Does the Intensity of Therapy Correspond to the Severity of Acute Respiratory Distress Syndrome (ARDS)? Authors: Domenico Nocera, Stefano Giovanazzi, Tommaso Pozzi, Valentina Ghidoni, Beatrice Donati, Giulia Catozzi, Rosanna D’Albo, Martina Caronna, Ilaria Grava, Gaetano Gazzè, Francesca Collino, Silvia Coppola, Simone Gattarello, Mattia Busana, Federica Romitti, Onnen Moerer, Michael Quintel, Luigi Camporota, Luciano Gattinoni Published: 2024 ABSTRACT Objectives: The intensity of respiratory treatment in acute respiratory distress syndrome (ARDS) is traditionally adjusted based on oxygenation severity, as defined by the mild, moderate, and severe Berlin classifications. However, ventilator-induced lung injury (VILI) is primarily determined by ventilator settings, namely tidal volume, respiratory rate, and positive end-expiratory pressure (PEEP). All these variables, along with respiratory elastance, are included in the concept of mechanical power. The aim of this study is to investigate whether applied mechanical power is proportional to oxygenation severity. Methods: We analyzed 291 ARDS patients (71 mild, 155 moderate, and 65 severe). We defined low, middle, and high mechanical power by dividing the entire population into tertiles with a similar number of patients. In each oxygenation class, we measured computed tomography (CT) anatomy, gas exchange, respiratory mechanics, mechanical power, and mortality rate. Results: ARDS severity was proportional to lung anatomy impairment, as defined by quantitative CT scans (i.e., lung volume and well-aerated tissue decreased across the ARDS classes, while respiratory elastance increased, as did mortality). Mechanical power, however, was similarly distributed across the severity classes, as the decrease in tidal volume in severe ARDS was offset by an increase in respiratory rate. Within each ARDS class, mortality increased from low to high mechanical power (roughly 1% for each J/min increase). Conclusions: Both lung severity and mechanical power independently impact mortality rates. It is tempting to speculate that ARDS severity primarily reflects the natural course of the disease, while mechanical power primarily reflects the risk of VILI. 1. Introduction Traditionally, respiratory treatment in ARDS (acute respiratory distress syndrome) has focused primarily on achieving and maintaining adequate oxygenation. Despite over 50 years [1] of evolving strategies, the intensity of respiratory support is still largely guided by the degree of oxygenation impairment. Recent ARDS guidelines from the ESICM (European Society of Intensive Care Medicine) [2] and the ATS (American Thoracic Society) [3] continue to recommend interventions based on the severity of oxygenation impairment (mild, moderate, or severe) with indications of strategies, such as high PEEP (positive end-expiratory pressure), prone positioning [4], and extracorporeal support [5,6], all based on these oxygenation thresholds. However, there is growing recognition that the potential harms of mechanical ventilation are more closely related to the management of CO2 (carbon dioxide) clearance and respiratory mechanics, rather than on the oxygenation levels alone. Strategies like permissive hypercapnia [7] and lung-protective ventilation [8] aim to minimize lung damage caused by mechanical factors. The risk of ventilator-induced lung injury (VILI) is largely influenced by the interaction between underlying lung pathology [9,10], and the consequent reduced “baby lung” size, the high elastance, and the ventilator settings chosen by the clinician [11]. This interaction can be summarized in the concept of “mechanical power” [12], which integrates the primary factors predisposing the lungs to injury (e.g., high elastance) with the ventilator settings (e.g., tidal volume, respiratory rate, and PEEP). If the degree of oxygenation impairment was directly proportional to the risk of VILI, as measured by mechanical power, tailoring the intensity of therapy based on oxygenation levels or mechanical power would yield similar outcomes. However, our recent study involving patients with moderate to severe ARDS found that the intensity of mechanical power was similar across different oxygenation severity categories. This led us to conclude that the risk of VILI is more closely linked to mechanical variables than to oxygenation levels. In the present study, we extended our analysis to include patients with mild ARDS, focusing on mortality rates. The primary hypothesis is that oxygenation impairment reflects the natural progression of ARDS and its associated mortality risk, while mechanical power, primarily dependent on ventilator settings, has a stronger influence on mortality across all severities of ARDS (mild, moderate, and severe). If this hypothesis holds true, it suggests that the choice of respiratory therapy should be based less on the severity of oxygenation impairment and more on the mechanics of the respiratory system. 2. Methods 2.1. Study Population The study population includes 291 patients with ARDS, enrolled between May 2003 and July 2024, from San Paolo University Hospital of Milan (Italy), University Medical Center Göttingen (Germany), Ospedale Maggiore Policlinico of Milan (Italy), Azienda Ospedaliera San Gerardo Monza (Italy), and Pontificia Universidad Catolica de Chile (Chile). Patients were classified based on their severity. Quantitative CT scan taken in standard conditions and respiratory and gas exchange variables collected within 24 h from entry were prerequisites for enrollment and study inclusion. Patient with respiratory failure due to COVID-19 were excluded from this analysis due to their prevalent vascular pathology in the early stage of the disease. This study is a retrospective analysis of the combined Milan and Göttingen databases. The ethics committee was notified and permission to use the data was granted (Göttingen Antragsnummer 14/12/12). 2.2. Study Design Population Stratification The patient population was stratified into the following groups:ARDS SeverityPF classes ARDS SeverityPF classes ARDS SeverityPF classes Using baseline records, we classified the patient population according to the Berlin definition into the following three severity groups: mild ARDS (P/F ratio between 201 and 300 mmHg, n = 71), moderate ARDS (P/F ratio between 101 and 200 mmHg, n = 155), and severe ARDS (P/F ratio ≤ 100 mmHg, n = 65). ARDS SeverityMP classes ARDS SeverityMP classes The mechanical power, expressed as J/min, was calculated using the following formula:(1)Mechanical Power=0.098∗RR∗∆V2∗12∗ ELrs+RR ∗ 1+I:E60∗I:E∗ Raw+∆V ∗ PEEP where 0.098 is the conversion factor from liters*cmH2O to joule, RR is the respiratory rate, TV is the tidal volume, PPeak is the peak pressure, and ∆PAW is the plateau pressure minus the positive end expiratory pressure. We measured the distribution of MP in the whole population (see Supplementary Figure S1), which was divided in three tertiles with similar numbers of patients. The tertiles limits were <16.1 J/min (low tertile), the second tertiles 16.1–21.8 J/min (middle tertile) and the third ≥ 21.9 J/min (high tertile). These thresholds were then applied in the ARDS severityPF classes to define how many patients were included in each MP threshold interval for each class. 2.3. Outcome Measurements All the variables were measured during the first 24 h after entry into the ICU. Respiratory and gas exchange variables were measured at baseline, i.e., at the ventilatory setting decided by the attending physician substantially in agreement with the protective lung strategy (lower tidal volume). The CT scan (computed tomography scan) variables were collected immediately afterwards, but in standard static conditions, at 5 and 45 cmH2O pressure. The following variables were collected:Gas exchange: Arterial PaO2 (partial pressure of arterial oxygen), PaCO2 (partial pressure of arterial carbon dioxide), PaO2/FiO2, and pH were measured, and the ventilatory ratio was calculated using the following equation: [minute ventilation (L/min) × PaCO2 (mmHg)]/(predicted body weight × 100 × 37.5 mmHg).Ventilatory setting: tidal volume/ideal body weight, minute ventilation, FiO2 (fraction of inspired oxygen), respiratory rate, and positive end expiratory pressure were recorded or calculated.Respiratory system mechanics: peak, plateau, and driving pressures (plateau pressure—PEEP) were measured. Respiratory system elastance was computed as the ratio driving pressure and tidal volume. The mechanical power was normalized to the total lung capacity (TLC).CT scan analysis and derived variables: CT images were processed using custom-designed software (Maluna2020®) to calculate various CT-derived variables, including the following: lung weight; total gas volume; overinflated tissue (ranging from −900 HU to −1000 HU); normally inflated tissue (ranging from −500 HU to −900 HU); poorly aerated tissue (ranging from −100 HU to −500 HU); non-aerated tissue (ranging from +100 HU to −100 HU); total gas volume was measured as the sum of tissue volume (total lung tissue) and total lung capacity. Gas exchange: Arterial PaO2 (partial pressure of arterial oxygen), PaCO2 (partial pressure of arterial carbon dioxide), PaO2/FiO2, and pH were measured, and the ventilatory ratio was calculated using the following equation: [minute ventilation (L/min) × PaCO2 (mmHg)]/(predicted body weight × 100 × 37.5 mmHg). Gas exchange: Arterial PaO2 (partial pressure of arterial oxygen), PaCO2 (partial pressure of arterial carbon dioxide), PaO2/FiO2, and pH were measured, and the ventilatory ratio was calculated using the following equation: [minute ventilation (L/min) × PaCO2 (mmHg)]/(predicted body weight × 100 × 37.5 mmHg). Ventilatory setting: tidal volume/ideal body weight, minute ventilation, FiO2 (fraction of inspired oxygen), respiratory rate, and positive end expiratory pressure were recorded or calculated. Ventilatory setting: tidal volume/ideal body weight, minute ventilation, FiO2 (fraction of inspired oxygen), respiratory rate, and positive end expiratory pressure were recorded or calculated. Respiratory system mechanics: peak, plateau, and driving pressures (plateau pressure—PEEP) were measured. Respiratory system elastance was computed as the ratio driving pressure and tidal volume. The mechanical power was normalized to the total lung capacity (TLC). Respiratory system mechanics: peak, plateau, and driving pressures (plateau pressure—PEEP) were measured. Respiratory system elastance was computed as the ratio driving pressure and tidal volume. The mechanical power was normalized to the total lung capacity (TLC). CT scan analysis and derived variables: CT images were processed using custom-designed software (Maluna2020®) to calculate various CT-derived variables, including the following: lung weight; total gas volume; overinflated tissue (ranging from −900 HU to −1000 HU); normally inflated tissue (ranging from −500 HU to −900 HU); poorly aerated tissue (ranging from −100 HU to −500 HU); non-aerated tissue (ranging from +100 HU to −100 HU); total gas volume was measured as the sum of tissue volume (total lung tissue) and total lung capacity. CT scan analysis and derived variables: CT images were processed using custom-designed software (Maluna2020®) to calculate various CT-derived variables, including the following: lung weight; total gas volume; overinflated tissue (ranging from −900 HU to −1000 HU); normally inflated tissue (ranging from −500 HU to −900 HU); poorly aerated tissue (ranging from −100 HU to −500 HU); non-aerated tissue (ranging from +100 HU to −100 HU); total gas volume was measured as the sum of tissue volume (total lung tissue) and total lung capacity. 2.4. Statistical Analysis Continuous variables are presented as means with standard deviations, while categorical variables are shown as percentages. Linear regression was used to assess the relationship between continuous variables (see Supplementary Materials). To compare groups, we used a one-way ANOVA or the Kruskal–Wallis test for continuous data, depending on suitability, with Bonferroni corrections applied. Mortality rates were analyzed using the Pearson’s chi-square test. Post hoc comparisons between pairs of groups were carried out using either Student’s t-tests or Wilcoxon tests, as appropriate. A p-value of less than 0.05 was considered statistically significant. All analyses were performed using RStudio (RStudio Team 2020; RStudio: Integrated Development for R, RStudio, PBC, Boston, MA, URL: http://www.rstudio.com/ (accessed on 20 October 2024)). 3. Results In Table 1, we report the demographic characteristics of the study population. As shown, patients with mild, moderate, and severe ARDS had similar ages, body weights, and overall illness severity, as assessed by SAPS II (Simplified Acute Physiology Score). The mortality rate significantly increased with increasing ARDS severityPF. In Table 2, we present the main lung CT anatomical and physiological differences observed in ARDS severityPF classes. As shown, the amount of lung edema, as assessed by lung weight, progressively increased with a decreasing PaO2/FiO2 ratio, while the size of the “baby lung”, assessed by total gas volume, progressively decreased. Non-aerated and well-inflated lung tissue followed similar trends to lung weight and gas volume, respectively, while poorly aerated and poorly inflated lung tissue showed no significant differences. At similar minute ventilation, PaCO2 and the ventilatory ratio increased when the PaO2/FiO2 ratio decreased. Notably, while the tidal volume pro kg was significantly lower throughout the three ARDS severityPF classes, the respiratory rate significantly increased; consequently, the MP was similar throughout the three classes, given the similar PEEP levels. In Figure 1, we display the distribution of mechanical power (MP) across the three ARDS severityPF classes, as defined by the Berlin criteria. The distribution of MP was similar, irrespective of P/F ratios. In fact, no significant correlation between MP and P/F ratio was found (p = 0.186, adjusted R2 < 0.01). Notably, within each class, MP distribution was similarly wide. No correlation was identified between the MP and P/F ratio pairs across the two variables. In Figure 2, we show the impact of MP on mortality within each of the three ARDS severityPF classes. As shown, an increase in MP was associated with a similar increase in mortality across all classes. Specifically, the increment in the mortality risk per unit increase in MP (measured in Joules/min) was 13%, 18%, and 12% in the mild, moderate, and severe ARDS severityPF classes, respectively. In the overall population, a 1-joule/min increase in MP roughly corresponded to a 1% increase in mortality. This figure illustrates that both ARDS severityPF classes and MP are associated with the mortality rate. A logistic regression analysis showed that these two variables were independently associated with outcomes (P/F p < 0.001; MP p = 0.016); see online supplement for details. Since MP is a composite variable encompassing factors related to ventilator-induced lung injury (VILI), including respiratory system elastance, tidal volume, respiratory rate, and PEEP, we analyzed MP by breaking it down into its main components within each ARDS severityPF class. The results are presented in Table 3. As shown, differences in MP within the ARDS severityPF classes were not primarily due to tidal volume (a cornerstone of lung-protective strategies) but were driven by differences in respiratory rate and PEEP. 4. Discussion In this study, we found that both ARDS severityPF classes and mechanical power were independently associated with patient outcome. For oxygenation, we maintained the traditional distinction between mild, moderate, and severe ARDS, following the thresholds commonly used in ARDS literature. For mechanical power, we arbitrarily divided the population into tertiles with similar numbers of patients. The thresholds we identified were low ≤ 16.1 J/min, middle between 16.2 and 21.8 J/min, and high ≥ 21.9 J/min. Interestingly, in a large database study where mechanical power was associated with mortality in 8207 patients, the threshold for increased mortality was 18 J/min [13]. This allows us to discuss both the characteristics of oxygenation impairment and mechanical power, as well as their interaction. 4.1. Oxygenation Impairment and ARDS Classes In this study, we found that the PaO2/FiO2 (P/F) ratio closely reflects the anatomical impairment across the three ARDS severityPF classes, particularly when comparing mild to moderate and severe ARDS. We observed fewer anatomical differences between moderate and severe ARDS. It is important to note that differences in the P/F ratio for similar anatomical impairment likely depend on the distribution of pulmonary blood flow to non-inflated tissue, which may be influenced by factors such as lung capillary collapse, microthrombosis, and the degree of hypoxic pulmonary vasoconstriction (HPV). Since our data were collected within 24 h of ARDS diagnosis, i.e., early in the disease course, it is reasonable to speculate that the relationship between P/F and anatomy primarily reflects the natural progression of the disease. At this early stage, patients treated with lung-protective strategies likely did not have sufficient time to develop extensive lung lesions. 4.2. Mechanical Power and VILI Mechanical power (MP) encompasses both anatomical lesions that reflect a prerequisite for VILI development, such as the small size of the “baby lung” and increased elastance, as well as ventilator settings chosen by the physician to maintain acceptable CO2 levels (typically below 60 mmHg). To achieve these goals while applying lung-protective strategies with low tidal volumes, the respiratory rate often must be increased. In epidemiological studies of randomized trials, MP was found to be associated with a greater risk of weaning failure [14] and morality [15] (with driving pressure and respiratory rate being the major components of this association) [15]. These data are presented for all ARDS classes without differentiating for severity. A different approach taking severity into consideration is by normalizing MP to static compliance of the respiratory system, which has been shown to be more closely related to outcomes in ARDS patients [16]. Our data suggest that this increase in the respiratory rate, along with a modest rise in PEEP to address oxygenation impairment, is the primary reason for the similar mechanical power values observed across different oxygenation classes. Indeed, the significant decrease in tidal volume from mild to severe oxygenation impairment is offset by significant increases in both PEEP and respiratory rate. As mechanical power is a summary variable, an increase in one component can be offset by a decrease in another. It is important to emphasize that, at least in experimental studies, the total mechanical power is what drives VILI, independent of the variations in individual components. In summary, mechanical power, more than oxygenation, is primarily driven by the need for CO2 clearance. 4.3. Interaction Between Oxygenation Classes and Mechanical Power Levels When we investigated the effects of mechanical power within each ARDS severity class, we observed a clear trend: mortality increased proportionally with the level of mechanical power applied. Notably, the difference in mortality within each ARDS oxygenation class—from the lowest to the highest mechanical power level—was roughly a 1% increase in mortality for every 1 J/min increment in mechanical power. It is important to stress that this increase is not detected by tidal volume alone but is strongly dependent on respiratory rate, a variable rarely considered when discussing the development and progression of VILI. 5. Study Limitations This study has several limitations. First, it is retrospective; however, it reflects the anatomical and physiological conditions at baseline before treatment. Second, it spans a wide timeframe, but we did not account for changes in therapeutic approaches over the past two decades. Our findings are representative of the standard approach, focused primarily on protective lung ventilation, where the goal was to maintain PaCO2 below 60 mmHg. 6. Conclusions and Future Perspectives Our findings suggest that respiratory therapy in ARDS should be tailored more to respiratory mechanics and CO2 clearance than to oxygenation levels alone. While the available evidence generally indicates that oxygenation impairment primarily reflects the anatomical and natural course of the disease, VILI is mainly driven by the ventilator settings chosen by physicians. We propose that a possible approach for patients admitted with ARDS should first involve assessing “baby lung” size (using imaging or gas dilution techniques), followed by the application of standard ventilation settings, such as 6 mL/kg tidal volume, 15 breaths per minute, 10 cmH2O of PEEP, and prone positioning. If, to achieve the desired PaCO2 and PaO2 levels, the mechanical power exceeds a threshold—tentatively 18 J/min—alternative forms of respiratory support, such as artificial organs, should be considered.