Effects of Peak Inspiratory Flow on Development of Ventilator-induced Lung Injury in Rabbits

The study was approved by the Laboratory Investigation Committee of Osaka University Medical School. Animals were cared for in accordance with the University’s standards for care and use of laboratory animals.

Adult male New Zealand white rabbits (weighing 2.8 ± 0.2 kg) were anesthetized via  an ear vein with an injection of 50–60 mg pentobarbital and were set in a supine position on a heating pad. The tracheostomy was performed using a midline incision of the neck after infiltration of local anesthesia (1.5–2 ml of 1% lidocaine) at the level of the second or third tracheal cartilage. After insertion of a 4 mm inner diameter endotracheal tube (Blue Line tracheostomy tube; SIMS Portex, Kent, UK) into the trachea, the trachea was tied to prevent air leak. Then the animals were paralyzed with pancuronium bromide and mechanical ventilation was started with a Siemens Servo 300 ventilator (Siemens-Elema AB, Solna, Sweden) with the following settings: PCV pressure 10 cm H²O; Fio²1.0; respiratory rate 20 breaths/min with inspiratory time (Tⁱ) at 50% of total cycle time (T^tot); and positive end-expiratory pressure (PEEP) 0 cm H²O. Anesthesia and muscle relaxation were maintained with pancuronium bromide (0.2 mg·kg⁻¹·h⁻¹) and sodium pentobarbital (5 mg·kg⁻¹·h⁻¹).

The internal carotid artery was cannulated with a 20-gauge catheter (Angiocath 20; Becton Dickinson Vascular Access, Sandu, Utah) and received a Multiparameter sensor (Paratrend 7+™; Diametrics Medical Ltd., High Wycombe, UK) through the catheter for continuous intraarterial blood gas monitoring of Pao², pH, Paco², and temperature. The data from this sensor was sampled at 1-min intervals and stored on a personal computer. Animals with Pao²less than 500 mm Hg with these settings of ventilator were excluded from this study. As a result a total of 24 animals were investigated.

Arterial pressure was measured with a pressure transducer (TRANSPAC Monitoring Kit; Abbott Critical Care Systems, North Chicago, IL). The signals were amplified (AP-641G, AR-601G; Nihon Kohden, Tokyo, Japan) and stored on the computer via  an analog-digital converter (DI-220, Dataq Instruments, Akron, OH). Lactated Ringer’s solution was administered at a rate of 10 ml·kg⁻¹·h⁻¹, and the rate was titrated to maintain blood pressure within the range of ± 10% of initial values.

Before starting the study, a functional check of the ventilator was made according to manufacturer instructions, and then the pressure and flow transducers of the ventilator were calibrated with independent devices. The signals for pressure, flow and V^Twere monitored with the ventilator and similarly stored via  the converter to avoid the dead space resulting from the sensors. Recording of signal data on the computer system was managed using data acquisition software (WinDaq, Dataq Instruments, Akron, OH). Subsequent data analysis was performed with dedicated software (WinDaq playback, Dataq Instruments, Akron, OH).

The animals were randomly assigned to one of three groups according to ventilatory strategy: the pressure regulated volume control (PC) group (n = 8) received pressure-regulated volume control mode with Tⁱat 20% of T^tot; the volume control with 20% Tⁱ(VC²⁰) group (n = 8) received volume-control mode with Tⁱat 20% of T^tot; and the volume control with 50% Tⁱ(VC⁵⁰) group (n = 8) received volume-control mode with Tⁱat 50% of T^tot. Other, common, settings for each animal were as follows: V^T, 30 ml/kg; respiratory rate, 20 breaths/min; and PEEP, 0 cm H²O. CO²gas was added to maintain Paco²at 35–45 mm Hg. The inspiratory fraction of CO²was kept constant throughout the study.

Every 30 min, we measured intrinsic-PEEP levels using the expiratory hold button of the ventilator to perform end-expiratory airway occlusion. Similarly, using the inspiratory hold button of the ventilator, plateau pressure (P^plateau) was also measured every 30 min. To calculate the static compliance of the respiratory system (C^rs), end-inspiratory occlusion was repeated three times. Each occlusion was held for at least 2.5 s. From the recorded curves, C^rswas calculated as V^T/(P^plateau− intrinsic-PEEP).

All animals were ventilated for 6 h and then killed with a large dose of pentobarbital sodium. Then the animals were exsanguinated as much as possible, the chest was carefully opened, and the lungs and trachea were removed en bloc. The right lungs were used for comparative measurement of wet and dry weights and the left for histologic analysis. The right lungs were weighed immediately after excision and put in an oven (Sanyo, Osaka, Japan) at 37°C for 48 h and weighed again to obtain the dry weight for calculation of wet to dry weight ratio. The left lungs were fixed by intratracheal insufflation with 10% formalin at a hydrostatic pressure of 15 cm H²O and the specimens were floated in a fixative for 48 h. The lungs were cut into slices 5 mm thick in a coronal fashion from apex to base. Samples were systematically taken at a distance of 1.5 cm from the hilum in the lower lobe and embedded in paraffin. From each sample, 3-μm slices were stained with hematoxylin and eosin and Masson trichrome for microscopic examination. Experimental group information was unknown to the microscopist who evaluated the samples. Criteria were based on total score of four parameters: alveolar congestion, hemorrhage, infiltration or aggregation of neutrophils in airspaces or vessel walls, and thickness of alveolar wall/hyaline membrane formation by using a damage scale of 0 to 4 as follows: 0, minimal or no damage; 1, mild damage (less than 25% of the field); 2, moderate damage (25% to 50% of the field); 3, severe damage (50% to 75% of the field); 4, maximal damage (more than 75% of the field).9Therefore, a total score of 0 presented normal histology and a score of 16 presented maximal damage. The examination was again performed after 6 months. Difference of total score between the two examinations was less than 10% on each slide; mean values of the two examinations were used in the results.

Results are presented as mean ± SD. We used two-way analysis of variance for repeated measures to determine the statistical significance of group differences in blood gases and respiratory parameters at different time points as well as differences among measurement points in each group. One-way analysis of variance was used to determine the statistical significance of group differences in lung weight. When statistical significance was indicated, it was further examined by post hoc  analysis (Tukey honest significant difference test). Lung injury scores were analyzed by Kruskal-Wallis nonparametric analysis of variance followed by the Mann–Whitney U  test. A statistical software package (STATISTICA 5.5; StatSoft Inc., Tulsa, OK) was used, and significance was set at P < 0.05.

In 24 rabbits randomly allotted to each of the three groups we found that at the beginning of mechanical ventilation, each group had similar baseline characteristics for body weight, Pao², Paco^2,pH, and mean arterial pressure (table 1).

Figure 1shows typical flow and airway pressure tracings of each group at the beginning of mechanical ventilation. For each animal throughout the protocol, V^Twas maintained at 30 ml/kg. Peak inspiratory pressure (PIP) for the VC²⁰group was significantly higher than for the PC group and the VC⁵⁰group (P < 0.05) at the start of mechanical ventilation, and PIP for the VC⁵⁰group was significantly lower than for the PC group and the VC²⁰group at end of the protocol (P < 0.05). PIP for the PC group and the VC²⁰group at the end of the protocol significantly increased compared with those at start of mechanical ventilation. Mean airway pressure (Paw mean) for the VC²⁰group was significantly lower than for the PC group and the VC⁵⁰group throughout the 6 h of ventilation (P < 0.05). Paw mean for the three groups at end of the protocol significantly increased compared with those at start of mechanical ventilation (P < 0.05). Intrinsic-PEEP did not occur in any animals. In the PC group, peak inspiratory flow at end of the protocol significantly increased compared with that at start of mechanical ventilation (P < 0.05). Mean flow of the VC⁵⁰group was significantly lower than the other groups (P < 0.05), and mean flows of the PC group and the VC²⁰group were similar. Tidal volumes of the three groups did not differ significantly throughout the 6 h of ventilation. After 6 h, Pao²and pH for the PC group and the VC²⁰group decreased significantly and Paco²for the PC group increased significantly compared to those at the start (P < 0.05) (table 1).

At the start of mechanical ventilation, P^plateauand C^rsof the three groups did not differ significantly; after 3.5 h of mechanical ventilation C^rsof the PC group decreased significantly (P < 0.05) (fig. 2). This decrease in C^rswas followed by increased P^plateauin the PC group, which was significantly higher after 4 h of mechanical ventilation than at the start (P < 0.05) (fig. 3).

Figure 4shows the wet to dry ratio of each group. After 6 h ventilation the lungs of the PC group had more water content than the other groups (P < 0.05).

The lungs of the PC group exhibited gross damage, including edema with distinct petechial hemorrhages on the surface. The lungs of the VC²⁰group were less damaged than those in the PC group. In the VC⁵⁰group most lungs were pink and neither edematous nor hemorrhagic. Lung injury scores were larger in the PC group than in the VC²⁰group and the VC⁵⁰group (P < 0.05) (fig. 5). Microphotographs representative of the lungs in the three groups are shown in figure 6A, B and C. Alveolar and interstitial edema, alveolar hemorrhage, and infiltration of neutrophils are observed in the PC group (fig. 6A). Alveolar wall thickness is observed in some of the fields in the VC²⁰group (fig. 6B) and almost no histologic abnormality is observed in the VC⁵⁰group (fig. 6C).

Major findings of the current study are as follows: 1) PCV was associated with more lung damage, manifested by hypoxemia and histologic damage, than the low flow of volume-control ventilation (VCV) with the same V^Tand the same P^plateau; 2) between the two settings of VCV (VC²⁰and VC⁵⁰groups), there was no difference in oxygenation and histologic damage of the lungs.

These results may suggest that there is some threshold in peak inspiratory flow that becomes injurious to the lungs in ventilating with large V^T. A possible explanation of the difference in lung injury between the PC group and the VCV groups is the shear stress during inspiration on the airways and alveoli, although this was not proved in our study.

A few studies have reported the effect of peak inspiratory flow on VILI. Comparison of findings is difficult, however, because the ventilatory settings were not the same. Investigating the effect of peak inspiratory flow on VILI in sheep, Rich et al.  7compared two levels of PCV (20 and 45 cm H²O) at 5 breaths/min and 15 breaths/min of respiratory rates with VCV of 15 l/min of peak inspiratory flow at 5 breaths/min. VCV induced the least damage even at high PIP; they concluded that slow peak inspiratory flow protected against VILI. Although PIP was 50 cm H²O in both VCV and PCV of 45 cm H²O, in neither group of animals was V^Tcontrolled. It is possible that P^plateaumight have been higher in PCV. With 20 cm H²O of PCV, lung injury was less, even though the mean flow of PCV 20 cm H²O and 45 cm H²O was the same.

In the current study, respiratory rate and V^Twere same for all animals. Meanwhile in the PC group and the VC²⁰group, Tⁱwas same, and the flow waveform was square in the VC²⁰group and the VC⁵⁰group. In the PC group, the flow waveform was decelerating pattern. We did not compare decelerating waveform in VCV, so it still remains to be clarified whether peak inspiratory flow itself or flow waveform is the more important factor in VILI. There are some reports relating inspiratory flow profiles to gas exchange, work of breathing, and cardiovascular functions;8,10,11however, few studies have investigated the relationship between VILI and inspiratory flow profiles. Peevy et al.  have investigated the effect of inspiratory flow on microvascular injury in isolated perfusion lungs.12Using different ventilatory rates, PIP, Tⁱ, and V^Tfor different groups, they found that a combination of high PIP and high peak inspiratory flow induced the most severe microvascular damage. Within similar ranges of PIP, high peak inspiratory flow induced more microvascular damage. Consequently, their results suggest that not only overdistension but also peak inspiratory flow range are important in VILI. Their results suggest that high peak inspiratory flow and inspiratory waveform in the PC group might be important to develop VILI.

For PCV we used the pressure-regulated volume control of the Siemens Servo 300. This unique technology maintains preset V^Tby controlling pressure control levels. If the mechanics of the patient vary, the ventilator decreases or increases pressure control levels to maintain a constant V^T. In the current study, the pressure control level increased gradually toward the end of the protocol because the lungs were injured and had decreased compliance. Higher inspiratory pressure resulted in more lung injury. If we had chosen simple PCV, V^Tdecreased along with the development of lung injury and increased Paco². As a result, the lung damage could have been less. Even with the pressure-regulated volume control, Paco²increased significantly in the PC group, probably owing to increased dead space in the injured lungs. On the contrary, it is possible that high inspiratory pressure generated by pressure-regulated volume control ventilation maintained Pao²at high level. In that case, Pao²could have decreased more quickly if we chose simple PCV.

Many animal studies have provided evidence that PEEP protects against VILI.13,14At slow peak inspiratory flow, inspiratory time was longer and expiratory time was consequently shorter. This may create intrinsic-PEEP even in healthy lungs and protect the lungs against VILI. To confirm that intrinsic-PEEP was not the reason for lung protection in the VC⁵⁰group in the current study, we measured intrinsic-PEEP every 30 min using the expiratory hold button and confirmed for all the animals that intrinsic-PEEP was not occurring. Another possibility is the difference in Paw mean. The Paw mean in the VC²⁰group was the lowest one and there was statistically no difference in Paw mean between the PC group and the VC⁵⁰group. We cannot explain the result of our experiment by Paw mean.

This study was performed using a normal rabbit lung model, which does not reflect the same pathophysiology observed in humans or in ALI/ARDS.15,16To induce the lung injury, we chose high V^T(30 ml/kg), which could cause injuries solely as the result of overdistension without involving shear forces.17,18Although the V^Trange was much higher than that used in clinical settings, many previous animal studies also chose high tidal volumes.19,20These studies have been substantially contributing to understanding of the mechanisms of VILI and decreasing of the morbidity and mortality of ALI/ARDS. Even with high V^T, low peak inspiratory flows of VC²⁰and VC⁵⁰were associated with less lung damage. This shows that peak inspiratory flow actually plays a role in the development of VILI, as well as high V^T, body position,19atelectasis,20PEEP,13and inflammatory cytokines.9,21–23 

In the current study, we revealed that high peak inspiratory flow in PCV induced significantly more severe lung damage than lower peak inspiratory flows in the VC²⁰and the VC⁵⁰groups. Oxygenation deteriorated significantly in the PC group. In contrast, oxygenation impairment was absent in the VC²⁰and the VC⁵⁰groups after at 6 h of mechanical ventilation. Histologic examination discovered diffuse lung damage, with the most severe damage in the PC group. Our results may suggest that there is some threshold in flow, which becomes injurious to the lungs in ventilating with large V^T.

In conclusion, when an injurious V^Tis delivered at a higher peak flow, the deterioration in gas exchange, respiratory mechanics, and lung histology appears to be more marked than when it is delivered at a lower peak flow in an animal model.