Rationale: Unphysiologic strain (the ratio between tidal volume and functional residual capacity) and stress (the transpulmonary pressure) can cause ventilator-induced lung damage.
Objectives: To identify a strain–stress threshold (if any) above which ventilator-induced lung damage can occur.
Methods: Twenty-nine healthy pigs were mechanically ventilated for 54 hours with a tidal volume producing a strain between 0.45 and 3.30. Ventilator-induced lung damage was defined as net increase in lung weight.
Measurements and Main Results: Initial lung weight and functional residual capacity were measured with computed tomography. Final lung weight was measured using a balance. After setting tidal volume, data collection included respiratory system mechanics, gas exchange and hemodynamics (every 6 h); cytokine levels in serum (every 12 h) and bronchoalveolar lavage fluid (end of the experiment); and blood laboratory examination (start and end of the experiment). Two clusters of animals could be clearly identified: animals that increased their lung weight (n = 14) and those that did not (n = 15). Tidal volume was 38 ± 9 ml/kg in the former and 22 ± 8 ml/kg in the latter group, corresponding to a strain of 2.16 ± 0.58 and 1.29 ± 0.57 and a stress of 13 ± 5 and 8 ± 3 cm H2O, respectively. Lung weight gain was associated with deterioration in respiratory system mechanics, gas exchange, and hemodynamics, pulmonary and systemic inflammation and multiple organ dysfunction.
Conclusions: In healthy pigs, ventilator-induced lung damage develops only when a strain greater than 1.5–2 is reached or overcome. Because of differences in intrinsic lung properties, caution is warranted in translating these findings to humans.
Mechanical ventilation can cause lung damage, perhaps proportionally to the applied strain and stress. Accordingly, use of small tidal volume (rough surrogate of strain) has been recommended not only in diseased but even in healthy lungs.
Ventilator-induced lung damage does not develop proportionally to the applied strain and stress, but only when a critical threshold is reached or overcome. On average, in healthy pigs, mechanical ventilation does not induce gross lung damage unless the applied strain is greater than 1.5–2 (corresponding to tidal volumes far greater than 20 ml/kg), that is, when the strain–stress relationship ceases to be linear. Mechanical ventilation seems to be reasonably safe in healthy lungs, unless very large tidal volumes are used.
Unfortunately, neither tidal volume per kilogram of body weight nor plateau pressure are adequate estimates of strain and stress, mainly because of the large interindividual variability in functional residual capacity and chest wall elastance, respectively (2).
Mechanical ventilation of healthy lungs with high tidal volume and plateau pressure (rough surrogates of strain and stress) had proved harmful (6). However, it remains unclear whether lung damage develops proportionally to the applied strain–stress, in a sort of continuum, or only when a critical threshold is reached or overcome. To clarify this issue, we investigated the occurrence of lung damage in healthy pigs undergoing prolonged mechanical ventilation with tidal volumes ranging from 9 up to 51 ml/kg of body weight, corresponding to strains between 0.45 and 3.30.
The aim of the present work was to answer the following questions. Does ventilator-induced lung damage occur only when a critical strain–stress threshold is exceeded? What is the relationship between this threshold (if any) and lung anatomy and physiology (i.e., volumes and mechanics)? In this paper we describe our experimental findings and briefly discuss their potential translation to humans.
The study was conducted in accordance with international recommendations (7) and approved by the Institutional Review Board. Twenty-nine healthy pigs underwent surgical preparation under general anesthesia and paralysis. In 22 animals, respiratory system and lung volume–pressure curves were obtained starting from functional residual capacity (airway pressure = 0 cm H2O), in steps of 100 ml. Strain–stress curves were derived once functional residual capacity was known (see below). Overall mean pressure–volume curves were obtained averaging the volumes calculated from individual fittings at pressure intervals of 0.5 cm H2O.
Strain was defined as the ratio between tidal volume and functional residual capacity and stress as the transpulmonary (airway minus esophageal) pressure at zero flow. Specific lung elastance was measured as both the transpulmonary pressure required to double functional residual capacity (i.e., strain = 1) and the slope of the linear part of the stress–strain curve.
Lung computed tomography (CT) was performed at 45 cm H2O (arbitrarily defined as total lung capacity) and 0 cm H2O (functional residual capacity) of airway pressure. Quantitative analysis was performed with dedicated software (SoftEFilm, www.elekton.it).
Each pig was then mechanically ventilated (Engstrom Carestation; GE Healthcare) for 54 hours or until death if this occurred earlier (as in 12 animals). Tidal volume was set to produce a predefined initial strain ranging from 0.45 up to 3.30, and kept constant thereafter. A considerable number of animals were ventilated with intermediate values after a preliminary analysis suggested the presence there of a threshold. Respiratory rate was always 15 breaths per minute, oxygen inspiratory fraction 0.5, and the inspiratory-to-expiratory time ratio 1:2 (1:3 if intrinsic positive pressure tended to develop). Because we wanted to define better the pathogenesis of ventilator-induced lung injury, animals were ventilated with no positive end-expiratory pressure. Doing so, we maximized the chance of observing lung damage while avoiding any potential protection caused by positive end-expiratory pressure (8).
In 20 animals, dead space was used to ensure an initial arterial carbon dioxide tension (PaCO2) of 35–45 mm Hg. In nine animals, any initial drop in PaCO2 was not corrected to investigate the possible harmful effect of hypocapnia (9). However, because normocapnic and hypocapnic animals behaved the same, they were analyzed together (see below). Saline and norepinephrine were used according to a standardized protocol, with a target mean arterial pressure greater than 60 mm Hg.
Respiratory system mechanics, blood gas analysis and hemodynamics were assessed every 6 hours. Serum cytokines (IL-6, IL-8, IL-10 and tumor necrosis factor-α) were measured every 12 hours. Urinary output and sodium and potassium concentrations were monitored with a dedicated device, K.IN.G (collaboration of Orvim, Paderno Dugnano, Italy, and Kardia, Milan, Italy) (10). Blood laboratory examinations were done at the beginning and end of the study.
At autopsy, lungs were excised and weighed. Lung damage was evaluated in terms of weight gain (i.e., edema formation) between start and end of the experiment. Cytokines (as above) were measured on left lung bronchoalveolar lavage fluid. Five tissue fragments from the right lung were used to calculate the wet-to-dry weight ratio. The residual right lung was fixed in formalin and underwent histologic examination (11).
Data are reported as mean ± standard deviation, unless otherwise stated. Continuous variables were compared with the Wilcoxon test because of the small sample size. Categorical variables were compared with Fisher exact test. Statistical analysis was performed with R-software (www.R-project.org). For more information on methods, see the online supplement.
In Figure 1 we plot the individual lung weight change as a function of duration of mechanical ventilation. As shown, two clusters of animals can be clearly identified, suggesting a threshold phenomenon: those that markedly increased their lung weight (ventilator-induced lung edema [VILE] group) and those that did not (No-VILE group). On average, VILE pigs were ventilated with a tidal volume of 38 ± 9 ml/kg, corresponding to an initial strain of 2.16 ± 0.58, whereas No-VILE pigs received a tidal volume of 22 ± 8 ml/kg, corresponding to a strain of 1.29 ± 0.57.

Figure 1. Ventilator-induced lung edema (VILE) and duration of mechanical ventilation. Lung weight change (left y axis) as a function of duration of mechanical ventilation (x axis). Lung weight changes were computed as final (measured with a balance after autopsy) minus initial (measured with computed tomography) lung weight. White circles refer to pigs that developed lung edema (VILE group); black circles refer to those that did not (No-VILE group). Average (± SD) baseline strains applied to the two groups are represented on the right y axis (P < 0.001, Wilcoxon test).
[More] [Minimize]Baseline CT lung characteristics of the two groups of animals were similar, as shown in Table 1. Differently from normal human subjects, no dependent atelectasis was observed by the time CT scan was performed, despite anesthesia and paralysis. On average, the ratio between total lung and functional residual capacities was 3.5 ± 0.5, corresponding to a strain of 2.5 ± 0.5.
Characteristic | Airway Pressure | No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value |
---|---|---|---|---|
Pig weight, kg | 21 ± 3 | 22 ± 2 | 0.71 | |
Total lung volume, ml | 0 cm H2O | 677 ± 96 | 702 ± 107 | 0.65 |
45 cm H2O | 1,583 ± 224 | 1,712 ± 195 | 0.07 | |
Lung gas volume, ml | 0 cm H2O | 378 ± 86 | 389 ± 85 | 0.62 |
45 cm H2O | 1,268 ± 222 | 1,378 ± 174 | 0.11 | |
Lung weight, g | 0 cm H2O | 299 ± 31 | 314 ± 36 | 0.16 |
45 cm H2O | 314 ± 34 | 334 ± 41 | 0.17 | |
Over inflated tissue, % | 0 cm H2O | 0 ± 0 | 0 ± 0 | 0.85 |
45 cm H2O | 3 ± 4 | 3 ± 4 | 0.81 | |
Poorly inflated tissue, % | 0 cm H2O | 32 ± 15 | 35 ± 12 | 0.38 |
45 cm H2O | 11 ± 1 | 12 ± 2 | 0.35 | |
Well inflated tissue, % | 0 cm H2O | 63 ± 15 | 60 ± 12 | 0.40 |
45 cm H2O | 82 ± 7 | 80 ± 7 | 0.56 | |
Not inflated tissue, % | 0 cm H2O | 5 ± 2 | 5 ± 2 | 0.65 |
45 cm H2O | 5 ± 3 | 5 ± 3 | 0.78 | |
Lung density, HU | 0 cm H2O | −549 ± 58 | −546 ± 48 | 0.81 |
45 cm H2O | −797 ± 34 | −803 ± 21 | 0.91 | |
Superimposed pressure, cm H2O | 0 cm H2O | 5 ± 1 | 5 ± 1 | 0.29 |
45 cm H2O | 2 ± 0 | 2 ± 0 | 0.95 |
In Figure 2 we describe the individual changes in lung weight as a function of the applied strain. As shown, three different intervals can be identified. Strains lower than 1 (corresponding to tidal volumes of 18 ± 3 ml/kg of body weight) did not produce any increase in lung weight. Conversely, strains greater than 2.1 (tidal volumes of 38 ± 6 ml/kg of body weight) invariably resulted in a dramatic increase in lung weight. Intermediate values of strain proved either safe or injurious. Accordingly, the critical threshold for the development of ventilator-induced lung edema cannot be defined as a single, clear-cut, value, but rather corresponds to an interval (or zone) ranging from 1 to 2. This interval might be further restricted to 1.5–2 by considering as an outlier the single, never replicated, hypocapnic animal that developed lung edema at a strain slightly above 1 (see Figure E1 in the online supplement).

Figure 2. Lung weight change and strain in normocapnic and hypocapnic animals. Individual changes in lung weight as a function of the strain initially applied, in hypocapnic (white squares; n = 9) and normocapnic (black squares; n = 20) pigs. Based on visual inspection, lung weight gain (i.e., lung edema formation) occurred in normocapnic animals ventilated at strain between 1.5 and 2 or higher. Apart from one isolated case (not replicated in four subsequent experiments), hypocapnic animals grossly behaved the same.
[More] [Minimize]In Figure 3A, we report the average pressure–volume curve of the study population, whereas the corresponding stress–strain relationship is shown in Figure 3B. The slope of the linear part of the stress–strain curve (i.e., the specific lung elastance) averaged 5.4 ± 2.2 cm H2O (about half of that recorded in humans) (2). The stress–strain relationship starts to lose its linearity at a strain between 1.5 and 2, corresponding to a transpulmonary pressure around 10 cm H2O, suggesting that at this level some lung units reach their own total capacity.

Figure 3. Mechanical properties of pig respiratory system and lung. (A) Average pressure–volume curve computed from the fittings of individual recordings, available in 22 pigs. Data are presented as mean ± SE. (B) Average stress–strain relationship computed from the fitting of individual recordings, available in 22 pigs. Stress was defined as the transpulmonary pressure, here presented at intervals of 0.5 cm H2O. Strain was calculated as the volume corresponding to each level of stress (according to individual fitting equations) divided by individual functional residual capacity. The straight line represents the linear part of the stress–strain relationship. Data are presented as mean ± SE.
[More] [Minimize]In Tables 2 and 3 we report the main ventilatory and respiratory data recorded from VILE and No-VILE groups at the beginning and end of experiments. Specific lung elastance, computed as the transpulmonary pressure when strain was 1, was similar to the slope of the linear part of the stress-strain curve, depicted in Figure 3. By the end of the experiments, VILE pigs had a remarkably greater deterioration of respiratory mechanics and gas exchange than No-VILE pigs. Of note, initial plateau pressure was usually below 30 cm H2O even in animals that finally developed lung edema.
No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value | ||
---|---|---|---|---|
Strain, tidal volume/FRC | 1.29 ± 0.57 | 2.16 ± 0.58 | <0.001 | |
Tidal volume/body weight, ml/kg | 22 ± 8 | 38 ± 9 | <0.0001 | |
Tidal volume, ml | 461 ± 175 | 816 ± 203 | <0.0001 | |
End-inspiratory lung volume, ml | 839 ± 194 | 1,205 ± 251 | <0.0001 | |
Specific lung elastance, cm H2O | 5 ± 2 (n = 12) | 6 ± 3 (n = 10) | 0.54 | |
Mean airway pressure, cm H2O | Start of experiment | 7 ± 3 | 12 ± 3 | |
End of experiment | 8 ± 3 | 17 ± 4 | ||
Δ (End − Start) | 1 ± 3 | 6 ± 3 | <0.001 | |
Plateau pressure, cm H2O | Start of experiment | 16 ± 5 | 26 ± 7 | |
End of experiment | 20 ± 8 | 41 ± 7 | ||
Δ (End − Start) | 4 ± 5 | 16 ± 5 | <0.001 | |
Δ Transpulmonary pressure, cm H2O | Start of experiment | 8 ± 3 (n = 13) | 13 ± 5 (n = 13) | |
End of experiment | 13 ± 7 (n = 13) | 29 ± 6 (n = 13) | ||
Δ (End − Start) | 5 ± 6 (n = 13) | 17 ± 4 (n = 13) | <0.001 | |
Δ Esophageal pressure, cm H2O | Start of experiment | 9 ± 2 (n = 13) | 13 ± 4 (n = 13) | |
End of experiment | 8 ± 2 (n = 13) | 12 ± 3 (n = 13) | ||
Δ (End − Start) | −1 ± 1 (n = 13) | −1 ± 3 (n = 13) | 0.75 | |
Respiratory system elastance, cm H2O/L | Start of experiment | 37 ± 9 | 32 ± 7 | |
End of experiment | 45 ± 13 | 52 ± 9 | ||
Δ (End − Start) | 8 ± 11 | 20 ± 10 | <0.01 | |
Lung elastance, cm H2O/L | Start of experiment | 16 ± 5 (n = 13) | 15 ± 4 (n = 13) | |
End of experiment | 28 ± 11 (n = 13) | 37 ± 9 (n = 13) | ||
Δ (End − Start) | 12 ± 14 (n = 13) | 22 ± 9 (n = 13) | 0.01 | |
Chest wall elastance, cm H2O/L | Start of experiment | 20 ± 7 (n = 13) | 17 ± 5 (n = 13) | |
End of experiment | 18 ± 8 (n = 13) | 15 ± 4 (n = 13) | ||
Δ (End − Start) | −2 ± 4 (n = 13) | −1 ± 3 (n = 13) | 0.52 | |
Change in lung weight, % | −21 ± 14 | 91 ± 26 | <0.0001 | |
Wet-to-dry ratio | 4.8 ± 0.7 | 7.6 ± 1.1 | <0.0001 | |
Hours of experimental mechanical ventilation | 54 ± 2 | 28 ± 17 | <0.0001 | |
54-h survival, % | 15/15 (100%) | 2/14 (14%) | <0.0001 |
No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value | ||
---|---|---|---|---|
PaO2/FiO2 | Start of experiment | 414 ± 107 | 419 ± 66 | |
End of experiment | 395 ± 90 | 146 ± 49 | ||
Δ (End − Start) | −19 ± 139 | −272 ± 59 | <0.0001 | |
PaO2, mm Hg | Start of experiment | 207 ± 54 | 209 ± 33 | |
End of experiment | 198 ± 45 | 73 ± 25 | ||
Δ (End − Start) | −9 ± 69 | −136 ± 29 | <0.0001 | |
PaCO2, mm Hg | Start of experiment | 33 ± 12 | 34 ± 11 | |
End of experiment | 28 ± 11 | 47 ± 14 | ||
Δ (End − Start) | −5 ± 4 | 13 ± 8 | <0.0001 | |
pHa | Start of experiment | 7.57 ± 0.15 | 7.53 ± 0.15 | |
End of experiment | 7.53 ± 0.07 | 7.25 ± 0.16 | ||
Δ (End − Start) | −0.05 ± 0.09 | −0.28 ± 0.14 | <0.0001 | |
BEa, mmol/L | Start of experiment | 6.2 ± 3.8 | 4.2 ± 4.3 | |
End of experiment | −0.6 ± 6.2 | −7.2 ± 6.3 | ||
Δ (End − Start) | −6.8 ± 6.5 | −11.4 ± 8.4 | 0.33 | |
Arterial lactate, mmol/L | Start of experiment | 0.8 ± 0.4 | 1.1 ± 0.9 | |
End of experiment | 0.5 ± 0.2 | 4 ± 3.7 | ||
Δ (End − Start) | −0.3 ± 0.3 | 2.9 ± 3.7 | <0.001 | |
SvO2, % | Start of experiment | 62 ± 6 (n = 10) | 66 ± 9 (n = 8) | |
End of experiment | 60 ± 6 (n = 10) | 40 ± 19 (n = 8) | ||
Δ (End − Start) | −2 ± 10 (n = 10) | −27 ± 19 (n = 8) | <0.01 |
The most relevant hemodynamic variables are summarized in Table 4. The VILE-group had a larger decrease in mean arterial pressure than the No-VILE group, despite higher heart rate, greater vasoactive support, and positive water balance. Metabolic acidosis, hyperlactatemia, and mixed venous blood desaturation, signs of hemodynamic inadequacy, were only observed in the VILE group. However, changes in central venous and pulmonary artery occlusion pressures, and cardiac output, did not differ between groups. Animals that finally developed lung edema increased their body weight (although the relative increase in lung weight was always larger), and had lower urinary output and greater sodium retention.
No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value | ||
---|---|---|---|---|
Mean arterial pressure, mm Hg | Start of experiment | 89 ± 18 | 81 ± 12 (n = 13) | |
End of experiment | 80 ± 14 | 47 ± 10 (n = 13) | ||
Δ (End − Start) | −9 ± 25 | −34 ± 17 (n = 12) | <0.01 | |
Heart rate, beats/min | Start of experiment | 111 ± 25 | 124 ± 31 (n = 13) | |
End of experiment | 82 ± 18 (n = 14) | 159 ± 45 (n = 12) | ||
Δ (End − Start) | −25 ± 29 (n = 14) | 33 ± 64 (n = 11) | <0.01 | |
Central venous pressure, mm Hg | Start of experiment | 4 ± 2 (n = 14) | 5 ± 2 (n = 10) | |
End of experiment | 3 ± 2 (n = 14) | 7 ± 2 (n = 9) | ||
Δ (End − Start) | 0 ± 1 (n = 13) | 2 ± 2 (n = 8) | 0.04 | |
Pulmonary artery occlusion pressure, mm Hg | Start of experiment | 7 ± 2 (n = 8) | 8 ± 4 (n = 8) | |
End of experiment | 5 ± 2 (n = 7) | 9 ± 6 (n = 4) | ||
Δ (End − Start) | −2 ± 2 (n = 7) | 5 ± 9 (n = 3) | 0.06 | |
Mean pulmonary pressure, mm Hg | Start of experiment | 14 ± 3 (n = 7) | 21 ± 7 (n = 6) | |
End of experiment | 13 ± 3 (n = 6) | 29 ± 11 (n = 5) | ||
Δ (End − Start) | −1 ± 5 (n = 6) | 14 ± 13 (n = 3) | 0.09 | |
Cardiac output, L/min | Start of experiment | 2.5 ± 0.6 (n = 8) | 2.5 ± 0.5 (n = 9) | |
End of experiment | 2 ± 0.6 (n = 8) | 1.9 ± 0.7 (n = 10) | ||
Δ (End − Start) | −0.3 ± 1.1 (n = 6) | −0.9 ± 0.8 (n = 8) | 0.37 | |
Norepinephrine infusion, μg/kg/min | 0.13 ± 0.24 | 0.84 ± 0.42 | <0.001 | |
Water balance, ml | −363 ± 643 (n = 11) | 910 ± 828 | <0.001 | |
Change in pig weight, kg | 0.3 ± 1.9 (n = 14) | 1.5 ± 1.2 | 0.02 | |
Urinary sodium concentration, mEq/L | Start of experiment | 66 ± 33 (n = 13) | 70 ± 35 (n = 13) | |
End of experiment | 121 ± 20 (n = 10) | 44 ± 48 (n = 12) | ||
Δ (End − Start) | 61 ± 40 (n = 10) | −31 ± 49 (n = 12) | <0.001 | |
Urinary potassium concentration, mEq/L | Start of experiment | 70 ± 51(n = 13) | 77 ± 36 (n = 12) | |
End of experiment | 22 ± 12 (n = 10) | 56 ± 28 (n = 12) | ||
Δ (End − Start) | −59 ± 58 (n = 10) | −22 ± 52 (n = 12) | 0.16 | |
Urinary output, ml/h | Start of experiment | 90 ± 70 (n = 13) | 97 ± 42 | |
End of experiment | 68 ± 30 (n = 11) | 20 ± 23 | ||
Δ (End − Start) | −19 ± 85 (n = 11) | −77 ± 54 | 0.02 |
In Table 5 we present the major findings of blood laboratory examination, focusing on markers of organs and systems function. Changes in serum urea and creatinine, electrolytes, albumin, bilirubin, transaminases, and troponin were more pronounced in the VILE group, possibly indicating renal, hepatic, and myocardial dysfunction or injury. Hemoglobin concentration significantly increased in the VILE group despite a positive water balance, suggesting a systemic capillary leak syndrome.
No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value | ||
---|---|---|---|---|
Urea, mg/dl | Baseline | 21 ± 5 (n = 14) | 23 ± 8 (n = 12) | |
End of experiment | 43 ± 27 (n = 14) | 67 ± 18 (n = 12) | ||
Δ (End − Baseline) | 22 ± 30 (n = 14) | 44 ± 23 (n = 12) | <0.01 | |
Creatinine, mg/dl | Baseline | 0.9 ± 0.2 (n = 14) | 0.9 ± 0.2 (n = 12) | |
End of experiment | 1.1 ± 0.6 (n = 14) | 2 ± 0.7 (n = 12) | ||
Δ (End − Baseline) | 0.2 ± 0.6 (n = 14) | 1.1 ± 0.7 (n = 12) | <0.01 | |
Albumin, g/dl | Baseline | 2.5 ± 0.3 (n = 12) | 2.9 ± 0.4 (n = 12) | |
End of experiment | 1.8 ± 0.2 (n = 12) | 1.9 ± 0.3 (n = 12) | ||
Δ (End − Baseline) | −0.7 ± 0.3 (n = 12) | −1 ± 0.3 (n = 12) | 0.04 | |
Total bilirubin, mg/dl | Baseline | 0.1 ± 0.1 (n = 14) | 0.1 ± 0.0 (n = 12) | |
End of experiment | 0.1 ± 0.1 (n = 14) | 0.2 ± 0.1 (n = 12) | ||
Δ (End − Baseline) | 0 ± 0.1 (n = 14) | 0.1 ± 0.1 (n = 12) | 0.01 | |
AST, IU/L | Baseline | 45 ± 11 (n = 14) | 44 ± 9 (n = 12) | |
End of experiment | 101 ± 148 (n = 14) | 445 ± 445 (n = 12) | ||
Δ (End − Baseline) | 55 ± 146 (n = 14) | 400 ± 443 (n = 12) | <0.001 | |
ALT, IU/L | Baseline | 51 ± 8 (n = 14) | 56 ± 14 (n = 12) | |
End of experiment | 42 ± 12 (n = 14) | 54 ± 19 (n = 12) | ||
Δ (End − Baseline) | −9 ± 13 (n = 14) | −2 ± 15 (n = 12) | 0.05 | |
Potassium, mEq/L | Baseline | 4.1 ± 0.4 (n = 14) | 3.8 ± 0.5 (n = 12) | |
End of experiment | 3.6 ± 0.5 (n = 14) | 5.7 ± 1.3 (n = 12) | ||
Δ (End − Baseline) | −0.6 ± 0.8 (n = 14) | 1.9 ± 1.4 (n = 12) | <0.001 | |
Sodium, mEq/L | Baseline | 139 ± 3 (n = 14) | 140 ± 3 (n = 12) | |
End of experiment | 142 ± 3 (n = 14) | 142 ± 3 (n = 12) | ||
Δ (End − Baseline) | 3 ± 4 (n = 14) | 1 ± 4 (n = 12) | 0.42 | |
Chloride, mEq/L | Baseline | 104 ± 3 (n = 14) | 104 ± 2 (n = 12) | |
End of experiment | 115 ± 3 (n = 14) | 112 ± 5 (n = 12) | ||
Δ (End − Baseline) | 12 ± 4 (n = 14) | 8 ± 5 (n = 12) | 0.1 | |
Total calcium, mEq/L | Baseline | 9.6 ± 0.4 (n = 12) | 10.1 ± 0.5 (n = 11) | |
End of experiment | 8.4 ± 0.4 (n = 12) | 8.4 ± 0.6 (n = 11) | ||
Δ (End − Baseline) | −1.2 ± 0.7 (n = 12) | −1.7 ± 0.8 (n = 11) | 0.28 | |
Phosphate, mEq/L | Baseline | 8.7 ± 0.9 (n = 12) | 8.9 ± 1.7 (n = 11) | |
End of experiment | 6.3 ± 0.7 (n = 12) | 11.2 ± 3.4 (n = 11) | ||
Δ (End − Baseline) | −2.4 ± 1.2 (n = 12) | 2.3 ± 3.3 (n = 11) | <0.001 | |
Magnesium, mEq/L | Baseline | 1.8 ± 0.2 (n = 12) | 1.8 ± 0.2 (n = 11) | |
End of experiment | 1.6 ± 0.2 (n = 12) | 2 ± 0.4 (n = 11) | ||
Δ (End − Baseline) | −0.2 ± 0.2 (n = 12) | 0.2 ± 0.4 (n = 11) | 0.01 | |
Troponin T, ng/ml | Baseline | 0.02 ± 0.01 (n = 12) | 0.01 ± 0.00 (n = 12) | |
End of experiment | 0.01 ± 0.01 (n = 12) | 0.14 ± 0.34 (n = 12) | ||
Δ (End − Baseline) | 0.00 ± 0.01 (n = 12) | 0.13 ± 0.34 (n = 12) | <0.01 | |
Hemoglobin, g/dl | Baseline | 9.2 ± 0.5 (n = 13) | 10 ± 1.0 (n = 13) | |
End of experiment | 8.7 ± 1.4 (n = 13) | 11.6 ± 2.4 (n = 13) | ||
Δ (End − Baseline) | −0.5 ± 1.2 (n = 13) | 1.6 ± 2.8 (n = 13) | 0.02 | |
Platelets, 106/ml | Baseline | 530 ± 147 (n = 13) | 532 ± 132 (n = 13) | |
End of experiment | 257 ± 137 (n = 13) | 160 ± 105 (n = 13) | ||
Δ (End − Baseline) | −273 ± 97 (n = 13) | −373 ± 150 (n = 13) | 0.02 |
As shown in Table 6, serum and bronchoalveolar lavage fluid IL-6 levels were significantly higher in the VILE, compared with the No-VILE, group. No significant differences were observed for IL-8, tumor necrosis factor-α, and IL-10. Leukocytosis and coagulopathy tended to be more evident in the VILE group.
No Ventilator-induced Lung Edema (n = 15) | Ventilator-induced Lung Edema (n = 14) | P Value | ||
---|---|---|---|---|
Serum TNF-α, pg/ml | Baseline | 113 ± 47 (n = 12) | 88 ± 43 (n = 13) | |
End of experiment | 77 ± 31 (n = 12) | 83 ± 59 (n = 13) | ||
Δ (End − Baseline) | −36 ± 45 (n = 12) | −5 ± 41 (n = 13) | 0.08 | |
Serum IL-6, pg/ml | Baseline | 49 ± 91 (n = 12) | 88 ± 183 | |
End of experiment | 18 ± 36 (n = 12) | 680 ± 711 | ||
Δ (End − Baseline) | −30 ± 73 (n = 12) | 592 ± 720 | <0.001 | |
Serum IL-8, pg/ml | Baseline | 169 ± 96 (n = 6) | 71 ± 54 (n = 4) | |
End of experiment | 47 ± 48 (n = 6) | 149 ± 240 (n = 4) | ||
Δ (End − Baseline) | −122 ± 123 (n = 6) | 78 ± 196 (n = 4) | 0.07 | |
Serum IL-10, pg/ml | Baseline | <4 | <4 | |
End of experiment | <4 | <4 | ||
Δ (End − Baseline) | ||||
BALF TNF-α, pg/ml | 202 ± 252 (n = 13) | 46 ± 71 | 0.07 | |
BALF IL-6, pg/ml | 211 ± 364 (n = 13) | 2,198 ± 1,685 | <0.0001 | |
BALF IL-8, pg/ml | 1,569 ± 1,475 (n = 13) | 725 ± 1,092 | 0.1 | |
BALF IL-10, pg/ml | <4 | <4 | ||
Prothrombin time International Normalized Ratio | Baseline | 1 ± 0 (n = 7) | 1 ± 0 (n = 8) | |
End of experiment | 0.9 ± 0.1 (n = 7) | 1.4 ± 0.7 (n = 8) | ||
Δ (End − Baseline) | 0 ± 0.1 (n = 7) | 0.4 ± 0.7 (n = 8) | 0.13 | |
Activated partial thromboplastin time ratio | Baseline | 0.6 ± 0.1 (n = 11) | 0.6 ± 0.1 (n = 11) | |
End of experiment | 0.6 ± 0.1 (n = 11) | 0.8 ± 0.3 (n = 11) | ||
Δ (End − Baseline) | 0 ± 0.1 (n = 11) | 0.2 ± 0.2 (n = 11) | 0.15 | |
Fibrinogen, mg/dl | Baseline | 182 ± 31 (n = 8) | 188 ± 24 (n = 10) | |
End of experiment | 317 ± 63 (n = 8) | 337 ± 95 (n = 10) | ||
Δ (End − Baseline) | 135 ± 66 (n = 8) | 150 ± 85 (n = 10) | 0.9 | |
D-dimer, μg/L | Baseline | 148 ± 73 (n = 8) | 215 ± 123 (n = 10) | |
End of experiment | 136 ± 86 (n = 8) | 9,963 ± 10,943 (n = 10) | ||
Δ (End − Baseline) | −12 ± 84 (n = 8) | 9,748 ± 10,898 (n = 10) | <0.0001 | |
White blood cells, ml−1 | Baseline | 13,727 ± 3,902 (n = 13) | 13,820 ± 4,781 (n = 13) | |
End of experiment | 12,787 ± 5,439 (n = 13) | 20,327 ± 11,277 (n = 13) | ||
Δ (End − Baseline) | −940 ± 5,500 (n = 13) | 6,507 ± 12,239 (n = 13) | 0.14 | |
Body core temperature, °C | Baseline | 39.7 ± 2.4 | 39.7 ± 1.8 | |
End of experiment | 38.5 ± 1.1 | 39.5 ± 1.3 | ||
Δ (End − Baseline) | −1.2 ± 2.8 | −0.3 ± 1.8 | 0.23 |
Macroscopically, lungs of animals from the No-VILE group looked pink and normally inflated, with only marginal small areas of atelectasis. In contrast, lungs of animals from the VILE group were purple and congested (Figure 4). Histologic findings are summarized in Table 7. Based on blinded qualitative assessment, histologic alterations (overall injury severity score, hyaline membrane, and alveolar hemorrhage formation) tended to be more diffuse and severe in the VILE group. For additional results, see the online supplement.

Figure 4. Macroscopic lung appearance after autopsy. In left panel, lungs of an animal from the No ventilator-induced lung edema (VILE) group (ventilated for 52 h with a tidal volume = 29 ml/kg body weight, corresponding to a strain = 1.87) are presented. Initial and final lung weights were 355 and 267 g, respectively. In right panel, lungs of an animal from the VILE-group (died after 20 h of mechanical ventilation with a tidal volume = 45 ml/kg, corresponding to a strain = 2.11). Initial and final lung weights were 246 and 513 g, respectively.
[More] [Minimize]Controls (n = 3) | No Ventilator-induced Lung Edema (n = 12) | Ventilator-induced Lung Edema (n = 13) | P Value | |
---|---|---|---|---|
Total histologic score | 12 (12–12) | 18 (16–19) | 22 (19–25) | 0.04 |
Emphysematous change | 2 (2–2) | 2 (2–3) | 2 (1–2) | 0.03 |
Interstitial congestion | 2 (2–2) | 2 (1–2) | 2 (2–3) | 0.22 |
Alveolar hemorrhage | 1 (1–1) | 1 (0–1) | 2 (1–2) | 0.02 |
Alveolar neutrophil infiltration | 0 (0–0) | 2 (1–3) | 2 (1–4) | 0.58 |
Alveolar macrophage proliferation | 1 (1–1) | 2 (1–2) | 2 (2–2) | 0.72 |
Alveolar type II pneumocytes proliferation | 1 (1–1) | 2 (2–2) | 2 (1–2) | 0.72 |
Interstitial lymphocytes proliferation | 1 (1–1) | 2 (1–2) | 1 (1–2) | 0.58 |
Interstitial thickening | 3 (3–3) | 2 (2–3) | 2 (2–3) | 0.86 |
Hyaline membrane formation | 0 (0–0) | 0 (0–1) | 3 (2–4) | <0.001 |
Interstitial fibrosis | 1 (1–1) | 2 (1–2) | 2 (1–2) | 1 |
Organization of alveolar exsudate | 0 (0–0) | 1 (0–1) | 1 (1–2) | 0.05 |
Our results demonstrate that mechanical ventilation can induce edema formation in healthy lungs only when global strain reaches or overcomes a critical interval, reasonably extending from 1.5 and 2. A similar threshold seems to be related to lung anatomy, because it roughly corresponds to the point where the stress–strain relationship loses, on average, its linearity. We also found that the specific lung elastance of pigs is nearly half of that in humans (2). Caution is therefore warranted in translating our current findings from animals to humans, especially when trying to define the stress corresponding to the injurious strain. Finally, we observed that when the strain threshold was overcome, systemic inflammation and multiple organ dysfunction developed.
In the past, ventilator-induced lung injury had been induced in healthy animals by applying tidal volumes ranging from 40 ml/kg (e.g., in sheep) down to 19 ml/kg (e.g., in mice) of body weight (12, 13). Mechanical ventilation with a tidal volume of 12 ml/kg of ideal body weight had proved harmful in patients with acute lung injury (14) and might even harm the healthy lung of surgical patients (15). Along this line, it had been claimed that a safe plateau pressure may not even exist, because mortality seems to linearly increase with end-inspiratory airway pressure in patients with acute lung injury (16). Therefore, even plateau pressure as low as 10 cm H2O, normally reached during anesthesia in adults, might still carry some harm. The bulk of these data may indicate that ventilator-induced lung injury develops proportionally to the applied tidal volume (strain) and plateau pressure (stress).
Our results, however, suggest that in healthy pigs ventilator-induced lung edema is a threshold phenomenon. In fact, lung weight markedly increased only when the anatomic and physiologic limits of the lungs were reached or exceeded for a rather prolonged period of time. The threshold for harm seems to correspond to the region where the stress–strain relationship loses its linearity and starts an exponential growth, suggesting that some lung regions reach their own total capacity and cannot expand any further (i.e., the collagen fibers of the lung skeleton of these regions get fully unfolded) (Figure 3). Surprisingly, as long as healthy lungs were ventilated within the linear portion of the stress–strain curve (i.e., within the limits of their anatomic and physiologic boundaries), no grossly detectable lung damage occurred. In fact, the use of tidal volumes as high as 27 ± 3 ml/kg (strain of 1.5) for 2.5 days, with no positive end-expiratory pressure, did not induce any remarkable alteration in gas exchange, respiratory system mechanics and organs function, and lung appearance at autopsy (Figure 4).
When the strain–stress threshold was overcome, the consequences were devastating. Twelve out of 14 pigs of the VILE group versus 0 out of 15 pigs in the No-VILE group died before the scheduled 54 hours (Table 2). Respiratory system mechanics, gas exchange, hemodynamics, inflammation and distal organs damage (Tables 2–5) markedly deteriorated in animals ventilated at a strain above the critical interval (1.5–2) but remained quite normal in those ventilated at lower strain.
We may try to hypothesize the sequence of events that may have finally led to the development of lung damage in the VILE group. When large strains are applied to the pulmonary extracellular matrix, the anchored epithelial and endothelial cells get locally deformed. Secondary stress fatigue (17, 18), microcapillary rupture (19, 20) or inflammation (21) may then result in increased lung capillary permeability, edema formation, lung mechanics, and gas exchange deterioration. Based on this model, any functional or anatomic damage to lung capillaries (and any change in hemodynamics) is likely to play a critical role in the pathogenesis of ventilator-induced lung damage (22, 23). It is worth emphasizing that lung injury became clinically evident after a variable, but still considerable, amount of time. As detailed in the online supplement, in most of the cases it was not possible to predict whether VILE would have developed by the end of the experiment during the first 12–24 hours. This fact underlines the importance of performing long-term experiments in properly designed animal intensive care units (24). Whatever the time period required, once lung mechanical impairment began, clinical deterioration proceeded rapidly and dramatically. The edema-induced collapse likely reduced the regions open to ventilation, increasing the global strain (constant tidal volume but decreased functional residual capacity) and creating inhomogeneities within the lung parenchyma. The interfaces between open and closed lung regions may have then acted as “stress raisers,” locally multiplying the applied insult (3).
The development of lung edema was associated with a remarkable systemic inflammatory response and multiple organ dysfunction. Animals in the VILE group had increased levels of proinflammatory cytokines in serum and bronchoalveolar lavage fluid, activated coagulation, and signs of systemic capillary leak syndrome (25). Interestingly, the derangement in hemodynamics was more likely caused by inflammation rather than by impeded venous return, as suggested by grossly preserved cardiac filling pressures and output.
How to translate these findings to humans with either healthy or injured lungs is a matter of speculation. Two concepts must be highlighted. First, the strain–stress values we have identified as injurious are those at which the lung approaches its maximal capacity. Second, the relationship between total lung and functional residual capacities is quite similar in pigs and humans (26). Based on these two facts, we suspect that even in healthy humans, the critical threshold for the development of lung edema may correspond to a strain interval between 1.5 and 2, where the strain–stress relationship ceases to be linear (similarly to pigs). However, the corresponding stress, would be markedly higher in humans than in pigs because of interspecies differences in specific lung elastance (∼13 cm H2O and 5 cm H2O, respectively) (5, 27). That the specific lung elastance in pigs is almost half of that in humans may be explained by the different proportion of elastin and collagen in lung parenchyma (28). Accordingly, even plateau pressure levels that may be considered tolerable in humans (<30 cm H2O) proved deleterious in these animals (Table 2). Although plausible, this model is flawed by many assumptions and needs to be properly validated.
To create a new hypothesis, we might carry on speculating on how these data might apply to patients with acute respiratory distress syndrome (ARDS). If one considers the lung open to ventilation during ARDS as a small lung with rather normal mechanical properties (the so-called “baby lung”) (29), then the application of a tidal volume as low as 6 ml/kg will result in an injurious strain only when the baby lung volume is exceptionally low. A harmful strain, therefore, would never (or only rarely) be reached in clinical practice, even in patients with acute lung injury. This hypothetical model is far from being proven from our present data. However, we would like to emphasize the apparent striking discrepancy between similar translational results (if proved valid) and the available evidence indicating that ventilator-induced lung injury does occur in real life, even when tidal volume and plateau pressure are largely lower than those proved dangerous in this work. To reconcile these data we have two possible, nonmutually exclusive, explanations. First, an open but inflamed parenchyma may be more “fragile” (30) and may become damaged by stress and strain much lower than the ones capable of injuring a healthy lung. Second, the ARDS lung is abnormally inhomogeneous with larger interfaces acting as “stress raisers” (2). In this case, even globally low strain–stress may be locally multiplied so as to reach the critical threshold found in normal lungs.
The authors are grateful to Fabio Ambrosetti for valuable technical support; to Paolo Taccone, Marco Lattuada, and Alberto Zanella for precious contribution to study design; and to Professor Peter M. Suter for kindly revising the manuscript. A.P., M.C., A.S., T.L., D.F., and L.G. designed the study, analyzed the data, and wrote the manuscript. A.P., A.S., T.L., C.M., M.C., S.C., G.C., and S.G. ran the experiments. M.C., C.M., D.F., M.C., L.L., and M.L. performed and analyzed lung CT scans. O.L. performed histologic examination of the lungs. S.M. measured serum and bronchoalveolar lavage fluid cytokines. E.R. performed the blood laboratory examinations. P.C. implemented the software for data storage and analysis and provided technical assistance.
1. | Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323. |
2. | Chiumello D, Carlesso E, Cadringher P, Caironi P, Valenza F, Polli F, Tallarini F, Cozzi P, Cressoni M, Colombo A, et al. Lung stress and strain during mechanical ventilation for acute respiratory distress syndrome. Am J Respir Crit Care Med 2008;178:346–355. |
3. | Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970;28:596–608. |
4. | Caironi P, Cressoni M, Chiumello D, Ranieri M, Quintel M, Russo SG, Cornejo R, Bugedo G, Carlesso E, Russo R, et al. Lung opening and closing during ventilation of acute respiratory distress syndrome. Am J Respir Crit Care Med 2010;181:578–586. |
5. | Gattinoni L, Carlesso E, Cadringher P, Valenza F, Vagginelli F, Chiumello D. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003;47:15s–25s. |
6. | Kolobow T. Volutrauma, barotrauma, and ventilator-induced lung injury: lessons learned from the animal research laboratory. Crit Care Med 2004;32:1961–1962. |
7. | Institute of Laboratory Animal Resources, Commission on Life Sciences, National Research Council. Guide for the care and use of laboratory animals. Washington: National Academy Press; 1996. |
8. | Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inflation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974;110:556–565. |
9. | Mascheroni D, Kolobow T, Fumagalli R, Moretti MP, Chen V, Buckhold D. Acute respiratory failure following pharmacologically induced hyperventilation: an experimental animal study. Intensive Care Med 1988;15:8–14. |
10. | Caironi P, Langer T, Taccone P, Bruzzone P, De Chiara S, Vagginelli F, Caspani L, Marenghi C, Gattinoni L. Kidney instant monitoring (K.IN.G): a new analyzer to monitor kidney function. Minerva Anestesiol 2010;76:316–324. |
11. | Tsuno K, Miura K, Takeya M, Kolobow T, Morioka T. Histopathologic pulmonary changes from mechanical ventilation at high peak airway pressures. Am Rev Respir Dis 1991;143:1115–1120. |
12. | Kolobow T, Moretti MP, Fumagalli R, Mascheroni D, Prato P, Chen V, Joris M. Severe impairment in lung function induced by high peak airway pressure during mechanical ventilation. An experimental study. Am Rev Respir Dis 1987;135:312–315. |
13. | Sibilla S, Tredici S, Porro A, Irace M, Guglielmi M, Nicolini G, Tredici G, Valenza F, Gattinoni L. Equal increases in respiratory system elastance reflect similar lung damage in experimental ventilator-induced lung injury. Intensive Care Med 2002;28:196–203. |
14. | The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–1308. |
15. | Schultz MJ, Haitsma JJ, Slutsky AS, Gajic O. What tidal volumes should be used in patients without acute lung injury? Anesthesiology 2007;106:1226–1231. |
16. | Hager DN, Krishnan JA, Hayden DL, Brower RG. Tidal volume reduction in patients with acute lung injury when plateau pressures are not high. Am J Respir Crit Care Med 2005;172:1241–1245. |
17. | Marini JJ, Hotchkiss JR, Broccard AF. Bench-to-bedside review: microvascular and airspace linkage in ventilator-induced lung injury. Crit Care 2003;7:435–444. |
18. | West JB, Tsukimoto K, Mathieu-Costello O, Prediletto R. Stress failure in pulmonary capillaries. J Appl Physiol 1991;70:1731–1742. |
19. | Marini JJ. Microvasculature in ventilator-induced lung injury: target or cause? Minerva Anestesiol 2004;70:167–173. |
20. | Hotchkiss JR, Simonson DA, Marek DJ, Marini JJ, Dries DJ. Pulmonary microvascular fracture in a patient with acute respiratory distress syndrome. Crit Care Med 2002;30:2368–2370. |
21. | Dos Santos CC, Slutsky AS. The contribution of biophysical lung injury to the development of biotrauma. Annu Rev Physiol 2006;68:585–618. |
22. | Broccard AF, Hotchkiss JR, Kuwayama N, Olson DA, Jamal S, Wangensteen DO, Marini JJ. Consequences of vascular flow on lung injury induced by mechanical ventilation. Am J Respir Crit Care Med 1998;157:1935–1942. |
23. | Broccard AF, Vannay C, Feihl F, Schaller MD. Impact of low pulmonary vascular pressure on ventilator-induced lung injury. Crit Care Med 2002;30:2183–2190. |
24. | Kolobow T. The (ir)relevance of short term studies. Int J Artif Organs 1990;13:1–2. |
25. | Mandava S, Kolobow T, Vitale G, Foti G, Aprigliano M, Jones M, Muller E. Lethal systemic capillary leak syndrome associated with severe ventilator-induced lung injury: an experimental study. Crit Care Med 2003;31:885–892. |
26. | Weibel ER. The pathway for oxygen. Structure and function in the mammalian respiratory system. Cambridge, MA: Harvard University Press; 1984. pp. 282–283. |
27. | De Robertis E, Liu JM, Blomquist S, Dahm PL, Thorne J, Jonson B. Elastic properties of the lung and the chest wall in young and adult healthy pigs. Eur Respir J 2001;17:703–711. |
28. | Mercer RR, Russell ML, Crapo JD. Alveolar septal structure in different species. J Appl Physiol 1994;77:1060–1066. |
29. | Gattinoni L, Pesenti A. The concept of “baby lung.” Intensive Care Med 2005;31:776–784. |
30. | Hernandez LA, Coker PJ, May S, Thompson AL, Parker JC. Mechanical ventilation increases microvascular permeability in oleic acid-injured lungs. J Appl Physiol 1990;69:2057–2061. |