Mechanical Power and Development of Ventilator-induced Lung Injury

VENTILATOR-INDUCED lung injury (VILI) has been described in various clinical¹  and experimental settings,²  and several factors have been considered as possible triggers for VILI. Tidal volume (TV) and plateau pressure²  have been studied most as surrogates of strain and stress, respectively.³  Strain is the TV related to the lung size capable of being ventilated. Stress is the pressure that develops in the lung structure when a force is applied and is equal to the transpulmonary pressure applied to the lung. Other factors such as temperature,⁴  respiratory rate,^5–7  and flow^8,9  have been described as cofactors for VILI. All these factors together generate the energy applied to the respiratory system, which, expressed per minute, is the mechanical power. Mechanical power acts directly on the lung skeleton, that is, the extracellular matrix, deforming the epithelial and endothelial cells anchored to it.¹⁰  Depending on the amount of mechanical power, the alterations to the lung parenchyma may range from mechanical rupture^11,12  to an inflammatory reaction due to activation of macrophages,¹³  neutrophils,¹⁴  and endothelial and epithelial cells.^15,16 

In this study, we varied the power applied to the respiratory system by changing the respiratory rate while keeping the TV (strain) and transpulmonary pressure (stress) constant, which we knew was lethal at 15 breaths/min.^17–19  Our aim was to identify a power threshold for VILI. In a further series of experiments, we tested the power threshold hypothesis by applying power above and below the threshold and using computed tomography (CT) scans to measure VILI.¹⁹ 

We studied 24 healthy female piglets (21 ± 2 kg), using 15 piglets to investigate the power threshold for VILI and 9 for confirmatory experiments. This study was approved by the Italian Board of Health (Rome, Italy) and complied with international recommendations.²⁰  The piglets were instrumented with a tracheal tube, esophageal balloon, central venous catheter, arterial line, and urinary catheter (see Supplemental Digital Content 1, https://links.lww.com/ALN/B259, for details).

Five groups of piglets (three piglets per each group) were ventilated in prone-Trendelenburg position²¹  at the respiratory rates of 3, 6, 9, 12, or 15 breaths/min. Other respiratory variables were identical in all groups: namely, the inspiratory oxygen fraction (50%), the TV (38 ml/kg—strain greater than 2.5,¹⁷  which had been previously shown to be lethal in 54 h^17–19 ), the inspiratory:expiratory ratio (1:2), and end-expiratory pressure at 0 cm H₂O. This ventilatory setting generates inspiratory flows ranging from a minimum of 0.11 ± 0.02 l/s in piglets ventilated at 3 breaths/min to a maximum of 0.57 ± 0.08 l/s in piglets ventilated at 15 breaths/min. In these experiments, we did not attempt to control arterial carbon dioxide tension or pH because we had previously shown that these did not change the volume/strain threshold for VILI in this model.¹⁷ 

After computing the possible mechanical power threshold for VILI (see Supplemental Digital Content 1, https://links.lww.com/ALN/B259), we performed a second series of nine experiments with mechanical power below (four piglets) and above (five piglets) the threshold. In order to exclude any main effect of the respiratory rate on VILI development, we set the respiratory rate at 35 breaths/min in all the confirmatory experiments. In five of these confirmatory experiments (three above and two below the mechanical power threshold), we kept the Paco₂ in the physiological range by using an external supply. The four piglets treated with mechanical power below the threshold (8 ± 2 J/min) were ventilated for 54 h with Fio₂ 0.5, TV 11 ± 0.22 ml/kg, and respiratory rate 35 breaths/min. The five piglets treated with mechanical power above the threshold (22 ± 5 J/min) were ventilated with Fio₂ 0.5, TV 22 ± 4 ml/kg, and respiratory rate 35 breaths/min. Neither 11 nor 22 ml/kg TV induced lethal VILI when delivered at a respiratory rate of 15 breaths/min for 54 h.¹⁷

At baseline and every 6 h or less, if changes occurred (e.g., increase in peak/plateau pressure despite tracheal suctioning, decrease in peripheral saturation, and unexpected arterial hypotension or hypertension), we took two CT scans (end-expiration and end-inspiration pause), acquired dynamic and static pressure–volume curves of the respiratory system, and collected a complete set of physiological variables. A static pressure–volume curve was plotted in 100-ml steps (inflation and deflation) with a supersyringe, and a dynamic curve was acquired during tidal ventilation.

The study ended after 54 h of mechanical ventilation or if whole-lung edema developed, defined as densities occupying all CT scan lung fields. Piglets were euthanized with a KCl bolus and autopsied, and samples for lung pathology were collected.

Measured/computed variables were as follows: respiratory variables (peak, plateau, end-expiratory airway, and esophageal pressures; TV; and chest wall and lung elastances), gas-exchange variables (arterial and central venous Po₂, Pco₂, and pH), lactates, hemoglobin concentration and saturation, and hemodynamics (arterial and central venous pressure, urinary output, fluid intake, vasoactive drugs, and fluid balance) (see Supplemental Digital Content 1, https://links.lww.com/ALN/B259, for details).

We measured the following during the end-inspiratory pause (5-s duration):

• Peak pressure: highest pressure recorded

• P1: pressure recorded when inspiratory flow reached zero

• P2: pressure recorded at the end of the inspiratory pause

Stress relaxation is the observed decrease in stress in response to the same amount of strain generated in the structure and was defined as the difference between P1 and P2 (cm H₂O).

CT scan variables were as follows: lung weight; lung gas volume; overinflated, normally inflated, poorly inflated, and noninflated lung tissue; and lung recruitment (percentage of lung tissue that regains inflation between end-expiration and end-inspiration pause).^22–24  Lung inhomogeneities were computed as previously described.²⁵  The lung was divided into six fields (apex–hilum–base and dependent/nondependent), and we visually classified the CT scan damage as follows¹⁹ : grade 0, the baseline scan; grade 1, regions of density clearly distinguishable from the remaining parenchyma; grade 2, density occupying at least one lung field; and grade 3, density occupying all six fields (see Supplemental Digital Content 1, https://links.lww.com/ALN/B259, for details).

The data were analyzed in terms of power delivered to the respiratory system, that is, the amount of energy delivered to the respiratory system by the ventilator in the time unit (J/min). The delivered energy per breath (airways + lung) was defined as the area between the inspiratory limb of the Δ-transpulmonary pressure (x)–volume curve and the volume axis (y) and was measured in joule (see Supplemental Digital Content 1, https://links.lww.com/ALN/B259, for computation of Δ-transpulmonary pressure–volume curve).

Sample size was determined on the basis of previous experience. We estimated a power threshold using the data from a previously published study.¹⁷  We estimated a lethal TV threshold of 750 ml (approximately 34 ml/kg) when delivered at a respiratory rate of 15 breaths/min. As we set a TV of approximately 38 ml/kg, similar to the one used in a previous study,¹⁹  we expected that piglets ventilated at respiratory rates of 12 and 15 breaths/min would develop VILI but that those ventilated at respiratory rates of 3 and 6 breaths/min would not. We were unable to make any prediction about the piglets ventilated at 9 breaths/min, but we estimated that—if the study hypothesis was confirmed—with at least six piglets treated below the threshold and six treated above the threshold, it would be sufficient to demonstrate changes in most of the CT scan and physiological variables.¹⁷ 

The relationship between mechanical power and severity of CT scan damage was assessed with the Spearman correlation. The effect of respiratory rate on physiological and CT scan variables at baseline was assessed with linear regression. In both cases, mechanical power was used as a continuous variable, CT scan damage was assessed as a categorical variable, and changes in oxygenation, lung elastance, and total lung weight were assessed as continuous variables. Changes in physiological variables between the start and the end of the confirmatory experiments in piglets ventilated with mechanical power above and below the threshold were compared using an unpaired t test. A P value of less than 0.05 was considered to indicate statistical significance. All tests were two tailed. Statistical analysis was done using R software.²⁶ 

The baseline physiological and CT scan variables at 3, 6, 9, 12, and 15 breaths/min are summarized in table 1. Owing to the different total ventilation, there were differences only in Paco₂ and pH, whereas oxygenation, mechanics, and CT scan-related variables were similar at all respiratory rates. pH was acidotic at the respiratory rate 3 breaths/min and alkalotic at 6, 9, 12, and 15 breaths/min. The transpulmonary mechanical power increased with the respiratory rate (mechanical power = delivered energy per breath × respiratory rate). Whereas the respiratory rate rose five-fold from 3 to 15 breaths/min, the mechanical power increased approximately 11 times from 2 ± 0.2 to 22 ± 2 J/min due to the effect of flow (table 2).

The effects of the mechanical power on VILI are presented in figure 1A, which shows the severity of lung damage in relation to the mechanical power. Up to a mechanical power of approximately 12 J/min, CT scans showed mostly isolated densities, whereas above approximately 12 J/min, all piglets developed whole-lung edema. Figure 1B illustrates the nine confirmatory experiments performed below and above the VILI threshold. Piglets with normocapnia developed VILI when ventilated above the threshold. Table 3 summarizes the differences in the main physiological and CT scan features at the beginning and at the end of the study. Piglets ventilated with a mechanical power above the threshold had a larger increase in lung weight at the end of the study (412 ± 55 vs. 20 ± 47 g, P < 0.0001), greater lung elastance (59 ± 7 vs. 16 ± 9 cm H₂O/l, P < 0.001), and larger drop in oxygenation (−332 ± 59 vs. −20 ± 81 mmHg, P < 0.001).

Higher power at the beginning of mechanical ventilation was associated with increases in lung weight (r² = 0.41, P = 0.001) and elastance (r² = 0.33, P = 0.003) and a decrease in Pao₂/Fio₂ (r² = 0.40, P < 0.001) at the end of the experiment (fig. 2). All piglets ventilated with higher power developed full-blown (all-field lung) edema. The experiment ended after 34 ± 13 h.

In piglets treated with mechanical power below the threshold, the energy delivered by a single breath remained constant up to the end of the study. In contrast, in piglets where the mechanical power at baseline was greater than the threshold, the energy delivered by a single breath rose during the study (fig. 3). The increases in transpulmonary energy per breath were positively correlated with the worsening of lung strain (r² = 0.14, P < 0.0001), collapse and decollapse (r² = 0.23, P < 0.0001), and lung inhomogeneity (r² = 0.37, P < 0.0001) (fig. 4).

The transpulmonary energy per breath was associated with “stress relaxation,” that is, the drop in airway pressure during an end-inspiratory pause (see fig. 11; Supplemental Digital Content 1, https://links.lww.com/ALN/B259).

Lungs of animals that did not develop ventilator-induced lung edema appeared macroscopically pink and normally inflated, with only small areas of atelectasis. In contrast, lungs of piglets that developed VILI were purple and congested. Blinded qualitative assessment indicated that histological changes (hyaline membranes, ruptured alveoli, and interstitial and intraalveolar infiltrate) tended to be more severe in piglets that developed ventilator-induced lung edema. However, the piglets that did not develop edema also presented relevant histological alterations (see table 7 and fig. 22 in the Supplemental Digital Content 1, https://links.lww.com/ALN/B259).

In this study, we hypothesized that, for VILI to occur, what matters is the amount of energy delivered to the respiratory system in the unit of time (joule/min), that is, the mechanical power. In these healthy piglets, widespread edema developed only when the delivered transpulmonary mechanical power exceeded 12.1 J/min.

Mechanical ventilation, depending on the ventilator setting and the lung mechanics, is based on different combinations of TV, airway pressures, flows, and respiratory rates. We quantified these variables, together, as mechanical power. TV 2.5 times the functional residual capacity (38 ml/kg) resulted in lethal widespread edema within 54 h if delivered at a respiratory rate of 15 breaths/min,^17–19  with lesions starting at the interfaces between structures with different elasticity (stress raiser).¹⁹  We postulated that VILI was not due to the TV per se but due to the mechanical power, that is, the product of TV, plateau pressure, and respiratory rate. To prove this theory, in the first 15 experiments, we lowered the respiratory rate: the TV that was lethal at 15 breaths/min was not lethal at 3, 6, and 9 breaths/min. This established the mechanical power threshold at approximately 12 J/min. The idea that lowering the respiratory rate might be beneficial, permitting lung rest and protection, has already been used clinically²⁷  and experimentally,^28,29  together with extracorporeal carbon dioxide removal. Those studies, however, did not focus directly on VILI as an endpoint. There are only a few reports of studies on the effect of respiratory rate on VILI, most of them finding that the higher the rate, the worse the damage.^5–7 

If the power is “excessive,” then the chemical bonds of the polymers composing the extracellular matrix may be disrupted. This can be considered a kind of “extracellular matrix fatigue” in analogy with material fatigue, where microfractures result from cyclic application of mechanical power above a given threshold.¹⁰  Furthermore, the microfractures of the polymers of the extracellular matrix produce low-molecular-weight hyaluronans, which may act as a trigger for both an inflammatory reaction³⁰  and repair processes.^30,31 

Our nine confirmatory experiments corroborated the mechanical power hypothesis. TVs of 11 and 22 ml/kg at a respiratory rate of 35 breaths/min produced mechanical power below and above the threshold, respectively. The piglets above the threshold developed VILI, whereas those below the threshold did not. These findings suggest that neither the TV alone (38, 22, and 11 ml/kg) nor the respiratory rate (15 or 35 breaths/min) is the cause of VILI, which instead is induced by their combination when the mechanical power produced is higher than a certain threshold. In this experimental setting, quite likely because of the wide mechanical power variations induced, we could not find any difference with or without control of the Pco₂. Increases in mechanical power were associated with a deterioration of the main CT scan variables, such as intratidal opening and closing,^32,33  lung strain,³  and extent of lung inhomogeneity.²⁵  The damaged lung needed increased mechanical power to be ventilated at constant TV: the reduction of inflated lung, the decrease in open airways, and the increased collapse all lead to a rise in pressures and flows. This will result, for a given TV, in an increase in the delivered mechanical power, accelerating the vicious cycle that might partly explain the exponential increase of lung damage previously observed when we studied the time course of VILI.¹⁹ 

We may wonder to what extent the results of this study can be translated to a clinical scenario. The TV that we used largely exceeded that adopted in clinical practice (38 ml/kg as opposed to 6 ml/kg)¹ ; however, in acute respiratory distress syndrome, the tissue open to ventilation becomes progressively less in proportion to severity. Consequently, with lower TV, levels of strain may be similar.³  Moreover, stress raisers may worsen the whole situation.^25,34  The basic concept of the lung-protective strategy, in our opinion, is to provide the lowest possible mechanical power while maintaining the lung as homogenous as possible by prone positioning³⁵  and “adequate” positive end-expiratory pressure.³⁶  Although the TV concept is well accepted in the lung-protective framework,³⁷  less attention has been paid to the respiratory rate. Too high a respiratory rate may well cause damage, even if the TV is at least partly reduced. High-frequency oscillations,³⁸  as an example, may at first glance seem a safe form of support, as the delta-volume is low. However, the energy required to oscillate the system may be so high as to exceed the energy threshold for VILI. Therefore, in our opinion, paying attention to mechanical power might help extending our focus on VILI, taking into account not only the TV and driving pressure, as recently suggested,³⁹  but also the flows and the respiratory rate and—maybe more important—their combination. Thus, any reduction in any component of the cyclic mechanical power should lower the risk of VILI.

The authors thank Patrizia Minunno (Dipartimento di Anestesia, Rianimazione ed Emergenza Urgenza, Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Milan, Italy) for valuable logistic support; Fabio Ambrosetti (Dipartimento di Fisiopatologia Medico- Chirurgica e dei Trapianti, Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Università degli Studi di Milano, Milan, Italy) for technical assistance; Greta Rossignoli, M.D., and Cristina Rovati, M.D. (Dipartimento di Fisiopatologia Medico- Chirurgica e dei Trapianti, Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Università degli Studi di Milano), for assistance in the experiment; and Massimo Monti, M.D., and Manuela Chiodi, M.D. (Dipartimento di Fisiopatologia Medico-Chirurgica e dei Trapianti, Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Università degli Studi di Milano), for technical assistance with computed tomography scan analysis.

Supported by European Society of Intensive Care Medicine 2014 Clinical Research Award (Bruxelles, Belgium; to Dr. Cressoni) and by institutional funds.

Dr. Cressoni, Mr. Cadringher, and Dr. Gattinoni hold an Italian patent for determination of lung inhomogeneities (0001409041) and have applied for European and U.S. patents. The other authors declare no competing interests.