Introduction

Hypoxemic respiratory failure is the leading cause of intensive care unit (ICU) admission in COVID-19, the majority of subjects meeting Acute Respiratory Distress Syndrome criteria (C-ARDS) [1]. Initially, it was observed that many patients presented a disparity between well-preserved lung mechanics and severe hypoxemia [2] and 2 different phenotypes in C-ARDS were proposed, which should be managed with different ventilatory strategies [2]. However, this was not confirmed in posterior published data, remaining recommendations to treat C-ARDS accordingly ARDS ventilation evidence-based [3]. Several hypotheses were proposed to the wide range of respiratory system compliance (Crs) observed in many C-ARDS series, including hypoxemia due to impaired perfusion in patients with higher compliance or lungs with high recruitability and lower compliance [2].

Optimal positive end-expiratory pressure (PEEP) has been pursued [4] and the question of how to recognize patients that get benefit from higher PEEP levels has led to new technologies like Electrical Impedance Tomography (EIT), a bedside tool to monitor ventilation distribution, allowing PEEP titration to reduce both collapse and hyperdistention [5].

Airway driving pressure (dPaw) is a simple parameter to monitor on the ventilator and, when diminished with increased PEEP was associated with reduced mortality risk in ARDS [6]. In C-ARDS, a lower dPaw was associated with better survival [7, 8].

In the present study, we aim to describe the profile of PEEP-induced changes in dPaw during a PEEP adjustment procedure as aid for individualized protective ventilation, including a group where it was done together with an EIT monitor.

Methods

Patients

In this prospective observational physiologic study, adults patients admitted to the ICU of three hospitals with C-ARDS confirmed by positive nasopharyngeal polymerase chain reaction for SARS-CoV-2 and receiving invasive mechanical ventilation (MV) ≤ 48 h were analyzed. Patients with barotrauma assessed by computed tomography (CT), chronic pulmonary disease, and increased intracranial pressure were excluded.

Mechanical ventilation settings

After analgesia and sedation adjustment, all subjects were initially ventilated in volume-controlled ventilation, tidal volume of 6 mL/kg with constant inspiratory flow, plateau pressure ≤ 30 cmH2O, FiO2 and PEEP adjusted to keep SaO2 > 90% based on the ARDSNetwork table [9] and respiratory rate to maintain normal partial pressure of carbon dioxide (PaCO2). Fluids and vasopressors were provided to maintain mean arterial pressure above 60 mmHg and, neuromuscular blocking used to avoid ventilatory asynchronies.

PEEP adjustment procedure

After initial ventilatory settings, PEEP was reduced, 2 cmH2O every thirty seconds [10], from 20 until 6 cmH2O while dPaw was assessed in each step, and the lowest PEEP that minimized dPaw (PEEPmin_dPaw) was identified. The posterior PEEP adjustment was at the discretion of the clinical team responsible for patient care.

EIT assessment

In one of the hospitals, patients were investigated by EIT (Enlight 1800, Timpel, São Paulo, Brazil) during the PEEP adjustment procedure. Regional variations in impedance (∆Z) during ventilation, map the Vt distribution in the lung and creates a PEEP titration tool which was used to assess PEEP-induced pulmonary hyperdistention and collapse and its effects on dPaw during the PEEP adjustment procedure. The EIT optimal PEEP (PEEPEIT) was defined as the PEEP that represents the best compromise between hyperdistention and collapse estimated [5, 11].

Evaluation of dPaw vs PEEP curve profile

After the PEEP adjustment procedure, each dPaw vs PEEP curve was recorded and retrospectively classified into one of three categories according to the difference between the minimum dPaw [12] and the dPaw at the lowest (ΔdPlow) and highest (ΔdPhigh) PEEP [4]. If ΔdPlow < 0.2 × ΔdPhigh, the curve was classified as J-shaped; if ΔdPhigh < 0.2 × ΔdPlow, the curve was classified as inverted-J-shaped; otherwise, the curve was U-shaped.

Statistical analysis

Results are reported without imputation as mean (standard deviation), or count (percentage), after testing for normality using the Shapiro–Wilk test. One-way ANOVA was used for the comparison between the three groups. A Bonferroni-Holm post hoc test was applied to correct multiple testing. Hyperdistention and collapse curves at different PEEP levels were assessed by computing areas under the curves (AUCs) [13] by adding the areas under each pair of consecutive observations:

$${\text{AUC}} = \frac{1}{2}\mathop \sum \limits_{1}^{8} \left( {{\text{PEEP}}_{i + 1} - {\text{PEEP}}_{i} } \right) \times \left( {Y_{i + 1} + Y_{i} } \right),$$

where Y was the estimated hyperdistention or collapse. The AUCs were compared only between U-shaped and J-shaped PEEP vs dPaw groups, because just one patient with Inverted-J shape had EIT measurement.

Statistical analysis was performed in R (The R Foundation, Vienna, Austria), and a p < 0.05 was considered significant.

Results

Between Jul 27th, 2020, and Feb 24th, 2021, a total of 184 patients were included, and a PEEP adjustment procedure was performed before 48 h on invasive MV. Table 1 shows clinical characteristics in each curve profile dPaw vs PEEP. Patients with inverted J-Shaped dPaw versus PEEP profile presented significantly higher body mass index (BMI) (Table 1) and lower partial pressure of arterial oxygen and fraction of inspired oxygen ratio (PaO2/FiO2) and Crs at baseline (Table 2).

Table 1 Characteristics of patients with C-ARDS enrolled in the PEEP titration
Full size table
Table 2 Respiratory mechanics and EIT data
Full size table

Respiratory mechanics and PEEP titration

Based on the analysis of the dPaw vs PEEP profile, most of the COVID-19 patients (n = 126) exhibited a J-shaped dPaw vs PEEP profile with dPaw starting to increase for PEEPs ≥ 7.5 ± 1.9 cmH2O, only a few COVID-19 patients had mostly inverted-J profiles (n = 18), usually requiring higher levels of PEEP (PEEPmin_dPaw ranging from 14 to 20 cmH2O) (Table 2, Fig. 1). Only 21.7% of COVID-19 patients presented the U-shaped profile with the PEEPmin_dPaw ranging from 10 to 14 cmH2O.

Fig. 1
figure 1

Respiratory system mechanics associated with the percentage of collapse and hyperdistention at different levels of PEEP. In panels A, D, and G, data were obtained by electrical impedance tomography, where ● is the respiratory system compliance; Δ is the percentage of collapse and □ is the percentage of overdistension. Panels B, C, E, F, H, and I show the percentage change in driving pressure obtained by a mechanical ventilator for a representative patient (B, E, H) and all patients (C, F, I). Panels A–C correspond to the category of patients with J-shaped curves; panels D–F correspond to the category of patients with U-shaped curves, and panels G–I correspond to the category of patients with inverted J-shaped curves

Full size image

The J-shaped dPaw vs PEEP profile was associated with increased hyperdistention, and collapse reduction as PEEP increased and, in this group, PEEPmin_dPaw was lower than PEEP based on the ARDSNetwork table (Table 2). At the range of the PEEPmin_dPaw both hyperdistention and collapse were minimized independent of the dPaw vs PEEP profile (Table 2, Fig. 1).

Discussion

Our study interpreted the dPaw vs PEEP curve profile among C-ARDS patients. The main findings were: (1) 90% of C-ARDS-19 patients presented a J- or U-shaped dPaw vs PEEP curve profile usually requiring PEEPs < 12 cmH2O to minimize dPaw; (2) PEEPs > 15 cmH2O would be necessary in only 10% of C-ARDS, and those patients presented an inverted-J dPaw vs PEEP curve profile and higher BMI; and (3) PEEPmin_dPaw was associated with a reduction of both alveoli collapse and hyperdistention. All these patients averaged PaO2/FiO2 below 150 which there is evidence of benefit from using higher levels of PEEP in ARDS [14].

ARDS and C-ARDS are heterogeneous conditions with uncertainty about to set PEEP [2, 3] commonly based by oxygenation targets [9]. However in C-ARDS this strategy frequently resulted in worse lung mechanics [15], and cardiac output impairment [16].

Our EIT data and an experimental CT study [4] show that, at constant VT, dPaw and compliance respond to both hyperdistention and collapse. 126/184 of our patients presented a J-shaped curves, with the largest hyperdistention AUC, where increasing PEEP to improve oxygenation may not work. In U-shaped curves the balanced risk of collapse and hyperdistention was obtained with about 12 cmH2O PEEP. In these two groups, higher PEEPs would carry a greater risk of iatrogenesis. Finally, patients with an inverted-J-shaped required higher PEEPs to minimize dPaw and presented higher BMI and lower initial PaO2/FiO2 ratio. In the only patient with this profile on EIT, PEEP decreased collapsed areas without increasing hyperdistention up to 20 cmH2O. The interpretation of the PEEP with respiratory system mechanics or with the amount of recruitment and overdistension on EIT seems to give the same information.

At least one-third of patients were obese in C-ARDS different cohorts [3, 7, 8], even though the effect of obesity on respiratory mechanics is well known, a relationship between BMI and compliance has not been described as an explanation, at least in part, for the COVID-19 different phenotypes. Obesity reduce Crs with the major contribution coming from the lung and not the chest wall [17] in spite of no significant association between compliance and BMI has been detected in a large cohort study of C-ARDS [18]. Mezidi et al. comparing a group of obese vs non-obese in C-ARDS patients monitoring esophageal pressure in a decremental PEEP trial demonstrated a significant difference in PEEP level for the same transpulmonary driving pressure (∆PL) and dPaw [19]. ∆PL also did not enhance significant information concerning the prediction of outcome in ARDS patients compared to dPaw itself [20].

Limitations

The observational nature of this study is its major limitation, and although data were acquired prospectively, they were interrogated retrospectively. The heavy workload upon COVID-19 pandemic made impossible to perform a clinical trial comparing clinical outcomes considering the observed profiles. The small proportion of patients investigated with EIT did not allow an appropriate comparison between the two methods, but data suggest a similar result to obtain the best PEEP for protective ventilation with a much simpler bedside procedure.

Conclusion

The dPaw vs PEEP curve is a feasible method and provides individualized information. A range of compliance and PEEPmin_dPaw was observed in all 3 groups and its interpretation suggested that just in a minority of C-ARDS patients, higher PEEP improves compliance, and even in these cases, it appears that obesity, together with disease severity, determines this behavior. The overall influence of personalizing PEEP on clinical outcomes remains to be determined.