Mechanical power of ventilation and driving pressure: two undervalued parameters for pre extracorporeal membrane oxygenation ventilation and during daily management?

In 1998 Dreyfuss et al. [9] reported by application of an experimental animal model, that high inflation pressure due to ventilation with high tidal volumes results in increased pulmonary edema and was reduced by straps around the chest and abdomen [10]. Apart from the avoidance of high volume ventilation, the application of positive end exspiratory pressure (PEEP) of 10 cm H₂0 reduced peri-vascular and alveolar edema reflecting the high relevance between over-distension and low end-expiratory lung volume for ventilator-induced lung injury (VILI) [10, 11]. These findings were approved in large human clinical trials and are the basis of the current ARDS guidelines which subsequently recommend P_plat < 30 cmH₂O, PEEP > 5 cmH₂O and V_t 6 ml/kg of predicted body weight (BPW) [Fig. 1] [13]. However, the Lung safe study prospectively analyzed more than 29,000 patients in 50 countries and revealed sub-optimal application of lung protective ventilation [12]. Particularly, P_plat was considered in only 40% of the ARDS patients and of these only two-thirds were ventilated in a lung protective mode (tidal volume ≤ 8 ml/kg BPW and P_plat ≤ 30 cmH₂O) [12, 13]. Indeed, it is not surprising, that these single values might not be adequate for all ARDS patients. While some patients suffer severe carbon dioxide retention at V_t ventilation, others may be intolerant to high PEEP levels due to circulatory insufficiency or do not benefit from increased PEEP due to limited recruitability. Otherwise, inexactitude realization of lung protective ventilation might have serious consequences particularly for the patients with moderate-severe to severe ARDS. Lung lesions are distributed unequally and injured lung tissue or atelectasis coexists with aerated or normal lung tissue [14]. This is accompanied by a marked heterogeneity in ventilation. Particularly within the border areas between aerated and atelectatic regions up to four to five times increased stretching forces were suggested by a mathematical model application [15]. Subsequently, the potentially injurious ventilator settings were applied to a progressively smaller and more inhomogeneous “baby lung”.

The applied pressure to support the delivery of the V_t is defined as driving pressure, which represents the strain applied to the lung during each ventilatory cycle. Driving pressure comprises the difference between the airway pressure at the end of the inspiration (P_plateau) and PEEP [16‐18]. The quotient between V_t and driving pressure represents the static compliance of the respiratory system. Finally, the driving pressure reflects the V_t in relation to the compliance of the respiratory system (C_RS) which is associated with ARDS severity as it reflects the proportion of lung availability for ventilation. In patients suffering ARDS, C_RS was reported to be directly related to lung functional size [19‐21].

However, the clinical benefit of moving a mere V_t to a C_RS-based ventilation strategy is currently discussed. Amato et al. [22] suggested that the driving pressure was strongly associated with mortality and a decrease due to changed ventilator settings was associated with improved survival. Interestingly, this correlation was also persistent during lung protective ventilation. Recently Haudbourg et al. [19] reported that a driving pressure guided ventilation strategy with target levels between 12 and 14 cm H₂O required V_t adoptions in 90% of the patients. In contrast, earlier reports suggested no significant advantage of the driving pressure concept compared to the P_plateau in view of mortality [13, 23]. However, available data is very limited and based on retrospective and observational designs or with very limited encompassed patients. Moreover, the transpulmonary driving pressure (the difference between P_plateau minus PEEP and P_{esophageal-Plateau} minus P_{end-expiratory oesophageal}) which particularly includes chest wall elastance was reported to better reflect lung stress [16, 24]. Finally, VILI was suggested to be triggered by mechanical stress and strain which is determined by V_t and endexspiratory lung volume (corresponding to higher respiratory system elastance)—both parameters are represented by the driving pressure [25]. However, driving pressure is physiologically and mathematically coupled with V_t, elastance and subsequent disease severity [26, 27]. Therefore Goligher et al. [26] analyzed the relationship between the respiratory elastance and mortality for the higher and lower V_t strategy arms. The absolute risk reduction associated with a lower V_t ventilation strategy increased progressively with increased elastance. In conclusion, driving pressure should be monitored during daily routine practice in ARDS patients and critically evaluated for V_t reductions below 6 ml/kg PBW when exceeding 15 cm H₂O. Of course, the threshold is currently a matter of debate and remains to be evaluated within clinical trials. In this regard, clinicians have to be aware that PEEP changes might subsequent influence elastance (increase: overdistension; decrease: lung recruitment). Finally, clinical trials which evaluate the elastance based on very low V_t ventilation strategies potentially facilitated by extracorporeal CO₂ removal strategies are urgently needed to optimize lung protection in ARDS patients.

While the relevance of the static ventilation parameter including V_t, PEEP, P_plateau and driving pressure is well established, current increasing evidence suggests a relevant contribution of the dynamic ventilation parameter including respiratory rate, inspiratory and expiratory airflow to VILI (Fig. 2) [27, 28]. Subsequently, the concept of mechanical power defined as the product of respiratory rate and total inflation energy gained attention for ventilation monitoring. Inflation energy comprises three components: (i) the power to overcome airway resistance during gas movements; (ii) the power to inflate the lung and chest wall movements, and (iii) the power to overcome end-exspiratory pressure-related recoil of the lung and respiratory system (Fig. 3) [29]. When these parameters are multiplied by the respiratory rate, mechanical power applied to the respiratory system per minute results [24]. Particularly the respiratory rate is an undervalued parameter during clinical practice. However, recent evidence revealed two ventilation strategies (High V_t; P_Plateau 34 cm H₂O; driving pressure 29 cm H₂O versus RR 40 pb; P_Plateau 17 cm H₂O; driving pressure 9 cm H₂O) caused the same degree of lung lesion after 48 h which suggests that also increased respiratory rate might cause increase lung injury in a specific damaged area [28, 30].

Neto et al. [31] analyzed 8207 patients receiving invasive ventilation for at least 48 h hours and suggested that mechanical power in the second 25 h of ventilation might be independently associated with increased mortality of critically ill patients, a lower number of ventilator free days and survival at day 28. Concordantly, Umer et al. [32] reported that the cumulative exposure to higher intensities of mechanical ventilation was harmful and that a significant increase in the hazard of death was found to be associated with each daily increment in driving pressure and mechanical power. In contrast, Coppola et al. reported that mechanical power resulting from airway pressure and from transpulmonary pressure were assessed and not related to the outcome of ARDS patients [33]. However, both the normalization to compliance and to well-inflated tissue independently increased the intensive care mortality of 1.78 and 2.64 times for one unit increase [33].

An experimental animal setting confirmed that high mechanical power ventilation is associated with increased levels of interleukin 6 and amphiregulin expression and correlated well with diffuse alveolar damage score and club cell protein 16 extression [34]. Importantly, mechanical power combines the effects of different variables and changing of none variable may not necessarily protect the lungs, as it may increase mechanical power delivered to the lung [35]. In detail, a decreased V_t might not necessarily be translated into lung protection if respiratory rate is increased for compensation or PEEP increases may not be protective if not accompanied by declined driving pressure [30, 31, 35]. Particularly, the ARDS is very heterogenous and PEEP is regularly applied to reduce inhomogeneity. However, Maiolo et al. reported variable effects of increased airway pressure and reported an increased inhomogeneity of 20% in mild ARDS while this effect was less pronounced or even negative in severe ARDS [39]. In agreement, further evidence suggests worsened outcome due to high pressure recruitment or high PEEP levels [37].

However, the threshold of optimized mechanical power is currently a matter of debate. Based on the previous study by Neto et al. [31] 25 J/min may discriminate between higher and lower lung damage, and iso-mechanical power results in similar degree of lung damage independent of whether the reason was high V_t, respiratory rate or PEEP [28, 38]. This might be explained by an application of healthy animals but also due to exceeding the threshold causing maximal lung damage [28]. This hypothesis is in line with Cressoni et al. [39] reporting 12 J/min as threshold for mechanical power to induce VILI. Finally experimental and clinical trials to determine the optimal threshold for lung protective ventilation remain to be initiated.

The inhomogeneity of the lung is associated with inhomogeneous distribution of forces and obviously of mechanical power which might subsequent result in the prior lung dependent reason for the progression of VILI [40]. Indeed, differences in elasticity were suggested to concentrate the applied forces by doubling [41]. For a determined mechanical power, intensity of ventilation is increased in lung tissue with fewer ventilated areas and at the interface between lung areas with different mechanical properties [15, 42]. Apart from the dependence of the lung surface, VILI might also be impaired by the open/closed interfaces which were associated with increased [(18F)FDG] uptake and subsequent increased proportion of lung condition severity [40, 43]. Finally, this suggests the total area of the well-inflated lung tissue as well as the inhomogeneous poorly inflated or uninflated lung tissue important potential parameter for normalization of mechanical power [41]. Recently, Coppola et al. reported that normalization of mechanical power as well as respiratory system compliance were prior compared to the mere values in prediction of mortality [33]. Normalization was based on a whole lung CT scan at 5 cm H₂O of PEEP and performed after a recruitment maneuver and lung gas volume and amount of well-inflated tissue were computed [33, 41].

Whether an earlier time point for initiation of ECMO therapy might an option, is currently under investigation (NCT04341285). Indeed, after implementation of the veno-venous EMCO therapy a so called “lung rest” with limited inspiratory plateau pressure (P_plat) < 25 cm H₂O which is highly recommended by the ELSO is mostly feasible [6]. Further reductions in the P_plat below 20 cm H₂O were reported to be associated with fewer VILI and improved patient outcome [45, 47]. Ultra-protective- or even near-apneic –ventilation during ECMO support was reported to attenuate lung injury due to decreased V_t and driving pressure [45, 48]. Concordantly, Araos et al. [49] reported, that near-apneic ventilation caused histologic less lung injury compared to both, a non-protective and conventional ventilatory strategy in an experimental ARDS ECMO pig model. However, others detected no superiority of ultra-protective ventilation strategies during vvECMO [46, 50]. Particularly, near apneic ventilation strategies are associated with a risk of atelectasis with subsequent secondary infections and severe induction of ventilation/perfusion mismatch unless PEEP is appropriately increased to keep part of the lung open [51]. Additionally, ultra-protective ventilation requires deep sedation which is necessary to avoid patient—ventilator asynchrony. Which ventilation mode and subsequent lung unloading is necessary to secure recovery, healing and repair need to be determined by clinical trials.

An international multicenter prospective cohort study revealed, that most high ECMO volume centers prefer a “lung rest” pressure-controlled lung protective ventilation strategy [46]. In contrast, pre-ECMO ventilation intensity was less considered and mechanical power prior vvECMO implementation was with 26.1 ± 12.7 J/min particularly high [46]. This might be explained, that EMCO therapy is complex, labor-intensive, expensive and moreover a highly invasive procedure. Therefore some centers apply ECMO therapy as a kind of “last rescue” procedure for a severely hypoxemic population after several trials of optimal conventional ventilation, prone positioning and neuromuscular blockers have failed. However, increased duration of mechanical ventilation before ECMO therapy initiation might be associated with higher mortality rates after ECMO therapy. Recent logistic regression analyses revealed greater delay from endotracheal intubation to ECMO initiation is independently associated with 6 month mortality [17]. Otherwise, the mere duration of ventilation prior to ECMO implantation was not associated with increased mortality other reports [4]. More important seems to be the intensity of ventilation and particularly even short durations of high intensity ventilation might cause lung injury [32]. Subsequently, an individualized comprehensive twice daily analysis of ventilation parameters particularly mechanical power during the pre-ECMO period is highly recommended. If sufficient oxygenation and/or decarboxylation cannot be achieved with at least moderate ventilation intensity, a MP and/or DP limited ventilation strategy with subsequent lung unload by vvECMO should be critically evaluated.

The extended reference equation to calculate MP by Gattinoni et al. represents the most precise calculation [52, 53]. However, some variables like airway and tissue resistance or elastance of the respiratory system are complex to measure within the clinical routine setting [53]. Moreover, the application of this formula requires muscle relaxation and volume-controlled ventilation of the patients [53]. Meanwhile, several simplifications were developed for an application during the daily routine. For pressure-controlled ventilation two accurate equations were suggested, but require the knowledge of resistances and respiratory system compliance which are not determined within the daily routine [53‐55]. However, Becher et al. recently suggested a simplified equation for MP calculation in pressure-controlled ventilated patients and Chiumello et al. reported bedside calculations of MP during volume- and pressure-controlled mechanical ventilation [55, 56]. Although this equations might be associated with a small bias of overestimation, it seems to be accurate and easy applicable during the daily routine [53]. Finally, the equations might also be applicable to spontaneous ventilation, but studies of accuracy and potential simplifications are currently lacking [57]. The current major limitation is, that airway pressure, flow and esophageal pressure are affected counter-directionally by the action of the ventilator and the respiratory muscles [57‐59].

Another disadvantageous of MP is the requirement for normalization on well aerated lung tissue. However, most vvECMO centers conduct CT scans as standardized diagnostic procedure. Alternately, normalization might be based on the lung compliance with a decreased predictive performance compared to the well-aerated tissue and the acceptance of a potential collinearity between MP and compliance [33]. A MP and DP-based ventilation strategy is shown in Fig. 2. Extended and simplified MP equations for volume—and pressure-controlled ventilation are shown in Figs. 3 and 4.

The relevance of DP and MP for vvECMO indication is just beginning. The key question clinical trials should address is the time point at which a lung unload and vvECMO support is necessary to avoid further ventilator-induced lung injury e.g., is the prize of sufficient oxygenation reasonable? With other words, threshold levels for pre-defined equations in dependence of ventilation strategy have to be determined. If adaptions of the ventilator strategy do not result in decreased MP, vvECMO implantation should be re-checked critically. However, the relative relevance of the MP components on VILI are currently not completely clear. Indeed, mathematically, MP increases with the exponent of 2 of Vt, the exponent of 1.4 of the RR and the exponent of 1 of PEEP. Whether an optimal composition of these parameters might be preferable to reduce VILI should be evaluated during the pre-ECMO ventilation strategy [60, 61].

However, in the case of ARDS, the development of innovative trial design is generally associated with several challenges. First, molecular biology of ARDS is very heterogeneous between patients with considerably differences in biomarkers of key pathways including inflammation, coagulation, and alveolar epithelial injury [62‐64]. Therefore, a rigorously phenotyped patient large collective seems to be essential [62, 64]. Recently, the development of assay platforms based on protein-based enrichments strategy was suggested by Beitler et al. which might be most efficient to overcome this task [62, 65].

Second, outcome definition of ARDS in clinical trials is challenging [62]. Although mortality is frequently used in clinical trials as endpoint, ARDS-related risk of mortality differs considerably between patients and diseases. Moreover, endpoints other than mortality have to address death as a competing risk. However, this might be resolved by the application of a ranked composite score which compares each patient to all other patients encompassed in the study by vital status and subsequently, only if both patient in pair survive by the second to be analyzed outcome parameter. [62, 66, 67]

Third, VILI-related lung injury and effect of different ventilator strategies on VILI are not simply detected at bedside and separated form disease-related lung injury [68]. Frequently, global respiratory mechanics and the degree of ventilator support required were applied to characterize ARDS suffering patients [62]. However, particularly to separate disease from ventilator-related lung injury and to evaluate thresholds of DP and MP for ECMO indication, a standardized preclinical animal approach with an option of post-mortal lung injury characterization and associated biomarker and proteome analysis seems a reasonable approach. However, particularly lung compliance, airflow and VILI-induced inflammation seem to be highly species dependent [69]. In contrast to humans, most of the elastic recoil measured in intact mice can be attributed to the lung as chest wall and further thoracic structures are very compliant [69, 70]. Moreover, species dependent differences in inflammation and innate immunity were reported [71]. Exemplarily, Toll like receptor 4 from humans and mice recognize different lipopolysaccharide [72] and mice lack the CXCL8 gene coding IL8 [73, 74], a potent neutrophil chemotactic factor with a key role in the pathogenesis of ARDS. Although some this limitations might be overcome by the application of a large animal model, currently none of these models adequately reproduce the full characteristics of human ARDS. Therefore, subsequent bench to bedside approaches are indispensable.