Main

Neonates undergoing mechanical ventilation often have vigorous spontaneous respiratory activity. It has been shown that synchronizing ventilator cycles to spontaneous breathing by patient-triggered ventilation or respiratory mechanical unloading may improve gas exchange (1, 2) and reduce ventilator pressure requirements (2, 3). Potential disadvantages of neuromuscular blockade in ventilated neonates include atrophy of respiratory muscles, reduced cardiac output and blood pressure (4), and lower functional residual capacity (5).

PLV, also known as perfluorocarbon-associated gas exchange, initially described by Fuhrman et al. (6), uses conventional gas tidal breathing into a lung partially filled with perfluorochemical fluids. PLV has been shown to decrease alveolar surface tension, resulting in alveolar recruitment, which may improve lung mechanics (79) and reduce ventilation-perfusion mismatch, resulting in improved gas exchange (10, 11). Using various animal models of lung disease, it has been shown that PLV may result in improved gas exchange and pulmonary mechanics (8, 9, 1219) and cause less barotrauma as assessed by lung histology (15). However, PLV has been tested almost exclusively in pharmacologically paralyzed subjects and not during preserved spontaneous respiratory activity. Most recently, it has been shown that synchronized mechanical ventilation is feasible in spontaneously breathing piglets with healthy lungs during PLV (20).

This study is the first step of a number of ongoing studies evaluating the feasibility of spontaneous breathing assisted by respiratory mechanical unloading during PLV in animals without and with different lung diseases. The aim of these studies is to explore physiologic advantages of preserved spontaneous ventilation during PLV.

Respiratory mechanical unloading, also called proportional assist ventilation, is a recently developed mode of ventilatory support (2123) that adjusts its support in proportion to the patient's effort. During respiratory mechanical unloading, there is no fixed target airflow, volume, or Paw. However, the Paw applied is in proportion to the patient's airflow and to the inspired tidal volume at any point in time within each respiratory cycle. Thus, the depth, timing, and pressure profile of each ventilator breath is controlled by the patient (21, 22). This ventilatory mode was chosen because it provides potential advantages compared with other assisted modes of ventilation, such as improved patient comfort, preservation of the subject's own reflex and control mechanisms, and a lower peak as well as mean Paw to maintain ventilation (2, 23). Combination of PLV and respiratory mechanical unloading may be useful to improve and to compensate for decreased lung compliance in subjects with lung disease and during the weaning process from PLV. Respiratory mechanical unloading is highly dependent on the patient's own respiratory control. The focus was to study the interaction between the animals and the ventilator. Because respiratory control mechanisms may be affected by lung diseases, we used animals with healthy lungs as a first step before studying animals with lung disease. We hypothesized that respiratory mechanical unloading, when combined with PLV, would result in stable ventilation and gas exchange in spontaneously breathing animals.

METHODS

This animal research protocol was approved by the Animal Care Committee of the government agencies of Baden-Wuerttemberg. The study was designed as a cohort study using each animal as its own control. The animals were initially supported with respiratory unloading and then with unloading combined with PLV.

Ventilator.

A Stephanie infant ventilator (Stephan Medizintechnik GmbH, Gackenbach, Germany) was used throughout the study. In addition to conventional ventilation, this ventilator provides negative ventilator resistance and elastance, also called resistive and elastic unloading. This device has been described in detail elsewhere (21, 2426). It is a servocontrolled system with a pneumotachograph (dead space, 0.6 mL) placed between the endotracheal tube connector and the ventilator circuit. The system continuously receives the flow signal of the animal's spontaneous breathing from the pneumotachograph. This signal is processed by microcomputer using special algorithms to control a rapid valve that determines the Paw applied at the endotracheal tube. The pressure applied per unit airflow determines the degree of resistive unloading (Kr). The pressure applied per unit of inspired volume determines the degree of elastic unloading (Ke). Thus, the delivered Paw is a weighted sum of the resistive and the elastic components at any point in time during a spontaneous breathing cycle MATH where V˙ is flow and V is volume.

Animal preparation.

Ten female adult New Zealand White rabbits with a body weight of 3063 ± 149 g (mean ± SD) were given 0.2 mg/kg atropine and anesthetized with ketamine (15–40 mg/kg) and xylazine (1.5–4 mg/kg) i.v. After supine positioning, animals were intubated using a 3.0 or 3.5 mm cuffed endotracheal tube, and the cuff was inflated to prevent leaks. A rectal temperature probe (Siemens Sirecust 302, Erlangen, Germany) was placed, and a core temperature of 38 to 39.5°C was maintained using a heating blanket and an overhead warmer (Babytherm 8000, Draeger, Luebeck, Germany). Anesthesia was maintained with a continuous infusion of ketamine (50–130 mg/kg/h). The dose was adjusted to maintain anesthesia deep enough to prevent spontaneous movements other than respiration. During the surgical instrumentation, the animals were placed on volume-controlled, synchronized intermittent positive-pressure ventilation with the following settings: assist/control mode; FiO2, 0.21–0.4; tidal volume, 10 mL/kg; PEEP, 0.4–0.6 kPa; inspiratory time, 0.4 s; and minimum respiratory rate, 15–20/min, which, in case of poor respiratory effort, was adjusted to maintain normoventilation (PaCO2, 4.7–6.0 kPa). This ventilator mode ensured adequate ventilation in case respiratory effort would be impaired secondary to deep anesthesia. Dextrose 5% with 35 mmol/L Na, 18 mmol/L K, and 1 U/mL heparin was administered at 5 mL/kg/h into a peripheral vein. A 3.5F arterial femoral line was inserted for continuous blood pressure monitoring and sampling for blood gas analyses and was continuously flushed with heparinized (1 U/mL) normal saline solution at a rate of 3 mL/h.

Airflow was measured using the pneumotachograph of the ventilator. Tidal volume was calculated by integrating flow. Paw was measured at the connector of the endotracheal tube, and Pe was measured through a fluid-filled 5F feeding tube with its tip placed into the distal esophagus. All pressure transducers (Sorenson Transpac tranducers, Abbott Critical Care Systems, North Chicago, IL) were calibrated using a water manometer. Immediately before data acquisition, correct placement of the esophageal tube was checked by performing end-inspiratory airway occlusions and comparing Paw and Pe. A ΔPe/ΔPaw ratio of 1.00 ± 0.05 was accepted (27). Otherwise, the esophageal catheter was repositioned until correct placement was confirmed. The catheter was continuously flushed with water (3 mL/h) to avoid bubble formation. Arterial Hb oxygen saturation (SpO2) was measured transcutaneously with a Nellcor N 200 pulse oximeter (Nellcor Inc., Hayward, CA). All signals were digitized at a frequency of 100 Hz and recorded simultaneously using a data acquisition system (DATAQ Instruments, Inc., Akron, OH).

Protocol.

After instrumentation, animals were switched to the unloading mode of ventilation using the same PEEP and an FiO2 of 1.0. Resistive unloading was adjusted to compensate approximately for the expected resistance of the endotracheal tube, which was estimated by measuring peak airflow and by using previously published data (28). Elastic unloading was adjusted to maintain a PaCO2 within the target range. After allowing the animal 30 min to adjust to the new mode of ventilation, data were recorded for 60 min, and arterial blood gases were measured at the end of this GV period.

Next, 30 mL/kg of prewarmed (38°C) PFC (Rimar 101, Miteni, Italy) was instilled continuously at a rate of 1 mL/kg/min into the endotracheal tube without disconnecting the ventilator. The degree of resistive unloading was increased during the administration of the PFC to compensate for increased airway resistance caused by the presence of liquid in the airways during filling. The degree of elastic unloading was readjusted to maintain a PaCO2 within the target range. Specifically, the degree of elastic unloading was increased if respiratory rate increased during liquid filling by >50% or increased to >100 breaths/min, because preliminary experience has shown that this increase would be very likely to be associated with CO2 retention. The degree of elastic unloading was decreased if tidal volume was >10 mL/kg, because it has been our experience that this would be associated with low respiratory rates, low PCO2, and apnea in these animals without lung disease. Arterial blood gases were used to confirm PaCO2 values being within the target range.

The filling condition was ascertained by disconnecting the animal from the ventilator and performing a slight thoracic compression to observe a meniscus at the endotracheal tube. Immediately after filling, data were recorded again for 60 min, and arterial blood gases were measured at the end of this PLV period. No refill was performed during the PLV period. Sequence randomization was not feasible because complete removal of the liquid from the lung would not have been possible within the time frame of the study.

Data analysis.

Primary outcome measures for the comparison of the two ventilatory modes (GV versus PLV) were V˙E and its variability. Variability of V˙E was measured as the coefficient of variation, calculated as the SD of V˙E data, obtained from 1-min intervals, divided by the mean V˙E for each animal. Secondary outcome measures were tidal volume as obtained by integration of the expiratory flow signal from all breaths, and respiratory rate as measured using flow and Pe traces. The means and coefficients of variation of data obtained from 1-min intervals of these variables were compared between both ventilation modes (GV versus PLV). Peak Pe deflections were measured from 20 randomly distributed breaths from each recording period to calculate mean peak Pe and coefficients of variation. Mean Paw was calculated as the integral of Paw divided by the recording time. Lung compliance and airway resistance were calculated by a program based on the equation of motion (29). Work of breathing of the lung was calculated as the area given by the integral of inspiratory transpulmonary pressure over volume. Transpulmonary pressure was defined as Paw minus Pe. The power of breathing was calculated as work of breathing over time. Heart rate and mean ABP were measured from the ABP trace using the complete recording periods. Arterial blood gases were drawn at the end of each ventilation mode. The time with SpO2 <85% was measured as the percentage of the complete recording period.

Statistics.

Two-tailed paired t tests were used to compare matched data. If data were not normally distributed, Wilcoxon signed rank tests were used instead. We considered p < 0.05 to indicate statistical significance. Values are expressed as mean ± SD or as median and range.

RESULTS

Instillation of PFC was well tolerated in all animals, without apparent adverse respiratory or hemodynamic consequences except for ventilation problems in one animal, secondary to an inadvertent overfill with liquid, which were abolished immediately by increasing PEEP transiently from 0.6 to 0.8 kPa. Coughing was induced in three animals during filling but disappeared immediately after an additional i.v. injection of 5–30 mg of ketamine. The ketamine dose was similar during both ventilatory conditions (72 ± 17 versus 77 ± 20 mg/kg/h; GV versus PLV). The degree of resistive unloading was similar during both ventilatory conditions (2.5 [1.0–3.7] kPa/L/s), and the degree of elastic unloading was 0.01 (0–0.02) versus 0.02 (0.01–0.03) kPa/mL during GV versus PLV.

Figure 1 shows representative traces of airflow, Pe, Paw, and ABP of one of the animals during both ventilatory conditions. A decrease in Pe was followed instantaneously by an increase in Paw, indicating that patient effort and ventilator pressure were in phase during both GV and PLV.

Figure 1
figure 1

Representative traces of airflow, Pe, Paw, and ABP of one of the animals during GV and PLV. A decrease in Pe was followed almost instantaneously by an increase in Paw, indicating that patient effort and ventilator pressure were in phase during both ventilatory conditions.

Table 1 shows the effects of filling the lung with liquid on ventilatory and hemodynamic variables. Compared with GV, V˙E was significantly larger during PLV, whereas there was no difference in tidal volume. Mean respiratory rate was also significantly higher during PLV. Therefore, the larger V˙E during PLV was entirely caused by the higher respiratory rate. Mean Paw was slightly higher during PLV, indicating more ventilatory support during PLV. Lung compliance was decreased, whereas airway resistance and work and power of breathing were increased during PLV compared with GV. There were no statistically significant differences between the means of peak Pe deflections, heart rate, and mean ABP between ventilation modes.

Table 1 Effect of PLV on V˙E, tidal volume, respiratory rate, peak Pe deflection, mean Paw, lung compliance, airway resistance, work and power of breathing, heart rate, and mean ABP Values are mean ± SD. *p < 0.05; †p < 0.01.

Variability data expressed as coefficients of variation for V˙E, tidal volume, respiratory rate, and peak Pe deflections for both ventilatory conditions (GV versus PLV) are shown in Table 2. Coefficients of variation were similar for all variables, suggesting no difference in variability of these variables between both ventilatory conditions.

Table 2 Coefficients of variation for V˙E, tidal volume, respiratory rate, and peak Pe deflections for both ventilatory modes Values are median (range).

Arterial blood gas values, drawn at the end of each ventilation mode, are shown in Table 3. There were no differences for pH, base excess, and PaCO2, but PaO2 was higher during GV than during PLV. SpO2 was always between 99% and 100% during both ventilation modes in all animals.

Table 3 Arterial blood gases drawn at the end of each ventilation mode Values are mean ± SD. *p < 0.05. Abbreviation: BE, base excess.

DISCUSSION

The principal finding of this study is that ventilation and gas exchange are maintained in spontaneously breathing rabbits with normal lungs partially filled with PFC. V˙E was larger during PLV versus GV, whereas tidal volume remained unchanged. Despite increased V˙E during PLV, PaCO2 values were unchanged after transition to PLV in our animals. This observation is consistent with findings from other investigators, who found increased PaCO2 values after transition from GV to PLV in piglets with healthy lungs when V˙E was controlled for by using volume-controlled ventilation (30). Because PaCO2 remained unchanged in our study, only changes in ventilation-perfusion mismatch, an increased arterial-alveolar CO2 gradient, or an increased metabolic rate during PLV can explain the difference in V˙E. The latter seems to be unlikely, because, at least for total liquid ventilation it has been shown that oxygen consumption and CO2 production do not change during transition from gas to liquid ventilation (31). However, there is evidence of ventilation-perfusion heterogeneity and diffusion limitation in animals with healthy lungs during PLV (3234). Mates et al. (32) found that O2 and CO2 exchange is impaired in proportion to the volume of added PFC fluid into the lung of normal piglets. This may entirely explain the higher respiratory rate and V˙E observed in our study and also the higher mean Paw because of the inherent coupling of ventilatory support and patient effort of our ventilator system. Although any increase in ventilatory support over time potentially may promote ventilator-associated lung injury, the magnitude of the observed difference in mean Paw is probably not clinically significant.

The decreased lung compliance and increased resistance during PLV is explained by the high density and viscosity of the PFC in comparison to gas. Similar changes in pulmonary mechanics have been observed during volume-controlled mechanical ventilation after intratracheal PFC administration in healthy animals and therefore do not seem to be specific for the type of ventilatory support used in our study (35). The changes in pulmonary mechanics resulted in increased work of breathing. Power of breathing increased even more, because respiratory rate increased as well to maintain PaCO2. Improvement of lung mechanics resulting in a decreased work of breathing can be expected only in subjects with severe pulmonary disease in which PLV may lead to significant recruitment of alveoli. In this case, a lower degree of ventilatory support, resulting in a lower mean Paw, may be expected during PLV compared with GV.

The decreased PaO2 during PLV is consistent with results of other studies (6, 30, 32, 33, 35) and may be explained by impaired diffusion, ventilation-perfusion heterogeneity, and shunt in animals with healthy lungs. Because animals were ventilated with an FiO2 of 1.0 and had no lung disease, we expected a PaO2 of >40 kPa with an SpO2 of 99–100% throughout the study, provided ventilation was stable. Inasmuch as ventilation was stable during both ventilation modes, episodes of hypoxemia did not occur.

Heart rate and ABP were unchanged after transition from GV to PLV, which is consistent with observations of other investigators (6, 14, 30, 35, 36).

Shaffer and Moskowitz (37) have performed early experiments in dogs with a demand-controlled device using the Pe of the animal as the respiratory input signal to control pumps and valves of a total liquid-ventilation system. Although these animal experiments demonstrated that this system can provide adequate control of gas exchange, there was a tendency toward progressive hypercarbia over time. Most recently, Bendel-Stenzel et al. (20, 38) have published studies in which they assessed the dynamics of spontaneous breathing during patient-triggered PLV in piglets without lung disease and with ARDS, showing basically that spontaneous breathing and triggering a synchronized infant ventilator is feasible with a liquid-filled lung. As we found in our study, these authors also showed that the animals were able to self-regulate their respiratory rate and minute ventilation to maintain physiologic blood gases during all tested ventilatory modes such as assist/control, regular IMV, and synchronized IMV, with assist/control being the most efficient mode in terms of CO2 elimination. The unloading mode of ventilation used in our study is to some extent comparable with the assist/control mode of patient-triggered ventilation, because every spontaneous breath is supported by the ventilator system. However, the major difference is that during the unloading mode, the pressure profile within each ventilator breath is in proportion to the volume and flow produced by the subject's own spontaneous breathing at any point in time of each breath. In the studies mentioned above, the animals triggered ventilator breaths, which were characterized by a preset peak pressure and inspiratory time. However, currently no studies are available to determine whether any of the ventilatory modes such as mechanical unloading, IMV, synchronized IMV, assist/control, or pressure support ventilation have any long-term physiologic or clinical advantages in spontaneously breathing animals with a partially liquid-filled lung.

All changes observed between the ventilatory modes in our study may, at least in part, be related to sequence effects. However, this is unlikely, inasmuch as the observed changes can easily be explained on the basis of known physiologic mechanisms that typically occur during PLV. Carryover effects are also unlikely, as the filling procedure allowed a washout period of as long as 30 min between modes. Furthermore, the main purpose of the study was to demonstrate stability of the target variables.

In conclusion, PLV combined with respiratory mechanical unloading achieved stable ventilation and gas exchange in spontaneously breathing animals without lung disease. Therefore, spontaneous breathing with a partially liquid-filled lung seems to be feasible, allowing avoidance of paralysis, which is associated with side effects. However, PLV combined with respiratory mechanical unloading resulted in impaired lung mechanics and increased work of breathing in animals without lung disease. Therefore, there is no direct clinical application for this mode of ventilation in most subjects without severe lung disease. However, our findings may be of clinical relevance for future applications of liquid ventilation such as pulmonary administration of drugs (39, 40) or serving as a carrier for gene targeting of bronchial or alveolar cells (41). Further studies should clarify whether PLV combined with respiratory mechanical unloading may help to recruit lung volume and improve gas exchange in spontaneously breathing animals with lung disease, and whether or not spontaneous breathing has circulatory or other advantages compared with controlled mechanical ventilation in paralyzed subjects with a partially liquid-filled lung.