Following the initial observation in animal models as early as 1967, a key paper in 1974 demonstrating that high inspiratory pressures caused lung injury [1] (Fig. 1), crystallized the recognition that mechanical ventilation itself can harm the lung. Experimental models were also instrumental in showing that cyclic lung re-opening and closing creates shear stress with alveolar injury, which could be prevented by the application of positive end-expiratory pressure (PEEP) [2]. The use of computed tomography showed the response to PEEP in terms of recruitment of lung volume, giving rise to the ‘baby lung’ concept in acute respiratory distress syndrome (ARDS) [3]. Experimental models were also helpful in determining that it was not inspiratory pressure per se, but rather over-distention that caused harm leading to the concept of ‘volutrauma’ [4]. These findings led to the ‘open lung’ concept [5], which argues that a ventilation approach that prioritizes maintenance of lung volume would avoid the lung injury resulting from shear stress. Also, the concept of ‘biotrauma’ was demonstrated in an experimental model [6], whereby high-stretch ventilation generates the release of inflammatory mediators from the lung, leading to systemic inflammation and distal organ injury, particularly in the presence of zero PEEP. Two years later, these findings were validated clinically in ARDS patients.

Fig. 1
figure 1

Contributions over time to the knowledge of how to apply protective ventilation

Full size image

In the early 1990s, these laboratory insights were followed up by observations that low tidal volume ventilation while disregarding the concurrent hypercapnia, seemed to improve outcome [7]. This concept of ‘permissive hypercapnia’ underlined the rationale of limiting intensity of mechanical ventilation and inspired the design of the pivotal ARDS Network ARMA trial [8]. In this trial, a ‘lung-protective’ strategy [tidal volume (VT) of 6 ml/kg and a plateau pressure of 25–30 cmH2O] was compared with traditional VT of 12 ml/kg. A low VT strategy reduced absolute mortality by 9%, ushering in the era of ‘lung protective’ ventilation in ARDS. It took roughly another 15 years before it was demonstrated that application of low VT also benefitted patients without ARDS.

Our understanding of what constitutes ‘protective ventilation’ continued to evolve. The clinical use of oesophageal balloon catheters allowed estimation of trans-pulmonary pressure by quantifying the contribution of extra-thoracic elements (chest wall, abdomen) to ventilation pressures. In a small physiologic trial in ARDS patients, the use of oesophageal pressures to set PEEP significantly improved oxygenation and compliance compared to a standard approach with a PEEP table [9]. The concept of lung ‘stress and strain’ followed, as trans-pulmonary pressure could be used to assess the force applied to the lung during mechanical ventilation — the so-called lung parenchymal stress. Lung strain refers to the associated deformation of the lung in response to the applied stress and is computed as the change in lung volume as a proportion of functional residual capacity (ΔV/FRC). A key insight is that injurious levels of stress and strain may occur even during application of standard ‘protective’ ventilatory settings. Bedside application of this concept can be done with the ‘stress index’, which is assessed by examining the pressure–time curve during a constant inspiratory flow [10]. Additionally, it was demonstrated that driving pressure (ΔP = plateau pressure – PEEP) was a more robust determinant of mortality risk than inspiratory pressure or VT, and that decreases in ΔP resulting from optimization of ventilator settings were associated with increased survival in a mediation analysis [11].

The optimal goals of setting of the ventilator continue to be refined as new insights emerge. The ‘open lung’ concept was challenged using an experimental model of ARDS, in which combination of low tidal volume and low PEEP with associated low driving pressure and alveolar collapse, resulted in a reduction of stress and lung injury [12]. This strategy of keeping the collapsed lung closed is also known as ‘permissive atelectasis’.

Taken together, these concepts move beyond standard protective ventilation settings toward individualized mechanical ventilation. By now it is clear that both static (tidal volume, PEEP, plateau pressure, trans-pulmonary driving pressure) and dynamic parameters (respiratory rate, inspiratory and expiratory airflow) are implicated in the pathophysiology of ventilator induced lung injury (VILI). This is reflected in the concept of ‘mechanical power’ which aims to quantify the total amount of energy delivered during mechanical ventilation, defined as the sum of the degree of strain per cycle and the duration of exposure. The accepted goal is now to impart the least mechanical power possible to mitigate the risk of VILI [13], while achieving adequate gas exchange. The vulnerability of lung tissue to mechanical stress and strain also plays a role. Since lungs behave as a viscoelastic system, the components of the extracellular matrix require time to adapt to each ventilator variable, suggesting that setting adjustments should preferably be done in small increments or decrements. Of note, these insights have evolved mostly from studies in patients with injured lungs. Thereby, the optimal settings for those with normal compliant lungs may differ from those with ARDS. Perhaps in line with this, is the observation in the ARDSnet tidal volume study, that there was no benefit of lower VT in those patients with the highest compliance.

As lung injury evolves, the insult inflicted by mechanical ventilation can change over time, requiring constant re-evaluation of settings. Closed-loop ventilation modes automatically adjust settings based on physiologic signals and patient activity, thus requiring fewer manual interventions while delivering safe ventilation [14].

Implications for clinical practice

As VILI importantly impacts outcome of mechanically ventilated patients, even in those without lung injury, it follows that those caring for the critically ill should apply protective ventilatory settings to the best of their abilities. This is feasible, without the need for advanced monitoring systems, but by appreciation of the concepts as summarized here. The field of protective ventilation will continue to evolve as we generate new insights in the laboratory and in the clinics. A promising novel innovation is the use of in silico computational models of critically ill ARDS patients [15], which most likely will inform us on how to further refine ventilator settings. In addition, such refinements can potentially be done around the clock, instead of only during ward rounds. Moreover, functional imaging modalities may be useful to help set the ventilator. Although robust data of benefit are still lacking, electrical impedance tomography, for instance, can provide imaging of regional air distribution in the lungs.

Our exciting journey continues, and we are not there yet!