrical stimulation imposed during early expiration in anesthetized animals prolongs the expiratory time of the perturbed cycle (8-11).

Perhaps, then, mechanical feedback linked to ventilator frequency may be an important parameter for resetting respiratory rhythm; and if the resetting effect is sufficiently strong, the increased ventilator frequency may cause apnea by “superseding” the inherent respiratory rhythm, as suggested by Puddy and coworkers (12). The importance of the duration of repeated rhythmic sensory inputs, such as occurs with controlled mechanical ventilation, has not been studied directly but the duration of sustained inhibitory input, as with tetanic or repeated phasic electrical stimulation of the superior laryngeal nerve (SLN), has been shown to be an important determinant of the degree of phase resetting (3, 10).

Our aim was to determine the parameters of mechanical ventilation required to cause a resetting of the respiratory rhythm and cessation of respiratory motor output, both during and after mechanical ventilation. To this end we studied sleeping dogs, using measurements from indwelling EMGs of the diaphragm (EMGdi) and of the expiratory triangularis sterni muscle (EMGts) to determine the effects of phase, frequency, amplitude, and duration of mechanical ventilation on respiratory motor output. We also determined the influence of a changing background PaCO2 on the effectiveness of mechanical ventilation at varying frequencies and Vt in causing apnea.

Six adult female mixed-breed dogs weighing between 20 and 25 kg were studied. The surgical and experimental protocols of this study were approved by the Animal Care and Use Committee of the University of Wisconsin (Madison, WI).

Surgery and Measurements

Sterile surgical techniques were used to create a permanent tracheostomy and to implant electrodes for recording of EMGdi, EMGts, electroencephalogram (EEG), and an arterial catheter for blood sampling. The dogs were premedicated with acepromazine (0.5 mg/kg, subcutaneous), induced with thiamylal sodium (20 mg/kg, intravenous) and maintained with 1% halothane in O2 by means of an anesthesia machine–ventilator.

Midline cervical incision and removal of the ventral aspect of four or five cartilaginous rings was performed to create the chronic tracheostomy. Bipolar Teflon-coated multistrand stainless steel wire electrodes (AS637; Cooner Wire, Chatsworth, CA) were sewn into the crural diaphragm and, in four of the six dogs, trangularis sterni (fourth or fifth interspace) muscles for measurement of EMG activity. All EMG and electro-oculogram (EOG) electrodes were tunneled subcutaneously to the cephalad portion of the back, where they were exteriorized. The raw EMG data were filtered (30 Hz–1 K/Hz), amplified, rectified, and moving time averaged with a time constant of 100 ms. A five-lead EEG/EOG montage was implanted subcutaneously. These electrodes were connected to amplifiers (BMA-831; CWE, Ardmore, PA) and filtered at 10 Hz for the EOGs and at 1–50 Hz for the EEG. These techniques have been described in detail previously (13). In a separate surgical session, a catheter for measurement of arterial blood gases was placed into the femoral artery in three dogs. The catheter was also tunneled subcutaneously and exteriorized to the cephalad portion of the back and was filled with heparin (10,000 U/ml) and flushed daily. Analgesics and antibiotics were administered during the postoperative periods. The dogs were allowed a recovery period of at least 4 wk.

The dogs were intubated with a cuffed endotracheal tube (o.d. 12.0 mm) via the chronic tracheostomy. The endotracheal tube was connected to a thermostatted (37° C) pneumotachograph system (model 3700 [Hans Rudolph, Kansas City, MO]; model MP-45, ± 2 cm H2O, [Validyne, Northridge, CA]) for measuring airflow and was calibrated daily with five known flow rates. Tracheal pressure (Ptr) was measured with a pressure transducer (model MP-45, ± 2 cm H2O; Validyne) connected to the tracheal cannula and was calibrated before each study by applying eight known pressures. End-tidal Pco 2 (PET CO2 ) and Po 2 (PET CO2 ) were detected by sampling from the tracheal cannula and analyzing via mass spectrometry (MGA 1100; Perkin-Elmer, Norwalk, CT). Two three-ported connections and three low-resistance one-way valves (model 1400; Hans Rudolph) were attached to the pneumotachograph. A mechanical ventilator (model MA-1; Puritan-Bennett, Boulder, CO) was used to apply lung inflation at a preset volume and flow. To avoid transient hypocapnia, the dead space in the ventilator circuit was filled in advance with 2–4% CO2 in air. The effective dead space of this system was approximately 40 ml.

All variables were recorded on a 12-channel polygraphic recorder (ES 2000; Gould, Rolling Meadows, IL) and were sampled via an analog-to-digital converter and stored on the hard disk of a personal computer for subsequent analysis by means of custom-written software developed in our laboratory. The sampling rate for all signals was 128 Hz.

Staging of Sleep State

Standard canine criteria were applied to identify the sleep stages (14). Nonrapid eye movement (NREM) sleep was defined as a synchronized low-frequency (< 10-Hz) EEG associated with an absence of rapid eye movements. EEG arousal was defined as a desynchronization and speeding (> 10 Hz) of the EEG for > 3 s. All trials that had arousals and/or sleep state change during the control or experimental periods were excluded from further analysis.

Experimental Protocol

Studies were performed on five separate days on each dog during NREM sleep. The dogs were trained to lie quietly on a bed located in an air-conditioned (19–22° C), sound-attenuated chamber. The animals were unrestrained during the experiments and the body position in which they chose to sleep was not restricted. Throughout all experiments, the behavior of the dogs was monitored by an investigator seated within the chamber and also by closed-circuit television. The dogs breathed through their tracheostomy throughout the experiment.

Study 1: Single-breath Trials

Once a stable breathing pattern was observed during NREM sleep, single lung inflations delivered by mechanical ventilator were applied during expiration in five dogs. To minimize FRC changes, all the single lung inflation studies were performed only when spontaneous breathing frequency (fb) of the dog was less than 20 breaths/min. Single lung inflations of four different preset volumes (mechanical Vt = 50, 75, 100, and 150% of spontaneous breathing), with a mean inspiratory flow rate at 40 L/min and inspiratory time (Ti) averaging 28% of spontaneous Ti were applied randomly at various times during expiration. The increased fraction of inspired CO2 (Fi CO2 ) contained in the ventilator dead space prevented hypocapnia when the ventilator volume was delivered. Multiple trials, with at least a 2-min interval between each trial, were performed over several days in each of five dogs. Te was measured from the end of the inspiratory flow signal of the spontaneous breath preceding the machine breath to the onset of the next EMGdi, and compared with control Te.

Study 2: Isocapnic Controlled Mechanical Ventilation (CMV): Effects of Ventilator Frequency, Vt, and Duration of CMV

In four dogs, three different protocols of isocapnic CMV trials were performed. In these trials, PET CO2 was held constant (PET CO2 0.3 to 1.2 mm Hg > spontaneous breathing) by raising the level of inspired CO2 during CMV. First, we determined the effects of ventilator frequency (fb = +1, +3, +5 > eupnea) at constant Vt (120% > eupnea) for 2 min each, on eliminating phasic EMGdi and thereby causing “passive” mechanical ventilation. CMV was then abruptly discontinued (during expiration) in order to determine the duration of post-CMV apnea, as measured from the end of the last CMV inspiratory flow signal to the onset of the next EMGdi. Only trials without evidence of EEG arousal were accepted to determine post-CMV apnea duration.

Next, in three dogs six sustained passive CMV trials were conducted at three Vt amplitudes (120 to 200% of spontaneous Vt) and five durations (3 to 25 ventilator cycles) in order to determine the independent and interactive effects of ventilator Vt and CMV duration on the post-CMV apnea length. Once a stable breathing pattern (fb < 20 breaths/min) was observed during NREM sleep, the first mechanical ventilator breath of CMV was applied during expiration.

Study 3: Effects of CMV on the CO2 Apneic Threshold

The protocol was designed to determine the effects of hyper- and hypocapnia imposed during CMV at varying Vt on post-CMV apnea length. In five dogs, five trials were conducted in which CMV was applied for 1.5 to 2 min at fb 1–2 breaths/min > control, Vt twice control, and PET CO2 2–4 mm Hg > control. In two of the dogs, 95 and 102 trials per dog were completed in order to quantify the effect of variations in Vt and PET CO2 on EMGdi during and after CMV. CMV (mechanical fb = 15 breaths/min, mean flow rate of 50 L/min) was carried out for 1.8 ± 0.4 min or 25 ± 2 ventilator cycles at each of three mechanical Vt levels (Vt = 120, 160, and 200% of spontaneous breathing). During CMV, PET CO2 was held constant at several different levels of PET CO2 (range = ± 5 mm Hg of eupneic PET CO2 ) by altering the level of inspired CO2.

PET CO2 versus Arterial Pco 2 during Spontaneous Breathing and Isocapnic CMV

PET CO2 was measured in all trials as an estimate of arterial Pco 2. To determine any systematic differences between the two measurements, we completed nine trials in each of three dogs under varying conditions of isocapnic CMV. In each dog, arterial blood was sampled repeatedly during 2 min of eupnea (four samples) and during 3 min of isocapnic CMV (six samples) (mechanical fb = 15 or 20 breaths/min) at each of three mechanical Vt levels (Vt = 120, 160, and 200% of spontaneous breathing). PET CO2 was averaged over several breaths to obtain the PET CO2 per blood sample. Arterial blood gases were analyzed with a blood gas analyzer (ABL-300; Radiometer, Copenhagen, Denmark). The blood gas analyzer was calibrated before use with tonometered blood.

Statistical Analysis

For the single-breath inflation studies one-way repeated measures analysis of variance (ANOVA) was used in combination with post-hoc paired analysis (Dunnett test) to compare the mean Te obtained from several repeat trials between the immediately preceding eupneic control breaths and that following single-breath lung inflations applied during each of three specific times of expiration.

To determine the effects of CMV duration and Vt on post-CMV apnea length, a two-factor (Vt and CMV duration) ANOVA was used to compare mean apnea lengths across the three increased Vt levels at any given duration of CMV and across the six different durations of CMV at any given Vt. Linear regression was used to correlate changes in PET CO2 during CMV, with apnea length following CMV, at each of three levels of Vt. A one-way ANOVA was used to compare the mean end tidal-to-arterial Pco 2 differences during spontaneous breathing versus isocapnic CMV.

End tidal-to-arterial Pco 2 Differences during CMV

The data for six sessions of eupnea and nine trials of CMV in each of three dogs showed that end-tidal Pco 2 exceeded arterial Pco 2 by 1.6 ± 0.5 mm Hg during eupnea (p < 0.05) and that the end tidal–arterial Pco 2 difference during CMV at increased Vt averaged 1.5 ± 0.4 mm Hg at a ventilator frequency of 15 breaths/min and 1.7 ± 0.5 mm Hg at fb = 20 breaths/min (p < 0.05). These end tidal–arterial Pco 2 differences were similar during eupnea and CMV (p > 0.10). Thus, we concluded that maintaining PET CO2 at normocapnic levels during CMV also meant that arterial Pco 2 was simultaneously maintained. These data in sleeping dogs confirm those obtained during normocapnic mechanical ventilation with increased Fi CO2 in healthy humans (4, 15).

Eupnea during Sleep

Several minutes of eupneic breathing in NREM sleep preceded each trial of mechanical ventilation. The mean breathing patterns and PET CO2 for each dog over all of these control periods are shown in Table 1.


DogTi (s)Te (s)Vt (liters)PET CO2
11.3 ± 0.33.1 ± 1.10.32 ± 0.0738.3 ± 1.0
21.7 ± 0.33.6 ± 1.20.35 ± 0.0536.4 ± 0.5
31.3 ± 0.12.0 ± 0.30.32 ± 0.0236.4 ± 0.9
41.5 ± 0.53.0 ± 0.80.39 ± 0.1838.2 ± 0.7
51.5 ± 0.13.9 ± 0.50.31 ± 0.0341.3 ± 1.2
61.6 ± 0.24.8 ± 1.10.34 ± 0.0541.7 ± 1.3

*Each value represents the mean ± SD.

Effects of Single Ventilator Breaths

The effects of delivering ventilator breaths at different volumes and phases of expiration in NREM sleep are illustrated in Figures 1A–1C. When a ventilator breath was applied in the first 50% of expiration, the resumption of diaphragm EMG activity was delayed and Te was prolonged (Figure 1A); and this effect on Te prolongation was increased when the Vt of the ventilator breath was augmented (Figure 1C). However, when a ventilator breath was applied in the latter 20% of expiration, Te was shortened slightly (Figure 1B).

The group mean data are shown in Figure 2A (n = 5) and the variable effects of single ventilator breaths on Te are shown for all trials in a single animal in Figure 2B. The data were quite consistent in showing that when the ventilator breath was applied during the initial 25–65% of expiration, (1) Te was prolonged significantly (p < 0.05) when the Vt of the breath was 75 to 150% of eupneic Vt; and (2) the greater the Vt the greater the Te prolongation. Added ventilator breaths of Vt = 50% of eupneic Vt did not significantly prolong Te (p > 0.10). When ventilator breaths of any Vt were applied during the latter 25% of expiration, Te remained unchanged. Application of ventilator breaths in the 60 to 75% range of Te had highly variable effects, causing Te to prolong (significantly so only at the highest Vt), to shorten, or to remain unchanged from eupnea (see Figure 2B). In all cases, these single ventilator breaths affected the timing only of the perturbed breath, as mean Ti and Te of the subsequent spontaneous breath were returned to eupneic control values (p > 0.10; data not shown).

Controlled Mechanical Ventilation: Effects of Frequency, Vt, Duration, and PET CO2

Effects of CMV frequency. In three dogs, when Vt was held constant at 20% > eupnea, and the frequency of CMV was increased 1, 3, or 5 breaths/min > eupnea for 2-min periods and normocapnia was maintained, EMGdi was eliminated and significant postventilator apnea ensued (see Figure 3). EMGdi inhibition and postventilator apnea both occurred when fb was increased only 1 breath/min > eupneic control. Increasing fb of CMV further to 3 and 5 breaths/min > eupneic control had no further effect on the duration of postventilator apnea in any of the three dogs (p > 0.10). The effects of ventilator frequency were not systematically studied in the remaining three dogs; however, they also silenced their EMGdi during and after CMV when their ventilator frequency was increased 1 to 3 breaths/min > eupnea.

Effects of volume and duration of CMV. These effects were tested in 3 dogs, using 75, 94, and 103 trials per dog. In Figures 4A and 4B, polygraph records of CMV at fixed increases in Vt (60% > eupnea) and fb (15/min) are shown for durations of 5 and 15 ventilator cycles. Note that the EMGdi stayed silent throughout the CMV and that an apnea ensued after cessation of CMV. The postventilator apnea duration was longer after 15 versus 5 ventilator cycle trials.


Vt(liters)EMGdi(% Control )PET CO2 (mm Hg)
Pre- versus postnormocapnic CMV0.32 ± 0.050.22 ± 0.08100 ± 1467 ± 2541 ± 0.744 ± 1.5
Pre- versus posthypercapnic CMV0.33 ± 0.050.30 ± 0.10100 ± 1872 ± 1941 ± 1.746 ± 2.8

*  All values represent means ± SD.

p < 0.05, compared with eupneic control.

p < 0.01, compared with eupneic control.

Mean data in these three dogs displayed in Figure 5 illustate the interaction between duration and volume of CMV on postventilator apnea length. Note that for any given duration of CMV the postventilator apneic period was longer as the CMV Vt increased; and that at any given Vt during CMV the longer the duration of CMV the longer the apnea, at least up to 10 cycles (or 50–60 s) of CMV. The duration of postventilator apnea increased substantially after the initial 3 to 5 ventilator cycles but beyond about 10 cycles or 1 min of CMV at any Vt, prolonging the duration of CMV had little or no further effect.

Tonic expiratory muscle EMG. As shown by the representative polygraph recording in Figure 6, during prolonged CMV at increased fb the silence of EMGdi was accompanied by a tonic expiratory muscle EMG activity of the triangularis sterni, which persisted throughout the period of passive CMV and during the apneic period after the period of CMV. A brief slowing of the EMGts frequency occurred with each ventilator cycle; and then the EMGts became purely tonic during the post-CMV apneic period and returned to rhythmic behavior in the first active respiratory cycle after apnea termination. The tonic EMGts activity was observed to coincide with a silent EMGdi during and after CMV in all 24 trials in 2 dogs. In both of these dogs, phasic EMGts activity was observed during eupnea while sleeping in the lateral recumbent position.

Effects of ΔPET CO2 at varying Vt during CMV. The influence of increasing PET CO2 above eupneic Pco 2 during CMV at fb 15 breaths/min and with Vt two times eupnea is shown in a polygraph record in Figure 7. Note in the example, when PET CO2 was raised 2–4 mm Hg greater than eupnea, coincident with the onset of CMV, EMGdi was silenced within 3 or 4 CMV cycles, remained silent throughout 2 min of “passive” CMV, and apnea occurred on termination of the CMV.

In each of five dogs, CMV was applied during NREM sleep for five trials of 1.5 to 2 min, at fb = 1 to 2 breaths/min > eupnea, Vt two times eupnea, and PET CO2 2–4 mm Hg > eupnea (data not shown). These animals all responded in a manner similar to that shown in the example in Figure 7; that is, EMGdi became silent within the initial 2 to 4 cycles of CMV, and remained silent throughout the remaining period of CMV and for a period of 13 ± 3 s after the cessation of CMV.

In Figure 8, individual values and regression lines are shown for two dogs in whom multiple trials were performed to test the relationship between PET CO2 and post-CMV apneic length at 3 different Vt levels with fb = 15 breaths/min. Postventilator apneas occurred when PET CO2 was maintained normocapnic, hypercapnic, and hypocapnic during CMV (± 5 mm Hg ΔPET CO2 from eupnea). At any given increase or decrease in PET CO2 , post-CMV Te prolongation was greater the higher the Vt maintained during the CMV. Thus, the minimum mean change in PET CO2 required to cause significant post-CMV apnea (i.e., apneic threshold) was 0 to +2 mm Hg, +3 to +5 mm Hg, and +6 to +8 mm Hg > eupneic PET CO2 , respectively, when Vt was held during the CMV at 120, 160, and 200% > eupnea.

Postapneic Reductions in Spontaneous Vt and EMGdi

After iso- or hypercapnic CMV of more than 1 cycle in duration, apnea ensued and then even the spontaneous breaths that followed these apneas showed reduced Vt and EMGdi despite further increases in PET CO2 (see examples in Figures 4A and 4B and in Figure 7). As shown in Table 2, for isocapnic CMV trials, the first postapneic spontaneous breath showed a mean PET CO2 that had risen 3 mm Hg > eupnea and a Vt and EMGdi that averaged 68 and 67%, respectively, of eupneic control (all p < 0.05). For hypercapnic CMV trials, the first postapneic spontaneous breath showed a PET CO2 that had risen 5 mm Hg > eupnea and a Vt and EMGdi that averaged 90 and 72%, respectively, of eupneic control (all p < 0.05).

Our findings demonstrate that increasing the frequency of normocapnic mechanical ventilation as little as 1 breath/min above average eupneic frequency causes complete cessation of EMGdi and tonic activation on EMGts. This silencing effect of increased ventilator frequency of EMGdi occurred within a few ventilator cycles, was sustained for several minutes of CMV, and on cessation of CMV, apneas persisted for durations of three to four times eupneic Te, even when PaCO2 was held above normocapnic levels. Once inspiratory motor output was eliminated via increased ventilator frequency, the magnitude of ventilator volume and the duration of mechnical ventilation were the major determinants of the length of postventilator apneas.

Effects of Ventilator Frequency on Silencing EMGdi

The marked effects of increasing ventilator frequency on resetting the respiratory rhythm and silencing EMGdi, even in the face of substantial hypercapnia, contrast with the effects of increasing ventilator volume (or applying pressure assist) by itself. When Vt was increased by (subject-initiated) assist control mechanical ventilation, the amplitude of respiratory motor output decreased by 50% or more but breath timing did not change appreciably (2, 4, 5) (also see the Introduction). Only when the increased ventilator Vt was extended into early electrical expiration was Te prolonged (16, 17). These timing effects of increased Vt were small and returned immediately to control on cessation of ACMV (4). Similar findings were obtained during controlled isocapnic mechanical ventilation held at eupneic breathing frequencies (7).

How does an increased ventilator frequency silence EMGdi? The findings from the single-breath ventilator applications (see Figures 1 and 2) show that presenting the breath during early expiration was critical to prolonging Te and that the greater the volume of the single breath, the greater the prolongation of Te. Knox (9) described this early expiration period as the “inflation-sensitive” phase of expiration. However, the resetting effect of the single ventilator breaths on respiratory rhythm did not carry over to the next respiratory cycle, so why was the EMGdi silenced over several repeat cycles of CMV? One possible explanation is that each repeated ventilator Vt was fortuitously delivered precisely during the inflation-sensitive phase of each respiratory cycle; but this seems unlikely given the highly variable effects on timing if the delivered breath happened to fall slightly later in neural expiration (see Figure 2B).

Rather, it seems more likely that there was a cumulative carryover of an inhibitory influence, which in effect widened the inflation-sensitive phase so that the excitatory threshold for the next normally occurring stimulus to inspiration was raised, thereby prolonging duration of the subsequent cycle. Thus, the next ventilator Vt always occurred during the widened inflation-sensitive phase. These proposed cumulative carryover effects of repeated ventilator volumes were manifested in the apneas observed after termination of CMV. As shown in Figure 5, a sharp increase occurred in postventilator apnea length as the duration of CMV (at elevated frequency) was increased beyond 1 or 2 ventilator cycles. Similarly, in anesthetized animals, inhibitory aftereffects of electrical vagal or superior laryngeal nerve stimulation on phrenic apnea became significant after two or more successive stimuli; and apnea length after withdrawal of these stimuli varied in proportion to stimulus duration (3, 8, 10, 18). Furthermore, although increasing ventilator volume by itself had no appreciable effect on breath timing (1, 4), we observed that once a high ventilator frequency had silenced EMGdi, the magnitude of the ventilator Vt had profound effects on the length of the postventilator apnea.

In summary, to the extent that the length of the postventilator apnea can be used as a window to quantify the strength of inhibition of respiratory motor output that accumulated during CMV, our findings would confirm theoretical predictions that phase, strength, and duration of consecutive mechanical sensory inputs were all major determinants of the magnitude of phase resetting and apnea during mechanical ventilation.

Additional Effects of Chemoreception and State on Mechanical Ventilation-induced Apnea

We observed the sustained inhibitory effects of CMV at high frequency on respiratory motor output, primarily under conditions of normocapnia and during NREM sleep; however, additional sensory inputs will certainly influence the degree of inhibition, or indeed whether it is even manifested. Adding chemoreceptor stimuli via increased Fi CO2 caused significant shortening of the postventilator apnea and even prevented this apnea when passive CMV was conducted near eupneic Vt (see Figure 8). However, the combination of high Vt plus elevated frequency of CMV apparently presented sufficiently strong inhibitory inputs to silence EMGdi throughout and for significant periods after CMV, even when PaCO2 was 5 mm Hg above the eupneic PaCO2 (or about 7–10 mm Hg above the apneic threshold normally elicited by hypocapnia alone in NREM sleep [19]).

State of consciousness is an additional factor influencing resetting of breathing rhythm, as behavioral inputs to the respiratory controller associated with the wakeful state may override any reduction or silencing of respiratory motor output associated with assist control or controlled mechanical ventilation at high frequency, even when a substantial hypocapnia is allowed to occur (12, 20-22). REM sleep also presents additional random excitatory inputs that might “break through” any coexisting feedback inhibition (19, 23).

Mechanisms Responsible for Elimination of the EMGdi and the Inhibitory Memory Effect of CMV

Puddy and coworkers (12) theorized that apnea during mechanical ventilation was due to both tidal volume and frequency settings that were greater than the specific subject's tidal volume and frequency demands and not due to the existance of built-in inhibitory neural mechanical feedback that suppresses ventilatory drive. Certainly, our own findings in sleeping humans and dogs, using high Vt with ACMV (4) or presently with increasing ventilator frequency via CMV, are consistent with this generalization that if more Vt or fb is supplied than required, the subject will reduce the amplitude of inspiratory motor output (in the case of increased Vt) or completely eliminate the EMGdi (in the case of increased fb). However, the “decision” to turn down or reset the respiratory motor output because of ventilatory supply exceeding requirement surely must be an informed one. Accordingly, when PaCO2 is held at or above normal, the means of informing the respiratory controller of the mechanical status of the lung and chest wall must require neurally mediated sensory information, probably from several sites of mechanoreception (see below). Thus, sustained inhibition of respiratory motor output during high-frequency normocapnic mechanical ventilation in sleep was clearly achieved via nonchemical, neural–mechanical inhibitory influences.

The exact source(s) of this sustained mechanical feedback inhibition are not clear. Certainly, the phase resetting caused by single ventilator breaths can also be elicited by stimulation of vagal (9), intercostal (11), or SLN (10) afferent inputs and influenced by lesioning of the dorsolateral rostral pontine areas (24). However, the sustained inhibition of respiratory motor output (and its aftereffects) caused by mechanical ventilation at high frequency and Vt in awake dogs and humans did not require intact vagal or intercostal afferents (2, 25); although the inhibitory effects of ventilator frequency, per se, were shown to differ between intact and lung transplant patients (26). Multiple and redundant feedback pathways are likely involved, especially in the waking state; perhaps further study of these potential pathways should be carried out in sleep, during which a single feedback pathway may have greater relative significance.

The mechanisms involved in the inhibitory aftereffects of passive normocapnic CMV are also not precisely known; however, an extensive literature describes changes in synaptic efficacy that persist beyond the duration of a sensory stimulus (27). Mifflin has used high-frequency stimulation of the carotid sinus or vagus nerves together with intracellular and extracellular recordings from neurons in the nucleus tractus solitarius (NTS) in the anesthetized rat to show that potentiation of the synaptic input to cells in the NTS will enhance an inhibitory outflow and reduce the response of NTS neurons to subsequent afferent chemoreceptor inputs (28). The temporal patterning of visceral afferent inputs to these NTS neurons has also been shown to be important in determining the synaptic response from these cells (29). Further uses of these types of preparations will permit identification of the types of NTS cells that express inhibitory aftereffects in response to a variety of repeated afferent inputs of varying amplitude, phase, and duration and their effects on the frequency of phrenic motor nerve activity. It also remains to be determined whether the persistent tonic expiratory EMG activity that coincided with the silent EMGdi during the postventilator period (see Figure 6) truly represents reciprocal inhibition of inspiratory neurons by expiratory cells at the level of the medullary rhythm generator (30-32).

Relevance of Findings to Human Breathing

The strong inhibitory influences of mechanoreceptor feedback, as shown by the present findings, are clearly applicable to the control of breathing during mechanical ventilation; but we are not sure what implications these findings have for the neuromechanical regulation of spontaneous respiration. The findings pertaining to postventilator apnea support the concept of strong central “inhibitory memory” effects imposed by repetitive mechanoreceptor feedback. Indeed, these aftereffects likely underlie the observation that central sleep apneas are often prolonged well beyond the normal threshold for chemoreceptor excitation. Apparently, once the medullary rhythm generator is silenced by mechanical or chemical mechanisms, the respiratory rhythm is not easily restored.

Several findings in humans parallel our current findings in dogs. First, we previously eliminated respiratory motor output by isocapnic CMV at high frequency and Vt in both intact and vagally denervated humans and dogs during wakefulness (2) and sleep (1, 33). Second, our preliminary data in sleeping humans showed Te prolongation when single ventilator breaths were applied during the “inflation-sensitive” phase of expiration (34). Third, in sleeping humans, prolonged mechanical ventilation at increased Vt was shown to suppress active inspiration until PET CO2 had risen well above eupneic levels (35). Fourth, evidence in a single human subject (36) has demonstrated tonic activation of expiratory intercostal muscle EMG during apnea. In summary, on the basis of these similarities, we propose that our current findings in dogs are likely to be applicable to the control and resetting of respiratory motor output during mechanical ventilation in humans. On the other hand, we do not yet know whether sleeping humans will show similar sensitivities to inhibition of respiratory motor output and its aftereffects via ventilator frequency, Vt, and duration as presently shown in the dog.

The authors are indebted to Kathy Henderson for technical support, to Patricia Kalscheur for manuscript preparation, and to Steve Mifflin and Leon Glass for helpful feedback.

Supported by the NHLBI; Peter Eastwood was supported by a fellowship from the National Health and Medical Research Council of Australia.

1. Leevers AM, Simon PM, Dempsey JAApnea after normocapnic mechanical ventilation during NREM sleep. J Appl Physiol77199420792085
2. Leevers AM, Simon PM, Xi L, Dempsey JAApnea following normocapnic mechanical ventilation: a demonstration of control system inertia. J Physiol (London)4721993749768
3. Bruce EN, Daubenspeck JA. Mechanisms and analysis of ventilatory stability. In: Dempsey JA, Pack AI, editors. Regulation of breathing. New York: Marcel Decker; 1995. p. 285–314.
4. Wilson CR, Satoh M, Skatrud JB, Dempsey JANon-chemical inhibition of respiratory motor output during mechanical ventilation in sleeping humans. J Physiol5181999605618
5. Morrell MJ, Shea SA, Adams IL, Guz AEffect of inspiratory pressure support upon breathing in humans during wakefulness and sleep. Respir Physiol9319935770
6. Kimoff RJ, Kozar LF, Yasuma F, Bradley TD, Phillipson EAEffect of inspiratory muscle unloading on arousal responses to CO2 and hypoxia in sleeping dogs. J Appl Physiol74199313251336
7. Henke KG, Arias A, Skatrud JB, Dempsey JAInhibition of inspiratory muscle activity during sleep: chemical and non-chemical influences. Am Rev Respir Dis1381988815
8. Sammon M, Romaniuk JR, Bruce ENBifurcations of the respiratory pattern produced with phasic vagal stimulation in the rat. J Appl Physiol751993912926
9. Knox CKCharacteristics of inflation and deflation reflexes during expiration in the cat. J Neurophysiol361973284295
10. Lewis J, Bachoo M, Polosa C, Glass LThe effects of superior laryngeal nerve stimulation on the respiratory rhythm: phase resetting and after effects. Brain Res51719894450
11. Kitano S, Komatsu ACentral respiratory oscillator: phase–response analysis. Brain Res43919881930
12. Puddy A, Patrick W, Webster K, Younes MRespiratory control during volume-cycled ventilation in normal humans. J Appl Physiol80199617491758
13. Chow CM, Xi L, Smith CA, Saupe KW, Dempsey JAA volume-dependent apneic threshold during NREM sleep in the dog. J Appl Physiol76199423152325
14. Phillipson E, Murphy E, Kozar LRegulation of respiration in sleeping dogs. J Appl Physiol401976688693
15. Simon PM, Skatrud JB, Badr MS, Griffin D, Iber C, Dempsey JARole of airway mechanoreceptors in the inhibition of inspiration during mechanical ventilation in humans. Am Rev Respir Dis144199210331041
16. Younes M, Polacheck JCentral adaptation to inspiratory inhibitory, expiratory prolonging vagal input. J Appl Physiol59198510721084
17. Tobert DG, Simon PM, Stroetz RW, Hubmayr RThe determinants of respiratory rate during mechanical ventilation. Am J Respir Crit Care Med1551997485492
18. Lawson EERecovery from central apnea: effect of stimulus duration and end-tidal PCO2 partial pressure. J Appl Physiol531982105109
19. Xi L, Smith CA, Saupe KW, Henderson K, Dempsey JAEffects of rapid-eye movement sleep on the apneic threshold in dogs. J Appl Physiol75199311291139
20. Skatrud JB, Dempsey JAInteraction of sleep state and chemical stimuli in sustaining rhythmic ventilation. J Appl Physiol551983813822
21. Corfield D, Morrell MJ, Guz AThe nature of breathing during hypocapnia in awake man. Respir Physiol1011995145160
22. Fink BR, Hanks EC, Ngal EH, Pappen EMCentral regulation of respiration during anesthesia and wakefulness. Ann NY Acad Sci1091963892899
23. Orem JExcitatory drive to the respiratory system in REM sleep. Sleep191996S154S156
24. Knox CK, King GWChanges in the Breuer–Hering reflexes following rostral pontine lesion. Respir Physiol281976189206
25. Simon PM, Griffin D, Landry D, Skatrud JInhibition of respiratory activity during passive ventilation: a role for intercostal afferents. Respir Physiol9219935364
26. Lofaso F, Simoneau G, Laduirie FLR, Cerriva J, Chapelia A, Brenot F, Dartevelle P, Herve PFrequency of mechanical ventilation and respiratory activity after double lung transplantation. Respir Physiol921993319328
27. Zucker SShort-term synaptic plasticity. Amer Rev Neurosci1219891331
28. Mifflin SWShort-term potentiation of carotid sinus nerve inputs to neurons in the nucleus of the solitary tract. Respir Physiol1101997229236
29. Mifflin SConvergent carotid sinus and superior laryngeal nerve afferent inputs to neurons in the NTS. Am J Physiol2711996R870R880
30. Sears TA, Berger AJ, Phillipson EAReciprocal tonic activation of inspiratory and expiratory motoneurons by chemical drives. Nature2991982728730
31. Oku Y, Dick TE, Cherniack NSPhase-dependent dynamic responses of respiratory motor activities following perturbation of the respiratory cycle in the cat. J Physiol4611993321337
32. Orem J, Vidruk EHActivity of medullary neurons during ventilator-induced apnea in sleep and wakefulness. J Appl Physiol841998922932
33. Manchanda S, Leevers AM, Wilson CR, Simon PM, Skatrud JB, Dempsey JAFrequency and volume thresholds for inhibition of inspiratory motor output during mechanical ventilation. Respir Physiol1051996116
34. Satoh M, St. Croix CM, Skatrud JB, Dempsey JALung inflation during expiration resets respiratory rhythm in sleeping humans [abstract]. Am J Respir Crit Care Med1591999A783
35. Simon P, Dempsey JA, Landry DM, Skatrud JBEffect of sleep on respiratory muscle activity during mechanical ventilation. Am Rev Respir Dis14719933237
36. Bainton CR, Kirkwood PA, Sears TAOn the transmission of the stimulating effects of carbon dioxide to the muscles of respiration. J Physiol2801978249272
Correspondence and requests for reprints should be addressed to Jerome A. Dempsey, Ph.D., Department of Preventive Medicine, University of Wisconsin, 504 N. Walnut Street, Madison, WI 53706-2368. E-mail:


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record