Introduction

In clinical practice the monitoring of airway pressure is used as a marker of lung stress, based on the assumption that airway pressure adequately reflects transpulmonary pressure (i.e. the difference between the airway and pleural pressure) [1]. However, as the ratio between chest wall and lung elastance is highly variable in patients with acute respiratory distress syndrome (ARDS) [1–3], the transpulmonary pressure may be variable at a constant airway pressure in these patients [higher or lower lung stress and ventilator-induced lung injury (VILI) for similar applied airway pressure] [4, 5].

Pleural pressure cannot be measured directly in the clinical setting, which has led to esophageal pressure being used as a reliable surrogate for more than 50 years [6–8]. Two strategies have been developed to compute transpulmonary pressure by the esophageal pressure technique. The first one computes the changes in airway and esophageal pressure during ventilation, estimating the fraction of pressure which distends the lung and chest wall, respectively [1, 4]. In this method, transpulmonary pressure is measured as the change in airway and esophageal pressure due to tidal inflation and positive end-expiratory pressure (PEEP) from end-inspiration to atmospheric pressure (i.e. release-derived transpulmonary pressure). In the second strategy, which is a partially modified method of the first, the transpulmonary pressure is calculated as the product of the end-inspiratory airway pressure and the ratio of lung to respiratory system elastance (i.e. elastance-derived transpulmonary pressure) [9, 10]. The ratio of lung and respiratory system elastance is estimated as the change in airway and esophageal pressure during a tidal volume inflation (from PEEP to end-inspiration). The elastance-derived method assumes that in each patient the changes in esophageal and airway pressure are linear during tidal volume inflation and by PEEP. However, previous studies have shown that this assumption is not always valid [3, 11–13].

Although the reduction of tidal volume is commonly indicated in ARDS, the optimal PEEP level in patients with ARDS has not been defined [14, 15]. Among the different criteria for bedside PEEP selection which have been proposed [14, 15], Talmor et al. found a better oxygenation and a trend in reduction in mortality by selecting PEEP according to the end-expiratory transpulmonary pressure [16]. In Talmor et al.’s study the transpulmonary pressure was computed as the difference in absolute airway and esophageal pressure (directly derived transpulmonary pressure), assuming that esophageal pressure reflects the effective pleural pressure [7, 17]. However, the validity of using directly measured esophageal pressure is questionable in supine mechanically ventilated patients because the value obtained can be falsely high due to the mediastinum weight, abdominal pressure and position of the catheter in the relevant part of the thorax resulting in very low transpulmonary pressure [18–20]. For these reasons several correction factors for directly measured esophageal pressure have been proposed [19, 21].

The aim of this study was to compare the different methods described in literature for estimating end-inspiratory and end-expiratory transpulmonary pressure in a population of patients with ARDS.

Materials and methods

Study population

Data on 44 patients affected by ARDS according to the “Berlin criteria” [22], who had previously been enrolled in a clinical trial, were analyzed [23]. Patients were classified into mild, moderate and severe ARDS according to the PaO2/FiO2 (partial pressure arterial oxygen/fraction of inspired oxygen) ratio obtained at a PEEP of 5 cmH2O [22].

Study design

Patients were sedated and paralyzed and maintained in supine position. The clinical characteristics of the patients, respiratory variables and ventilator settings were recorded before the study. All patients were ventilated in volume-controlled mode with a tidal volume of 6–8 mL/kg of predicted body weight throughout the study protocol. The oxygen fraction and respiratory rate were maintained unchanged for the entire study. Immediately before each step of the PEEP trial, a recruitment maneuver was performed to standardize the lung volume history [24]. The recruitment maneuver was performed with pressure control ventilation at a PEEP of 5 cmH2O, with a plateau pressure of 45 cmH2O, inspiration to expiration ratio of 1:1 and a respiratory rate of 10 bpm for 2 min [24]. After the recruitment maneuver, 5 and 15 cmH2O of PEEP were randomly applied. Respiratory mechanics and blood gas analyses were measured after 20 min at each PEEP level. At the end of the study, lungs were allowed to deflate from end-inspiratory pressure down to atmospheric pressure (i.e. at functional residual capacity). The patient was subsequently transported to the radiological department for a computed tomography (CT) scan.

Measurements

Respiratory mechanics

The respiratory flow rate was measured with a heated Fleisch pneumotachograph (Fleish 2, Fleisch, Switzerland). Volume was measured as integration of the flow. Airway pressure was measured proximally to the endotracheal tube with a dedicated pressure transducer (model MPX 2010 DP; Motorola, Schaumburg, IL). Esophageal pressure was measured with a radio-opaque balloon (SmartCath Bicore; GE Healthcare Life Sciences, Little Chalfont, UK) inflated with 1.0–1.5 mL of air connected to a pressure transducer. All traces were sampled at 100 Hz and processed on a dedicated data acquisition system (Colligo and Computo, www.elekton.it). The esophageal balloon was positioned in the lower third of the esophagus between a depth of 35–40 cm. A similar catheter was positioned in the stomach to measure the gastric pressure. The amount of gas in the balloon was periodically checked throughout the experiment.

During an airway occlusion the concordant changes of airway and esophageal pressure were verified to check the correct position of the balloon [4]. The static airway plateau pressure and esophageal pressure were measured at end-inspiration, end-expiration at PEEP and end-expiration after a release maneuver disconnecting the patient from the ventilator [4]. The occlusion was maintained until both airway and esophageal pressure decreased from a maximum to a plateau value [25]. The expired volume from PEEP to atmospheric pressure was the gained lung volume due to PEEP application. The gastric pressure was measured during the release maneuver. An example record of flow, airway and esophageal pressure is given in Fig. 1.

Fig. 1
figure 1

Tracings of flow, airway pressure (Paw) and esophageal (Pes) pressure during an end-inspiratory occlusion and a release maneuver in a representative patient. Pplat plateau pressure, Pmax maximum pressure

Full size image

Elastance

  • Elastance related to tidal volume Elastance of the respiratory system (Ers), lung (El) and chest wall (Ecw) during tidal inflation was calculated as the changes in airway, transpulmonary and esophageal pressure between end-inspiration and PEEP divided by the inspired tidal volume (Fig. 1) [25].

  • Elastance related to PEEP The Ers, El and Ecw due to PEEP were calculated as the changes in airway, transpulmonary and esophageal pressure between PEEP and atmospheric pressure divided by the expired volume from PEEP to achieve atmospheric pressure (Fig. 1) [25].

End-inspiratory transpulmonary pressure

  • Elastance-derived end-inspiratory transpulmonary pressure Elastance-derived end-inspiratory transpulmonary pressure was calculated as: Airway pressure at end-inspiration × El/Ers.

  • Release-derived end-inspiratory transpulmonary pressure Release-derived end-inspiratory transpulmonary pressure was calculated at: (Airway pressure at end-inspiration−atmospheric pressure)−(esophageal pressure at end-inspiration−esophageal pressure at atmospheric pressure) [4].

End-expiratory transpulmonary pressure

  • Release-derived end-expiratory transpulmonary pressure Release-derived end-expiratory transpulmonary pressure was calculated as: PEEP−(esophageal pressure at PEEP−esophageal pressure at atmospheric pressure) [4].

  • Directly measured end-expiratory transpulmonary pressure Directly measured end-expiratory transpulmonary pressure was calculated as: Airway pressure at PEEP−esophageal pressure at PEEP.

An implicit assumption of all transpulmonary pressure measurement methods is that pleural pressure at zero end-expiratory pressure (ZEEP) is zero. In fact, we defined the change in esophageal pressure between ZEEP and any end-inspiratory pressure as being equal to the airway pressure change multiplied by the ratio between chest wall elastance and respiratory system elastance; it follows that transpulmonary pressure at ZEEP results in zero as there is no airway pressure change. In reality, the absolute value of transpulmonary pressure at ZEEP is slightly negative. Consequently, both release-derived and elastance-derived transpulmonary pressures must be regarded as delta-transpulmonary pressures starting from the transpulmonary pressure at functional residual capacity and that the transpulmonary pressure at functional residual capacity can not be directly measured.

Lung CT and quantitative analysis

Each patient received two lung CT scans in static condition at 5 cmH2O of PEEP and 45 cmH2O of airway pressure, using the following parameters: 110 mAs, tube voltage of 120 kV, rotation time of 0.5 s, collimation at 128 × 0.6 mm, pitch of 0.85 and reconstruction matrix of 512 × 512. In each of the CT slices, lung profiles were manually delineated and analyzed with a dedicated software package (Soft-E-Film; www.elekton.it).

We assumed that lung parenchyma is composed of two compartments with very different density values: air [density −1,000 Hounsfield Units (HU)] and lung tissue with a density close to that of water (0 HU). Based on these assumptions it is possible to compute the fractions of gas and tissue in each voxel and, knowing the voxel volume, it is also possible to measure lung weight and lung gas volume [26]. Lung recruitability was computed as previously described [24].

Statistical analysis

Data are expressed as mean ± standard deviation (SD) or as the median with interquartile range. Comparisons were performed using Student’s t test when the variables were normally distributed or with the Mann–Whitney rank sum test when variables did not appear to be normally distributed on graphic inspection. The agreement between results was assessed using the Bland–Altman analysis [27] and linear regression. The bias was computed as the mean difference between the two methods and the limits of agreement as 1.96 SD. The percentage error was calculated as the SD of the bias divided by the mean [28]. Statistical significance was defined as P < 0.05. Analysis was performed with SAS 9.2 statistical software (SAS Institute Inc, Cary, NC) and SigmaPlot 11.0 (Systat, Chicago, IL).

Results

The main characteristics of the whole study population and of patients categorized according to the severity of ARDS are summarized in Table 1. Among the 44 patients with ARDS enrolled in our study, ten (22 %), 26 (59 %) and eight (18 %) presented mild, moderate and severe ARDS, respectively. The overall mortality in the intensive care unit (ICU) was 41 %. The baseline ventilator settings included a tidal volume of 6.6 ± 1.5 mL of predicted body weight and a respiratory rate of 16.7 ± 4.9 bpm. The directly measured esophageal pressure averaged 11.9 ± 4.1 cmH2O at atmospheric pressure without any difference according to the severity of ARDS.

Table 1 Baseline characteristics of the whole patient population
Full size table

End-inspiratory transpulmonary pressure

The average elastance and release-derived transpulmonary pressure were 14.4 ± 3.7 and 14.4 ± 3.8 cmH2O at 5 cmH2O of PEEP (P = 0.32) and 21.8 ± 5.1 and 21.8 ± 4.9 cmH2O at 15 cmH2O of PEEP, respectively (P = 0.98). These two parameters were significantly related (r 2 = 0.98, P < 0.001 at 5 cmH2O of PEEP; r 2 = 0.93, P < 0.001 at 15 cmH2O of PEEP) [Electronic Supplementary Material (ESM) Fig. E1]. In the Bland–Altman analysis the bias and agreement bands for the elastance and release-derived transpulmonary pressure were 0.06 (agreement bands −0.74 to 0.87) at 5 cmH2O and –0.02 at 15 cmH2O of PEEP (agreement bands −2.65 to 2.60) (Fig. 2). The percentage error was 5.6 and 12.0 %, respectively.

Fig. 2
figure 2

Bland–Altman analysis of end-inspiratory transpulmonary pressure computed by the elastance- and release-derived method at 5 cmH2O of positive end-expiratory pressure (PEEP) (upper panel) and 15 cmH2O of PEEP (lower panel). x-axis Mean of the two measurements, y-axis difference between the elastance and release method. Bias (limits of agreement): 0.06 (agreement bands −0.74 to 0.87) at PEEP 5 cmH2O and −0.02 (agreement bands −2.65 to 2.60) at PEEP 15 cmH2O. Percentage error: 5.6 % at PEEP 5 cmH2O and 12.0 % at PEEP 15 cmH2O. Green solid horizontal line Mean Bias, red broken horizontal line limits of agreement

Full size image

The mean ratio between lung elastance and respiratory system elastance due to tidal inflation and PEEP were 0.75 ± 0.11 and 0.75 ± 0.12 cmH2O at 5 cmH2O (P = 0.86) and 0.75 ± 0.15 and 0.75 ± 0.12 cmH2O at 15 cmH2O of PEEP, respectively (P = 0.87) (ESM Fig. E2). The ratio between lung elastance and respiratory system elastance due to tidal inflation and PEEP were related both at 5 and 15 cmH2O of PEEP (r 2 = 0.62, P < 0.00 at 5 cmH2O of PEEP; r 2 = 0.42, P < 0.001, at 15 cmH2O of PEEP) (Fig. 3 upper and lower panel).

Fig. 3
figure 3

Linear regression between the ratio of lung elastance and respiratory system elastance due to PEEP and tidal volume at 5 cmH2O (upper panel) and 15 cmH2O of PEEP (lower panel). Upper panel y = −0.212 + 1.025x, r 2 = 0.62 P < 0.001, lower panel y = 0.186 + 0.755x, r 2 = 0.42, P < 0.001

Full size image

End-expiratory transpulmonary pressure

The mean directly measured and release-derived transpulmonary pressure were −8.0 ± 3.8 and 3.9 ± 0.9 cmH2O at 5 cmH2O of PEEP and −1.2 ± 3.2 and 10.6 ± 2.2 cmH2O at 15 cmH2O of PEEP. These parameters were not related (r 2 = 0.070, P = 0.082 at 5 cmH2O of PEEP; r 2 = 0.10, P = 0.528 at 15 cmH2O of PEEP) (Fig. 4). The release-derived end expiratory transpulmonary pressure in the individual patients at 5 and 15 cmH2O of PEEP is shown in ESM Fig. E3.

Fig. 4
figure 4

Linear regression between the end-expiratory transpulmonary pressure computed by the directly measured and release-derived method at 5 cmH2O of PEEP (upper panel) and 15 cmH2O of PEEP (lower panel). Upper panel y = −3.498 −1.173x, r 2 = 0.07 P = 0.082; lower panel y = 0.266–0.138x, r 2 = 0.01, P = 0.528. Open circles Patients with primary ARDS, filled circles patients with secondary ARDS

Full size image

Directly measured esophageal pressure and lung function

The distribution of directly measured esophageal pressure at 5 cmH2O of PEEP in the whole population is shown in Fig. 5, which shows that this parameter was not significantly correlated with lung weight, lung recruitability, amount of not aerated lung tissue, hypoxemia, chest wall elastance and gastric pressure (ESM Figs. E4–E9).

Fig. 5
figure 5

Distribution (histogram) of directly measured esophageal pressure at 5 cmH2O of PEEP

Full size image

Discussion

The major findings of this study are that (1) the elastance- and release-derived end-inspiratory transpulmonary pressure were quite similar; (2) the release-derived and directly measured end-expiratory transpulmonary pressure were not related; (3) the directly measured esophageal pressure was not related to lung weight, amount of non-aerated tissue, chest wall elastance and gastric pressure.

In ARDS patients the mechanical ventilation should provide adequate gas exchange and minimize the VILI, reducing both lung stress at end-inspiration and alveolar collapse at end-expiration. In clinical practice, airway pressure is commonly used as a reliable marker of lung stress [1]. However, in their study of 50 patients with ARDS and 30 controls, Chiumello et al. [4] found that plateau pressure and tidal volume were inadequate surrogates for lung stress and strain as the ratio between lung elastance and respiratory system elastance was highly variable (range 0.33–0.95). As the “baby lung” size (i.e. the functional residual capacity) in patients with ARDS is also highly variable, it follows that for the same applied tidal volume the strain variability is large. Therefore, low or high tidal volume, such as 6 and 12 mL/kg, respectively, could produce similar stress and strain in a notable fraction of patients in each subgroup.

The impossibility to directly measure pleural pressure in clinical practice has led to the proposal that esophageal pressure be used as an adequate surrogate due the passive behavior of the esophagus [7, 18, 30, 31]. The esophageal lumen is separated from the pleural space by the muscular wall of the esophagus and mediastinal soft tissue. If all these tissues were flaccid, pressure changes in the pleural space would be transmitted unattenuated to the esophageal lumen. It follows that the changes in esophageal pressure should reflect the changes in pleural pressure. A study in dogs has shown that esophageal pressure reflects pleural pressure only at the level of the mid-lung [37]. It is also known that esophageal pressure can be influenced by the shape of the pressure–volume curve, lung volume, weight of the heart, reactivity of the esophageal smooth muscle wall and mechanical properties of the balloon [18]. However due to the possible difficulties in esophageal catheter positioning, interpretation of the data and the relatively paucity of available catheters, the measurements of esophageal pressure at bedside is seldom performed [32].

Elastance versus release-derived end-inspiratory transpulmonary pressure

The airway pressure applied to the respiratory system is used to inflate the lung and chest wall; thus, the elastance of the lung and chest wall determines how much of this pressure is used in each compartment [1]. Accordingly, the transpulmonary pressure is computed as the total difference in airway and esophageal pressure between end-inspiration and atmospheric pressure, thus taking into account the effect of PEEP and tidal volume. This method cannot measure the transpulmonary pressure at functional residual capacity since only delta-transpulmonary pressure can be computed. The pleural pressure at functional residual capacity is slightly negative (directed outwards) to counterbalance the retracting force of the lung and keep it expanded [1, 4].

A simplified method which computes transpulmonary pressure as the product of the changes in airway pressure during tidal inflation and the ratio between lung and respiratory system elastance has been introduced [9, 10]. This method assumes that the lung and respiratory system pressure–volume curves are linear in the range of PEEP and the tidal volume used in the clinical setting.

The computed values of transpulmonary pressure by the elastance and release methods were quite similar in our study, with a mean percentage of error of 5 and 12 %, respectively, which is clinically acceptable. This similarity can be due to a good relationship between the ratio of lung and respiratory system elastance as a result of a change in tidal volume and to PEEP, suggesting that in the range of PEEP and tidal volume used, the respiratory mechanics of the lung and chest wall are similar. In their published study, Chiumello et al. reported that a linear fitting was adequate to describe the pressure–volume curve in 75 % of patients with ALI/ARDS and in 78 % of control subjects (r 2 > 0.95), suggesting that the assumption of linearity of the pressure–volume curve is adequate in most patients [4]. We can conclude that transpulmonary pressure, which is used as a marker of end-inspiratory stress, can be satisfactorily estimated by the elastance method, which does not require any disconnection from the ventilator, thereby avoiding possible risks of lung derecruitment and hypoxemia due to the loss of PEEP.

Directly measured and release-derived end-expiratory transpulmonary pressure

In contrast to the reduction in tidal volume which has been extensively shown to reduce the mortality in patients with ARDS, several trials comparing high and low levels of PEEP have failed to find significant benefits of higher PEEP. Consequently, the optimum level of PEEP is still being debated [14, 33]. If the main beneficial effect of PEEP should be to keep the recruited lung open, an identical level of PEEP could be adequate in a given patient, whereas in a different patient it might not be sufficient to prevent alveolar collapse due to extremely variability in chest wall elastance among the subjects. Working under the assumption that directly measured esophageal pressure is similar to pleural pressure and reflects collapse of the lung, Talmor et al. showed better oxygenation and respiratory mechanics using a PEEP selected to minimize alveolar collapse at end-expiration by maintaining a directly measured transpulmonary pressure ranging between 0 and 10 cmH2O compared to PEEP selected according to gas exchange [16]. Unfortunately, Talmor et al.’s study did not use CT to demonstrate a reduction of lung collapse; rather they used the increase in oxygenation as the primary endpoint, which can be related to factors independent of lung recruitment (i.e. perfusion redistribution) [34].

In our study, directly measured and release-derived transpulmonary pressure were not related, and at both levels of PEEP tested the directly measured transpulmonary pressure were significantly lower than the release-derived transpulmonary pressure. The former was calculated as simply the difference between PEEP and esophageal pressure, with the assumption that the pleural pressure is equal to the esophageal pressure. However, this assumption is highly controversial due to the elastance of the esophageal catheter, tone of the esophageal wall, weight of the mediastinum organ and patient position, all of which could alter the measured value so that absolute esophageal pressure would not reflect pleural pressure [18, 35, 36]. Furthermore, in an experimental model the esophageal pressure was found to be nearly similar to that of the pleural space measured in the middle part of the lung, while it overestimated and underestimated the pleural pressure in the nondependent and dependent lung regions, respectively [37]. In contrast, the variations in pleural pressure were similar to those in esophageal pressure at each lung level, as previously observed by other investigators [7, 38].

When the PEEP selected accordingly to the directly measured and elastance-derived transpulmonary pressure was compared, the values differed by as much as 10 cmH2O in a given patient [39]. Several factors for correcting the directly measured esophageal pressure have been proposed with the aim of reducing such bias [19, 21]. However, Guerin et al., comparing the directly measured transpulmonary pressure corrected for a fixed value of 5 cmH2O to the release transpulmonary pressure, still found that the directly measured transpulmonary pressure was significantly lower and that the two parameters were not related [40]. Similarly, in a previous study we showed that setting PEEP equal to the directly measured esophageal pressure (i.e. to obtain a end-expiratory transpulmonary pressure equal to zero) was not related to lung recruitment [23].

In our study the directly measured esophageal pressure was highly variable between the subjects and was not related to lung weight, chest wall elastance and amount of lung collapse, suggesting that the use of directly measured transpulmonary pressure to set PEEP for avoiding alveolar collapse is highly questionable.

Conclusions

The elastance-derived method may be used instead of the release-derived method to estimate the end-inspiratory transpulmonary pressure, thereby avoiding ventilator disconnection, PEEP loss and derecruitment. The directly measured end-expiratory transpulmonary pressure is not related to the release-derived transpulmonary pressure. All methods of estimating transpulmonary pressures are based on assumptions; thus, the decision on which method to use should be guided by further clinical trials.