Recent studies in healthy mice and rats have reported that positive pressure ventilation delivered with physiological tidal volumes at normal end-expiratory volume worsens lung mechanics and induces cytokine release, thus suggesting that detrimental effects are due to positive pressure ventilation per se. The aim of this study in healthy animals is to assess whether these adverse outcomes depend on the mode of mechanical ventilation.
Rats were subjected to 4 h of spontaneous, positive pressure, and whole-body or thorax-only negative pressure ventilation (N = 8 per group). In all instances the ventilatory pattern was that of spontaneous breathing. Lung mechanics, cytokines concentration in serum and broncho–alveolar lavage fluid, lung wet-to-dry ratio, and histology were assessed. Values from eight animals euthanized shortly after anesthesia served as control.
No evidence of mechanical ventilation–dependent lung injury was found in terms of lung mechanics, histology, or wet-to-dry ratio. Relative to control, cytokine levels and recruitment of polymorphonuclear leucocytes increased slightly, and to the same extent with spontaneous, positive pressure, and whole-body negative pressure ventilation. Thorax-only negative pressure ventilation caused marked chest and lung distortion, reversible increase of lung elastance, and higher polymorphonuclear leucocyte count and cytokine levels.
Both positive and negative pressure ventilation performed with tidal volumes and timing of spontaneous, quiet breathing neither elicit an inflammatory response nor cause morpho-functional alterations in normal animals, thus supporting the notion of the presence of a critical volume threshold above which acute lung injury ensues. Distortion of lung parenchyma can induce an inflammatory response, even in the absence of volotrauma.
Large tidal volumes used in positive pressure ventilation in normal lungs can cause ventilator-induced lung injury
Anesthetized rats subjected to positive pressure or negative pressure ventilation with the breathing pattern occurring before mechanical ventilation did not experience ventilator-induced lung injury, unless marked chest distortion was produced, supporting the hypothesis that with normal lung expansion, there is a critical strain threshold associated with ventilator-induced lung injury
POSITIVE pressure mechanical ventilation (PPMV) is widely used as a life-saving intervention, as well as a valuable tool during anesthesia. Nonetheless, this intervention is regarded as potentially harmful even in healthy lungs, when ventilator-induced lung injury can arise from overstretching of lung parenchyma, leading to surfactant inactivation, failure of the alveolar–capillary barrier, and edema,1–3 or abnormal stresses due to cyclic opening of closed airways, inducing epithelial necrosis and sloughing and airway–parenchymal uncoupling.4–6 Furthermore, ventilator-induced lung injury is usually,6–8 though not always,3,9 accompanied by the release of proinflammatory cytokines, possibly because of cellular mechano-transduction processes.10,11
If overstretching and cyclic airways closing and opening were the major triggers of the pathological cascade of ventilator-induced lung injury, then a safe PPMV would be obtained by using appropriate tidal volumes (VT) and end-expiratory lung volumes (EELV). This task could be challenging or even impossible in the clinical settings, where lung parenchyma is often heterogeneous.12,13 However, no harm should result from the PPMV of normal lungs if VT and EELV are similar to those of spontaneous, quiet breathing. Indeed, the conclusion has been drawn that ventilation becomes injurious to normal lungs only above a relatively high, species-dependent VT threshold, usually greater than 15 ml/kg.3
In contrast to this conclusion, recent studies indicate that a safe threshold in terms of VT and EELV does not exist for PPMV. In healthy rats, an increased expression of proinflammatory cytokines has been observed in macrophages of broncho–alveolar lavage fluid (BALF) after 2 h of PPMV with a VT of 10 ml/kg,14 and disruption of the extracellular matrix, with perivascular space engorgement, cuff formation, and substantial alterations of lung mechanics was found after 4 h of PPMV with VT of 8 ml/kg and zero end-expiratory pressure.15 In normal mice, PPMV with a VT of 7.5 to 8 ml/kg and positive end-expiratory pressure of 2 to 4 cm H2O for 4 to 6 h caused an increased expression in lung tissue homogenates, and higher concentration in serum and BALF of several inflammatory cytokines,16–18 besides recruitment of pulmonary granulocytes,16,18 moderate lung edema,16,17 and increased permeability of the alveolar–capillary barrier.18 These studies suggest that even the ventilatory pattern adopted by the spontaneously breathing animal becomes injurious if produced in the anesthetized, paralyzed animal by means of conventional positive pressure ventilators. Indeed, a recent study has shown that in surfactant-depleted, paralyzed rabbits negative pressure ventilation improves oxygenation and lessens parenchymal and airway injury relative to conventional PPMV.19
With the aim of assessing whether mechanical ventilation with the VT of spontaneous, quiet breathing causes per se detrimental effects on lung functions in normal animals, the outcomes of prolonged spontaneous ventilation (SV) in anesthetized rats have been compared with those obtained with PPMV and negative pressure mechanical ventilation of corresponding duration, negative pressure having been applied either to the whole body (NPWBMV), as in the iron lung, or to the thorax only (NPTOMV), as with cuirass or poncho-like devices.
Materials and Methods
Forty male Sprague–Dawley rats (weight range, 365 to 440 g) were premedicated with diazepam (10 mg/kg) and anesthetized with an intraperitoneal injection of pentobarbital sodium (40 mg/kg) and chloral hydrate (170 mg/kg). The animals were kept supine and they breathed room air at zero end-expiratory pressure throughout the experiment. The rectal temperature was maintained at 37°C with a heating pad. The trachea was cannulated, and a balloon-tipped catheter placed in the lower third of the esophagus, the appropriate positioning of the balloon being tested with the occlusion method.20 A pentobarbital (approximately 4 mg/ml) saline solution was continuously infused (approximately 8 ml kg−1 h−1) into a tail vein.
The animals were handled according to the National Institute of Health guiding principles. The study was approved by Ministero della Salute, Rome, Italy.
Procedure and Data Analysis
Airflow, tracheal (Ptr), and esophageal pressure signal were measured, digitized, and stored as previously described.5,6 Transpulmonary pressure (PL) was obtained as Ptr-esophageal pressure, and volume changes (ΔV) by integration of the digitized airflow signal.
The whole experimental procedure is shown in figure 1. After a 10-min period of SV (point A), the animals were either killed immediately with an overdose of anesthetics (C group) or randomly assigned to one of the four ventilation modes (SV, PPMV, NPWBMV, and NPTOMV group; eight rats each). Animals were paralyzed (fig. 1, squares) with pancuronium bromide (1 mg/kg), and ventilated with the pattern observed during spontaneous breathing. Adequateness of anesthesia was repeatedly checked from the absence of sudden changes of heart rate and paw pinch reflex in the spontaneously breathing animals.
Positive pressure ventilation was performed using a specially designed, computer-controlled ventilator.5 Negative pressure ventilation with the required VT and timing was obtained by means of an adjustable negative pressure source and computer-controlled valve group connected to perspex chambers containing either the whole body or its upper part only, in which case a rubber diaphragm was made to fit the animal’s body just below the xiphoid process. In spontaneously breathing animals, application of the diaphragm alone had little constraining effects, as lung elastance increased by approximately 2%. Ports in the chamber walls provided the connections to the tracheal cannula, esophageal balloon, venous catheter, and rectal probe.
To ensure the same volume history and alveolar recruitment, two inflations at Ptr of approximately 20 cm H2O were performed before the collection of mechanical measures (fig. 1, hatched areas). Lung dynamic elastance (EL) and resistance (RL) were computed using the subtraction method for SV, NPWBMV, and NPTOMV groups,21 and the rapid end-inflation occlusion method for PPMV group,5 whereas chest-wall elastance (Ew) was computed according to the subtraction method in all but the SV group. The difference between the end-expiratory and residual volume was assessed by connecting the expiratory port to a drum in which pressure was set at −10 cm H2O and used as an index of EELV. After the 4-h ventilation period, all animals were subjected to PPMV for about 7 min (crossed area in fig. 1), for the assessment of pulmonary and chest-wall quasistatic elastance (Est,L and Est,w, respectively), interrupter resistance of the lung (Rint,L; essentially airway resistance), respiratory viscoelastic resistance (Rvisc,rs), and time constant (τvisc,rs) according to the rapid end-inflation occlusion method, as previously described.5 Two inflation pressure–volume curves were also obtained by slowly inflating (approximately 1 ml/s) the respiratory system to a Ptr of 40 cm H2O.
After collection of 1.5 to 2 ml of blood from the heart (point B in fig. 1), the animals were killed with an overdose of anesthetic, the main left bronchus was cannulated, and the left lung removed, weighed immediately, lavaged with 4.3 ml/kg of normal saline in two aliquots, left overnight in an oven at 120°C, and weighed again to compute the wet-to-dry ratio (W/D). The effluents were pooled, centrifuged (Harrier 18/80; Sanyo Gallenkamp PLC, Loughborough, United Kingdom) at 2,000 rpm for 10 min, and the supernatant was frozen and stored at −20°C.
Cytokine (tumor necrosis factor-α, interleukin [IL]-1β, IL-6, and IL-10, macrophage inflammatory protein-2) analysis and assessment of albumin concentration was carried out in duplicate in blinded fashion on BALF and serum, as previously described.6 Sandwich enzyme immunoassay of E-Selectin was performed using goat anti-rat E-selectin antibodies (R&D Systems Inc., Minneapolis, MN). Absorbance was read at 405 nm. A pool of sera from 20 anesthetized healthy rats was used as the standard control, and the readings from individual specimens were expressed as percent of control.
The right lung was fixed by intratracheal instillation of a 8% formaldehyde, 0.1% glutaraldehyde solution with the pressure maintained at 20 cm H2O for 24 h, and processed for histological examination according to standard methods, as previously described.5,6,22 Histologic evaluation was performed in a blind fashion by a single observer by using an image analysis system (IMAQ; National Instruments, Austin, TX). The following measures were obtained (1) mean linear intercept (Lm), a measure of air-space enlargement, and its coefficient of variation as a measure of dispersion; (2) percent ratio of abnormal to total (normal and abnormal) bronchiolar–alveolar attachments, and distance between normal attachments as indices of airway–parenchymal mechanical uncoupling; (3) percent ratio of lesioned (epithelial necrosis and sloughing) to total membranous bronchioles, the bronchiolar injury score, as an index of small airway injury; (4) percent ratio of cuffed to total arterioles (50 to 250 μm in diameter) as an additional index of edema; and (5) polymorphonuclear leukocyte count in the alveolar septa as an index of parenchymal inflammation.
Analyses were performed using SPSS 18.0 (SPSS Inc., Chicago, IL). Results are presented as mean ± SD for mechanical parameters, W/D and BALF to serum albumin concentration ratio, and as median and range for cytokine and E-selectin concentration, and histologic parameters. The sample size was based on preliminary observations of the variables examined in previous studies15–17 (W/D, polimorphonuclear count, dynamic PL). Comparisons of pressure–volume curves, breathing patterns, and mechanical parameters during prolonged ventilation with various modes were performed using mixed-, between-, and within-group factorial ANOVA for repeated measurements. One-way ANOVA was used to detect between-group differences for final measures of respiratory mechanics, W/D and BALF to serum albumin concentration ratios. In all instances, two-tailed testing was performed. Analysis of nonparametric variables was made with the Kruskal–Wallis and Mann–Whitney test. When post hoc tests were performed, the Bonferroni correction was applied. The level for statistical significance was taken at P value 0.05 or less.
Breathing pattern (VT, TI, and TE) and lung mechanics (EL and RL) did not differ statistically among the various groups, during the initial period of spontaneous breathing (table 1). In 40 animals, VT, both absolute and per kg body weight, TI, TE, EL, and RL averaged 2.52 ± 0.25 ml, 6.2 ± 0.6 ml/kg, 0.23 ± 0.01 s, 0.44 ± 0.07 s, 1.89 ± 0.41 cm H2O/ml, and 0.111 ± 0.026 cm H2O s ml−1, respectively.
Prolonged and Final Ventilation Period
Lung dynamic elastance was similar in the SV, PPMV, and NPWBMV groups, but larger in the NPTOMV group by approximately 60% at all points in time (P = 0.001); it increased progressively in all groups by the same relative amount (P < 0.001), which averaged 17 ± 15% (fig. 2). In contrast, RL was constant and similar in all groups (fig. 2; P = 0.342, 0.538, and 0.591 for time, time group, and group, respectively). Ew was also time independent; it did not differ significantly between the PPMV and NPWBMV group (0.81 ± 0.27 and 0.72 ± 0.25 cm H2O/ml; P = 0.515), but was markedly larger in the NPTOMV group (1.29 ± 0.39 cm H2O/ml; P = 0.005). EELV was similar in all groups (P = 0.873), and time independent (P = 0.878), averaging 1.96 ± 0.28 ml and 1.99 ± 0.28 ml at the beginning and end of the ventilation period, respectively.
When respiratory system mechanics was reassessed during the final PPMV period (fig. 1, crossed area), no statistically significant differences in Est,L, Est,w, Rint,L, Rvisc,rs, and τvisc,rs were found among the four groups (table 2). Similarly, the inflation pressure–volume curves of the respiratory system, chest wall, and lungs of the various groups were almost superimposed (fig. 3).
W/D Ratio, Albumin BALF to Serum Ratio, and Cytokines
No difference in the W/D ratio and that between BALF and serum albumin concentration occurred among groups (P = 0.846 and P = 0.169, respectively), indicating that interstitial or alveolar edema was absent. The mean values of these variables were 4.24 ± 0.25% and 0.70 ± 0.48%, respectively.
Cytokine concentrations in serum and BALF and serum levels of E-selectin are shown in figure 3. Serum concentration of IL-1β, IL-6, and IL-10 in the SV, PPMV, NPWBMV, and NPTOMV groups was significantly higher than that of C group, whereas that of IL-1β and IL-10 was higher in the NPTOMV than SV, PPMV, and NPWBMV groups, in which it was similar. In BALF, no statistically significant changes from control levels were observed, except for IL-1β and IL-6 concentration in NPTOMV group, which were increased significantly. Serum E-selectin concentration was similar in C, SV, PPMV, and NPWBMV, but significantly higher in the NPTOMV group.
The results of histologic measurements are reported in table 3. None of the measured variables differed significantly among the various groups, except polimorphonuclear leukocyte count. Indeed, the number of polymorphonuclear leukocytes in the alveolar septa per unit length of the alveolar wall was similar in groups SV, PPMV, and NPWBMV (P = 0.318), lower in group C (P = 0.004 or less), and higher in group NPTOMV (P < 0.001) than in SV, PPMV, and NPWBMV groups. In all groups, no signs of focal alveolar collapse, edema, or hemorrhage were present.
This study in healthy rats has investigated the effects of different types of prolonged ventilation that occurred with the same VT, timing, and EELV of the anesthetized, spontaneously breathing animal, showing that: (1) compared with spontaneous breathing, PPMV causes neither lung damage nor inflammatory reaction; (2) NPWBMV is equivalent to PPMV; and (3) NPTOMV induces lung distortion and a mild inflammatory response.
No change in lung mechanical properties occurred because of prolonged SV, PPMV, or NPWBMV (figs. 2 and 3; table 2), though differences could have resulted from lung distortion due to different changes between spontaneous and positive pressure ventilation of the vertical gradient of PL,23 rib cage dimensions in anesthetized rabbits and dogs,24 and regional diaphragmatic displacement in anesthetized humans.25 On the contrary, NPTOMV should have caused a kind of cylindrical deformation of the lungs,26 as the rib cage expanded and the abdominal content shifted cranially with each inflation; indeed, EL increased immediately with NPTOMV, remained increased throughout the 4 h of mechanical ventilation (fig. 2), and returned to normal on removal of distortion (table 2). Similar distortion should also occur in dogs and humans, as can be inferred from the smaller pressure changes required to produce equal changes in EELV with positive end-expiratory pressure than grid and wrap devices or cuirass.27–29 Furthermore, lung distortion can induce cyclic closing and opening of small airways despite preserved EELV, with possible epithelial cell damage and cytokine release.4–6,30
In contrast with the findings of Mead and Collier31 in dogs and present results in rats (fig. 2), Moriondo et al.15 have reported that for the same VT, Pl swings are fivefold larger in mechanically ventilated than spontaneously breathing rats, implying a nearly seven times greater EL with PPMV. These results are difficult to explain. Such very large swings of PL were observed only in surfactant-depleted, open-chest rats and rabbits during PPMV at zero end-expiratory pressure, and were associated with markedly greater W/D and BALF to serum albumin concentration ratio, heavy signs of airway and parenchymal damage, and prominent inflammatory response.6,32
The effects produced by NPWBMV never differed from those with PPMV (table 3 and fig. 4), as predictable on the basis of the similar increases of PL and Pw, which also ensure similar cardiovascular responses.33 Indeed, the differences between negative and positive pressure ventilation reported by Grasso et al.19 were not reproduced by a recent study.34
PPMV and NPWBMV neither impaired the integrity of the alveolar–capillary barrier, because both the W/D and BALF to serum albumin concentration ratio were similar to those of the C and SV group, nor caused histologic evidence of airway, parenchymal, and vascular damage (table 3). Serum levels of IL-1β, IL-6, and IL-10 were, however, higher in the PPMV and NPWBMV than C group, but similar to those of the SV group (fig. 4), consistent with previous findings of similar tumor necrosis factor α, IL-6, IL-10, and keratinocyte-derived chemokine concentration in serum or lung homogenates of spontaneously breathing and mechanically ventilated rats.35 The same applies to recruitment of polymorphonuclear leukocytes in the alveolar septa (table 3), suggesting that inflammatory response during SV, PPMV, and NPWBMV is related to anesthesia, surgery, and/or animal’s instrumentation.36 Hence, the comparison of cytokines levels between control and mechanically ventilated animals can lead to an erroneous evaluation of the proinflammatory effects of mechanical ventilation. This may have happened in the studies on mice, where the SV group was lacking,16 or different doses of anesthetics had been administered.17 Furthermore, it seems unlikely that the discrepancy between these and present studies can be attributed to the somewhat longer period of mechanical ventilation used in mice,16,17 because increased expression of inflammatory cytokines in BALF macrophages of rats and increase of pulmonary and systemic cytokine levels in mice are already observed within 120 and 60 min from the onset of PPMV with noninjurious tidal volumes.14,16 At any rate, the kind of response to PPMV observed in mice vanished completely after 24 h.16
In contrast with PPMV and NPWBMV, NPTOMV elicited a significant increase of polymorphonuclear leukocyte count in the alveolar septa (table 3), as well as of IL-1β, IL-10, and E-selectin concentration in serum and IL-1β and IL-6 concentration in BALF relative to corresponding levels in the SV group (fig. 4). The presence of IL-1β and IL-6 in BALF was probably the consequence of local production in the airspaces, because their BALF and serum concentrations did not correlate (P > 0.70), consistent with the preserved integrity of the alveolar–capillary barrier. Cytokines release with NPTOMV cannot be attributed to anesthesia, surgery, animal’s instrumentation, and breathing pattern, because they were the same in all groups. A possible explanation can be provided by the presence with NPTOMV of lung distortion, leading to proinflammatory cytokine release via cellular mechanotransduction.10,11 This inflammatory response, however, had no apparent, short-term consequences, because none of the functional and morphological variables differed among all groups at the end of the experiments, (tables 2 and 3). However, it remains to be seen whether with a more protracted NPTOMV, the inflammatory reaction can eventually cause mechanical and/or histologic alterations.
Interestingly, serum E-selectin, an endothelial cell adhesion molecule mediating leukocyte rolling and phagocyte chemotaxis,37 was increased with NPTOMV only, and its levels did in fact correlate with the degree of lung distortion (P = 0.035) indexed by the ratio between EL at the onset of NPTOMV and the preceding SV period. This suggests that distortion could have activated the endothelial cells by changing their shape either directly or through alterations of the blood flow regime in their proximity.38,39
In conclusion, both prolonged positive and negative pressure ventilation performed in normal animals with the VT and timing of spontaneous, quiet breathing neither elicit a proinflammatory response nor cause functional and morphological alterations. In view of the adverse effects of conventional mechanical ventilation with large tidal volumes, this supports the notion of the presence of a critical volume threshold above which acute lung injury ensues. However, an inflammatory reaction with cytokines production and/or release can be induced by distortion of the pulmonary parenchyma, even in the absence of volotrauma in normal lungs, as it occurs with poncho-like devices. This could have important implications in the presence of pathological, heterogeneous alterations of lung mechanics, when distortion can be induced by conventional positive pressure ventilation in spite of the use of low tidal volumes.
Support was provided solely from institutional and/or departmental sources.
The authors declare no competing interests.