Comments
Abstract
Rationale: Pressures and volumes needed to induce ventilator-induced lung injury in healthy lungs are far greater than those applied in diseased lungs. A possible explanation may be the presence of local inhomogeneities acting as pressure multipliers (stress raisers).
Objectives: To quantify lung inhomogeneities in patients with acute respiratory distress syndrome (ARDS).
Methods: Retrospective quantitative analysis of CT scan images of 148 patients with ARDS and 100 control subjects. An ideally homogeneous lung would have the same expansion in all regions; lung expansion was measured by CT scan as gas/tissue ratio and lung inhomogeneities were measured as lung regions with lower gas/tissue ratio than their neighboring lung regions. We defined as the extent of lung inhomogeneities the fraction of the lung showing an inflation ratio greater than 95th percentile of the control group (1.61).
Measurements and Main Results: The extent of lung inhomogeneities increased with the severity of ARDS (14 ± 5, 18 ± 8, and 23 ± 10% of lung volume in mild, moderate, and severe ARDS; P < 0.001) and correlated with the physiologic dead space (r2 = 0.34; P < 0.0001). The application of positive end-expiratory pressure reduced the extent of lung inhomogeneities from 18 ± 8 to 12 ± 7% (P < 0.0001) going from 5 to 45 cm H2O airway pressure. Lung inhomogeneities were greater in nonsurvivor patients than in survivor patients (20 ± 9 vs. 17 ± 7% of lung volume; P = 0.01) and were the only CT scan variable independently associated with mortality at backward logistic regression.
Conclusions: Lung inhomogeneities are associated with overall disease severity and mortality. Increasing the airway pressures decreased but did not abolish the extent of lung inhomogeneities.
Ventilator-induced lung injury (VILI) is a well-known side effect of mechanical ventilation. Pressures and volumes needed to induce VILI in healthy animals are far greater than pressure and volumes applied in clinical practice. A possible explanation may be the presence of local pressure multipliers (stress raisers).
Lung inhomogeneities were quantified and correlated with all the variables describing acute respiratory distress syndrome severity and outcome. Lung inhomogeneities may act regionally as stress raisers by multiplying the applied pressures by near 2. Increasing airway pressure decreased but did not abolish the lung inhomogeneities determined by CT scan.
Ventilator-induced lung injury (VILI) is a well-recognized side effect of mechanical ventilation when applied to patients with acute respiratory distress syndrome (ARDS) (1, 2), chronic obstructive pulmonary disease exacerbation (3), and even to patients with healthy lungs during surgical procedures (4–6). The primary cause of VILI is an unphysiologic strain and stress applied to the extracellular lung matrix, which may be considered as the “lung skeleton” (7, 8). However, to damage the healthy lung with mechanical ventilation, in experimental settings it is necessary to apply tidal volumes as high as 30–70 ml/kg (9–11), whereas only 12 ml/kg are known to induce harm in ARDS lung (12, 13). Several hypothesis can be made to explain this discrepancy. First, in injured lung further damage develops at lower stress and strain (second hit hypothesis) (14). Second, the “baby lung” is so small that even low tidal volumes lead to overdistension of the ventilatable lung parenchyma (15). Third, in an injured inhomogeneous lung parenchyma the stress and strain are unevenly distributed with localized increase of stress, as described on a theoretical background by Mead and coworkers (16). According to this, the regional inhomogeneity may act as “stress raisers.” If so, a transpulmonary pressure that is safe in a homogeneous lung, if multiplied sufficiently by regional stress raisers, may locally reach harmful levels for the lung structure.
On this background, we aimed to quantify the lung inhomogeneity of patients with ARDS by computed tomography (CT). To do so we analyzed whole-lung parenchyma by comparing a given lung region (approximately acinar dimension) with its neighboring regions: if the gas fraction (a surrogate of strain) was the same in the region studied and its neighbors, their ratio would have been equal to 1 (perfect homogeneity), whereas if the gas fraction was greater in the neighboring regions (ratio >1) we assumed that these regions were more strained and the region in study could act on its neighbors as a stress raiser. In this paper we describe the quantitative distribution of the lung inhomogeneities and their relationship with ARDS severity (17) and with the main relevant physiologic variables and outcome in a population of 148 patients with ARDS.
Some of the results of these studies have been previously reported in the form of an abstract (18).
We retrospectively analyzed a database of 148 patients previously enrolled in four clinical trials (see online supplement for detailed information) who underwent a whole-lung CT scan at different pressure levels. Patients were classified as mild, moderate, and severe ARDS according the Berlin definition (17). A control group of 100 spontaneous breathing subjects without ARDS who underwent lung CT scan at end inspiration (19) was studied for comparison.
Lung profiles were manually drawn in each CT section, and images were analyzed with dedicated software (Soft-EFilm; www.elekton.it, Milan, Italy). Lung parenchyma was classified according to its gas/tissue ratio into not-inflated, poorly inflated, well-inflated, and overinflated tissue (20). Lung recruitability was defined as the fraction of lung tissue that is not inflated at 5 cm H2O positive end-expiratory pressure (PEEP) and regains inflation at 45 cm H2O airway pressure.
A homogeneous lung would have the same gas/tissue ratio in all its regions. If a lung region expands less than the neighboring regions, the neighboring regions are more strained to compensate for the nonexpanding or less-expanding region. Therefore, the less expanded and nonexpanded regions act as stress raisers and are characterized by an increased local pressure/stress. Accordingly, the presence of neighboring regions with different inflation through the lung parenchyma indicates the presence of stress raisers. We measured the lung inhomogeneities by computing, voxel by voxel, the ratio between the gas fraction of each voxel (weighted by the surrounding voxels using a Gaussian filter with the size of one acinus, 2.41 mm) and a spherical crust starting at 2.41 mm (radius of an acinus at FRC) and ending at 3.675 mm (equal to the radius of an acinus at FRC × 0.5) (7). If the inflation is the same (homogeneity) the ratio is equal to 1; if the inflation of the surrounding regions is greater than the region of interest (i.e., more strained) the ratio between the two is greater than 1 and was taken as a measure of stress raiser (see Figure 1). However, our assessment of inhomogeneity is always relative (between two neighboring units) because we lack an absolute reference standard. See the online supplement for details.
Figure 1. Two-dimensional representation of the algorithm used for the lung inhomogeneity computation. The grid represents the individual voxels seen from above and below. Circle radii and voxels are represented in scale (each voxel is 0.73 × 0.73 mm). The central voxel is represented in black and the two radii determining the spherical crust are indicated. All the voxels included, even partially, in the spherical crust were included in the analysis. The stress raiser was computed as the maximum of the vector of ratios between the gas fraction of the voxels included in the spherical crust and the gas fraction of the central voxel.
[More] [Minimize]A representative false-color map of lung inhomogeneities in healthy subjects and patients is shown in Figure 2. We considered as pathologic lung inhomogeneities the regions showing a value greater than the 95th percentile of the control group (1.61). We defined as “extent” of the lung inhomogeneities the fraction of lung volume including pathologic lung inhomogeneities and as “intensity” the average value of the lung inhomogeneities included in that fraction.
Figure 2. False-color images of the lung inhomogeneities superimposed to computed tomography (CT) scan images of representative subjects. Light blue = stress raisers < 1.61, orange = stress raisers between 1.61 and 3, red = stress raisers > 3. (A) Representative image of a control subject taken at end-inspiration. (B–D) Representative images of patients with mild, moderate, and severe acute respiratory distress syndrome (ARDS) taken a 5 cm H2O positive end-expiratory pressure (PEEP). (B) Representative CT slice of the lung obtained at the level of lung hilum at 5 cm H2O PEEP. ARDS developed in the patient after an episode of necrotizing fasciitis (PaO2/FiO2 240; Paco2 38.1 mm Hg; lung recruitability 11.4%; Vd/Vt 44%; lung inhomogeneity extent 18.3% at 5 cm H2O PEEP). (C) Representative CT slice of the lung obtained at the level of lung hilum at 5 cm H2O PEEP. ARDS was caused by pneumonia in a bone marrow transplant recipient (PaO2/FiO2 137; Paco2 51 mm Hg; lung recruitability 10%; Vd/Vt 60%; lung inhomogeneity extent 32% at 5 cm H2O PEEP). (D) Representative CT slice of the lung obtained at the level of lung hilum at 5 cm H2O PEEP. ARDS was caused by viral pneumonia (PaO2/FiO2 66; Paco2 47.8 mm Hg; lung recruitability 10%; Vd/Vt 66%; stress raisers extent 32% at 5 cm H2O PEEP). Differences of light blue intensity are caused by the overlapping between the underlying CT scanning images and the false-color map of lung inhomogeneities.
[More] [Minimize]Physiologic and CT scan data between mild, moderate, and severe ARDS according to the Berlin definition (17) were compared with one-way analysis of variance or Kruskal-Wallis test, as appropriate. Multiple comparisons were performed with the Bonferroni correction. Relationships between physiologic and CT scan variables and lung inhomogeneities were assessed with linear regression. The relationship between lung inhomogeneities and outcome was studied with multivariate logistic regression. The R software was used for statistical analysis (21).
Figure 3 shows the frequency distribution of lung inhomogeneities in patients with ARDS and control subjects. As shown, lung inhomogeneities greater than the 95th percentile of the control group (1.61) that we arbitrary considered abnormal were 5 ± 2% in the control patients and 18 ± 8% in the patients with ARDS. Moreover, lung inhomogeneities in healthy subjects seem to be primarily located at the pleural surface and at the interface with bronchi and vessels, whereas in patients with ARDS they are distributed through the lung parenchyma primarily at the interface between “healthy” and “diseased” lung regions, of different densities (Figure 2).
Figure 3. Frequency distribution of lung inhomogeneities in patients with acute respiratory distress syndrome (ARDS) and healthy control subjects. White bars represent healthy control subjects; black bars represent patients with ARDS. Statistical analysis was performed with two-way analysis of variance on ranked data using as classification factors the lung inhomogeneity class (0–1.61, 1.61–2, 2–2.5, >2.5) and the healthy/ARDS status of the subject; both the classification factors and the interaction were highly significant (P < 0.0001); we performed post hoc tests comparing healthy control subjects with patients with ARDS within each lung inhomogeneity class using the Bonferroni correction. *P < 0.05 healthy subjects versus patients with ARDS. Only 2 ± 1% of the lung volumes in patients with ARDS and 0.0008 ± 0.00052% of healthy subjects had lung inhomogeneities below 1.
[More] [Minimize]Table 1 shows the characteristics of the study population, divided into mild, moderate, and severe ARDS using the PaO2/FiO2 threshold defined in Berlin, where PaO2/FiO2 (17) was measured at standard PEEP of 5 cm H2O. As shown there was a progressive deterioration of gas exchange from mild to severe ARDS, whereas the anthropometric characteristic, the causes leading to ARDS, and respiratory mechanics were similar in the three subgroups. In Table 2 we report the main CT scan variables measured at 5 cm H2O PEEP in mild, moderate, and severe ARDS. As shown, from mild to severe ARDS there is a progressive decrease in well-aerated tissue (the baby lung) (22) associated with a significant increase of noninflated tissue, whereas the amount of overinflated and poorly inflated tissue were similar in the two groups. In Table 3 we report the average values of the lung inhomogeneities in mild, moderate, and severe ARDS; their extent; and their intensity. As shown, the mean values of lung inhomogeneities significantly increased from mild to severe ARDS. Interestingly, however, the intensity of the lung inhomogeneities was similar in mild, moderate, and severe ARDS. What significantly changed was the extent of the lung inhomogeneities, which significantly increased from mild to severe ARDS, from 14 to 23%. Therefore, we may estimate that abnormal lung inhomogeneities represent 14–23% of the lung volume and it is possible to speculate that in these fractions of lung volume the stress may be almost double the transpulmonary pressure applied to the whole lung (multiplication factor ∼1.88 ± 0.09).
| Mild ARDS (33 Patients) | Moderate ARDS (94 Patients) | Severe ARDS (21 Patients) | P Value | |
|---|---|---|---|---|
| Age, yr | 56.8 ± 16.4 | 60.0 ± 16.5 | 59.3 ± 16.8 | 0.64 |
| Female sex, no. (%) | 11 (33) | 30 (32) | 8 (38) | 0.85 |
| Body mass index, kg/m2 | 26.3 ± 4.7 | 25.6 ± 4.8 | 28.0 ± 9.2 | 0.19 |
| SAPS II | 39.2 ± 12.2 | 42.5 ± 15.6 | 45.2 ± 13.3 | 0.32 |
| Tidal volume, ml/kg | 8.6 ± 1.6 | 8.4 ± 1.5 | 8.2 ± 1.6 | 0.59 |
| Plateau pressure, cm H2O | 18.5 ± 3.1 | 18.9 ± 3.8 | 19.6 ± 4.3 | 0.61 |
| Respiratory system compliance, cm H2O | 45 ± 14 | 43 ± 15 | 40 ± 18 | 0.53 |
| Minute ventilation, L/min | 9.3 ± 2.6 | 9.0 ± 2.3 | 10.1 ± 2.4 | 0.12 |
| PaO2/FiO2 | 247 ± 38*† | 146 ± 28* | 73 ± 14 | <0.0001 |
| Fraction of inspired oxygen, % | 40 ± 7*† | 50 ± 11* | 80 ± 15 | <0.0001 |
| Paco2, mm Hg | 39.9 ± 6.2*† | 43.9 ± 8.5* | 49.4 ± 8.6 | <0.001 |
| Arterial pH | 7.402 ± 0.075 | 7.388 ± 0.076 | 7.362 ± 0.073 | 0.17 |
| Cause of lung ARDS, no. (%) | 0.14 | |||
| Pneumonia | 8 (5) | 42 (28) | 14 (9) | |
| Sepsis | 14 (9) | 25 (17) | 3 (2) | |
| Aspiration | 2 (1) | 8 (5) | 1 (1) | |
| Trauma | 3 (2) | 7 (5) | 0 (0) | |
| Other | 6 (4) | 12 (8) | 3 (2) |
| Mild ARDS | Moderate ARDS | Severe ARDS | P Value | |
|---|---|---|---|---|
| Total lung tissue, g | 1,255 ± 263* | 1,446 ± 479* | 1,957 ± 604 | <0.0001 |
| Total gas, ml | 1,355 ± 505 | 1,241 ± 681 | 1,087 ± 769 | 0.35 |
| Lung recruitability, % | 8 ± 7*† | 14 ± 10* | 20 ± 13 | <0.0001 |
| Not inflated tissue, % | 31 ± 13*† | 40 ± 15* | 49 ± 17 | <0.0001 |
| Poorly inflated tissue, % | 31 ± 11 | 32 ± 12 | 32 ± 13 | 0.87 |
| Well-inflated tissue, % | 38 ± 11*† | 27 ± 13* | 18 ± 14 | <0.0001 |
| Overinflated tissue, % | 0 ± 0 | 0 ± 1 | 0 ± 2 | 0.66 |
| PEEP (cm H2O) | Healthy Subjects | Mild ARDS | Moderate ARDS | Severe ARDS | P Value | |
|---|---|---|---|---|---|---|
| Average lung inhomogeneities | 5 | 1.34 ± 0.09* | 1.38 ± 0.12 | 1.45 ± 0.16 | 0.01 | |
| 15 | 1.31 ± 0.10 | 1.36 ± 0.14 | 1.42 ± 0.16 | 0.09 | ||
| 45 | 1.15 ± 0.05 | 1.27 ± 0.09* | 1.31 ± 0.13* | 1.40 ± 0.14 | <0.001 | |
| Extent of lung inhomogeneities | 5 | 0.14 ± 0.05*† | 0.18 ± 0.08* | 0.23 ± 0.10 | <0.001 | |
| 15 | 0.12 ± 0.04 | 0.16 ± 0.08 | 0.20 ± 0.09 | 0.01 | ||
| 45 | 0.05 ± 0.02 | 0.10 ± 0.04 | 0.12 ± 0.07 | 0.18 ± 0.08 | <0.0001 | |
| Intensity of lung inhomogeneities | 5 | 2.62 ± 0.22 | 2.57 ± 0.18 | 2.53 ± 0.21 | 0.22 | |
| 15 | 2.76 ± 0.26 | 2.63 ± 0.25 | 2.60 ± 0.27 | 0.08 | ||
| 45 | 2.26 ± 0.24 | 2.88 ± 0.26 | 2.77 ± 0.22 | 2.71 ± 0.32 | 0.04 |
Figure 4 shows the distribution of the abnormal lung inhomogeneities (with values >1.61) along the sternum–vertebral axis; as shown, the greater concentration of the lung inhomogeneities in mild and moderate ARDS is in the most dependent lung regions, whereas it shifts to the middle lung in severe ARDS. This regional distribution should correspond to regional distribution of the VILI. The distribution of the lung inhomogeneities along the apicocaudal axis, in contrast, was similar in the mild, moderate, and severe ARDS, although quantitatively increasing from mild to severe ARDS (data not shown).
Figure 4. Distribution of lung inhomogeneities (with values > 1.61) along the sternum–vertebral axis. Bar height represents the median of the lung inhomogeneity extent along 10 sternum–vertebral segments; error bars represent the third quartile of lung inhomogeneity extent. The first bar (light gray) represents control subjects, whereas the black, dark gray, and white bars represent mild, moderate, and severe patients with acute respiratory distress syndrome (ARDS). To compare the distribution between mild, moderate, and severe ARDS we fit a Gaussian function along the median points [f = y0 + a × exp(−0.5 × [(x − x0)/b]2)]; no statistics were performed on control subjects that do not exhibit a particular sternum–vertebral pattern. Function peak value was at sternum–vertebral level 6.31 (standard error 0.18) for severe ARDS, 7.98 (standard error 0.13) for moderate ARDS, and 8.84 (standard error 0.08) for mild ARDS. The three Gaussian distributions computed on medians were compared with one-way analysis of variance (P = 0.0007) and post hoc test was performed with the Bonferroni correction. The position of the peak of mild ARDS was significantly different from severe ARDS (P = 0.008), whereas moderate and severe ARDS peaks (P = 0.05) and moderate and mild ARDS (P = 1) were not different.
[More] [Minimize]We found several correlations between the extent of lung inhomogeneities and physiologic variables as PaO2/FiO2 (PaO2/FiO2 = 208 − 279 × lung inhomogeneity extent; P < 0.0001; r2 = 0.13), physiologic shunt (shunt fraction = 0.26 + 0.68 × lung inhomogeneity extent; P < 0.0001; r2 = 0.15; shunt fraction was available in 140 patients), and plateau pressure (plateau pressure [cm H2O] = 15 + 22 × lung inhomogeneity extent; P < 0.0001; r2 = 0.21). The most striking correlation, however, was with the dead space fraction (Figure 5) (lung inhomogeneity extent = 0.41 + 0.93 × Vd/Vt; r2 = 0.34; P < 0.0001; physiologic dead space fraction was available in 124 patients).
Figure 5. Relationship between lung inhomogeneity extent and physiologic dead space measured at positive end-expiratory pressure (PEEP) 5 cm H2O. Physiologic dead space (%) = 0.41 + 0.93 × lung inhomogeneity extent; r2 = 0.34; P < 0.0001. Physiologic dead space was available in 124 patients.
[More] [Minimize]The lung inhomogeneity extent was associated with several CT scan variables: we found a positive correlation with lung weight (lung inhomogeneity extent = 0.088 + 6.26 × 10−5 × lung weight [g]; r2 = 0.16; P < 0.0001) (see Figure E7 in the online supplement) and a weak but significant correlation with noninflated tissue (lung inhomogeneity extent = 0.14 + 0.092 × fraction of not inflated tissue; r2 = 0.03; P = 0.02) (see Figure E8), whereas we measured a negative correlation with the fraction of well-inflated tissue (lung inhomogeneity extent = 0.29 – 0.39 × well-inflated tissue [fraction]; r2 = 0.47; P < 0.0001) (Figure 6A). A positive correlation, however, was found between the lung inhomogeneities and the poorly aerated tissue (lung inhomogeneity extent = 0.06 + 0.39 × poorly inflated tissue [fraction]; r2 = 0.35; P < 0.0001), as shown in Figure 6B.
Figure 6. (A) Relationship between lung inhomogeneity extent and the fraction of well-inflated tissue (computed tomography [CT] value between −500 and −900 HU) measured at positive end-expiratory pressure (PEEP) 5 cm H2O. Well-inflated tissue (fraction) = 0.51- 1.25 × lung inhomogeneity extent; r2 = 0.47; P < 0.0001. (B) Relationship between lung inhomogeneity extent and the fraction of poorly inflated tissue (CT value between −100 and −500 HU) measured at PEEP 5 cm H2O. Poorly inflated tissue (fraction) = 0.15 + 0.91 × lung inhomogeneity extent; r2 = 0.35; P < 0.0001.
[More] [Minimize]The lung inhomogeneity pattern at airway pressure of 5, 15, and 45 cm H2O is summarized in Table 3. As shown, the average values of the lung inhomogeneities slightly but significantly decreased. Moreover, their extent significantly decreased by 18 ± 18% increasing PEEP from 5 to 15 cm H2O and by 30 ± 22% increasing the airway pressure from 5 cm H2O to a plateau pressure of 45 cm H2O airway pressure, suggesting a more homogeneous lung at higher airway pressures. The lung inhomogeneity extent at PEEP 15 cm H2O and at 45 cm H2O airway pressure was highly correlated with the lung inhomogeneities extent at PEEP 5 cm H2O (lung inhomogeneity extent PEEP 15 cm H2O = −0.0035 + 0.85 × lung inhomogeneity extent PEEP 5 cm H2O; r2 = 0.83; P < 0.0001; Figure 7A) (lung inhomogeneity extent 45 cm H2O airway pressure = 0.0073 + 0.65 × lung inhomogeneity extent PEEP 5 cm H2O; r2 = 0.59; P < 0.0001; Figure 7B).
Figure 7. (A) Relationship between the lung inhomogeneity extent determined at positive end-expiratory pressure (PEEP) 5 cm H2O and at PEEP 15 cm H2O. Lung inhomogeneity extent at PEEP 15 cm H2O = −0.0035 + 0.85 × lung inhomogeneity extent PEEP 5 cm H2O; r2 = 0.83; P < 0.0001. (B) Relationship between the lung inhomogeneity extent measured at 5 cm H2O and at 45 cm H2O end inspiration. Lung inhomogeneity extent at 45 cm H2O end inspiration = 0.0073 + 0.65 × lung inhomogeneity extent PEEP 5 cm H2O; r2 = 0.59; P < 0.0001.
[More] [Minimize]Interestingly, 15 patients (17%) increased instead of decreased their lung inhomogeneity extent while increasing PEEP from 5 to 15 cm H2O and 11 patients (8%) going from 5 to 45 cm H2O end-inspiratory pressure. This was primarily related to the behavior of the poorly aerated tissue: in the 11 patients in whom the lung inhomogeneities unexpectedly increased increasing the pressure from 5 to 45 cm H2O the poorly aerated tissue also increased by 12 ± 8%. In contrast, in the remaining 130 patients where the lung inhomogeneities decreased increasing pressure, the poorly aerated tissue also decreased by 7 ± 10%. The changes of the lung inhomogeneity extent and the poorly aerated fraction changes with PEEP were significantly correlated over the range of pressures we explored (Δ lung inhomogeneity extent [5 – 45 cm H2O] = −0.045 + 0.21 × Δ fraction of poorly inflated tissue [5 – 45 cm H2O]; r2 = 0.21; P < 0.0001) (see Figure E9).
We found that the recruitability, measured between 5 and 45 cm H2O, was not significantly related with the extent decrease of lung inhomogeneities from 5 to 45 cm H2O (P = 0.09) (see Figure E10). However, if we consider that the relationship between recruitability and lung inhomogeneities changes only in the patients in whom the poorly aerated tissue decreased (104 patients out of 141) we found that the relationship becomes significant (Δ lung inhomogeneity extent = 0.031 + 0.36 × lung recruitability; r2 = 0.39; P < 0.0001) (see Figure E11). In contrast, no correlation between recruitability and the lung inhomogeneity extent was observed in the patients (37 patients) in which the poorly aerated tissue increased while increasing pressure (P = 0.30) (see Figure E12).
The lung inhomogeneity extent was significantly higher in the 51 patients who died compared with the 97 patients who survived (20 ± 9 vs. 17 ± 7%; P = 0.01). We performed a multivariate backward logistic regression model to study the association between CT scan variables that were different between survivors and nonsurvivors (lung recruitability, well-inflated tissue, lung inhomogeneity extent). We added in the logistic regression the fractions of poorly inflated and noninflated tissues, not significantly different between survivors and nonsurvivors, which were considered markers of overall severity. Among the CT-derived variables entered in the regression the only one independently associated with outcome was the extent of the abnormal lung inhomogeneities (P = 0.65 by the Hosmer-Lemeshow goodness-of-fit statistics; C = 0.62). The odds ratio for each percent point increase of the lung inhomogeneity extent was 1.062 (95% confidence interval, 1.015–1.11).
In this study we attempted to quantify the inhomogeneities of the lung parenchyma whose prevention and correction represents the theoretical basis of the open lung strategy as modeled by Mead and coworkers (16) and proposed by Lachmann (23). Briefly, when a force is applied to a theoretical homogeneous lung, it is evenly distributed through the lung skeleton (the extracellular matrix), so that each fiber bears the same load (8). If the lung is not homogenous (e.g., because of the presence of local atelectasis or consolidation), the fibers neighbor to the unexpanded lung region must carry the additional load of the nonexpanding fibers, therefore locally increasing their stress and strain. Mead and coworkers (16) estimated that the ratio between the volume of two neighboring regions, one totally expanded and the other totally collapsed, is 10:1 at 30 cm H2O transpulmonary pressure, and the local pressure (stress) is 30 × (V1/V0)2/3 = 30 × (10/1)2/3 = ∼140 cm H2O being the two-third power the scale factor to pass from volumes to areas. In this study, as measure of the strain (i.e., distention) we used the gas/tissue ratio computed from CT scan density (24); if a lung region does not expand or expands less, the neighboring regions carrying their load are more strained (and stressed) and present a higher gas/tissue ratio.
Unfortunately, although the stress caused by material inhomogeneity may be quantified in the engineering field, it is impossible to measure it directly in the lung (8). In particular alveolar level information cannot be obtained from CT scan measurements. Therefore, it is possible that the association we found between our measurements and the overall ARDS severity are simply explained by the presence of more or less edema in different lung regions independently on local pressure multiplication. It is worth remembering, however, that the theory on which the lung protective strategy is based is the presence of stress raisers. Indeed, with these limitations in mind, we proceed in our discussion assuming that the underlying disease generates lung inhomogeneities, which, in turn, act as a stress raiser, further disrupting lung parenchyma.
Because the normal lung parenchyma is inhomogeneous due to airways, vessels, and the visceral pleura, we used 100 subjects as control group (19) to define the control inhomogeneity. In these control subjects we found that the average 95th percentile of inhomogeneity was 1.61. Accordingly, we defined as abnormal lung inhomogeneities the regions with value greater than 1.61 (intensity) and we computed the fraction of lung volume represented by these regions (extent). We found that the lung inhomogeneities increased with ARDS severity, were positively associated with dead space fraction and poorly aerated tissue, decreased when PEEP was increased, and were independently associated with outcome.
As shown in Table 3, the average values of the lung inhomogeneity intensity were similar in mild, moderate, and severe ARDS and did not change significantly increasing pressure from 5 to 45 cm H2O. In its theoretical model Mead and coworkers (16) computed the multiplication factor between open and closed regions as high as (10/1)2/3 = 4.64. In this study we found that this average multiplication factor was (2.62)2/3 = 1.9; this suggests that at the interfaces of interest the applied transpulmonary pressure should induce a stress almost double. Although the intensity of the lung inhomogeneities was similar in mild, moderate, and severe ARDS what actually changed was their extent. In normal animals the applied transpulmonary pressure becomes lethal when it approaches the total lung capacity (9); by inference, one may speculate that in humans an “innocent” applied transpulmonary pressure of 12 cm H2O may locally result in a harmful stress of 12 × 1.9 (intensity) = 22.8 cm H2O, which is in the range of total lung capacity of humans (25), and that this stress occurs in fractions of lung volume that sharply increase with ARDS severity. That the intensity of lung inhomogeneities was similar in mild, moderate, and severe ARDS may appear surprising. However, if we consider, for analogy, a pulmonary unit collapsed or consolidated the Po2 of the blood leaving this unit (the intensity of local hypoxemia) is obviously equal to the venous Po2 in mild, moderate, and severe ARDS. What determines the severity of the total hypoxemia is the number of units (extent), which generate different total shunt. Analogously, the stress raisers may be considered as local points that multiply the pressure at similar level (intensity), whereas the overall severity depends on the number of such points (extent).
The physiologic dead space is likely the strongest predictor of the ARDS outcome because it is related, more than oxygenation, to the structural changes of the lung (26, 27). Of note, the only CT scan–derived variable significantly associated with dead space fraction was the lung inhomogeneity extent. This relationship has a sound anatomic and physiologic rationale because the inhomogeneity, and the dead space, is likely related to structural abnormalities of the ventilated lung fraction as capillary and alveolar wall destruction, presence of bullae or blebs, and fibrotic reactions. These structural abnormalities may determine either the dead space fraction or the gas/tissue ratio inhomogeneities. Not surprisingly we found an inverse relationship between lung inhomogeneity extent and the well-aerated tissue that appears homogeneously inflated. In contrast, the fraction of the poorly aerated tissue is well associated with the extent of inhomogeneity ratio. This is quite interesting because, as far as we know, the pathophysiologic meaning of the poorly aerated tissue has never been fully understood in ARDS (but the obvious fact that this tissue is associated with less aeration). It is possible, however, that the poorly aerated tissue may be relevant in the development of VILI. Other factors, however, should be considered because the CT scan simply models the lung as “blood flow less” organ. The lung microcirculation, as an example, is prone to rupture depending on the interplay between trans-pulmonary pressure and hemodynamic conditions (28–30); this interaction is most marked in the most dependent lung regions initially and thereafter at the interphase between aerated and nonaerated lung tissue where lung inhomogeneities are prevalent. These findings, however, are not in contrast with the hypothesis that lung inhomogeneity may lead to pressure multiplication because the introduction of unphysiologic strain may lead to capillary lesions.
The potential benefit of PEEP in the framework of lung protective strategy is to maintain open, at end-expiration, the lung parenchyma that has been recruited during inspiration phase, thus decreasing the lung inhomogeneity and the VILI (15, 16, 23). Actually, we found that the lung inhomogeneities decreased increasing PEEP and, in addition, were significantly but weakly related with lung recruitability. In individual patients, however, recruitability, PEEP, and lung inhomogeneity pattern were not associated. A possible explanation is the behavior of the poorly aerated tissue. In fact, increasing pressure may lead to an absolute decrease of the poorly aerated tissue, a powerful source of inhomogeneity, or may lead to its increase, depending on the characteristics of the lung. If we consider only the patients in whom the poorly aerated tissue decreased with pressures, greater was the recruitability, greater was the decrease of the lung inhomogeneities when airway pressure was increased (see online supplement). These results suggest that, in the individual patient, PEEP may increase, not modify, or decrease the lung inhomogeneities, although in most patients, its favorable effects prevail.
Mechanical ventilation is the common therapy of ARDS and over the decades huge efforts have been devoted to decrease its potential harms. Within the possible risk factors as FiO2 (31–33), respiratory rate (34), tidal volume (12), and PEEP (35–37), in the last years the greatest attention had been paid to the last two variables. The accepted view is that low tidal volume and higher PEEP may protect the lung by decreasing the stress and strain of the parenchyma: the low tidal volume by lowering the lung stress and strain and the higher PEEP by decreasing the lung inhomogeneity. In this paper we found that a consistent fraction of the lung, depending on ARDS severity, may be exposed to a stress twofold the transpulmonary pressure applied. We do not know the intensity or the extent thresholds for the lung inhomogeneities to become lethal because their effects may depend on the applied pressure, the respiratory rate, and the time of exposure. We found that if the PEEP-maintained lung recruitment leads, as a net balance, to an increase of well-aerated tissue the lung inhomogeneities should also decrease. However, if the PEEP-maintained lung recruitment primarily results in a greater amount of poorly aerated tissue, paradoxically, the lung inhomogeneities may increase. It is also possible that the modest effect of PEEP on lung inhomogeneity and stress raisers is because the recruited units kept open by PEEP have greater elastance than the units already open. This could explain why the inhomogeneity stays relatively high despite that relevant positive pressure levels are applied to expand the lung. In summary, mechanical ventilation, in some patients, could lead to more harm after PEEP increase. Therefore, the complex interaction between poorly aerated and well-aerated tissue and the possible increased specific elastance of the recruited lung regions may account for the modest results obtained with PEEP in all the clinical studies performed so far (38–40). Despite all the limitations of this study, the data suggest that the estimate of the lung inhomogeneity may open possible new scenarios in understanding the complex interaction between VILI, PEEP, and mechanical ventilation.
| 1. | Plötz FB, Slutsky AS, van Vught AJ, Heijnen CJ. Ventilator-induced lung injury and multiple system organ failure: a critical review of facts and hypotheses. Intensive Care Med 2004;30:1865–1872. |
| 2. | Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323. |
| 3. | Brochard L, Mancebo J, Wysocki M, Lofaso F, Conti G, Rauss A, Simonneau G, Benito S, Gasparetto A, Lemaire F, et al. Noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease. N Engl J Med 1995;333:817–822. |
| 4. | 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. |
| 5. | Della Rocca G, Coccia C. Acute lung injury in thoracic surgery. Curr Opin Anaesthesiol 2013;26:40–46. |
| 6. | Futier E, Constantin J-M, Paugam-Burtz C, Pascal J, Eurin M, Neuschwander A, Marret E, Beaussier M, Gutton C, Lefrant J-Y, et al.; IMPROVE Study Group. A trial of intraoperative low-tidal-volume ventilation in abdominal surgery. N Engl J Med 2013;369:428–437. |
| 7. | Haefeli-Bleuer B, Weibel ER. Morphometry of the human pulmonary acinus. Anat Rec 1988;220:401–414. |
| 8. | Roan E, Waters CM. What do we know about mechanical strain in lung alveoli? Am J Physiol Lung Cell Mol Physiol 2011;301:L625–L635. |
| 9. | Protti A, Cressoni M, Santini A, Langer T, Mietto C, Febres D, Chierichetti M, Coppola S, Conte G, Gatti S, et al. Lung stress and strain during mechanical ventilation: any safe threshold? Am J Respir Crit Care Med 2011;183:1354–1362. |
| 10. | 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. |
| 11. | 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. |
| 12. | 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. |
| 13. | Hager DN, Krishnan JA, Hayden DL, Brower RG, ARDS Clinical Trials Network. 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. |
| 14. | Hernandez LA, Coker PJ, May S, Thompson AL, Parker JC. Mechanical ventilation increases microvascular permeability in oleic acid-injured lungs. J Appl Physiol (1985) 1990;69:2057–2061. |
| 15. | Terragni PP, Rosboch G, Tealdi A, Corno E, Menaldo E, Davini O, Gandini G, Herrmann P, Mascia L, Quintel M, et al. Tidal hyperinflation during low tidal volume ventilation in acute respiratory distress syndrome. Am J Respir Crit Care Med 2007;175:160–166. |
| 16. | Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970;28:596–608. |
| 17. | Ranieri VM, Rubenfeld GD, Thompson BT, Ferguson ND, Caldwell E, Fan E, Camporota L, Slutsky AS; ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012;307:2526–2533. |
| 18. | Cressoni M, Amini M, Cadringher P, Chiurazzi C, Febres D, Gallazzi E, Marino A, Brioni M, Menga F, Cigada I, et al. Quantification of stress raisers in ARDS. Crit Care 2013;17(Suppl 2):P105. |
| 19. | Cressoni M, Gallazzi E, Chiurazzi C, Marino A, Brioni M, Menga F, Cigada I, Amini M, Lemos A, Lazzerini M, et al. Limits of normality of quantitative thoracic CT analysis. Crit Care 2013;17:R93. |
| 20. | Gattinoni L, Pesenti A, Avalli L, Rossi F, Bombino M. Pressure-volume curve of total respiratory system in acute respiratory failure. Computed tomographic scan study. Am Rev Respir Dis 1987;136:730–736. |
| 21. | Team RDC. R: a language and environment for statistical computing. Vienna, Austria: 2010 [accessed 2012 Jan 2]. Available from: http://www.R-project.org/ |
| 22. | Gattinoni L, Pesenti A. The concept of “baby lung”. Intensive Care Med 2005;31:776–784. |
| 23. | Lachmann B. Open up the lung and keep the lung open. Intensive Care Med 1992;18:319–321. |
| 24. | Gattinoni L, Pesenti A, Bombino M, Baglioni S, Rivolta M, Rossi F, Rossi G, Fumagalli R, Marcolin R, Mascheroni D, et al. Relationships between lung computed tomographic density, gas exchange, and PEEP in acute respiratory failure. Anesthesiology 1988;69:824–832. |
| 25. | Agostoni E, Hyatt R. Static behaviour of the respiratory system. In: Handbook of physiology. Bethesda, MD: American Physiological Society; 1986. pp. 113–130. |
| 26. | Kallet RH. Measuring dead-space in acute lung injury. Minerva Anestesiol 2012;78:1297–1305. |
| 27. | Gattinoni L, Bombino M, Pelosi P, Lissoni A, Pesenti A, Fumagalli R, Tagliabue M. Lung structure and function in different stages of severe adult respiratory distress syndrome. JAMA 1994;271:1772–1779. |
| 28. | 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. |
| 29. | West JB, Mathieu-Costello O. Stress failure of pulmonary capillaries: role in lung and heart disease. Lancet 1992;340:762–767. |
| 30. | 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. |
| 31. | Barber RE, Lee J, Hamilton WK. Oxygen toxicity in man: a prospective study in patients with irreversible brain damage. N Engl J Med 1970;283:1478–1484. |
| 32. | Singer MM, Wright F, Stanley LK, Roe BB, Hamilton WK. Oxygen toxicity in man. A prospective study in patients after open-heart surgery. N Engl J Med 1970;283:1473–1478. |
| 33. | Gould VE, Tosco R, Wheelis RF, Gould NS, Kapanci Y. Oxygen pneumonitis in man. Ultrastructural observations on the development of alveolar lesions. Lab Invest 1972;26:499–508. |
| 34. | Conrad SA, Zhang S, Arnold TC, Scott LK, Carden DL. Protective effects of low respiratory frequency in experimental ventilator-associated lung injury. Crit Care Med 2005;33:835–840. |
| 35. | 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. |
| 36. | Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Oliveira R, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338:347–354. |
| 37. | Villar J, Kacmarek RM, Pérez-Méndez L, Aguirre-Jaime A. A high positive end-expiratory pressure, low tidal volume ventilatory strategy improves outcome in persistent acute respiratory distress syndrome: a randomized, controlled trial. Crit Care Med 2006;34:1311–1318. |
| 38. | Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A, Ancukiewicz M, Schoenfeld D, Thompson BT; National Heart, Lung, and Blood Institute ARDS Clinical Trials Network. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004;351:327–336. |
| 39. | Meade MO, Cook DJ, Guyatt GH, Slutsky AS, Arabi YM, Cooper DJ, Davies AR, Hand LE, Zhou Q, Thabane L, et al.; Lung Open Ventilation Study Investigators. Ventilation strategy using low tidal volumes, recruitment maneuvers, and high positive end-expiratory pressure for acute lung injury and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2008;299:637–645. |
| 40. | Briel M, Meade M, Mercat A, Brower RG, Talmor D, Walter SD, Slutsky AS, Pullenayegum E, Zhou Q, Cook D, et al. Higher vs lower positive end-expiratory pressure in patients with acute lung injury and acute respiratory distress syndrome: systematic review and meta-analysis. JAMA 2010;303:865–873. |
Supported in part by an Italian grant provided by Fondazione Fiera di Milano for Translational and Competitive Research (2007 to L.G.) and by the ESICM Bernhard Dräger Award for Advanced Treatment of Acute Respiratory Failure.
Author Contributions: Conception and design, L.G., P.C., and M.C. Acquisition of data, D.C., C.C., M.C., E.G., A.M., M.B., G.B., and M.Q. Analysis and interpretation of data, L.G., M.C., C.C., M.A., A.M., M.B., and E.C. Drafting of the manuscript, L.G., M.C., C.C., M.A., and A.M. Critical revision of the manuscript for important intellectual content, all authors. Statistical analysis, L.G., M.C., C.C., M.A., A.M., and E.C.
Originally Published in Press as DOI: 10.1164/rccm.201308-1567OC on November 21, 2013
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Author disclosures are available with the text of this article at www.atsjournals.org.
