Does Iso-mechanical Power Lead to Iso-lung Damage?: An Experimental Study in a Porcine Model

All primary components of mechanical ventilation, namely tidal volume (V_T),^1–4  driving pressure,⁵  inspiratory flow,⁶  respiratory rate (RR),⁷  and positive end-expiratory pressure (PEEP),⁸  have been recognized as potential contributors to ventilator-induced lung injury. Taken together, they are components of a “summary variable”: the mechanical power.⁹  The computation of mechanical power is based on the motion equation¹⁰ : the total pressure present in the respiratory system at any given time is the sum of the pressure required to overcome the incremental elastic load (V_T · respiratory system elastance), the pressure required to move the gas through the respiratory system (flow · resistance), and the pressure stored in the respiratory system at end-expiration (PEEP). The sum of each of these pressures, multiplied by the V_T, equals the inflation energy spent during each breath. This inflation energy per cycle, multiplied by RR, determines the mechanical power, expressed in joules per minute (Supplemental Digital Content, https://links.lww.com/ALN/C261). While V_T and RR are unquestioned components of the mechanical power, more controversial has been the inclusion of PEEP in the mechanical power formula,¹¹  as no energy must be spent to maintain a given level of PEEP if the lung remains motionless. It must be pointed out, however, that to inflate the lung from a given level of PEEP requires an energy input relative to its relaxed baseline greater than the potential energy stored in the lung at that PEEP level. Indeed, the energy required equals the absolute pressure times the change in volume. In work specifically aimed to test PEEP as a component of mechanical power, we found a U-shaped relationship of PEEP to lung damage: harmful at zero end-expiratory pressure, harmless at 4 to 7 cm H₂O, and increasingly harmful over the range from 11 to 18 cm H₂O.⁸ 

The possible association between mechanical power and mortality/ventilator-induced lung injury has been found in critically ill patients receiving mechanical ventilation for at least 48 h¹²  and in patients without acute respiratory distress syndrome (ARDS),¹³  as well as in experimental animals with healthy¹⁴  or diseased lungs.¹⁵  Despite these promising signals of the usefulness of mechanical power as an indicator of damage risk, the relative weights of the mechanical power components in determining ventilator-induced lung injury has not been investigated. Indeed, the same value of mechanical power of inflation may be reached independently of the principal component variable being altered, V_T, RR, or PEEP. Yet within this framework of iso-mechanical power, it is unknown whether the overall lung damage inflicted by each component is similar, suggesting that the primary cause of ventilator-induced lung injury is the mechanical power per se, or whether ventilator-induced lung injury differs depending on the dominant component variable influencing power. The objective of this study was to answer this question by comparing in healthy piglets the individual effects of high V_T, high RR, and high PEEP, each delivered at two levels of mechanical power (15 and 30 J/min), on lung mechanics, hemodynamics, and gas exchange as well as on macroscopic and histologic lung anatomy.

The study, approved by local authorities (18/2795, LAVES, Oldenburg, Niedersachsen, Germany), included 42 healthy domestic piglets (body weight, mean ± SD, 24.2 ± 2.0 kg), which were maintained under total intravenous general anesthesia throughout the experiment (sufentanil, propofol, and midazolam). The animals were mechanically ventilated in volume-controlled mode, in prone position for 48 h at a constant inspired fraction of oxygen (0.40). They were instrumented with esophageal balloon, central venous, pulmonary artery, femoral artery, and urinary catheters. Crystalloids were infused as maintenance fluid, and norepinephrine and colloids were administered as needed to maintain a mean arterial pressure greater than 60 mmHg. Epinephrine was added to increase cardiac output if hemodynamic instability developed (for the details of experimental procedure see Supplemental Digital Content, https://links.lww.com/ALN/C261).

We studied healthy animals so that whatever damage occurred could be attributed uniquely to mechanical ventilation. Forty-two pigs were allocated among six groups (seven pigs each) defined by two levels of mechanical power (high, approximately 30 J/min, or low, approximately 15 J/min) and three ventilatory strategies, in which the dominant mechanical power component was V_T, RR, or PEEP. We decided to apply 15 and 30 J/min of mechanical power to confirm the results of a previous study, in which 25 ± 1.7 J/min over 48 h discriminated between lower and higher degrees of lung damage. In high V_T groups, we set V_T equal to functional residual capacity (FRC) times 2.5 (approximately 30 to 35 ml/kg). To deliver this V_T at 15 or 30 J/min, we arbitrarily chose to keep the PEEP constant at 5 cm H₂O and to vary the RR from 7 to 15 bpm. This applied V_T is known to produce lung damage in similar sized healthy pigs ventilated over extended periods.¹⁶  In the high RR group, we set the RR at 40 breaths/minute. To deliver this frequency at 15 or 30 J/min, we kept the PEEP constant at 5 cm H₂O, as the V_T was adjusted from 10 and 14 ml/kg. This frequency was arbitrarily chosen as a compromise to deliver potentially harmful RR, while minimizing the auto-PEEP. In the high PEEP groups, we set PEEP at 25 cm H₂O. We were obligated to use such a high level of PEEP so as to reach the targeted level of mechanical power, which increases only linearly with PEEP, instead of exponentially, as with V_T and RR.⁹  To deliver a mechanical power of 15 and 30 J/min in this group while maintaining a nonharmful V_T in healthy lungs, we maintained 13 ml/kg of V_T and varied the frequency from 11 to 21 bpm. The ventilatory variables and the settings of the injurious ventilation are presented in figure 1 and detailed in table 1.

The piglets were randomly allocated to one of the six groups and ventilated according to the predefined respiratory settings (V_T, RR, and PEEP) for 48 h. Every 6 h we measured ventilatory, hemodynamic, and gas exchange variables. We measured FRC every 12 h via multiple breath helium dilution in a closed system.¹⁷  The animals were euthanized at the end of the experiment by injection of pentobarbital and potassium chloride. Autopsy was performed and lung weight and wet-to-dry ratios measured. Coded samples for histology were taken from apical, middle, and basal lung sectors, both in dependent and nondependent lung regions. The experimental procedures for sampling and analysis are detailed in the Supplemental Digital Content text and figure 1 (https://links.lww.com/ALN/C261).

No a priori statistical power calculation was conducted; the sample size was based on our experience with previous experiments. For the sake of clarity, data are presented classifying the six groups either according to the mechanical power level, regardless of the ventilatory strategy, or according to the ventilatory strategy (high V_Tvs. high RR vs. high PEEP), regardless of the mechanical power (30 J/min = higher power vs. 15 J/min = lower power; hereafter we will refer to these two levels as high and low, respectively). Data are presented as mean and SD; No. refers to the number of animals. We tested the normal distribution of the variables with the Shapiro–Wilk test. To evaluate differences in continuous variables at a given time point, we used a two-sample independent t test for mechanical power level comparisons and independent sample one-way ANOVA for ventilatory strategy comparisons, Fisher’s exact test was used to compare categorical variables. To evaluate the different time courses of mechanics, hemodynamics, and gas exchange variables among groups (repeated measures, excluding baseline values), we used a linear mixed model: the fixed effects are the ventilatory strategy (high V_T, high RR, high PEEP groups), the mechanical power (high and low groups), and the interactions between each of them and time, while the animals are the random effect.This model ignores the missing data, but produces outputs with the same numbers of observations compared with the original dataset. This model allowed detection if the time course of the given variable was significantly different among the power levels and ventilatory strategies. A linear regression was used to follow the changes of a given variable within a ventilatory strategy with time. Post hoc comparisons were conducted using Tukey’s test. A P value less than 0.05 was considered statistically significant (two-tailed testing). Outliers were not excluded from the analyses. The end-experiment variables were the ones measured after 48 h (34 pigs) and the last measured before death in the pigs that died during the experimental procedure. Analyses were performed with R software version 3.5.2 (R Foundation for Statistical Computing, Austria, libraries Rcmdr, nlme, emmeans).

Mechanical, hemodynamic, and gas exchange variables differed strikingly among the three ventilatory strategies, primarily due to the setting applied. Their time courses throughout the experiment are reported in detail in the Supplemental Digital Content, figures 2 to 25 (https://links.lww.com/ALN/C261).

Altered mechanics were characterized by the obviously sudden increases of plateau pressure (from 14.0 ± 1.4 to 32.8 ± 4.5 cm H₂O), driving pressure (8.9 ± 1.4 to 27.6 ± 4.9 cm H₂O), and lung stress (6.8 ± 2.7 to 17.2 ± 4.7 cm H₂O). With time, all these variables worsened further. Indeed, the plateau pressure rose up to 37.8 ± 12.9 cm H₂O (P = 0.025; Supplemental Digital Content fig. 2, https://links.lww.com/ALN/C261), the driving pressure increased up to 32.5 ± 12.6 cm H₂O (P = 0.037; Supplemental Digital Content fig. 3, https://links.lww.com/ALN/C261), and the lung stress reached 22.2 ± 9.8 cm H₂O (P = 0.003; Supplemental Digital Content fig. 4, https://links.lww.com/ALN/C261). The baseline lung elastance was 19.4 ± 6.7 cm H₂O/l and rose to 24.9 ± 11.5 cm H₂O/l over the 48-h period (P = 0.002; Supplemental Digital Content fig. 5, https://links.lww.com/ALN/C261). Of note, all of these variables were worse when ventilated at 30 J/min compared to 15 J/min, although they did not reach statistical significance (Supplemental Digital Content figs. 2 to 5, https://links.lww.com/ALN/C261). The strain immediately rose from 0.9 ± 0.2 to 2.5 ± 0.3 and increased over time up to 3.4 ± 1.0 (P < 0.001; Supplemental Digital Content fig. 6, https://links.lww.com/ALN/C261). Interestingly, the specific lung elastance remained nearly constant throughout the experiment from 7.2 ± 2.4 cm H₂O to 6.5 ± 1.7 cm H₂O (P = 0.839; Supplemental Digital Content fig. 7, https://links.lww.com/ALN/C261). The FRC steadily decreased over time from 385 ± 69 to 294 ± 75 ml (P = 0.001; Supplemental Digital Content fig. 8, https://links.lww.com/ALN/C261).

The cardiac output significantly decreased from 4.4 ± 0.8 to 3.3 ± 0.8 l/min (P = 0.001; Supplemental Digital Content fig. 12, https://links.lww.com/ALN/C261). The mean arterial pressure slightly but significantly decreased (from 76 ± 8 to 61 ± 9 mmHg, P = 0.001; Supplemental Digital Content fig. 13, https://links.lww.com/ALN/C261), pulmonary artery (Supplemental Digital Content fig. 14, https://links.lww.com/ALN/C261) and wedge pressures (Supplemental Digital Content fig. 15, https://links.lww.com/ALN/C261) remained unmodified, and central venous pressures slightly but significantly increased (Supplemental Digital Content fig. 16, https://links.lww.com/ALN/C261). No statistical differences were found in the time course of these variables, whether delivered either at 30 or at 15 J/min. The end-experiment fluid balance, however, was approximately 57% higher in the high-power group compared to the low-power group (8.2 ± 2.5 vs. 5.3 ± 2.8 l; Supplemental Digital Content fig. 17, https://links.lww.com/ALN/C261; P < 0.001). The difference of norepinephrine infusion between the high and low groups did not reach the statistical significance (Supplemental Digital Content fig. 18, https://links.lww.com/ALN/C261). This hemodynamic setting and this fluid strategy were associated with a linear increase of extravascular lung water from 289 ± 45 to 441 ± 112 ml (P < 0.001; Supplemental Digital Content fig. 19, https://links.lww.com/ALN/C261).

At the setting of the injurious ventilation, Pao₂ (Supplemental Digital Content fig. 21, https://links.lww.com/ALN/C261) remained high, while the partial pressure of CO₂ rapidly decreased to about 25 mmHg in the low-power group and about 15 mmHg in the high-power group (Supplemental Digital Content fig. 22, https://links.lww.com/ALN/C261) with proportional rises in pH (Supplemental Digital Content fig. 23, https://links.lww.com/ALN/C261). The physiologic dead space fraction significantly decreased from 50 to 30% (P < 0.001; Supplemental Digital Content fig. 24, https://links.lww.com/ALN/C261). Over time, the Paco₂ remained constant (P = 0.420), while the pH tended to normalize (P < 0.001). The Pao₂ was always greater than 200 mmHg, but slightly decreased over time (P = 0.005). The shunt fraction held near constant throughout the experiment at approximately 5% (Supplemental Digital Content fig. 25, https://links.lww.com/ALN/C261).

After the application of the experimental ventilatory settings, the baseline plateau pressure (14.9 ± 1.4 cm H₂O), driving pressure (9.7 ± 1.2 cm H₂O), and lung stress (7.2 ± 2.1 cm H₂O) values immediately increased slightly, due to the appearance of auto-PEEP (up to 19.6 ± 3.4 cm H₂O, 11.3 ± 2.7 cm H₂O, and 7.3 ± 1.7 cm H₂O, respectively). Throughout the following 48 h, each of these variables sharply increased (P < 0.001 for all): plateau pressure up to 27.7 ± 7.0 cm H₂O (Supplemental Digital Content fig. 2, https://links.lww.com/ALN/C261), driving pressure up to 17.9 ± 5.0 cm H₂O (Supplemental Digital Content fig. 3, https://links.lww.com/ALN/C261), and lung stress up to 14.0 ± 4.7 cm H₂O (Supplemental Digital Content fig. 4, https://links.lww.com/ALN/C261). Lung elastance doubled over time from 22.2 ± 5.7 to 42.2 ± 18.7 cm H₂O (P < 0.001; Supplemental Digital Content fig. 5, https://links.lww.com/ALN/C261). The lung strain underwent a modest but significant increase from 1.0 ± 0.1 to 1.6 ± 1.0 after 48 h (P = 0.002; Supplemental Digital Content fig. 6, https://links.lww.com/ALN/C261). The specific lung elastance sharply increased after 24 h from a baseline value of 7.7 ± 2.6 up to 10.1 ± 3.9 cm H₂O (P < 0.001; Supplemental Digital Content fig. 7, https://links.lww.com/ALN/C261). The FRC significantly decreased with time (P < 0.001), similar to what was observed with the high V_T groups (Supplemental Digital Content fig. 8, https://links.lww.com/ALN/C261). Each of these variables behaved similarly at 15 and 30 J/min, with the exception of plateau pressure, which was higher in the high-power high RR group.

The cardiac output slightly decreased from 3.8 ± 0.7 to 3.2 ± 0.7 l/min (P = 0.054; Supplemental Digital Content fig. 12, https://links.lww.com/ALN/C261), as well as the mean arterial pressure (P < 0.001; Supplemental Digital Content fig. 13, https://links.lww.com/ALN/C261). In contrast, pulmonary artery (Supplemental Digital Content fig. 14, https://links.lww.com/ALN/C261), wedge (Supplemental Digital Content fig. 15, https://links.lww.com/ALN/C261), and central venous pressures (Supplemental Digital Content fig. 16, https://links.lww.com/ALN/C261) slightly but significantly increased (P = 0.008, 0.041, and 0.002 respectively). The fluid balance and norepinephrine infusion rate were similar to the high V_T groups (Supplemental Digital Content figs. 17 and 18, https://links.lww.com/ALN/C261). The extravascular lung water increased at lower pace compared to the high V_T groups, but after 48 h reached similar values (rising from 264 ± 35 to 423 ± 184 ml, P < 0.001; Supplemental Digital Content fig. 19, https://links.lww.com/ALN/C261).

Pao₂ (Supplemental Digital Content fig. 21, https://links.lww.com/ALN/C261) remained constant throughout the experiment (P = 0.938), while arterial partial pressure of CO₂ (Supplemental Digital Content fig. 22, https://links.lww.com/ALN/C261) and pH (Supplemental Digital Content fig. 23, https://links.lww.com/ALN/C261) decreased, similar to the high V_T groups. The dead space fraction was significantly higher throughout the experiment than in high V_T groups (48.1 ± 7.6 vs. 31.8 ± 11.7%; Supplemental Digital Content fig. 24, https://links.lww.com/ALN/C261), while the shunt fraction (Supplemental Digital Content fig. 25, https://links.lww.com/ALN/C261) remained similarly constant at 5%.

After applying 25 cm H₂O of PEEP, the plateau pressure sharply and immediately increased from 15.4 ± 2.4 up to 52.3 ± 5.4 cm H₂O, and the driving pressure rose from 10.2 ± 2.6 to 28.3 ± 4.9 cm H₂O, while the lung stress increased from 7.8 ± 2.8 to 32.1 ± 7.7 cm H₂O. No significant differences in these variables were found between higher- and lower-power categories. In contrast to the other two strategies, however, plateau pressure (Supplemental Digital Content fig. 2, https://links.lww.com/ALN/C261), driving pressure (Supplemental Digital Content fig. 3, https://links.lww.com/ALN/C261), and lung stress (Supplemental Digital Content fig. 4, https://links.lww.com/ALN/C261) progressively decreased until the end of the experiment, down to 44.8 ± 3.0, 20.3 ± 2.9, and 22.9 ± 5.7 cm H₂O, respectively (P < 0.001 for all). The lung elastance (Supplemental Digital Content fig. 5, https://links.lww.com/ALN/C261) immediately increased after the application of the high PEEP setting from 24.3 ± 8.2 to 64.4 ± 18.8 cm H₂O but later decreased down to 38.6 ± 14.1 cm H₂O (P < 0.001). The lung strain rose immediately from 1.0 ± 0.2 to 3.8 ± 0.6 after PEEP application, increased to a maximum of 5.7 ± 1.4 after 30 h, and stabilized to 5.2 ± 1.6 near the end of the experiment (Supplemental Digital Content fig. 6, https://links.lww.com/ALN/C261). The baseline specific lung elastance (7.8 ± 2.4 cm H₂O) after 24 h sharply decreased compared to the other ventilatory strategies, reaching a minimum of 4.5 ± 0.9 cm H₂O by the end of the experiment (P < 0.001; Supplemental Digital Content fig. 7, https://links.lww.com/ALN/C261). The FRC decreased earlier than in the other ventilatory strategies due to the low FRC recorded in the three pigs that developed pneumothorax; nonetheless, its values at the end of the experiment were similar to those of the other groups (from 351 ± 35 to 292 ± 67 ml; Supplemental Digital Content fig. 8, https://links.lww.com/ALN/C261).

The hemodynamic pattern in this group was sharply different from the two other ventilatory strategies: indeed, in pigs exposed to high mechanical power, the cardiac output remained remarkably constant from 3.9 ± 0.6 to 3.9 ± 1.0 l, whereas it rose from 4.1 ± 0.5 to 5.8 ± 1.9 l in pigs ventilated at lower mechanical power (P < 0.001; Supplemental Digital Content fig. 12, https://links.lww.com/ALN/C261). The average arterial pressure (Supplemental Digital Content fig. 13, https://links.lww.com/ALN/C261) was slightly but significantly lower in this group compared to the other ventilatory strategies (67 ± 9 mmHg vs. 73 ± 11 in high V_T strategy, P = 0.017, and 73 ± 11 mmHg in high RR strategy, P = 0.013). In contrast, pulmonary artery (Supplemental Digital Content fig. 14, https://links.lww.com/ALN/C261), wedge (Supplemental Digital Content fig. 15, https://links.lww.com/ALN/C261), and central venous pressures (Supplemental Digital Content fig. 16, https://links.lww.com/ALN/C261) were significantly 54%, 60%, and 60% higher than in the other two groups (P < 0.001). The fluid balance associated with this hemodynamic pattern was significantly greater than with the other two strategies (10.5 ± 1.6 l vs. 6.2 ± 1.2 l in the high V_T groups and 6.4 ± 1.5 l in the high RR groups, P < 0.001), but not statistically different between higher and lower power (Supplemental Digital Content fig. 17, https://links.lww.com/ALN/C261). In contrast, norepinephrine was given in significantly greater amounts only in the group with high PEEP delivered at low power (Supplemental Digital Content fig. 18, https://links.lww.com/ALN/C261). Epinephrine was needed in all the animals of this high PEEP group (Supplemental Digital Content fig. 20, https://links.lww.com/ALN/C261). The extravascular lung water increased similarly to the other two types of power-altering strategies (from 307 ± 55 to 415 ± 65 ml, P < 0.001; Supplemental Digital Content fig. 19, https://links.lww.com/ALN/C261).

Due to the ventilatory protocol, the partial pressure of CO₂ (Supplemental Digital Content fig. 22, https://links.lww.com/ALN/C261) was higher and the pH was lower (Supplemental Digital Content fig. 23, https://links.lww.com/ALN/C261) in this high PEEP strategy, particularly in the low-power group, where 25 cm H₂O of PEEP was associated with 0.32 l V_T at a RR of 11/min versus 0.32 l at a RR of 22/min. The Pao₂ was significantly lower in the low-power group compared to all the other groups, independent of strategy (P = 0.001; Supplemental Digital Content fig. 21, https://links.lww.com/ALN/C261). The dead space was significantly higher in the high-power group compared to the low-power group (P = 0.002; Supplemental Digital Content fig. 24, https://links.lww.com/ALN/C261). In general, dead space fraction was similar to that measured in high RR groups and significantly higher than measured in the high V_T groups. The average shunt fraction (Supplemental Digital Content fig. 25, https://links.lww.com/ALN/C261) was significantly higher in the high PEEP, low-power group than high PEEP, high-power group (8.1 ± 5.3 vs. 3.7 ± 2.1, P = 0.002).

No significant differences were found in lung weight among the three ventilatory strategies delivered at iso-mechanical power (fig. 2, right panel), averaging 26.6 ± 6.3, 23.7 ± 5.7, and 26.3 ± 4.9 g/kg in the high V_T, RR, and PEEP groups, respectively (P = 0.341). Similarly, the end-experiment lung weights were similar in pigs treated with high or low mechanical power, regardless of the ventilatory strategy, averaging 26.1 ± 6.3 and 24.9 ± 5.1 g/kg of body weight, respectively (P = 0.508; fig. 2, left panel). In table 2 we report lung, liver, bowel, kidney, and muscle wet-to-dry ratio, body weight gain, end-experiment fluid balance, pneumothorax, and death according to mechanical power level and ventilatory strategy. As shown, the wet-to-dry ratios of lungs, liver, bowel, and muscle were similar across the three ventilatory strategies, with the exception of the kidney wet-to-dry ratio, which was significantly greater in the high PEEP groups. No significant differences were found between high- and low-power groups. Notably, the animals treated with high PEEP had a body weight gain and cumulative fluid balance nearly double compared to the other groups. The composite adverse event markers (pneumothorax and death) were significantly worse in the high PEEP groups than in those of the other ventilatory strategies (Fisher exact test P = 0.045). In figure 3, upper panel, we summarize the histologic findings according to lung mechanical power: overall, no differences were found between mechanical power levels, with the exception of alveolar edema, which was significantly greater in the lower power groups. Similarly, no significant histologic differences were found among the three ventilatory strategies (fig. 3, lower panel). In figure 4, we report a representative histologic pattern we found with each ventilatory strategy delivered at different mechanical power. Histologic patterns that ranged from near-normal to near complete destruction of the lung structure coexisted in the same lung, regardless of the mechanical power or the ventilatory strategy (Supplemental Digital Content fig. 26, https://links.lww.com/ALN/C261).

The main findings of this paper were as follows: (1) different ventilatory strategies produced different time courses of mechanical, hemodynamic, and gas exchange variables, partly influenced by the mechanical power level; (2) at iso-mechanical power, the anatomical lung damage caused by high V_T, high RR, or high PEEP interventions were indistinguishable.

The application of high V_T immediately elevated stress and strain, proportional to the level of mechanical power. Alterations of lung structure and inflammatory reaction accounted for the increase in extravascular lung water and decreased FRC. Interestingly, the specific lung elastance remained constant in this group, as found in patients with ARDS, where the “baby lung” maintains relatively normal intrinsic elastic properties.¹⁸  In summary, the high V_T group showed the typical time course of ventilator-induced lung injury.

The ventilatory setting in this group was a combination of RR, V_T, and PEEP, which may be encountered in ARDS patients. The only immediate change of mechanical properties was attributable to an increase of 3 to 5 cm H₂O of auto-PEEP. All mechanical variables, however, sharply deteriorated thereafter, ending up with extravascular lung water and lung weight similar to that observed in the high V_T groups. After 24 h of treatment, the increase of specific lung elastance suggested a stiffening of the ventilatable lung, possibly due to the increased rate of stress and strain application and/or impaired surfactant functioning.^19,20 

The application of high PEEP with two levels of mechanical power induced tremendous stress and strain. All pressures and lung elastance decreased with time, as did specific lung elastance, suggesting a progressive relaxation of the lung structural framework. Relaxation was also indicated by the increased strain, despite decreased stress.

In summary, the mechanics of the high V_T groups were characterized by decreased lung volume with normal intrinsic elastic properties; the high RR groups by decreased lung volume with stiffening of the lung structure; and high PEEP groups by decreased lung volume with structural relaxation of the lung. Whatever the ventilatory strategy, the trigger of ventilator-induced lung injury must be the “fracture” of molecular bonds in the extracellular matrix in excess of repair capability^21–23  with consequent inflammatory edema.²⁴  We may then wonder how such different ventilatory strategies might lead to the structural changes of the extracellular matrix. No doubt that extremely high V_T, as used in this work, may be accompanied by sufficient energy to induce rupture, particularly at the interface of lung regions with different elasticity (stress risers).^25–27  The injuring mechanism of low-level repeated stresses at high frequency might arise from a sort of progressive structural fatigue, in which cumulative stress may result in molecular breakdown of matrix elements. High PEEP applies an increased baseline stress to the extracellular matrix. A further stress added by the V_T may require energy sufficient to fracture the molecular bonds of the polymers in the matrix. Therefore, although the mechanism by which a critical energy over time is delivered to the extracellular matrix may differ among the ventilatory strategies we tested, it is not entirely surprising that a similar amount of energy would eventually lead to similar lung damage. Actually, the histologic findings, which documented the regional heterogeneity of damage, were indistinguishable within the ventilatory strategies.

The functional/structural alterations of the extracellular matrix unavoidably lead to alterations of the capillary/alveolar barrier anchored to it. Indeed, edema is the most widely used variable to quantify ventilator-induced lung injury. We must consider, however, that for edema to develop, structural/functional alteration of capillary/alveolar barrier alone does not act in isolation; hemodynamics may be as crucial as the structural lung damage itself. At the two extremes, edema may develop without any anatomical alterations simply due to hydrodynamics,²⁸  while little or no edema would develop despite structural alteration if the capillary flow were reduced to near zero, as we have shown in experiments of regional perfusion block.²⁹  Accordingly, the ventilator-induced lung injury estimated by lung weight cannot be precisely interpreted without simultaneously considering the hemodynamic pattern. In this study we found a similar lung weight when the three ventilatory strategies were delivered at higher or lower levels of mechanical power. This unexpected finding indicates at first sight that 15 J/min are equally as damaging as 30 J/min. Alternatively, it is possible that more severe structural lung lesions induced by 30 J/min are associated with hemodynamic patterns less favorable to edema formation. The higher wedge pressure recorded at higher mechanical power (favoring edema formation, Supplemental Digital Content fig. 15, https://links.lww.com/ALN/C261) was associated with higher mean airway pressure (impeding edema formation, Supplemental Digital Content fig. 11, https://links.lww.com/ALN/C261). This combination may account for the greater extent of edema found in low-power groups, characterized by lower mean airway pressure. Also, the differences between this study and our previous one on mechanical power⁸  may be discussed in the framework lung structure/hemodynamic interaction. In that earlier study, despite higher mechanical power (26.6 ± 12.1 vs. 21.8 ± 8.3 J/min, P = 0.049), the lung weight was lower (15.6 ± 5.2 vs. 25.5 ± 5.7 g/kg, P < 0.001). In the current study, however, increased fluid administration was needed to keep the animals alive (10.1 ± 2.3 vs. 4.2 ± 0.9 l). Several factors may account for this, primarily a different study design. Indeed, by protocol, in this study we applied significantly greater stress and strain. Moreover, due to a change of institutional policy for animal management, the average dose of propofol was increased in the current experiment (10.7 ± 2 vs. 8.2 ± 3.3 mg · kg^-1 · h^-1P < 0.001). Summarizing, while 15 and 30 J/min of mechanical power might induce different degrees of structural lung damage, the end result of structure/hemodynamic interactions occurring during management is ultimately similar. Therefore, 15 J/min of mechanical power is still far from being a safe “threshold” for ventilator-induced lung injury in this model.

Of particular interest regarding the lung structure/hemodynamic interaction is the ventilator-induced lung injury induced by high PEEP. In our study, the end-inspiratory lung volumes exceeded baseline total lung capacity (3 × FRC), potentially overcoming the physical limits of the lung’s supporting architecture. Simultaneously, the animals of the high PEEP groups presented the most severe hemodynamic impairment and needed the greatest amount of fluid and cardiovascular support. The combination of higher fluid delivery and higher norepinephrine and epinephrine dosing led in this group to a hemodynamic pattern sharply different from the other two, characterized by higher cardiac output, lower systemic vascular resistances, and higher pulmonary artery and wedge pressures. It is likely that this distortion of normal hemodynamic profile accounts in part for the observed lung edema, extrapulmonary organ alterations (e.g., higher kidney wet-to-dry ratio) and worsened outcomes observed in this group of animals.

Apart from the concerns already expressed, we acknowledge that our model involved initially healthy pigs managed in their natural prone position. Our findings may well have differed if ventilation were applied to animals with preexisting acute lung injury or to those managed in the supine position. We believe that the greatest limitation, however, is that our study cannot elucidate two aspects of mechanical power, which are likely of paramount importance, i.e., its distribution in time and space. At the same mean power, its damaging intensity could be concentrated at different times of the respiratory cycle³⁰  and, in injured lungs, at the interface of lung units with different elastic properties (stress risers).^22,31  Moreover, as our findings suggest, using one high iso-power strategy might initiate injury of a different kind or at a different pace than another, but over extended time, their end results may be indistinguishable.

Conclusions drawn from experimental work designed to investigate mechanisms do not translate directly to the clinical scenario, as the interventions applied to prove or disprove our hypothesis were deliberately exaggerated. Few thoughtful physicians would routinely apply a V_T twice the FRC, 25 cm H₂O of PEEP, or a RR of 40 breaths per minute—especially not to nearly normal lungs. However, our results do provide key messages of clinical interest and potential impact: (1) that each major component of the ventilatory cycle may contribute impressively to lung damage; (2) that hemodynamics play a key role in the manifestations of ventilator-induced lung injury; and (3) that RR may strongly influence injury.

When discussing each of the variables separately, the unavoidable risk is to obscure the most important finding of the study: ventilator-induced lung injury primarily arises from a combination of variables and not from just one. Indeed, if we were to consider the single variables in isolation, from this study we should conclude that V_T values of 35 ml/kg and 10 ml/kg are equally dangerous, or (if one ignores the striking role of RR) that a driving pressure of 27 cm H₂O produces the same effects as a driving pressure of 10 cm H₂O. Our data strongly suggest that what really matters is the context in which the single variable operates, that all the variables must be considered together, and that their combined risk might be best quantified as mechanical power.

This work was made possible by the generous donation of Ilse Liselotte Munz to the Department of Anesthesiology of Göttingen. The authors thank Tim Behnemann, M.D.,and Julia Niewenhuys, M.D., from Department of Anesthesiology, Emergency and Intensive Care Medicine, University of Medicine Göttingen, Niedersachsen, Germany, for their support during the experiments.

Support was provided solely from institutional and/or departmental sources.

The authors declare no competing interests.