Main

Respiratory mechanical unloading (otherwise called negative ventilator elastance and negative ventilator resistance) has been proposed as a new mode of assisted mechanical ventilation(1–4). Compared with the conventional modes of patient-initiated mechanical ventilation, this technique may enable the clinician to improve the match of applied ventilator pressure waveforms and specific abnormalities in respiratory system elastance and/or resistance. Improving the match between ventilator pressures and lung mechanics may reduce the overall pressure needed for achieving a given ventilation and may therefore reduce the barotrauma associated with mechanical ventilatory support.

During elastic unloading, the ventilator pressure applied at the endotracheal tube connector (Paw) increases nearly instantaneously in proportion to the volume of each spontaneous inspiration and returns to a preset end-expiratory pressure (PEEP) during expiration. The clinician sets the degree of elastic unloading, i.e. the ratio (gain) of the increase in ventilator pressure per unit of tidal volume initiated by the patient. This ratio ΔP aw/V is the ventilator's negative elastance, which can be adjusted on a continuous scale at the ventilator front panel. In this mode, the ventilator pressure is servo-controlled to enhance changes in lung volume during spontaneous breathing and decreases the elastic work of breathing.

During resistive unloading, the ventilator pressure is servo-controlled in synchrony with the flow of spontaneous breathing so that it increases above the set PEEP in proportion to the inspiratory airflow and decreases below baseline in proportion to expiratory airflow. The clinician can set the degree of resistance compensation, i.e. the ratio (gain) of the increase in ventilator pressure per unit of airflow, which is initiated by the patient. The ratio ΔP aw/V˙ is the ventilator's negative resistance, which can also be adjusted on a continuous scale on the ventilator front panel. This type of ventilator pressure profile compensates for airway resistance and specifically decreases the resistive work of breathing.

Elastic and resistive unloading can be combined to reduce both the elastic and resistive work of breathing. The clinician can set two independent gain factors, a gain for elastic compensation and a gain for resistive compensation. This translates into a ventilator pressure, which is at any point in time proportional to the weighted sum of both the flow and volume of spontaneous breathing.

Thus, the patient can breath spontaneously during all three modes and controls tidal volume, airflow, and timing parameters of the breathing cycle. The ventilator enhances the effect of the patient's effort but it does not initiate a respiratory cycle or control the pattern of breathing. If there is no spontaneous effort, there will be no change in ventilator pressure.

However, if the gain of the assist is set inappropriately high, such that the negative ventilator elastance or resistance exceeds and overcompensates lung elastic recoil or airway resistance, this will result in an unstable positive feedback situation. Any small increase in lung volume or airflow will induce an increase in ventilator pressure so as to further increase lung volume or airflow, which in turn can result in even higher ventilator pressures. Such an inappropriately high gain may result in runaway instability. Overcompensation may occur if the clinician increases negative ventilator elastance or resistance above respiratory system elastance or resistance. It may result, however, from an improvement in respiratory system elastance or resistance when a preset gain of the assist is not reduced appropriately. These situations are potential safety hazards. To our knowledge, the physiologic effects of these types of overcompensation have not been studied systematically.

The aim of this study was to evaluate physiologic consequences of two different gain settings: 1) when the gain of the assist was adjusted to partially unload lung elastic recoil and/or airway resistance, or 2) when the gain of either was increased to an extend that the negative ventilator elastance or resistance supersede the respiratory system mechanics to cause overcompensation.

We specifically investigated short-term effects on the pattern of spontaneous breathing, electromyographic activity of the diaphragm(EMGdi), changes in end-expiratory lung volume, and systemic blood pressure fluctuations. We studied surfactant-depleted and normal rabbits to evaluate whether those effects occur with and without derangements in pulmonary mechanics.

METHODS

Animal preparation. Sixteen adult New Zealand White rabbits(mean body weight, 2710 g; range, 2000-4500 g) were anesthetized with 5 mg/kg xylazine and 25 mg/kg ketamine intramuscularly. Anesthesia was maintained by a continuous i.v. pentobarbital infusion of 15 mg kg-1 h-1 i.v. infusion fluids were given at 3 mL kg-1 h-1. Tracheostomy was performed on each animal, and the trachea was cannulated with an endotracheal tube (length, 150 mm; inner diameter, 3.0 mm), secured in place by a peritracheal ligature. The animals were then connected in the supine position to a ventilator. They received 100% O2 throughout the study with an end-expiratory pressure of 0.2 kPa before and 0.5 kPa after the lavage procedure. An arterial line was inserted into the right carotid artery for intermittent determination of arterial blood gases (Ciba-Corning Diagnostics Corp., Medfield, MA) and for recording of arterial blood pressure(Deltranâ„¢ pressure transducer, Utah Medical Products, Inc., Midvale, UT; cutoff frequency 70 Hz). To measure the EMGdi, bipolar stainless steel hook electrodes were inserted in the right hemidiaphragm via an abdominal incision. The EMGdi signals were amplified and filtered (Grass P15 Preamplifier, Grass Instrument Co., Quincy, MA; bandwidth 30-1000 Hz). A 2-cm latex balloon was advanced into the lower third of the esophagus to record esophageal pressure (Pes), referenced to atmospheric pressure. The balloon was connected with a 5 French polyethylene catheter to a Statham PD 23 pressure transducer (Gould, Cleveland, OH). Mercury-in-rubber strain gauge jacket plethysmography (model 270-A Plethysmograph, Parks Electronics Laboratory, Beaverton, OR) was used to assess the effects on end-expiratory lung volume(5) of changing the gain of elastic and/or resistive unloading.

Lung surfactant depletion was induced by lavage procedures(6) until the Pao2 remained stable below 33.3 kPa while the animal was mechanically ventilated with 100% O2 for at least 30 min. Once the target Pao2 level had been established, the study protocol was continued as soon as the animal restarted regular spontaneous breathing activity.

Ventilator. A Stephanie® infant ventilator (F. Stephan Medizintechnik GmbH, Gackenbach and Dresden, Germany) was used throughout the study. Technical details of this device have been described elsewhere(2–4). It is a servo-controlled system with sensors for Paw and the flow near the endotracheal tube. The flow sensor is a pneumotachometer with a 1.1. kPa L-1 s resistance at 5 L/min and a dead space of 0.9 mL(7). In addition to the conventional modes of mechanical ventilation, the Stephanie® infant ventilator provided elastic unloading, resistive unloading, or a combination of these modalities. The cutoff frequency of the system in the respiratory mechanical unloading modes was above 15 Hz.

Data acquisition and processing. The Paw, VË™, Pes, EMGdi, and plethysmography signals were digitized and stored on a disk using an AT CODAS board with a Toshiba portable personal computer and CODAS data acquisition software (DATAQ Instruments, Inc., Akron, OH). The sampling frequency on each channel was 1 kHz. WINDAQ playback software (DATAQ Instruments, Inc.) was used to review, process, and analyze the acquired waveforms. Processing included integration of the flow signal to obtain tidal volume. After applying a digital 60-Hz notch filter to the raw EMGdi signal, a filter was used to remove baseline drifts, and the signal was full-wave rectified and integrated. To analyze single breaths, the WINDAQ notepad feature was used to extract wave form data at specific time points such as the zero crossings of the flow tracing. These data were put into spreadsheets for calculation of derived parameters such as lung compliance.

Protocol. The following ventilator settings were applied for 10 min periods each: CPAP, resistive unloading, elastic unloading, and combined unloading. At the end of each period, all signals were recorded over at least 10 consecutive respiratory cycles, and arterial blood samples were obtained. To allow for sequence effects, the ventilatory modalities were applied in random order determined by choosing from a set of sealed envelops that contained the mode assignment. To preclude potential carryover effects, a CPAP period of at least 10 min was interposed between the test modes during the experiments in animals with normal lungs. This was not possible in surfactant-depleted animals as they were unable to endure multiple epochs of CPAP without any other assist. Also, surfactant-depleted animals were not exposed to partial resistive unloading alone as the pressure assist was too low with this setting. After data acquisition during resistive unloading, the gain of the assist was continuously increased for a short period of time so that the signals could be recorded during resistance over-compensation. After data acquisition during elastic unloading, the gain was also increased so that the signals could be recorded at full compensation of lung elastance and at elastic overcompensation.

Gain adjustment during elastic unloading. The real time computer display of the flow and the esophageal pressure waveforms was used to assist in adjusting the gains. Starting from zero (which is essentially CPAP), the gain of elastic unloading was gradually increased. This decreased the difference in Pes between the points of end-expiration and end-inspiration. When Pes at end-inspiration equals Pes at end-expiration, the gain of the elastic unload is high enough to completely offset tidal changes in lung elastic recoil. At this point, the negative ventilator elastance equals lung elastance. A further increase in gain overcompensated lung elastic recoil and caused the end-inspiratory Pes to exceed end-expiratory Pes. We set the gain just below full compensation for the protocol test settings of "appropriate elastic unloading."

Gain adjustments during resistive unloading. To find an"appropriate gain" for resistive unloading, we gradually increased the gain from zero while the VË™ and P waveforms were observed. The gain which completely compensated respiratory system resistance (the negative ventilator resistance which equaled the respiratory system resistance) can be recognized when it starts to cause oscillations(8). In our protocol, we adjusted the gain just below this level which was determined to be the setting of resistive unloading at the appropriate gain.

Data analysis. Data are given as mean ± SD. Unpaired two-tailed t tests were used to evaluate the differences in baseline parameters between rabbits with normal and those with surfactant-depleted lungs. One-way repeated measures ANOVA was used to test for differences in the outcome variables (inspiratory time, EMG activity per breath, and so forth) between the ventilator modalities (CPAP, resistive unloading, and so forth). To assess the variability of the measurement techniques, coefficients of variation were calculated between breaths within animals for each outcome variable separately for each ventilatory modality. One-way repeated measures ANOVA was used to test whether the variability was different between modes. Bonferroni t tests were applied to isolate differences between the modes whenever a difference was detected by ANOVA. p values <0.05 were considered significant.

RESULTS

All rabbits who maintained a regular spontaneous breathing activity after the initial surgical procedures or after surfactant depletion and during the following continuous hour-long infusion of pentobarbitol tolerated the respiratory mechanical unloading modes well. No ventilation-related acute complications such as air leak syndromes occurred during any of the test or baseline experiments including overcompensation. Three of the 16 animals had to be excluded because they had long apneic episodes and erratic breathing after the initial surgery. The protocol test settings for normal lungs were applied to six rabbits. One of them was also studied after surfactant depletion. In the remaining seven rabbits, surfactant depletion was attempted immediately after the initial surgery without the test runs in animals with normal lungs. Five animals completed the protocol for surfactant-depleted lungs; three never regained spontaneous respiratory activity after the surfactant washout procedure.

Performance of the ventilator. The ventilator reliably generated a pressure output that was in proportion to the airflow or/and the changes in lung volume. This was confirmed by comparing the ventilator's pressure output with the volume plethysmograph signal that was obtained independently from the airflow signal, which was the only part of the feedback loop between the animal and the ventilator. During elastic unloading, the Paw was in proportion and in phase with the plethysmographic volume signal (Fig. 1). During resistive unloading, Paw was in proportion and in phase with the first derivative of the plethysmographic volume signal.

Figure 1
figure 1

Two spontaneous breaths of a normal rabbit during elastic unloading. The perpendicular line marks the time when the airflow signal crosses zero from expiration to inspiration. Paw was measured at the endotracheal tube connector.

Full size image

Gain during mechanical unloading adjusted to partially or completely compensate lung elastance and/or airway resistance. During partial elastic unloading (appropriate gain), the ventilator decreased the change in Pes per unit of inspired volume. During full elastic unloading, there were no changes in Pes related to tidal volume, so that esophageal pressures were equivalent at end-expiration and at end-inspiration. Under these conditions, the Pes signal changed during inspiration in proportion to the VË™ signal reflecting airway resistance only. If airway resistance was additionally compensated by resistive unloading during full elastic unloading, Pes changes with tidal breathing were almost absent (Fig. 2).

Figure 2
figure 2

Spontaneous breathing during CPAP and during combined resistive and elastic unloading in a normal rabbit. It was attempted to set the gain of the assist close to full compensation of lung elastance and airway resistance (right tracings). BP, blood pressure.

Full size image

There were fluctuations in systemic blood pressure with tidal breathing during CPAP characterized by a postinspiratory rise in blood pressure and a subsequent during the remaining expiratory time and early inspiration. These fluctuations became progressively smaller with an increase in the gains of elastic and resistive unloading. They were virtually absent when lung elastance and airway resistance were fully compensated (Fig. 2).

The end-expiratory lung volume as observed on the volume plethysmographic tracings did not change more than 1 mL during the different unloading modalities compared with CPAP provided that the same end-expiratory pressure was applied. All measures of diaphragmatic muscle activity decreased with any combination of elastic and resistive unloading in both normal and surfactant-depleted lungs (Tables 1 and 2). There was a significant decrease in the inspiratory duration of EMGdi activity in normal lungs. Combined resistive and elastic unloading decreased the integrated EMGdi activity per breath significantly. The mechanical inspiratory time as measured from the pneumotach signal decreased in proportion to the EMGdi activity. There was no significant change in the total mechanical cycle time, although respiratory rates tended to be higher with the assist. Because tidal volume increased significantly during elastic unloading in animals with normal and surfactant-depleted lungs and tended to be higher in both during all other test settings, minute ventilation increased significantly in all test runs except for resistive unloading in normal lungs. Therefore, the quotient of ventilation per unit of integrated EMGdi activity increased during respiratory mechanical unloading. Compared with normal lungs, this increase was larger in animals with surfactant-depleted lungs.

Table 1 Baseline data as measured during CPAP
Full size table
Table 2 Effects of resistive unloading (RU), elastic unloading (EU), and combined unloading (RU + EU) on the pattern of breathing and the diaphragmatic EMG activity
Full size table

Peak inspiratory airflow did not change during elastic unloading but increased significantly during resistive unloading and during the combined assist in normal lungs. Peak expiratory airflow increased significantly with resistive unloading and decreased significantly during elastic unloading. We did not observe active expiration during the experiments.

The variability of all the reported measures as judged from coefficients of variation was not different between modes in rabbits with normal lungs(with the exception of peak expiratory airflow). We therefore averaged the variability data across the applied modes. This averaged between breaths variability was <5% for all reported variables except for the integrated EMG activity (6.09%). The variability of the peak expiratory airflow was slightly albeit statistically significantly higher during combined unloading compared with both CPAP and resistive unloading (2.76% versus 1.49% and 1.89%, respectively). These findings were similar in rabbits with surfactant-depleted lungs. The coefficients of variation were not significantly different between modes and <5% with two exceptions, the variability of the duration of inspiratory EMG activity was slightly higher during combined unloading compared with both CPAP and elastic unloading (6.5%versus 4.4% and 3.43%, p = 0.005), and the coefficient of variation of the integrated EMG activity was 5.84%.

Overcompensation of lung elastance. When we produced elastic overcompensation with a gradual increase in gain during elastic unloading, tidal volume increased progressively, whereas the respiratory rate decreased. The tidal volumes during overcompensation were associated with an increase in Pes, suggesting passive inflation of the lungs. During elastic overcompensation, the initial expiratory flow was low and then accelerating. There was a "turning point" obtained during increasing the gain from almost full elastance compensation to overcompensation. At this gain, the decelerating expiratory flow pattern turned into an accelerating expiratory flow pattern (Fig. 3).

Figure 3
figure 3

Recording of spontaneous breathing in a normal rabbit during elastic unloading. The gain of the assist was continuously increased up to full compensation and overcompensation of lung elastance.

Full size image

Compared with an active inhalation during CPAP or to elastic unloading with a moderate gain, passive lung inflation during elastic overcompensation had an opposite effect on systemic blood pressure: it was followed by a decrease in systemic blood pressure. The amplitude of this cyclic depression increased when the gain was increased into overcompensation(Fig. 3).

Overcompensation of airway resistance. Resistive overcompensation induced sinusoidal flow oscillations over time. These occurred first between unchanged respiratory efforts. The oscillatory amplitudes increased with higher gains of overcompensating resistive unloading, and the respiratory efforts declined. Gross resistive overcompensation with large tidal oscillations inhibited the diaphragmatic EMG activity completely (Fig. 4). There was, however, a transitional type of entrained EMGdi activity during oscillations with smaller amplitudes. This was characterized by a burst of EMGdi activity during each oscillatory cycle.

Figure 4
figure 4

Transition from CPAP to resistive unloading with a gradually increased gain up to resistive overcompensation in a surfactant-depleted rabbit.

Full size image

The end-expiratory lung volume did not change with overcompensation during resistive unloading, even when total inhibition of diaphragmatic muscle activity ensued. The oscillations occurred symmetrically above and below the end-expiratory lung volume that was recorded with CPAP if the mean airway pressure during oscillation was equivalent to the previously applied CPAP level (Fig. 4). These effects of elastic and resistive overcompensation occurred in animals with normal and with surfactant-depleted lungs.

DISCUSSION

Laboratory studies in small animals(3,4,9) have provided promising data on the use of respiratory mechanical unloading as a mode of assisted mechanical ventilation in models of neonatal respiratory diseases. However, no studies on safety aspects of this technique or a closely related method called "proportional assist ventilation"(10,11) have been published. In our study, we applied unloading in two different gain ranges, either in an appropriate range so as to partially compensate for lung elastance and/or airway resistance, or we set the gain deliberately "inappropriately high" to overcompensation which may induce a self-perpetuating increase in ventilator circuit pressure.

We have shown that a well controlled and stable breathing pattern ensues during respiratory mechanical unloading. In both normal and surfactant-depleted rabbits, the between breaths variability of all our outcome measures was small and remained largely unaffected by any type of the assist when the gain was set to partially relieve elastic and/or resistive work of breathing. Respiratory unloading enhanced the effect of diaphragmatic activity on ventilation, more so in surfactant-depleted lungs. During unloading, the total diaphragmatic activity was down-regulated in animals with normal and injured lungs. This, however, did not completely offset the increase in ventilation such that it was higher during the assist mode than during unassisted breathing. The end-expiratory lung volume remained unaffected by any combination of elastic or resistive unloading.

Effects of any combination of unloading on tidal breathing-related fluctuations in beat-to-beat arterial blood pressure may have clinical implications(12). It has been shown that elimination of fluctuating cerebral blood-flow velocity may reduce intraventricular hemorrhage in preterm infants(13). We have shown in this study that fluctuations in systemic blood pressure with tidal breathing decrease with an increased gain during unloading to the point of total abolition at full elastic and resistive unloading. In this experimental preparation, the recorded blood pressure fluctuations during CPAP and low levels of unloading did not simply reflect a direct transmission of pleural pressure variations to the blood pressure. Figure 2 shows that the cyclic decrease in blood pressure starts when there is no change in esophageal pressure at end-expiration, long before the inspiratory decrease in esophageal pressure begins. This suggests that the observed cyclic variations in blood pressure primarily reflect an inspiratory surge of blood volume into the pulmonary vascular bed, which is followed by a cyclic increase in left ventricular preload and ventricular output. This assumes that right ventricular output exceeded left ventricular output during inspiration, increasing pulmonary blood volume, whereas the pulmonary vascular bed emptied during expiration when left ventricular output exceeds right ventricular output. These cyclic changes disappeared at full compensation during resistive and elastic unloading. Contrary to this, cyclic passive inflations of the lung with an increased intrathoracic pressure during conventional mechanical ventilation induce a rhythmic depression in systemic blood pressure(14). Elastic overcompensation caused passive lung inflations with an increase in esophageal pressure. The inflations during elastic overassistance in our study were associated with a decrease in blood pressure similarly to the effects on blood pressure of conventional ventilation (Fig. 3).

We observed a real time display of the esophageal pressure wave form in conjunction with the airflow signal to determine to what extend the applied gain compensated lung elastic recoil. In the clinical situation, this may be the optimum method to evaluate the effects of different gains. However, esophageal pressure is not routinely monitored in ventilated newborns. It will therefore be helpful to look for indirect signs of overcompensation during respiratory mechanical unloading. In our study, elastic overcompensation typically caused a marked increase in tidal volume with a long end-inspiratory plateau. Even though this did not cause serious acute complications such as air-leak syndromes, it caused cyclic depression in arterial blood pressure and interfered with the rhythm of spontaneous breathing. To limit the effects of a sudden inadvertent elastic overcompensation, an upper circuit pressure limit and an inspiratory time limit must be in place during this type of assisted ventilation.

In this study, a borderline resistive overcompensation induced small amplitude oscillations. Such resonant oscillations will occur in any pneumatic, mechanical, or electrical system incorporating an analog of elastic and inertial properties, provided that a resistive component is either absent or fully compensated. In our experiments, small resonant oscillations were most helpful in identifying the range of borderline resistive overcompensation. These oscillations did not change the end-expiratory lung volume, did not change the rhythm of spontaneous breathing, and did not have deleterious effects on arterial blood pressure. However, a further increase in gain during resistance overcompensation increased the amplitude of these oscillations and finally caused a passive high frequency oscillatory ventilation which completely inhibited diaphragmatic activity. Preceding complete inhibition, there was a transient phenomenon of entrained diaphragmatic activity characterized by a burst of EMGdi activity during each oscillatory cycle. To the best of our knowledge, an entrainment of diaphragmatic activity during these oscillations of the respiratory system near its presumed resonant frequency has not been described before. It is, however, well known that endogenous biologic rhythms will synchronize to rhythmic external stimuli in a process called selective frequency entrainment(15,16).

Conclusion. We conclude that elastic and resistive unloading can enhance the effect of diaphragmatic muscle activity on ventilation. We are cautious, however, that overcompensation of lung elastance or airway resistance during unloading could result in large tidal volumes or in oscillations which could be harmful.