We conducted a pilot feasibility study (clinicaltrials.gov NCT02703012) including 20 adult ICU patients ventilated in pressure-controlled mode with no spontaneous breathing activity. All patients had ARDS according to the Berlin Definition [14]. Exclusion criteria were severe hemodynamic instability, thoracic skin lesions, pregnancy, severe chronic obstructive pulmonary disease, esophageal pathologies, presence of cardiac pacemaker, duration of ARDS more than 72 h and inspired oxygen fraction (FiO2) of more than 80%. Informed consent was obtained from the patients’ legal representatives.
Measurements
The EIT device (PulmoVista 500, Dräger, Lübeck, Germany) was connected to the ventilator (Evita XL or V500, Dräger). Synchronized ventilator and EIT data were recorded at sampling rates of 50 Hz. Hemodynamic data, air flow, airway pressure (Paw), esophageal pressure as well as inspired and expired O2 and CO2 were additionally recorded with an S/5 monitoring system (Datex-Ohmeda, Helsinki, Finland) and stored electronically. The validity of esophageal pressure measurements was confirmed using an expiratory hold maneuver with gentle manual chest compressions. Cardiac output was assessed by transpulmonary thermodilution (PiCCO, Pulsion, München, Germany), where available.
Study procedure
Adjustment of ventilator settings according to the ARDS Network protocol and according to the EIT protocol was performed in sequential order without randomization. During the first 2 h of measurement, VT, respiratory rate (RR) and PEEP were adjusted according to the recommendations of the ARDS Network protocol with VT of 6 ml/kg predicted body weight (PBW) and PEEP setting according to the lower PEEP/FiO2 table of the ARMA trial [15]. Subsequently, an arterial blood gas (ABG) sample was taken and the first assessment of SDRVD, stress and strain was performed. Ventilator settings were then optimized according to the EIT-based protocol once every 30 min for a total of 4 h. At the end of the 4 h period, another assessment of SDRVD, stress and strain was performed.
EIT protocol
Recruitability was assessed using a sustained-inflation maneuver with Paw of 40 mbar applied for a duration of 40 s or until a decrease in systolic arterial pressure by more than 20% was observed, followed by a PEEP increase of 3 mbar.
Regional Crs was assessed by dividing the EIT image in four horizontal regions of interest (ROIs) and by multiplying global Crs with the relative tidal impedance change in each of the ROIs. For assessment of recruitability, we analyzed changes in regional Crs occurring after a sustained-inflation maneuver with Paw of 40 mbar followed by a PEEP increase of 3 mbar. If, in any of the four ROIs, a regional increase in Crs by more than 3% (normalized to global Crs) was identified following the sustained-inflation maneuver and PEEP increase, the recruitment maneuver was classified as “successful” and the higher PEEP level was kept. Alveolar cycling and overdistension were analyzed by halving inspiratory driving pressure (ΔP) for diagnostic purposes for about three consecutive breaths. If a reduction in regional Crs by more than 3% (normalized to global Crs) was observed in any of the four ROIs during ventilation with lower ΔP, this was interpreted as alveolar cycling, and PEEP was increased by 3 mbar. An increase in regional Crs in any ROI by more than 3% (normalized to global Crs) with lower ΔP was interpreted as overdistension.
In this case, VT was decreased by 1 ml/kg PBW provided this did not lead to severe acidosis (pH < 7.2). PEEP was decreased by 2 mbar if no recruitability and no alveolar cycling had been identified during the last 2 h. The details of the EIT protocol are presented in Fig. 1 and in the Additional file 1.
Assessment of ventilation delay, stress and strain
Starting at the set PEEP level, a low-flow pressure–volume maneuver with an inspiratory flow of 6 l/min and an inspiratory VT of 12 ml per kg PBW was performed to allow assessment of SDRVD as described by Muders and coworkers [11]. SDRVD was calculated offline by analyzing the EIT data obtained during the low-flow pressure–volume maneuver with the “Diagnostics” view of the PC version of PulmoVista 500 Software 1.2 (Dräger Medical, Lübeck, Germany).
Subsequently, FiO2 was increased by 10% and decreased to its original value after 10 min to allow calculation of end-expiratory lung volume (EELV) according to [16]. Total inspiratory lung volume (Vinsp) was calculated by adding VT to EELV: Vinsp = EELV + VT.
For assessment of FRC, we performed an expiratory release maneuver by setting PEEP to zero and allowing complete exhalation of inspired air to ambient pressure for a duration of 10 s. Expired volume during this maneuver (release volume, Vrelease) constitutes the difference between Vinsp and the relaxation volume of the respiratory system. It was then used to calculate release-derived FRC (FRCrelease) by subtracting Vrelease from Vinsp: FRCrelease = Vinsp–Vrelease. Subsequently, global lung strain was calculated as the ratio of Vrelease to FRC: Strainrelease = Vrelease/FRC.
This approach may lead to an underestimation of actual FRC because of alveolar derecruitment that may occur during complete exhalation to ambient pressure. Therefore, we additionally calculated recruitment-adjusted FRC (FRCrecr) by first calculating the assumed PEEP volume (VPEEP) by multiplying PEEP with global Crs (VPEEP = Crs * PEEP) and subsequently subtracting VPEEP from EELV: FRCrecr = EELV–VPEEP.
Recruitment-adjusted strain (strainrecr) was then calculated as the ratio of end-inspiratory lung volume to FRCrecr: Strainrecr = Vinsp/FRCrecr.
For assessment of airway plateau pressure and transpulmonary plateau pressure (Paw,plat; Ptp,plat), we performed an end-inspiratory airway occlusion of 3–4 s. Airway driving pressure (ΔPaw) was calculated as the difference between Paw,plat and PEEP: ΔPaw = Paw,plat–PEEP. Total end-expiratory transpulmonary pressure (Ptp,exp) was calculated as the difference between PEEP and end-expiratory esophageal pressure and transpulmonary driving pressure (ΔPTP) was calculated as difference between Ptp,plat and Ptp,exp: ΔPTP = Ptp,plat—Ptp,exp. Respiratory system elastance (Ers) and lung elastance (Elung) were calculated from the ratio of ΔPaw and ΔPTP to expired VT. Stress was calculated from Paw,plat multiplied with the ratio between Elung and Ers: Stress = Paw,plat * Elung/Ers.
Specific lung elastance (Elung,spec) was calculated as the ratio between end-inspiratory stress and strainrelease: Elung,spec = Stress/Strainrelease.
Tidal power was calculated as described by van der Staay and Chatburn [17], describing inspiratory mechanical power without the resistive portion and the energy which escapes to atmosphere during expiration: tidal power = 0.098 * RR * ΔPaw * VT/2 (with RR = respiratory rate per minute; VT = tidal volume in litres, ΔPaw = airway driving pressure in mbar).
Of note, ventilation delay, stress and strain were not used to optimize ventilator settings but as physiological endpoints only.
End points and statistical analysis
The primary end point was the number of patients with stress below 27 mbar and release-derived strain below 2.0 after 4 h of ventilation according to the EIT-based protocol. Secondary endpoints included changes in SDRVD, Crs, ΔPaw and PaO2/FiO2. As exploratory endpoints, we analyzed changes in lung compliance (Clung), ΔPTP, Ptp,exp, tidal power, recruitment-adjusted strain and cardiac output, where available. Statistical analysis was performed with GraphPad Prism 5.0 (GraphPad, LaJolla, USA). Normal distribution was assessed with Shapiro–Wilk test. Continuous variables are presented as mean ± standard deviation if normally distributed or as median [interquartile range, IQR] if not normally distributed. Comparisons were performed with two-sided paired t test or Wilcoxon matched-pairs test, as appropriate.