Introduction

Artificial positive pressure ventilation is a non-physiological condition potentially leading to deleterious effects on lung tissue, known as ventilator-induced lung injury (VILI).^1,2 Numerous strategies have been undertaken to optimise standard ventilation methods, especially by decreasing driving pressure [ΔP = peak pressure (P_peak) − positive end-expiratory pressure (PEEP)]³ and adjusting PEEP to lower the risk of VILI. VILI is even more likely in lungs with an inhomogeneous structure, as often observed in acute respiratory distress syndrome (ARDS).⁴ An inherent problem of artificial ventilation is that ventilator settings based on simply predicted body weight are not sufficient to cope with patient-to-patient variability even when a lung-protective low tidal volume (V_T) ventilation strategy is applied.⁵

Flow-controlled ventilation (FCV) is a novel ventilation mode that could overcome some well-known difficulties of artificial ventilation.⁶ It provides continuous, stable and identical gas flow rates during both inspiration and expiration, establishing slow and steady changes of intrapulmonary pressure without any plateau phase: these stable flow rates may influence haemodynamics in a positive way. Precise inspiratory flow (and thereby volumetry) is achieved by mass flowmeters, whereas control of expiratory flow is established by the combination of high resistance to outflow by the respiratory circuit and adapted, jet-flow generated suction by using a so-called ejector (Bernoulli's principle). With identical ventilator settings FCV has already been shown to improve CO₂ removal and oxygenation compared with volume-controlled ventilation (VCV).^7–10 A unique feature of FCV is the measurement of intratracheal pressures, which allows ventilator settings to be individualised according to the patient's dynamic compliance. V_T can therefore be set closely within lung mechanic limits by adjustment of PEEP and P_peak to achieve the highest dynamic compliance. This compliance-guided FCV approach has been shown to improve ventilation efficiency and oxygenation compared with pressure-controlled ventilation (PCV) in lung healthy animals.⁶ However, the effects of an individualised ventilation approach in ARDS and possible concomitant haemodynamic effects have not been studied in detail so far.

The study objective was to investigate the effects of compliance-guided FCV compared with best clinical practice PCV on gas exchange and haemodynamic parameters using an established oleic-acid induced ARDS model in pigs, which not only accurately mimics diffusion impairment but also lung mechanic deterioration such as reduced lung compliance.

Methods

Ethics

The current study was approved (protocol no.: BMBWF-66.011/0105-V/3b/2019) by the Institutional Animal Care and Use Committee of the Medical University of Innsbruck, Innsbruck, Austria (Chairperson Dr A. Beierfuss) on 13 May 2019 and the Austrian Ministry of Science, Research and Economy, Vienna, Austria (Chairperson Dr S. Bader) on 1 July 2019 and includes a registration of the study protocol prior to study enrolment. The trial was carried out from October 2019 until February 2020 at the Experimental Research Unit of the Department of Anaesthesiology and Intensive Care Medicine at the Medical University of Innsbruck in accordance with current ARRIVE guidelines.¹¹

Experimental protocol

An institutional protocol,^6,12,13 which has previously been shown to guarantee an appropriate depth of anaesthesia without haemodynamic disturbance, was used for induction and maintenance of anaesthesia in 18 domestic pigs (see Supplemental Digital Content, https://links.lww.com/EJA/A813). The trial was performed in the supine position throughout the study. Prior to experimental protocol initiation randomisation to the intervention or control group (allocation 1 : 1) was performed and all animals received mechanical ventilation with VCV at an FiO₂ of 0.3, a V_T of 7 ml kg⁻¹ and a PEEP level of 5 cmH₂O at an I : E ratio of 1 : 1.5. Following animal preparation, a recruitment manoeuvre with an inspiratory hold at 30 cmH₂O over 20 s was performed and subsequently baseline measurements were obtained. After increasing the FiO₂ to 1.0 ARDS induction was performed by intravenous injection of oleic acid boli as previously described.¹⁴ For this purpose oleic acid (0.1 ml kg⁻¹) was mixed with 2 ml of porcine blood and 18 ml of saline. Boli of 0.5 to 1 ml were administered repeatedly until a stable Horowitz index below 150 was reached over a period of 30 min. Normovolaemia was maintained by infusion of balanced crystalloid solution (5 to 10 ml kg⁻¹ h⁻¹ Elomel iso; Fresenius Kabi Austria GmbH, Graz, Austria). Norepinephrine was administered continuously to maintain adequate mean arterial pressure (MAP) above 65 mmHg.

Subsequently, the animals were ventilated with FCV or PCV over an observation period of 2 h. In the intervention group FCV, the ventilator (Evone; Ventinova Medical B.V., Eindhoven, The Netherlands) was connected to the tracheal tube using a conventional tube adapter (CTA; Ventinova Medical B.V.) which incorporates a pressure measurement line with the tip advanced to the end of the tracheal tube for tracheal pressure measurement.

FCV was performed with PEEP and P_peak set using a previously described compliance-guided setting method.⁶ Ventilator flow was adjusted to normalise the PaCO₂ level to 4.7 to 6.0 kPa. The Evone ventilator has a technical inspiratory flow limit of 20 l min⁻¹ which corresponds to a maximum minute volume of 10 l min⁻¹ with an I : E ratio set to 1 : 1.

The control group PCV (Evita XL, Dräger, Lübeck, Germany) was performed with compliance-guided titration of PEEP and P_peak set to achieve a V_T of 6 ml kg⁻¹. The I : E ratio was set to 1 : 1.5 and the respiratory rate was adjusted to normalise the PaCO₂ level to 4.7 to 6.0 kPa. The maximum respiratory rate was limited to prevent air trapping (assessed by repeated intrinsic PEEP measurement manoeuvres, whereby intrinsic PEEP had to be within a <0.5 cmH₂O limit above the set PEEP).

At the end of the protocol, a computed tomography (CT) scan of the chest during an inspiratory and expiratory hold was performed (see Supplemental Digital Content, https://links.lww.com/EJA/A813). At the end of the study, animals were euthanised during deep anaesthesia by a bolus injection of potassium chloride. Measurement time points were defined as baseline (T0), 30 min after ARDS induction (T1) and timepoints every 15 min (T2 to T10) after the start of FCV or PCV (Fig. 1).

Respiratory and haemodynamic measurements

Respiratory and haemodynamic measurements were performed at each time point (T0 to T10). P_peak and PEEP were directly recorded and ΔP was calculated for FCV and PCV. To allow a comparison of the pressure amplitude occurring at the alveolar space ΔP was additionally corrected for the continuous gas flow in FCV [ΔP (alv) = ΔP − 2 × flow (l s⁻¹) × R).^15,16 In PCV animals, a regular check for intrinsic PEEP was performed to rule out air trapping, and the respiratory rate was limited if air trapping was found. Intrinsic PEEP cannot occur during FCV because of the active nature of the expiration phase, where the ventilator sucks gas until the set value of intratracheal PEEP has been reached. Respiratory rate, minute volume, V_T, dynamic compliance (C_dyn) and (total) resistance (R) were directly recorded from the ventilators. Applied mechanical power was calculated based on published surrogate formulas.¹⁷ Arterial and mixed-venous blood gas samples were obtained and pH, arterial partial pressure of CO₂ (PaCO₂) and O₂ (PaO₂), arterial oxygen saturation, mixed-venous oxygen saturation, lactate and haemoglobin concentration were measured (ABL800 Flex; Radiometer, Brønshøj, Denmark). The pulmonary shunt fraction was calculated from arterial and mixed venous blood gas samples. The Horowitz quotient (PaO₂ FiO₂⁻¹) was calculated.

Haemodynamic monitoring included heart rate (HR), MAP, central venous pressure (CVP), pulmonary capillary wedge pressure (PCWP) and mean pulmonary arterial pressure (MPAP). Cardiac output (CO) was measured and corresponding systemic and pulmonary vascular resistance (SVR, PVR) calculated via the pulmonary artery catheter after three separate injections of 10 ml of saline. Indices of CO, SVR and PVR were calculated using the predicted body surface area for pigs.¹⁸

For interpretation of CT findings non-aerated lung tissue was defined as absorption values between 100 and –100 Hounsfield units (HU), poorly aerated lung tissue as values between –101 and –500 HU, normally aerated lung tissue as values between –501 and –900 HU, and airway as well as overdistended lung tissue as values between –901 and –1000 HU.¹⁹

Blood serum samples taken at baseline, and at 1 and 2 h after ARDS induction were analysed using ProcartaPlex porcine Cytokine & Chemokine panel 1 (Affymetrix, eBioscience, Vienna, Austria) according to the manufacturer's instructions.

As two different ventilators and different ventilator settings had to be used, the study could only be blinded for the investigator who analysed the CT scans.

Statistical analysis

Based on the results of a significant difference in a previous pilot trial comparing 12 lung-healthy animals with the same primary endpoint measure Horowitz quotient⁶ and considering we were only looking for clinically relevant differences with an effect size more than 0.8, we concluded a sample size of 18 would be appropriate to comply with a higher variance and possible dropouts in this ARDS model.

A mathematician (TH) not involved in the study procedures performed the statistical analyses using R, version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria). For the characteristics of laboratory animals before the start of the experiment, continuous data are presented as median [IQR] and categorical variables as number (%). Effect size and precision are shown with estimated median differences between groups for continuous data and odds ratios for binary variables with 95% confidence intervals (CI). The Wilcoxon rank sum test and Fisher's exact test were applied to assess differences between the groups.

The course of haemodynamic parameters during the observation period are shown per group using the median course with corresponding 95% CI's. Differences between groups were assessed using linear mixed-effects models with random intercepts for time points and animals as well as the group as fixed effect.

All statistical assessments were two-sided, a significance level of 5% was used.

Results

After the enrolment of 18 animals, the experimental protocol was completed in 16 pigs (FCV n = 8, PCV n = 8). One animal in the FCV group was excluded because of a Horowitz quotient below 60 after ARDS induction due to an unintended over titration of oleic acid. One animal in the PCV group died after 75 min due to increasing deterioration of respiratory function and hypoxic right ventricular failure before the protocol could be completed.

Baseline characteristics were comparable between groups (Table 1).

The primary outcome measure Horowitz quotient was significantly higher (149 vs. 110, median difference (MD) 38.7 (95% CI, 8 to 70) PaO₂ FiO₂⁻¹; P = 0.027) in FCV compared with PCV (Fig. 2). PaCO₂ values were significantly lower in FCV compared with PCV (7.25 vs. 9.05, MD −1.8 (−2.87 to −0.72) kPa; P = 0.006). Pulmonary shunt fraction was significantly lower in FCV compared with PCV (8.2 vs. 22.3, MD −14.2 (−23.2 to −5.1)%; P = 0.008).

Compliance-guided PEEP titration performed in both groups led to comparable PEEP values. Tidal volume was also similar in both groups (Table 2). All other respiratory variables differed between groups: minute volume was significantly lower in FCV compared with PCV (8.4 vs. 11.9, MD −3.6 (−5.6 to −1.5) l min⁻¹; P = 0.005), as well as respiratory rate (30.0 vs. 43.6, MD −13.6 (−18.5 to −8.7) min⁻¹; P < 0.001). Compared with PCV, P_peak was significantly higher in FCV (28.2 vs. 24.1, MD 4.0 (1.0 to 7.0) cmH₂O; P = 0.021) as well as driving pressure (20.9 vs. 17.1, MD 3.8 (1.1 to 6.5) cmH₂O; P = 0.016). Calculated mechanical power was significantly lower in FCV (16.1 vs. 27.0, MD −10.9 (−16.8 to −5.1) J min⁻¹; P = 0.001).¹⁷

Haemodynamic variables revealed comparable MAP and MPAP values in both groups (Table 2), whereas CVP (8.7 vs. 6.1, MD 2.6 (0.8 to 4.4) mmHg; P = 0.013) and PCWP (10.1 vs. 7.7, MD 2.4 (0.5 to 4.3) mmHg; P = 0.029) were significantly higher in FCV. A lower HR (117 vs. 161, MD –44 (–79 to –10) min⁻¹; P = 0.025) and, consecutively, a significantly lower CI (4.5 vs. 8.6, MD −4.1 (–5.0 to –3.1) l min⁻¹ m⁻²; P < 0.001) were observed in FCV at a significantly higher SVRI (1593 vs. 861, MD 732 (455 to 1008) dyn s cm⁻⁵ m⁻²; P < 0.001) (Fig. 3) as well as PVRI (642 vs. 368, MD 274 (137 to 412) dyn s cm⁻⁵ m⁻²; P = 0.002) compared with PCV.

A significantly lower dose of norepinephrine was required to maintain MAP at 65 mmHg in FCV than in PCV (0.26 vs. 0.86, MD –0.61 (–1.12 to –0.09) μg kg⁻¹ min⁻¹; P = 0.037). Lactate levels and mixed-venous oxygen saturation (SvO₂) were comparable (Table 2), whereas pH was significantly higher in FCV compared with PCV (7.38 vs. 7.27, MD 0.11 (0.05 to 0.17); P = 0.003).

Measurement of cytokine levels IFN-α, IFN-γ, IL-12p40, TNF-α, IL-1β, IL-8, IL-4, IL-6 and IL-10 revealed similar values in both groups at timepoint T0, T2, T6 and T10 (see Supplemental Digital Content, Table S1, https://links.lww.com/EJA/A813).

Computed analysis of the CT scans performed at the end of the protocol revealed no significant differences in the amount of overdistended, normally aerated, poorly aerated or non-aerated lung tissue (Fig. 4). Calculated end-expiratory lung volumes from the CT images showed similar values in both groups (546.4 (FCV) vs. 578.1 (PCV), MD 21.4 (–141.7 to 211.8) ml; P = 0.955).

Discussion

The main finding of this experimental study was that individualised FCV improved oxygenation and CO₂ removal in this ARDS model despite a reduction of respiratory minute volume. Simultaneously, FCV animals showed improved haemodynamic stability compared with PCV animals.

The findings of improved oxygenation and concomitant reduction of respiratory minute volume are consistent with previous studies comparing FCV with VCV in a similar ARDS model⁸ and FCV to PCV in long-term ventilation.⁶ A possible explanation for the improved oxygenation may be the reduction of pulmonary shunt in FCV (Fig. 2), most likely resulting from a more homogeneous gas distribution during FCV.^6,10 In particular, inhomogeneous lungs with compartments having different time constants due to different resistance to flow, which is typically found in ARDS and accurately imitated by the oleic acid model,²⁰ could benefit from a more uniform gas distribution. A reduced preferential gas distribution may lead to an improved ventilation/perfusion ratio and thus decreased pulmonary shunt fraction as shown in this study (Fig. 2). This assumption is supported by the fact that the proportion of non-aerated lung areas diverged only slightly in the two groups. Indeed, the proportion of atelectasis does not necessarily have to be associated with pulmonary shunt, but may also be due to dysregulation of perfusion.²¹

Of note, FCV resulted in a significant reduction in PaCO₂ despite significantly reduced respiratory minute volume compared with PCV. At a comparable technical dead space of the different respiratory circuits (approximately 75 ml per breath) and comparable anatomical dead space (2 ml kg⁻¹) at a similar mean V_T, the significantly higher respiratory rate in the PCV group (43 vs. 30 min⁻¹ in FCV) accounts for approximately 1.5 to 2 l min⁻¹ more dead space ventilation. Flow physics suggests that ventilation would be improved in compromised, slow(er) lung compartments and this could explain the observed improvement in elimination of carbon dioxide: In FCV, where gas is continuously shifted in and out of the lungs at the lowest flow achieving the desired level of CO₂ elimination, slow(er) lung compartments with higher airway resistance will have more time to participate in shifting gas volumes. In contrast, the gas flow pattern observed in PCV with an initially high and then decelerating gas flow will preferentially serve fast(er) lung compartments.²² As a result, CO₂ exchange may be compromised in PCV being limited to the faster compartments, whereas in FCV a higher proportion of slower compartments may contribute to gas exchange as well.

Importantly, our findings need to be interpreted based on some principal considerations since the ventilation parameters of FCV and PCV differ substantially due to different flow profiles and technical layout of the ventilators. In PCV, alveolar PEEP was determined by an intermittent short-lasting expiratory hold to rule out intrinsic PEEP, whereas valid alveolar P_peak values were not obtainable because there was no inspiratory plateau pressure phase without any flow at high respiratory rate values. In contrast, the absence of any pause phase during the ventilation cycle in FCV necessarily results in a continuous pressure drop across the airway resistance during both inspiration and expiration.^15,16 In the present trial, correction for this effect leads to a significantly reduced calculated alveolar pressure amplitude [ΔP_(alv)] of 13.8 cmH₂O in FCV compared with 17.1 cmH₂O observed in PCV (P = 0.011, Table 2).

The feasibility of individualised ventilation with a PEEP set closely above the lower inflection point and a P_peak set closely below the upper inflection point of the pressure–volume curve, thus ventilating with maximum V_T within the range of optimal dynamic compliance may allow for improved ventilation efficiency, especially in ARDS.²³ Increasing V_T within lung mechanical limits allows a reduction in the proportion of dead space ventilation and, concomitantly, reduce mechanical power and dissipated energy for the same gas exchange. Applied mechanical power as a surrogate parameter for VILI has been found to be well correlated to outcome in ARDS.^24,25 In our study, estimating the applied mechanical power based on published formulas,¹⁷ reveals a considerably lower value of 16.1 J min⁻¹ for FCV compared with 27.0 J min⁻¹ for PCV (Table 2). However, dissipated energy may even more directly indicate potentially deleterious effects of mechanical ventilation.^26–28

The concept of individualised FCV is endorsed by a strong correlation between applied V_T and the corresponding measured dynamic compliance (effect size = 0.586; P < 0.001; Fig. 5). Hence, individualised FCV offers the possibility of automatically adjusting ventilation parameters according to lung compliance while the use of fixed tidal volumes is currently debated in different types of ARDS patients with and without COVID-19.^29,30 Indeed, the use of compliance-guided ventilator settings instead of a fixed tidal volume for FCV is the main difference to a previously reported ARDS study⁸ and was the rational to perform this trial.

In terms of haemodynamic parameters, the findings of this study were rather unexpected. MAP was kept above 65 mmHg by continuous administration of norepinephrine to guarantee adequate coronary perfusion. Compared with the FCV group, the required norepinephrine dose to reach haemodynamic stabilisation and similar MAP levels between groups was significantly higher in the PCV group and increased over time (Fig. 3). In addition, we observed a significant reduction of SVRI with PCV, which was accompanied by doubling of CI compared with baseline values. Potentially, the higher PaCO₂ level observed in PCV due to technical and physiological limits of mechanical ventilation may facilitate sympathoadrenal mechanism such as increase in HR and CI.³¹ However, even acute hypercapnia is usually well tolerated.³² Accordingly, observed changes in pH and PaCO₂ do not provide a satisfactory explanation for the extent of systemic perturbations. Interestingly, a previous study in lung healthy pigs already showed significantly higher but still physiological values for SVRI and PVRI in FCV compared with PCV.⁶ The haemodynamic findings of our ARDS trial, in view of similar findings in lung healthy animals,⁶ suggest that FCV may lead to some increase in tone of the vasculature, thereby counteracting systemic vasoplegia as an additional contributor to improved haemodynamics. Therefore it could be hypothesised that FCV may be beneficial for haemodynamic stability compared with PCV, where fast intrathoracic pressure changes lead to systemic perturbations. A previous ARDS study⁸ using FCV did not report data on PVR nor SVR nor on the required vasopressor doses to obtain haemodynamic stabilisation; our study is therefore the first one reporting these findings. The mechanism behind this finding is not clear at this time and needs to be further investigated.

The current study has several limitations. First, considering the different ventilator configuration, dynamic compliance and total resistance differ substantially in FCV and PCV. Therefore, it is not meaningful to directly compare C_dyn and R values. In consequence, the values displayed on the two ventilators were only reported for interest, but without intergroup analysis. In addition, whereas intrinsic PEEP is not possible in FCV, in PCV there may be a small but clinically irrelevant intrinsic PEEP (<0.5 cmH₂O above the set PEEP according to the study protocol). A second limitation is the short duration of this experimental protocol, which may not reflect the full clinical picture of ARDS where changes in the respiratory condition of a patient may occur over several days. The short duration of the protocol, due to the used oleic acid induced ARDS model which is not stable over a longer period, may also impede interpretation of cytokine measurements and CT images obtained after 2 h. Although no significant differences in lung tissue aeration were found in this study, we would like to note that only the entire lung was examined. Regional differences in dependent and non-dependent sections were not specifically analysed. In addition, pigs have a higher metabolic rate and therefore require a higher respiratory rate and minute volume. Both exceed the typical range of adult humans, thus making direct transfer of the results to ARDS patients difficult.

In summary, in this oleic acid-induced ARDS model, individualised FCV significantly improved gas exchange compared with best clinical practice PCV. The haemodynamic findings of this study, however, need to be further investigated.

Supplemental Digital Content

• EJA_2023_02_09_MARTINI_EJA-D-22-00436_SDC1.docx; [Word] (24 KB)