Published ahead of print on April 24, 2003, doi:10.1164/rccm.200207-675OC
© 2003 American Thoracic Society
Controlled versus Assisted Mechanical Ventilation Effects on Respiratory Motor Output in Sleeping HumansDepartment of Population Health Sciences and Department of Medicine, John Rankin Laboratory of Pulmonary Medicine, University of Wisconsin-Madison, Veterans Administration Hospital Pulmonary Laboratory, Madison, Wisconsin Correspondence and requests for reprints should be addressed to Jerome A. Dempsey, Ph.D., Department of Population Health Sciences, The John Rankin Laboratory of Pulmonary Medicine, 504 North Walnut Street, Madison, WI 53726–2368. E-mail: jdempsey{at}wisc.edu
Central apneas occur after cessation of mechanical ventilation despite normocapnic conditions. We asked whether this was due to ventilator-induced increases in respiratory rate or VT. Accordingly, we compared the effects of increased VT (135 to 220% of eupneic VT) with and without increased respiratory rate, using controlled and assist control mechanical ventilation, respectively, upon transdiaphragmatic pressure in sleeping humans. Increasing ventilator frequency +1 per minute and VT to 165–200% of baseline eupnea eliminated transdiaphragmatic pressure during controlled mechanical ventilation and prolonged expiratory time (two to four times control) after mechanical ventilation. During and after assist control mechanical ventilation at 135–220% of eupneic VT, transdiaphragmatic pressure was reduced in proportion to the increase in ventilator volume. However, every ventilator cycle was triggered by an active inspiration, and immediately after mechanical ventilation, expiratory time during spontaneous breathing was prolonged less than 20% of that observed after controlled mechanical ventilation at similar VT. We conclude that both increased frequency and VT during mechanical ventilation significantly inhibited respiratory motor output via nonchemical mechanisms. Controlled mechanical ventilation at increased frequency plus moderate elevations in VT reset respiratory rhythm and inhibited respiratory motor output to a much greater extent than did increased VT alone.
Key Words: neuromechanical inhibition • resetting of respiratory rhythm • assisted mechanical ventilation • controlled mechanical ventilation • sleep Positive pressure mechanical ventilation applied to humans in non–rapid eye movement (NREM) sleep induces a loss of respiratory drive and an associated apnea after cessation of mechanical ventilation, despite normocapnic conditions (1–5). It is unclear whether these inhibitory effects are dependent on increases in VT (i.e., Hering Breuer mechanism) or are caused by a resetting of the inherent respiratory rhythm by the raised ventilator frequency (3, 6–8). On the one hand, in a sleeping canine model, small increases in respiratory frequency achieved via controlled normocapnic mechanical ventilation (CMV) resulted in complete abolition of diaphragm electromyogram (EMGdi) (3). On the other hand, in sleeping (5) or awake (9, 10) humans, increases in VT alone, as induced via normocapnic assist CMV (ACMV), resulted in reduced inspiratory effort with relatively little effect on breath timing. Our aim was to determine the relative contributions of increased respiratory rate and VT to the neuromechanical inhibition of respiratory motor output in the human. To this end, we compared the effects of CMV (at varying frequency and VT) and ACMV (at varying VT) on the amplitude and timing of transdiaphragmatic pressure (Pdi). We maintained normocapnia and used normoxic and hyperoxic backgrounds to exclude the influence of changing chemical drives and studied subjects during NREM sleep to avoid behavioral influences on respiratory motor output. Based primarily on animal studies (3, 11) and estimates of the sensitivity of the Hering Breuer reflex in the human (12–15), we hypothesized that resetting of the respiratory rhythm and complete inhibition of respiratory motor output via normocapnic mechanical ventilation in the human would require an increase in ventilator frequency combined with substantial elevations in VT.
Subjects Seven healthy adults, two women and five men, age of 36 ± 14 years, with no prior history of cardiopulmonary disease or sleep apnea participated in the study. Sleep architecture was staged during NREM sleep using standard electroencephalography (16). Each subject completed two to three nights of mechanical ventilation trials in stage II and stage III NREM sleep, and subjects were required to sleep on their back for the entire night. Each subject ingested 10 mg of zolpidem 10 to 20 minutes before the lights were turned out. In two subjects, we also completed several ventilator trials during NREM sleep before and after ingesting zolpidem. We observed similar effects of mechanical ventilation on Pdi in trials with and without the zolpidem in these two subjects when studied in the same sleep states.
Subject Consent, Preparation, Ventilator Circuit, and Measurements
CMV Each CMV trial was preceded by 2 to 3 minutes of stable eupneic breathing. The subject was switched into the ventilator circuit in the CMV mode during expiration and CMV continued for 60 to 90 seconds with FICO2 increased sufficiently to maintain PETCO2 normocapnic. Then the ventilator was switched off during early expiration and FICO2 returned immediately to room air levels near zero CO2. Expiratory time (TE) after the last CMV cycle was calculated as the time from the end of inspiration to the onset of the next spontaneous inspiratory effort, as determined from a negative shift in esophageal pressure (Pes) and an increase in Pdi, which exceeded the cardiogenic artifact.
ACMV
Data Analysis
Effects of CMV on Pdi and Breath Timing A polygraph recording from a typical CMV trial is shown for one subject (K.H.) in Figure 1A . This trial was conducted at an fR of three breaths per minute above the average spontaneous baseline frequency and a VT that was 200% of spontaneous baseline. Note the cardiogenic artifact superimposed on the Pes and Pga traces, which was especially prominent in this subject. At the onset of CMV, FICO2 was added to maintain PETCO2 1 to 1.5 mm Hg above baseline throughout the CMV trials. CMV caused an immediate positive shift in Pes and a marked reduction in Pdi, which was maintained over the 1-minute period of CMV. A transient increase in end-expiratory lung volume commonly occurred at the onset of CMV, but end-expiratory lung volume returned to or near baseline levels after the initial two to four ventilator cycles, as inferred from the constancy in end-expiratory Pes and Pga. At the cessation of CMV, a 17-second apnea ensued followed by resumption of spontaneous inspiratory efforts. Each of the initial two spontaneous breaths after the apnea had smaller VT and Pdi relative to baseline eupnea (see additional recordings during CMV at smaller VT settings in Figures E1 and E2 in the online supplement).
Single spontaneous and associated CMV breaths recorded at fast speed (28 mm/second) are shown in Figure 1B to illustrate the within-breath pressure changes. Note the typical development of negative Pes and positive Pga and Pdi during a spontaneous inspiration, as opposed to an unchanging Pes and Pdi during steady-state CMV at raised VT and frequency (see examples recorded at fast speeds in an additional subject and at two VT settings in Figures E3 and E4 in the online supplement).
Group mean values (n = 7) for
At the smallest (i.e., 35%) increase in VT during CMV and at +1 fR, nadir inspiratory  Pes shifted from negative values in spontaneous baseline breaths to +4 to 5 cm H2O during CMV; peak Pga increased 0.7 cm H2O above baseline, and Pdi fell to within ±1 cm H2O of zero Pdi. TE was not prolonged significantly after the final ventilator cycle, and Pdi and VT were reduced slightly but significantly below baseline on the first spontaneous breath after CMV.
At the two larger increases in VT during CMV (65% and 100% relative to spontaneous eupnea),
After the final CMV cycle with VT at two times eupnea, TE averaged 8–9 seconds or approximately 4.5 times the baseline spontaneous TE, and TE averaged 4–6 seconds when VT was held at 1.6 times eupnea (Figure 3)
. On the first spontaneous breath after these post-CMV apneas, the mean
In summary, normocapnic CMV at increased fixed fR and VT reduced
Effects of ACMV with Increased VT on Pdi and Breath Timing
Figure 5 shows group mean values for Pdi and VT during and after ACMV trials. PETCO2 averaged 0.4 ± 0.1 mm Hg greater than control (range = -0.9 to 2.8 mm Hg) over the three types of ACMV trials at increased VT. TI during ACMV remained similar to baseline spontaneous breathing; thus, the increases in VT were achieved by increases in inspiratory flow rate above baseline eupnea by 30% at the smallest increase in VT and by twofold at the highest VT. When VT was raised to 135% of baseline control via ACMV, TE was unchanged, and Pdi was reduced to 40% of baseline eupnea. During ACMV at the two higher VT settings, Pdi was reduced to less than 20% of eupnea (at 170% VT) and to -1 to -2 cm H2O (at 220% VT). Mean TE was prolonged to 128% and 134% of baseline eupnea during ACMV at 170% and 220% VT, respectively (p < 0.5).
On the first spontaneous recovery breath after cessation of ACMV, mean TE remained unchanged from baseline eupnea after the 135% VT trial but was prolonged to 120% of baseline TE (p < 0.05) after the 165 and 220% VT trials. These effects of ACMV on TE were less than 20–25% of the mean TE prolongation observed after CMV at comparable VT elevations (Figure 3). Pdi was 20–65% less than spontaneous control on the first two recovery spontaneous breaths and returned to near baseline levels by the third to sixth recovery breaths. Similarly, inspiratory flow rate and VT were also reduced in magnitude significantly below control during the first spontaneous recovery breaths (Figure 5), and these reductions were greatest when the VT delivered during the preceding period of ACMV equaled or exceeded 170% of eupneic control.
Effects of ACMV on Within-Breath Pes, Pm, and Pdi Figures 6A and 6B show a single spontaneous breath + continuous positive airway pressure and associated ACMV cycles recorded at high speed to illustrate the within-breath pressure changes. Note first the typical development of a negative esophageal and positive Pdi during a spontaneous inspiration. During ACMV, initial small changes occurred in Pm, Pes, and Pdi before the ventilator trigger. Then with the onset of ventilator-assisted inspiratory flow and positive Pm (1) at the 140% increase in VT, negative swings in Pes and positive Pdi were greatly reduced, relative to spontaneous, and (2) at the 200% increase in VT, Pes moved in a positive direction and Pdi in a slightly negative direction throughout inspiration.
Effects of Hyperoxia on Pdi and Breath Timing during CMV and ACMV In two subjects, inspired air was enriched with 60% O2 at a flow rate sufficient to raise PETO2 to 310–325 mm Hg throughout six trials of normocapnic ACMV at VT 180 to 200% of baseline eupnea and nine trials of normocapnic CMV at fR (+1fR) and VT 150 to 180% of eupnea (data not shown). With normocapnic hyperoxic ACMV, reductions in Pdi below spontaneous eupnea averaged 50–60% during the final 15 seconds and 30–40% on the first spontaneous breath after cessation of ACMV. These changes were within ±10% of those after normoxic normocapnic ACMV in these two subjects. Similarly, with hyperoxic normocapnic CMV trials, Pdi was reduced to 0 to 1 cm H2O during the final 15 seconds of CMV and TE prolonged four times eupneic control after CMV, as in the normoxic trials.
Our findings in sleeping healthy humans have shown that inspiratory motor output, as indicated by Pdi, is reduced during controlled, normocapnic mechanical ventilation when ventilator frequency is increased as little as one breath per minute above spontaneous eupnea. The complete elimination of inspiratory motor output also required that VT during CMV be raised 65% or more above eupnea. Prolonged TEs occurred after cessation of CMV, and their length increased in proportion to the magnitude of the VT maintained during the CMV but were independent of further increases in CMV frequency. During assisted normocapnic mechanical ventilation synchronous with inspiratory effort (ACMV), inspiratory motor output was never completely eliminated, and TE was only slightly but significantly prolonged. However, during ACMV at raised VT, Pdi was reduced well below spontaneous control; furthermore, in recovery from ACMV, the initial spontaneous efforts produced Pdi and VT, which were 20–70% below baseline eupneic control breaths. These reductions in spontaneous Pdi and VT in recovery from ACMV were greatest at the two highest VTs maintained during ACMV. Hyperoxia did not prevent the inhibitory effects of either CMV or ACMV on Pdi or apnea during or after mechanical ventilation. We attribute the resetting of respiratory rhythm and apnea caused by CMV to the neuromechanical effects of repeated and augmented lung inflation and the reduced Pdi during and after ACMV to the neuromechancial inhibitory effects on respiratory motor output associated with increased VT.
Use of Pdi to Assess Reduced Respiratory Motor Output during and after CMV and ACMV
First, the intrinsic mechanical properties of the diaphragm were changed during mechanical ventilation, both by the increased VT, and therefore shortened muscle length, and by the increased flow rate, and therefore increased velocity of muscle shortening. Based on phrenic nerve stimulation studies at varying lung volumes and flow rates in humans (18, 19), we estimate that our increases in VT and mean flow rates would have accounted for approximately 20% of the mean reduction observed in Pdi at the maximum VT and flow rate and 5–10% of the reduction in Pdi at the smaller increase in VT and flow rate. Second, phasic activation of expiratory muscles with relaxation of the diaphragm during subsequent inspiration could also have assisted in reducing diaphragmatic pressure development; however, this seems unlikely given the absence of changes in end-expiratory Pga or Pes throughout trials of ACMV. Third, changes in Pdi reflect primarily the activity of the diaphragm; thus, reduced or even eliminated Pdi during mechanical ventilation does not necessarily mean that motor output to all inspiratory pump muscles was similarly reduced. Finally, there were several circumstances during passive mechanical ventilation at high VT where the increases in Pes during inspiration exceeded that in Pga, resulting in a negative Pdi. In these circumstances, the transmission of changes in pleural pressure to the abdominal compartment was attenuated. This may be explained by the retention of passive tension in the diaphragm, as demonstrated by Froese and Bryan (20) in the paralyzed human diaphragm during mechanical ventilation at raised VT with subjects in the supine position. Alternatively, in the supine position, the abdominal contents may push on the diaphragm and impair transmission of all of the The recovery data obtained during spontaneous breathing in NREM sleep immediately after CMV or ACMV unequivocally demonstrate the significant inhibition of central respiratory motor output that occurred as a result of normocapnic mechanical ventilation. First, our interpretation that CMV completely eliminated central respiratory motor output to all inspiratory muscles is shown by the substantial apneas that consistently followed CMV conducted at increased frequency and at VT settings that were elevated 65% or more above eupnea. Furthermore, post-CMV apneic length was dependent on the magnitude of the VT maintained during CMV. Second, immediately after normocapnic ACMV, TE was prolonged, but only to one-fifth the TE observed after CMV at comparable levels of VT. However, we did consistently observe a significant reduction in Pdi and in VT in the initial two to four recovery spontaneous breaths when compared with the baseline spontaneous eupneic breaths, and these reductions after ACMV were greatest in trials with the higher VT settings delivered during the ACMV. Finally, we note that our interpretation of Pdi data during and after CMV in the human is consistent with previous studies in sleeping dogs, which showed that diaphragmatic EMG (as recorded from indwelling electrodes) was completely eliminated during and immediately after normocapnic CMV (at +1 fR) (3).
Past versus Present Findings Using CMV and ACMV Most of our findings in sleeping humans on CMV are similar to those in sleeping dogs (3), although there appears to be less volume-sensitive inhibition of respiratory motor output in the human versus the dog. For example, respiratory motor output in the human and dog was sensitive to both CMV frequency and VT; however, the dog required only a 10–20% increase in ventilator VT at raised frequency to cause complete cessation of EMGdi and substantial postventilator apnea, whereas the human required a much larger percentage increase in ventilator VT (Figure 3). At the same time, both species showed substantial VT-dependent apneas after normocapnic CMV.
ACMV effects.
Chemoreceptor versus Nonchemoreceptor Influences Third, our increases in FICO2 and VT likely had opposing effects on the amplitude of oscillations in arterial PCO2 and pH (33). The effect of changes in arterial PCO2 oscillation at constant mean PaCO2 on ventilatory control are controversial (34, 35); however, in studies which both measured and controlled these oscillations, no effect on respiratory motor output was discernable, at least when VCO2 was at resting levels (36). Certainly, we cannot attribute the reduced respiratory motor outputs after ACMV or CMV to alterations in CO2 oscillations because FICO2 was reduced to room air levels in recovery. We have additional types of evidence against a significant role for carotid chemoreceptors in accounting for reductions in respiratory motor output during isocapnic mechanical ventilation. First, we found equal amounts of resetting and/or inhibition of respiratory motor output during and after normocapnic CMV and ACMV in normoxia versus hyperoxia. Hyperoxia is known to markedly reduce the carotid chemoreceotor response to CO2 (37, 38). Second, in the sleeping dog, we found that bilateral denervation of the carotid bodies also did not influence the magnitude of inhibition of EMGdi during and after normocapnic CMV or ACMV (unpublished findings from the authors' laboratory) or pressure support ventilation (39). Perhaps medullary chemoreceptor PCO2 was reduced independently of arterial PCO2, but we know of no reason to expect such a dissociation to occur during normocapnic mechanical ventilation. Our study also provides evidence that the inhibitory influences of mechanical ventilation will occur even in the presence of mild but significant hypercapnia. First, the initial few spontaneous inspiratory efforts after the termination of apneas in recovery from normocapnic CMV showed a VT and Pdi that were consistently reduced significantly below eupneic baseline values, despite a PaCO2 that was undoubtedly significantly greater than eupnea. Second, during isocapnic ACMV at raised VT, we occasionally observed very small inspiratory efforts that failed to trigger the ventilator to deliver the desired VT (see the example in Figure 7 and, recorded at a faster speed, in Figure E5 in the online supplement). Note that despite the resultant transient increase in PETCO2 to 2–3 mm Hg above normocapnia after two successive untriggered efforts, the subsequent inspiratory effort still remained well below those in baseline eupnea. Finally, in sleeping dogs, raising the PETCO2 3–5 mm Hg above eupneic control during CMV at increased frequency and VT did not prevent the elimination of EMGdi or postventilator apneas (3). Similarly, in sleeping humans, inspiratory effort was eliminated via hypocapnic CMV at raised VT and did not return (as CMV was continued) until PETCO2 was increased an average of 6 mm Hg above spontaneous eupnea (40).
In summary, although we acknowledge the apnea and hypopnea-producing effects of even very small decrements in PaCO2 in sleeping humans (2, 23, 41), we believe the evidence, as outlined previously here, speaks strongly in favor of nonchemoreceptor mediation of the observed elimination or reduction in respiratory motor output during and after isocapnic CMV or ACMV.
Mechanisms of Neuromechanical Inhibition during CMV and ACMV The elimination of respiratory motor output in the human during and after CMV by means of increased ventilator frequency and VT represents a "resetting" of the respiratory rhythm. Two types of powerful influences are responsible for this resetting effect, as previously documented in anesthetized or sleeping dogs and cats (3, 43). First, application of single ventilator breaths during early expiration, that is, in the so-called inflation-sensitive phase of the respiratory cycle, causes TE prolongation (3, 11) whose duration is dependent on the magnitude of the ventilator volume (3). Second, although these TE prolonging effects lasted only for one respiratory cycle (3, 42), repeated ventilator cycles performed at an increased frequency above eupnea exerted a cumulative inhibitory effect on respiratory motor output, thereby lengthening the inflation sensitive phase of each respiratory cycle (3). This time-dependent cumulative inhibitory effect of CMV was manifested in postventilator apneas in both sleeping dogs (3) and humans (Figures 1 and 3) and was analogous to that originally reported in anesthetized piglets secondary to repeated electrical stimulation of the superior laryngeal nerve (44). In the sleeping dog, postventilator apneas first appeared after two or three CMV ventilator cycles, and apnea length increased progressively further as CMV was prolonged up through a duration of approximately 10 ventilator cycles (3). In the dog, tonic expiratory muscle EMG activity was also shown to occur simultaneously with a silent diaphragm EMG throughout the period of passive CMV, as well as during the subsequent apneic recovery period (3). In the anesthetized cat, Knox (11) showed that TE prolongation via application of single ventilator breaths during neural expiration required intact vagal feedback. However, in lung transplant humans who were awake and vagally blocked dogs, we showed postventilator apneas after normocapnic CMV at raised frequency and VT (45). Given the evidence cited previously here along with the potentially confounding effects of wakefulness, we propose that the role of vagal feedback in the resetting of respiratory rhythm during CMV needs testing in a sleeping state, where the effects of reflex inhibition may be more readily unmasked. In summary, present evidence using both CMV and ACMV confirms and extends earlier work (8) that strong and sustained inhibition of the drive to breathe may be exerted by nonchemical feedback influences in the sleeping ventilated human. Furthermore, these inhibitory influences persist beyond the termination of the mechanical perturbations and in the face of rising chemical stimuli. The influence of either mode of mechanical ventilation was critically dependent on the magnitude of the ventilator VT and likely on vagal feedback (discussed previously here). However, a raised fR (with CMV) combined with a moderately augmented VT showed substantially greater and more sustained inhibition of the amplitude and timing of respiratory motor output than did synchronous increases in VT alone (as with ACMV). We attribute this potent cumulative inhibitory effect on breath timing during CMV to repeated application of the ventilator volume during the progressively prolonged inflation-sensitive phase of the respiratory cycle.
The authors thank Ms. Patricia Kalscheur for manuscript preparation.
Supported by NHLBI, the Medical Research Service of the Department of Veteran's Affairs, an Allen+Hanburys Postdoctoral Research Fellowship, the Thoracic Society of Australia and New Zealand (A.R.), and National Heart, Lung, and Blood Institute-sponsored institutional training grant (H.H.). This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form July 9, 2002; accepted in final form April 23, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
z
ns, Switzerland) that allowed the control of VT, breathing frequency, and inspiratory time (TI). We used three VT settings, 1.3 to 2 times the eupneic VT, and two fixed ventilator frequencies (fR) (+1 and +3 breaths per minute above mean eupneic fR) during 1 minute of normocapnic CMV. The TI was set at the TI observed during spontaneous breathing; thus, increases in VT during CMV were achieved solely by increasing inspiratory flow rate. A square wave inspiratory flow pattern was selected. A servo-controlled flow valve in the ventilator opened during inspiration to supply gas to the subject and to control inspiratory flow rate, pattern, and duration. The ventilator delivered breath was terminated when the servo-controlled flow valve closed after delivering the preset VT. The ventilator delivered VT settings that were within 5 to 10% of the preset VT. 
E was increased 1.4 to 2 times baseline, and mean PETCO2 was maintained an average of 0.5 ± 0.1 mm Hg greater than eupnea over all CMV trials. 
