Introduction

Mechanical ventilation strategies aimed at avoiding tidal recruitment/overdistension seem to minimize ventilator-associated lung injury both in experimental and clinical acute lung injury (ALI) [1, 2]. In this context, the analysis of the shape of the dynamic airway pressure (Paw) curve may allow detection of tidal recruitment/overdistension and, therefore, be useful for positive end-expiratory pressure (PEEP) titration [3–6].

When the inspiratory flow is constant, the concavity of the Paw curve can be described by a dimensionless coefficient known as “stress index”. A stress index > 1.1 indicates tidal overdistension (upward concavity), whereas values <0.9 indicate tidal recruitment (downward concavity). Accordingly, values of stress index ≈ 1 are associated with protective ventilation [1, 7, 8].

Another proposed index to detect tidal recruitment/overdistension is the so-called %E 2, which represents the percentage of non-linearity of the elastance of the respiratory system (Ers). It has been suggested that %E 2 < 0% is associated with tidal recruitment, whereas %E 2 > 30% is associated with tidal overdistension [4, 5].

Alternatively, the linear Ers may be also useful to guide PEEP titration [9–11]. It was claimed that PEEP at which Ers is minimal corresponds to a compromise between the distribution of aeration and mechanical stress [9, 10].

In the present study, we investigated the ability of the stress index, %E 2 as well as Ers for guiding PEEP titration during a decremental PEEP trial to minimize mechanical stress while improving lung aeration assessed by computed tomography (CT).

Materials and methods

The materials and methods are described in detail in the Electronic Supplementary Material (ESM) accompanying this article, and briefly summarized here.

Experimental protocols

Two protocols were performed in different institutions and approved by the respective local animal care Committees in Brazil and Germany. Twelve mixed-breed female Landrace/Large White pigs were used. In six of them, ALI was induced by oleic acid (oleic acid injury) in Brazil, whereas surfactant depletion (lavage injury) was used in other six animals in Germany. Animals were anesthetized and ventilated with FIO2 equal to 1.0, PEEP of 5 cmH2O, inspiratory/expiratory ratio of 1:2 and tidal volume (V T) of 7 ml/kg. Respiratory rates were adjusted to achieve a PaCO2 between 35 and 45 mmHg at baseline and maintained unchanged thereafter.

After 15-min stabilization, baseline measurements were obtained. Before induction of injury with oleic acid (non-injury, n = 6), PEEP was sequentially set at 16, 12, 8, 6, 4 and 0 cmH2O (5 min/level), while other settings were kept constant. Then, PEEP was set at 5 cmH2O and ALI induced by central venous injection of oleic acid in aliquots of 0.05 ml kg−1 until PaO2/FiO2 < 200 mmHg (oleic acid injury, n = 6).

Lavage injury was induced by repetitive lung lavage with warmed saline (lavage injury, n = 6) [12] until PaO2/FiO2 < 200 mmHg. In both models, injury was considered stable if PaO2/FiO2 did not improve within 30 min.

Following injury, a recruitment maneuver was performed (40 cmH2O constant Paw, 30 s) and PEEP sequentially set at 26, 20, 16, 12, 8, 6 and 0 cmH2O (5 min/level).

Assessment of linear respiratory system elastance and airway pressure curve profile analysis

Paw and airway flow were recorded in a computer with a sample rate of 200 samples/s. Thereafter, the Paw was fitted to:

  1. 1.

    A linear single-compartment model (Eq. 1), for the estimation of the Ers [13];

    $$ {\text{Paw}} = {\text{Ers}} \cdot V + {\text{Rrs}} \cdot \dot{V} + {\text{EEP}} $$
    (1)

    where Rrs is the resistance of the respiratory system, \( \dot{V} \) is the airway flow, V is the volume and EEP is the end-expiratory pressure.

  2. 2.

    A volume-dependent single-compartment model (Eq. 2), which is an extension of the model presented in the Eq. 1, for the estimation of the %E 2 (Eq. 3) [3, 4];

    $$ {\text{Paw}} = \left( {E_{1} + E_{2} \cdot V} \right) \cdot V + {\text{Rrs}} \cdot \dot{V} + {\text{EEP}} $$
    (2)
    $$ \% E_{2} = 100 \cdot {{\left( {E_{2} \cdot V_{\text{T}} } \right)} \mathord{\left/ {\vphantom {{\left( {E_{2} \cdot V_{\text{T}} } \right)} {\left( {E_{1} + E_{2} \cdot V_{\text{T}} } \right)}}} \right. \kern-\nulldelimiterspace} {\left( {E_{1} + E_{2} \cdot V_{\text{T}} } \right)}} $$
    (3)

    where E 1 and E 2 · V are the volume-independent and dependent components of the Ers, respectively.

  3. 3.

    An airway pressure–time model (Eq. 4), for the estimation of the stress index [4];

    $$ {\text{Paw}} = a \cdot t_{i}^{b} + c $$
    (4)

    where a, b and c are constants, and t i is the inspiratory time related to the constant part of the inspiratory flow. The coefficient b is known as stress index.

A stress index = 1.0 and %E 2 = 0 indicate a straight Paw curve profile. The PEEP at which the Ers was minimal, the stress index was close to 1 and %E 2 to 0% was considered the optimal PEEP.

CT scan procedure and image analysis

Helical CT scans were obtained between the base of the heart and the right dome of the diaphragm. Images were reconstructed yielding five to seven slices of 1 mm thickness.

Static CT scans were obtained in both groups at each PEEP during end-expiratory and end-inspiratory breath holds (15–20 s). To achieve comparability of CT scans among PEEP levels, a new topogram was acquired at each PEEP and the juxta-diaphragmatic region selected. For analysis, we chose one slice at end-expiration and another one at end-inspiration showing the highest anatomical matching. After manual delineation of the lung parenchyma, the slice volume as well as the volumes of hyperinflated, normally-, poorly-, and non-aerated regions were calculated [14]. The fractions of recruitment and hyperinflation induced by PEEP, as well as of tidal recruitment and hyperinflation were calculated as described elsewhere [15–17].

Statistical analysis

Data are expressed as median and ranges. Association between the stress index, %E 2 and CT-scan evidences of tidal recruitment/hyperinflation were determined by linear regression. The Wilcoxon signed rank test was used to compare changes in variables investigated at each level of PEEP. The Bonferroni–Holm procedure was used for the adjustment of multiple comparisons. A P < 0.05 was considered statistically significant in all tests.

Results

The induction of ALI impaired oxygenation and increased Ers (P < 0.05, Table ESM-1 and Table ESM-2).

Non-injury

Table 1 shows the data for non-injured lungs. The reduction of PEEP decreased the amount of hyperinflated areas and increased normally aerated ones (P < 0.05). PEEP < 12 cmH2O increased poorly aerated areas, while non-aerated ones increased at PEEP < 4 cmH2O (P < 0.05).

Table 1 Lung mechanics and aeration distribution in non-injury
Full size table

The Ers exhibited a flat minimum from PEEP of 8–4 cmH2O. In this range a balance between the maximization of normally aerated regions and the minimization of non-, poorly-, and hyper-aerated regions was observed (Table 1).

Reduction of PEEP progressively decreased tidal hyperinflation and reduced stress index and %E 2. At PEEP levels ranging from 12 to 8 cmH2O, the stress index and %E 2 were not associated with cyclic collapse/reopening or overdistension. At PEEP levels <8 cmH2O, the stress index was <0.9, %E 2 < 0% and tidal recruitment increased (P < 0.05). As shown in Fig. 1, the stress index and %E 2 correlated with tidal hyperinflation (r 2 = 0.69, P = 0.0001) and tidal recruitment (r 2 = 0.35, P = 0.0001).

Fig. 1
figure 1

Relationship between values of stress index and %E 2 and amount of tidal hyperinflation (left column) and tidal recruitment (right column) measured at all levels of positive end-expiratory pressure (PEEP) in non-injured animals

Full size image

Figure 2 (left) shows CT-scan images (Panel A), Ers (Panel B), stress index (Panel C), %E 2 (Panel D), fractions of hyperinflation and recruitment (Panel E), tidal hyperinflation and recruitment (Panel F) at each PEEP in all non-injured lungs. A focal pattern of loss of aeration was observed at PEEP ≤ 4 cmH2O, whereas a balance between hyperinflation and recruitment occurred at PEEP of 12 and 8 cmH2O. Tidal hyperinflation was predominant in non-dependent lung regions.

Fig. 2
figure 2

Distribution of aeration as a function of PEEP in non-injured (left panels), oleic acid (midpanels) and lavage (right panels) injured animals. a Regions of interest in the CT scan images obtained in a representative animal at each PEEP level: red hyperinflation, blue normal aeration, light gray poor aeration, dark gray non-aeration. Note the focal loss of aeration in dependent lung regions at lower PEEP and the hyperinflation of non-dependent regions throughout inspiration; b–d elastance of the respiratory system, %E 2 and stress index. Open and filled symbols represent individual and median values, respectively; e median fractions of hyperinflation (FHyper, open red squares) and recruitment (FRec, open black circles) at end-expiration, i.e., induced by PEEP. Red and black dots represent individual values of fractional hyperinflation and recruitment, respectively; f median tidal hyperinflation (Tidal Hyper, filled red squares) and recruitment (Tidal Rec, filled black circles) (ml). Red and black dots represent individual values of hyperinflation and recruitment, respectively

Full size image

Oleic acid and lavage injury

Tables 2 and 3 show data for oleic acid and lavage injury, respectively. Reduction of PEEP progressively decreased hyperinflated and normally aerated areas, while increasing poorly and non-aerated ones. Ers showed a flat minimum from the PEEP of 16–12 cmH2O and from 12 to 8 cmH2O, in oleic acid and lavage injury, respectively. PEEP lower than the PEEP at which Ers was minimal resulted in a significant increase of poorly and non-aerated areas, and decrease of normally aerated areas (Tables 2, 3). Oleic acid injury showed less normally aerated areas at PEEP < 6 cmH2O and less hyperinflated areas at PEEP > 20 cmH2O, when compared to lavage injury (P < 0.05).

Table 2 Lung mechanics and aeration distribution in oleic acid injury
Full size table
Table 3 Lung mechanics and aeration distribution in lavage injury
Full size table

The stress index and %E 2 achieved values higher than 1.0 and 0%, respectively, in almost all PEEP levels, independently of the ALI model. No significant changes in stress index and %E 2 could be detected when PEEP was reduced. In lavage injury, %E 2 was higher compared to oleic acid injury (P < 0.05). Higher amounts of hyperinflated areas were also observed in lavage injury (Tables 2, 3). In lavage injury, stress index and %E 2 were inversely correlated with tidal hyperinflation (r 2 = 0.1 with P = 0.05, and r 2 = 0.23, P = 0.001, respectively). No correlation was observed between stress index and tidal recruitment (r 2 = 0.008, P = 0.6), but the %E 2 was significantly correlated with tidal recruitment in the lavage model (r 2 = 0.26, P = 0.0006). In oleic acid injury, stress index and %E 2 were inversely correlated only with tidal recruitment (r 2 = 0.41, P < 0.0001 and r 2 = 0.73, P < 0.0001, respectively).

Figure 2 shows CT-scan images (Panel A), Ers (Panel B), stress index (Panel C), %E 2 (Panel D), the fraction of hyperinflation and recruitment (Panel E), tidal hyperinflation and recruitment (Panel F) at each PEEP in all studied animals with oleic acid and lavage injury (middle and left, respectively). A focal pattern of loss of aeration was observed in both ALI groups. PEEP led to recruitment in dependent regions and hyperinflation in non-dependent ones, simultaneously. Tidal recruitment/hyperinflation occurred simultaneously at all levels of PEEP in both ALI models. In oleic acid injury, tidal recruitment was higher, especially at PEEP lower than 20 cmH2O.

In the ESM, dynamic CT movies in representative animals show the simultaneous occurrence of tidal recruitment and hyperinflation in the same lungs at different PEEP levels. Stress index and %E 2 indicated only overdistension in all situations.

Stress index versus %E 2

The degree of correlation between the stress index and %E 2 was r = 0.93 (r 2 = 0.87, P < 0.0001) for non-injured lungs, r = 0.76 (r 2 = 0.58, P < 0.0001) for oleic acid injury and r = 0.36 (r 2 = 0.13, P = 0.018) for lavage injury. The previously proposed threshold %E 2 > 30% for overdistension [4, 5] did not match with stress index > 1.1 and CT-scan measurements of hyperinflation. Values of stress index ranging from 0.9 to 1.1 corresponded to %E 2 in the range of −10 to 10% (Figure ESM 4).

Discussion

The main findings of the present study were that during low V T strategy: (1) in non-injured lungs, PEEP > 12 cmH2O induced hyperinflation at end-expiration and during tidal breathing (tidal hyperinflation), and increased Ers, stress index and %E 2; (2) in injured lungs, PEEP simultaneously recruited dependent lung regions and hyperinflated non-dependent ones, but stress index and %E 2 suggested tidal overdistension; (3) PEEP at which Ers was minimal always resulted in the best compromise between recruitment and hyperinflation.

Non-injured lungs

Significant amounts of hyperinflation and tidal hyperinflation in non-dependent lung regions were observed at PEEP > 8 cmH2O. In agreement with our previous works [9], the hyperinflation induced by PEEP was associated with an increase in Ers. When animals were ventilated with PEEP < 6 cmH2O, the increase of poorly aerated areas was also associated with an increase of Ers. Interestingly, tidal recruitment of dependent regions occurred simultaneously with some degrees of tidal hyperinflation of non-dependent regions.

Despite the complex aeration patterns, both stress index and %E 2 indicated cyclic collapse/re-opening. This confirms previous observations that, in non-injured lungs, tidal recruitment is associated with cyclic closure/re-opening of airways [18, 19], particularly at low PEEP [19]. Furthermore, the stress index and %E 2 correlated well with tidal hyperinflation.

The PEEP at which Ers was minimal corresponded to the best compromise between recruitment and hyperinflation [9], and was close to the PEEP at which the stress index and %E 2 indicated a straight Paw curve.

Hyperinflation and tidal hyperinflation occurred even at zero PEEP, explained by an increase in alveolar air content and/or reduction in regional perfusion, and/or changes in regional transpulmonary pressure gradient, favoring the increase of aeration in non-dependent lung zones [20].

Thus, our results suggest that Ers, stress index and %E 2 could be useful to guide PEEP titration in non-injured lungs, for example during general anesthesia.

Acute lung injury

At low PEEP levels, a focal loss of aeration was observed in dependent lung regions in both ALI models. However, compared to oleic acid injury, lavage injury showed higher amounts of normally aerated areas, favoring higher amounts of hyperinflated areas at high PEEP.

The association between the amount of normally aerated areas at zero PEEP and the potential of hyperinflation of these areas when PEEP is applied has been previously described in patients with ALI/ARDS presenting with a focal pattern of loss of aeration [21]. In both ALI models PEEP induced recruitment of poorly and non-aerated areas simultaneously to hyperinflation of previously aerated regions as also reported in human ALI/ARDS [14, 22, 23].

Similar to non-injured lungs, the PEEP at which Ers was minimal corresponded to the best compromise between recruitment and hyperinflation. In agreement with previous findings [9–11], high PEEP increased hyperinflation, whereas low PEEP increased poorly and non-aerated areas and reduced the amount of normally aerated areas.

The CT analysis showed a complex aeration pattern in both ALI models, with tidal recruitment and hyperinflation occurring simultaneously. Interestingly, both tidal recruitment and tidal hyperinflation increased as PEEP was reduced. In lavage injury they occurred almost in the same proportion, whereas in oleic acid injury tidal recruitment overweighed tidal hyperinflation. Since oleic acid injury presented less amounts of normally aerated areas compared to lavage injury, it is possible that more recruitment occurred during inspiration.

The finding that in injured lungs the stress index was uniformly higher than 1.0 and %E 2 higher than 0%, irrespective of PEEP level, is striking. Indices derived from the Paw curve reflect global Ers variation through inspiration. Since both ALI models showed a focal loss of aeration at low PEEP, hyperinflation of the more compliant non-dependent regions and recruitment of dependent regions occurred simultaneously. Furthermore, tidal hyperinflation seems to be associated with alveolar overdistension [21, 24, 25], resulting in progressive Ers increase [9, 10]. On the other hand, a similar increase in Ers can be expected during tidal recruitment of previously poorly and non-aerated areas.

By increasing PEEP, an almost linear increase in hyperinflation occurred and was associated with an almost linear decrease in poorly/non-aerated tissue. Similarly, increase of PEEP led to a progressive decrease in tidal recruitment and hyperinflation. Thus, we were unable to find a level of PEEP at which the recruitment was maximized and no hyperinflation occurred.

Our findings partially differ from those by Grasso et al. [26], who suggested that stress index is able to detect tidal recruitment/hyperinflation in experimental ALI. The most likely explanation is that we used low V T (7 ml kg−1), whereas Grasso et al. worked with high V T (12 to 18 ml kg−1). When high V T is used, a clear predominance of tidal recruitment, at low levels of PEEP, or hyperinflation, at high levels of PEEP, is possible. Our findings are supported by Terragni et al. [27], who proposed that simultaneous recruitment and hyperinflation may explain the relatively low specificity of the concavity of the Paw curve in detecting VILI. In addition, we computed tidal recruitment considering the tidal variation of poorly and non-aerated areas, whereas Grasso et al. [26] considered only the variations of non-aerated areas. It is possible that higher amount of hyperinflation than recruitment occurs when a recruitment maneuver is not performed [28]. However, as shown in Tables 2 and 3 the exclusion of poorly aerated areas computation would not change our overall main findings.

Limitations

First, the use of different experimental facilities could have influenced our results. However, measurements followed standard international procedures and were consistent among different ALI models, using identical CT-scan configurations as well as mechanical ventilation settings. Second, similar to others [26] we used only one juxta-diaphragmatic CT slice, making it difficult to compare the same areas at different pressures. However, we selected images with maximal anatomical-landmarks coincidence, and a new topogram was acquired at each step. Third, the thresholds for CT compartments we used likely resulted in underestimation of hyperinflation in ALI, but they correspond to those ones internationally recommended [21, 29–32]. Fourth, since measurements were taken during breath hold, it is possible that a redistribution of aeration occurred (pendelluft), but its effect on our measurements are likely negligible [33]. Fifth, it could be argumented that time intervals of 5 min at each PEEP level were not enough to allow stable measurements. However, recruitment and derecruitment become stable as fast as 4 s in both the oleic acid and surfactant depletion models of ALI [34]. Based on this, we believe that complete stabilization of lung compartments should have occurred at the end of each PEEP step. Sixth, our results apply only for ALI with focal loss of aeration. Thus, we cannot exclude that stress index and %E 2 may be useful to guide PEEP titration in presence of diffuse loss of lung aeration.

Conclusions

The PEEP at which Ers was minimal corresponded to the best compromise between recruitment/overdistension in non-injured and injured lungs. Stress index and %E 2 were able to identify tidal recruitment/overdistension only in non-injured lungs. Our results suggest that Ers can be superior to the stress index and %E 2 to guide PEEP titration in presence of focal loss of lung aeration.