Background
Acute respiratory distress syndrome (ARDS) is characterized by the development of bilateral acute lung inflammation and edema as a consequence of direct or indirect injury [
1]. Although inflammation is diffuse, lung edema is not homogeneously distributed and its impact on the mechanical properties of each lung region depends on several factors, such as gravity [
2] or distance from the pleural surface [
3]. The alveolar distending pressure, i.e., the transpulmonary pressure, presents, therefore, an uneven distribution in the lungs. When transpulmonary pressure (P
L) becomes negative, terminal airways and alveoli tends to collapse. Given the regional heterogeneity of P
L, the threshold at which each lung unit opens and closes (i.e., regional P
L = 0 cmH
2O) does not correspond to a univocal pressure measured at airway opening [
4]. Previous studies hypothesized [
5] and measured [
6] how: (1) recruitment and derecruitment are distributed along the entire pressure–pressure–volume curve and depend from gravity, (2) opening pressures are higher than closing pressure and (3) every patient has unique distribution of opening and closing pressures [
6]. Temporal heterogeneity of ARDS can further increase the variety of these physiological mechanisms.
The pressure–volume curve (PV) of the respiratory system has been used in the last decades to describe the effect of increasing and decreasing pressure in the respiratory system in a quasi-static condition. Its shape during inspiration and expiration is different, since the pressure needed to inflate is higher than the one at which collapse happens (hysteresis), probably for the dynamic action of the surfactant layer and its role in reducing superficial tension [
7]: Indeed, hysteresis is absent when water is used to expand isolated animal lungs [
8]. The lower inflection point (LIP) of the inspiratory PV curve indicates the pressure for terminal airways or alveoli to open, while the LIP of the expiratory limb represents the closing pressure. Despite its sound physiological basis, the use of the PV curve built from pressure and volume measured at airway opening expresses only the average behavior of the lung and it is not able to highlight the heterogeneity of regional lung mechanics [
9].
Knowing the patient’s regional distribution of opening and closing pressures can help in treating physicians to set mechanical ventilation and potentially improve the comprehension of regional pathophysiology and therefore lung protection. It is well known how the cyclical opening and closing, also called atelectrauma [
10,
11], can amplify the inflammatory reaction in ARDS, being one of the main determinants of ventilator-induced lung injury (VILI). Similarly, regional hysteresis could represent a simplified method to assess potential for lung recruitment at the bedside. So far, lung CT scan was used to highlight the distribution of opening and closing pressure, but this technique is not feasible at the bedside and exposes the patients to ionizing radiations [
6].
Electrical impedance tomography is a radiation-free technique which has been increasingly used in the last decade to monitor ventilation [
12‐
15]. Regional inspiratory pressure–volume curves have been previously built in ARDS patients, by integrating pressure signals and EIT images [
9]. Theoretically, the integration of EIT and airway pressure signal would allow to generate the inspiratory and expiratory pressure–volume curves of different functional lung units at the bedside and determine the regional behavior of opening and closing pressures, of the magnitude of hysteresis and, finally, of the risk of atelectrauma. Consequently, our hypothesis was that we would be able to detect regional opening/closing pressures using electrical impedance tomography and evaluate their gravity-dependent regional distribution in patients affected by ARDS.
Discussion
In the current study, we measured regional opening and closing pressure from pixel-level PV curves obtained by electrical impedance tomography. We described how the pressure determining opening and closing of alveolar units, the intensity of atelectrauma (i.e., the magnitude of atelectrauma index) and the separation between inspiration and expiration due to hysteresis are gravity-dependent, with worse scenario for the dorsal lung. Moreover, all these measures showed large inter-patient variability, indicating the need of bedside monitoring to appreciate the patient’s own regional characteristics.
Ventilator-induced lung injury can worsen ARDS through several mechanisms [
19]. In ARDS, lungs are characterized by increased lung weight and surfactant dysfunction, leading to heterogeneous distribution of lung edema and atelectasis [
7,
20]. During tidal ventilation, if the opening pressure of a lung unit is reached, the unit will open; during expiration, when the closing pressure is surpassed, the unit will close again. As opening is associated with high stress caused by the passage of the air bubble on the epithelial cells [
21], this phenomenon of cyclic opening and closing (atelectrauma) is a main determinant of VILI. This phenomenon of cyclic opening/closing of lung units is heterogeneous. We confirmed that OP
pixel and CP
pixel have a Gaussian distribution, with higher values for OP. Opening and closing pressures were also higher in the dependent lung, underlying the role of gravity in the distribution of lung edema and transpulmonary pressure [
22]. We disclosed values of opening pressure beyond 30 cmH
2O, confirming that recruitment is a continuous phenomenon [
23] during tidal breath insufflation [
3,
5].
These findings confirm previous description obtained using the lung CT scan [
6] and, more recently, EIT in ALI patients [
24]. Moreover, the distribution on OP and CP differed from patient to patient (see Additional file
1:
online supplement), underlying the need of individualized therapy when applying mechanical ventilation. In this context, this study supports the possibility to assess these phenomena at the bedside, avoiding transport to the radiology department and exposition to ionizing radiation.
Lung hysteresis is a known phenomenon characterized by the presence of a different volume at the same pressure during inspiration and expiration [
25]. Several mechanisms have been proposed to justify this behavior, including surfactant effect [
8,
26] and stress relaxation. Lung hysteresis indicates that higher energy is required to open the lung that to keep it open and that the extra amount of energy is dissipated between inspiration and expiration into the system [
27]. We found that also hysteresis is heterogeneous, with higher values in the dependent lung. This is probably correlated with lower initial alveolar volume and greater volume excursion in the dependent lung [
28] where the major part of tidal recruitment is thought to happen and where the atelectrauma index was higher. HysMAX, moreover, showed a good correlation with mean opening pressure. All these data confirm that HysMAX during two low-flow PV maneuvers reflects the extent of alveolar opening and closing and thus the recruitability, as previously found by Demory et al. [
29] and suggested by Koefoed-Nielsen et al. [
30,
31]. Moreover, we found that this phenomenon happens more in the dependent lung, where the atelectrauma index showed a higher value and hysteresis was higher confirming classical view of where atelectrauma is thought to happen [
12].
Minimizing VILI during mechanical ventilation can be crucial to improve the outcome of ARDS. Until now, no available mean exists to detect the risk of atelectrauma in different regions of the lung at the bedside in ARDS, since the pressure–volume curve of the respiratory system can be characterized by overlapping information in such heterogeneous diseases [
9]. Positive end-expiratory pressure can counteract the tendency of dorsal lung collapse, but the mechanical information coming from the ventilator (e.g., driving pressure, stress index) contains averaged information from areas with different mechanical behaviors and therefore is not useful to highlight this phenomenon. We showed that by combining pressure/volume curves and EIT it is possible to determine opening/closing pressure at the bedside. Their distribution was highly variable between lung regions and from patient to patient (Additional file
1: figure S2), and therefore, by using EIT, it would be possible to furtherly individualize protective mechanical ventilation to limit regional atelectrauma, instead of using average global indexes like driving pressure.
This technique, applied at the bedside, may increase the pathophysiological information conveyed by EIT. Indeed, by evaluating the percentage of lung units opening and closing one can 1) quantify the maximum risk of exposure of that specific patient to atelectrauma and 2) select a positive end-expiratory pressure that could potentially guarantee recruitment and counteract derecruitment of both the non-dependent and the dependent lung regions.
Our study has several limitations: First, we started the PV maneuver at PEEP = 5 cmH2O and not from functional residual capacity; this was done because a reduction in PEEP below 5 cmH2O could expose the patients to excessive derecruitment and hypoxemia. Second, we analyzed a relatively small number of patients, none with severe ARDS. These findings must be confirmed therefore in a larger and more severe population. Third, we referred to atelectrauma as the pixels opening/closing between 5 and 40 cmH2O, in order to characterize the physiology of each patient. However, atelectrauma is classically defined as intratidal opening/closing of alveolar unit and the intratidal difference in pressure (Pplat/PEEP) is usually lower that the explored one (5–40 cmH2O). Forth, no image registration process was used to track the moving parenchyma, as done for the analysis of terminal elements (alveolar units) in CT scan. In EIT imaging, the image is reconstructed in a 2D matrix according to the thorax dimension, and therefore, the pixel dimension varies according to inflation/deflation. This could overcome, at least partially, the problem of image registration seen in fixed pixel-size imaging techniques (e.g., CT scan). Fifth, EIT do not cover the entire lung area but only the tissue around the belt position and the EIT pixel can be characterized by an intrinsic heterogeneity that could not be highlighted by the technique. Finally, the amount of recruitment/derecruitment can be influenced dynamically by time and it could be underestimated by the quasi-static punctual evaluation of the pressure–volume relationship.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.