Comparison of Low Tidal Volume Ventilation and Neurally Adjusted Ventilation (NAVA) On

Electronic supplementary material

Neurally adjusted ventilatory assist decreases ventilator induced lung injury and non-pulmonary organ dysfunction in rabbits with acute lung injury

Running head:NAVA in rabbits with lung injury

Lukas Brander 1,7, Christer Sinderby 1,6, François Lecomte 1, Howard Leong-Poi 2,6, David Bell 3, Jennifer Beck 4,6, James N. Tsoporis 2, Rosanna Vaschetto 1, Marcus J. Schultz 1,5, Thomas G. Parker 2,6, Jesús Villar 6,8, Haibo Zhang 1,6, Arthur S. Slutsky 1,6

1 Interdepartmental Division of Critical Care Medicine, University of Toronto, Department of Critical Care Medicine, St. Michael’s Hospital, Toronto, Canada

2 Division of Cardiology, University of Toronto, St. Michael’s Hospital, Toronto, Canada

3 Department of Laboratory Medicine and Pathobiology, University of Toronto, St. Michael's Hospital, Toronto, Canada

4Department of Pediatrics, University of Toronto, Canada

5 Department of Intensive Care Medicine and Laboratory of Experimental Intensive Care and Anesthesiology, University of Amsterdam, The Netherlands

6 Keenan Research Center at the Li Ka Shing Knowledge Institute of St. Michael's Hospital, Toronto, Canada

7Department of Intensive Care Medicine, UniversityHospital - Inselspital, Bern, Switzerland

8CIBER de Enfermedades Respiratorias, Instituto de Salud Carlos III, Madrid, Spain; Translational Research on Organ Dysfunction, Research Unit, Hospital Universitario Dr. Negrin, Las Palmas de Gran Canaria, Spain

The work was performed at St. Michael’s Hospital and the University of Toronto,Toronto, Canada

This article is discussed in the editorial available at: doi: 10.1007/s00134-009-1632-z

Materials and methods

The study was performed according to the National Institutes of Health guidelines for the use of experimental animals. The protocol was approved by the Animal Care Committee, St. Michael’s Hospital, Canada.

Instrumentation, anaesthesia, and monitoring

30 adult male New Zealand white rabbits weighing 3.6–4.6 kg (Charles River Laboratories, St Constant, QC, Canada) were intramuscularly premedicated withketamine hydrochloride 35 mg/kg(de Wyeth-Ayerst Inc., Guelph, ON, Canada) and xylacin 10 mg/kg (Bayer Inc., Toronto, ON, Canada),followed by bilateral cannulation of the ear veins and arteries (22GA-Angiocath, Becton Dickinson Inc., Sandy, UT, USA), and by installation of a tracheal cannula (4.0 ID, Mallinckrodt Inc., St. Louis, MO, USA). All procedures were performed under sterile conditions. Anaesthesia was maintained with continuous intravenous infusion of ketamine hydrochloride 40 mg/kgh-1 and xylazine 4 mg/kgh-1. Ringer’s lactated solution (Baxter Corporation, Toronto, ON, Canada) 5 ml/kgh-1 was infused intravenously via an infusion pump (ColleagueTM, Baxter Healthcare Corporation, Deerfield, IL, USA) throughout the experiment. The volume of withdrawn blood was replaced by Ringer’s lactate solution.

No other fluids, including bicarbonates, and no vasoactive drugs were administered during the study.

Transcutaneous oxygen saturation was continuously measured by pulse oxymetry (NONIN 8600 VTM, Nonin Medical Inc., Plymouth, MN, USA) at the tail and the heart rate was calculated from the signal. Blood pressure was measured using quartz pressure transducers (TruWave PX12N, Edwards Lifesciences, Irvine, CA, USA) referencedto the midcardiac plane and displayed continuously on a multimodular monitor (S/5 Critical Care MonitorTM, Datex-Ohmeda, Helsinki, Finland). Rectal temperature was monitored continuously and body temperature was maintained at 39 +/- 0.5 °C using a heating lamp.

NAVA technology

NAVA was used as previously described [1-3]. The electrical activity of the crural part of the diaphragm (EAdi) was derived from an array of electrodes mounted on an 8F esophageal catheter. Proper positioning of the catheter in relation to the diaphragm was verified by online display of the electrocardiogram detected by the electrodes, and by correllograms of EAdi signals along the electrode array [4].

The EAdi signal was amplified, digitized, and processed with algorithms previously described [5,6], and was translated into a voltage every 16 ms which was then multiplied by a proportionality factor (NAVA level; cmH2O/unit of EAdi) throughout inspiration, and was used to control the pressure generated by the ventilator (Servo300, Maquet Critical Care, Solna, Sweden). Hence, with NAVA airway pressure (Paw) during inspiration is linearly proportional to EAdi (Paw = EAdi x NAVA level). Since with NAVA the command signal to the pressure generator of the ventilator is updated essentially in real time (every 16ms = 62.5 times per second), the pressure delivered to the animal’s airways is virtually synchronous and instantaneously proportional to respiratory demand as reflected by the EAdi (i.e. the intra-breath assist profile closelymirrors the profile of the EAdi).

The EAdi is influenced by facilitatory and inhibitory feedback loops that integrate information from mechano- and chemo-receptors as well as voluntary and behavioral inputs [1]. If the delivered assist exceeds the subject’s respiratory demand, EAdi will reflexively be down-regulated resulting in less assist for the same NAVA level. If the delivered assist falls short of the subject’s respiratory demand, EAdi will be up-regulated, resulting in delivery of more assist for the same NAVA level. Thus, the Paw delivered by the ventilator in response to a changed NAVA level always depends on the concurrent change in the EAdi level.

In the present study, NAVA was only applied during inspiration and therefore EAdi based trigger-on and cycling-off algorithms were implemented. EAdi based trigger-onwas set to initiate ventilatory assist when the EAdi exceeded the random noise-variability. Given that the variability of the noise level was low, the trigger threshold was set to a fixed level that permitted early detection of an EAdi increase without causing auto triggering when the diaphragm is inactive. EAdi based cycling-off was set to terminate ventilatory assist when the EAdi fell below 80% of peak inspiratory EAdi.

Experimental protocol (Figure E1)

A commercially available ventilator (Servo 300TM, Maquet Critical Care, Solna, Sweden) modified for NAVA was used. The animals were ventilated with a time cycled, volume targeted mode using a Vt of 6-ml/kg (VC 6-ml/kg), a fraction of inspired oxygen (FiO2) of 0.5, and a PEEP of 3 cm H2O during instrumentation and during the subsequent 30 minute stabilisation period. Stable mean arterial blood pressure (>60 mmHg), arterial oxygen tension (PaO2) to FiO2 ratio >400 mmHg, PaCO2 4.7-6.0 kPa (35-45 mmHg), and pH 7.30-7.50 were required at baseline before induction of lung injury to proceed with induction of lung injury.

Induction of lung injury

After induction of neuromuscular paralysis with pancuronium bromide 0.02 mg/kg (Sabex Inc., Boucherville, QC, Canada) and after EAdi was no longer detectible, hydrochloric acid (pH 1.5) was instilled intratracheally with the rabbit in the lateral position (0.75 ml/kg each side), followed by a ventilation pause at a positive airway pressure (CPAP) of 25 cm H2O for four seconds. The HCl instillation and the recruitment maneuver were repeated after five minutes.

After instillation of HCl, all animals were ventilated for another thirty minutes with VC (Vt 6 ml/kg; FiO2 0.5, PEEP 3 cm H2O). A PaO2/FiO2 ratio between 80 and 200 at 30 minutes after induction of ALI was required to proceed to randomization.

Randomization to NAVA, VC - ml/kg, or VC 1- ml/kg

Three out of 30 animals were sacrificed before randomization because their PaO2/FiO2 ratio was below the lower limit of 80. 27 animals were equally randomized (n=9 per group; concealed allocation) to 5.5 hours ventilation with one of the following 3 strategies:

- NAVA: No neuromuscular paralysis. NAVA was initiated as soon as the EAdi was sufficient to control the ventilator. We used a NAVA level of 0.5 cmH2O per unit of EAdi throughout the experiment in all animals based on our experience from previous studies using the same animal model where we demonstrated that the animals keep Vt constant at higher NAVA levels due to down regulation of their EAdi [2]. Note that employing a uniform NAVA level does not necessarily result in a uniform level of assist since the animals up- or down regulate their EAdi (and hence the pressure delivered) based on information from the respiratory system (e.g. from receptors sensitive to lung distension or to oxygen tension) on a breath by breath basis. With NAVA the ventilatory pattern including the Vt, the respiratory rate, and the intra-breath assist profile is entirely defined by the EAdi.

- VC 6-ml/kg or VC 15-ml/kg: In order to prevent lung recruitment with inspiratory muscle activity or with tonic muscle activity during expiration [1], muscle paralysis was continued after randomization in both volume controlled (VC) groups with an infusion of pancuronium bromide (0.25 mg/kgh-1) and additional boluses of 0.02 mg/kg if needed to suppress any activity of the continuously monitored EAdi. The pressure rise time was set at 5% and the end-inspiratory pause time was set at 10% of the breath cycle in both VC groups. An additional dead space of 25 ml was used in VC 15-ml/kgso that the ventilatory rate of both VC groups would be roughly the same. The ventilatory rate was adjusted to maintain PaCO2 levels between 4.7-6.0 kPa (35-45 mmHg).

Positive end-expiratory pressure (PEEP)

PEEP was adjusted similarly throughout the experiment in NAVA and VC 6-ml/kg aiming at the highest PEEP level possible while maintaining mean arterial pressure ≥60 mmHg. The PEEP was lowered in steps of 1 cmH2O if mean arterial pressure decreased below 60 mmHg. This approach was chosen in order to control for the effect of PEEP on cardiac function similarly in both groups. To prevent disparities between the groups (e.g. hemodilution, tissue edema, fluid overload, metabolic derangements) we did not administer additional fluid to maintain mean arterial pressure above 60 mmHg. In VC 15-ml/kg animals PEEP was maintained at 1 cmH2O throughout the protocol.

Fraction of inspired oxygen (FiO2)

FiO2was kept at 0.5 and increased up to 1.0 if needed to maintain oxyhemoglobin saturation in arterial blood (SaO2) above 90% if possible. Recruitment manoeuvres and suctioning of secretions were not performed throughout the study.

At the end of the protocol, the animals were sacrificed with an overdose of pentobarbital (intravenous bolus of at least 20mg/kg).

Measurements and sample collection

Respiratory, hemodynamic, and temperature measurements

Vt, ventilatory rate, Paw, and PEEPwere measured using the ventilator’s pneumotachograph and pressure transducers. Dynamic respiratory system compliance CRSdyn = Vt in ml / (peak airway pressure in cmH2O – PEEP in cmH2O) was assessed during neuromuscular paralysis in all animals before induction of ALI, immediately before randomization, and after completion of the protocol (NAVA animals were paralyzed for measuring CRSdyn after completion of the protocol) using uniform ventilator settings (Vt 6-ml/kg, PEEP 3 cmH2O) in all animals. Mean arterial pressure, heart rate, and rectal temperature were recorded immediately before induction of ALI and every 30 minutes thereafter.

Analyses in blood and urine samples

Arterial blood samples were drawn before induction of ALI and hourly thereafter at a FiO2 of 0.5 and processed immediately. Arterial blood gases (corrected for actual rectal temperature), and arterial lactate concentration was measured using standard electrodes (ABL 520, Radiometer, Copenhagen, Denmark). SaO2 and arterial hemoglobin concentration were analyzed using a spectrophotometry method (OSM 3, Radiometer, Copenhagen, Denmark) designed for rabbit blood. 4 ml arterial blood was withdrawn before as well as 3 and 6 hours after induction of ALI, immediately centrifuged and the supernatant was stored in aliquots at -80°C until further analyzed. The urinary bladder was emptied before induction of ALI by gently compressing the lower abdomen. Urin was collected after 3 hours by transcutaneous puncture and at the end of the protocol by direct puncture of the bladder (after median laparatomy) using a 23G needle to collect the entire urine produced during the experimental period.

Echocardiography and oxygen delivery

Echocardiography was performed immediately before induction of ALI and hourly thereafter using a Sonos 5500 system (Philips Ultrasound, Bothell, WA) equipped with a 8-12-MHz broad-band phased-array s12 transducer in 7 animals in each group (the equipment was not available in the other animals). Digital image frames and image loops were saved to magnetic-optical disk for off-line analysis (Medarchive Viewer, version 2.1). From a short-axis cross-sectional view of the left ventricle (LV) at the midpapillary muscle level, maximal LV end-diastolic dimension (EDD) and end-systolic dimension (ESD) were measured and LV fractional shortening (FS) was calculated as FS = [(EDD-ESD)/EDD].LV ejection fraction (LVEF) was calculated as LVEF=[(EDD3-ESD3)/EDD3]. With the ultrasound beam parallel to the long axis of the pulmonary artery (PA), the Doppler sample volume was placed above the pulmonary valve in the mid-lumen of the main PA, to record spectral profiles of blood flow velocity. From pulsed Doppler spectral recordings at maximum sweep speed, velocity time integral (VTI), peak velocity (Vmax), acceleration time (AcT), deceleration time (DcT), and flow period (FP) were measured. Mean acceleration (Acmean) was calculated as Vmax /AcT. PA diameter (d) was measured using the same view and PA cross section area was calculated as (d/2)2. Right ventricular (RV) stroke volume = (PA cross section area  VTI); and cardiac output (CO) = (RV stroke volume  heart rate) were calculated. Heart rate was calculated from the PA Doppler tracings for this equation.

Arterial oxygen content (CaO2 in [ml/L]) was calculated as: (1.36 [mlO2/g] x haemoglobin concentration [g/L] x SaO2 [%] / 100) + (0.03 [ml/L·mmHg-1] x PaO2 [mmHg]). Oxygen delivery (DO2 in [ml/min·kg-1]) was calculated as: CO [L/min·kg-1] x CaO2 [ml/L].

Broncho-alveolar lavage (BAL) fluid

After sacrificing the animals, the heart-lung block was removed with the lungs inflated at a CPAP of 20 cmH2O. The left main bronchus and the right main lower lobe bronchus were both tightly occluded before 20 ml of normal saline was instilled intratracheally. After applying a CPAP of 20 cmH2O for five seconds, the BAL fluid was aspirated, immediately centrifuged at 4 °C with 2000 rounds per minute (rpm), and the supernatant was stored in aliquots at -80°C until further analyzed.

Lung wet to dry ratio

Lung wet to dry ratio was assessed in all animals as well as in 6 healthy, non-ventilated controls. The right lower lung lobe was divided from ventral to dorsal into a caudal and a cranial portion. Both portions were further cut into a dependent (dorsal) and a non-dependent (ventral) portion. The caudal two portions were weighed before and after exposure to 40°C for 72 hours and lung wet-to-dry ratio = [(lung wet weight – lung dry weight) / lung dry weight] was calculated for dependent and non-dependent lung portions separately.

Lung and extra-pulmonary organ tissue

The cranial two portions of the right lower lung lobe and fragments of the heart (apex), right kidney (caudal third), small intestine, liver (middle lobe), and spleen were removed after exsanguination, stored in TritonX 0.2% (Sigma-AldrichCorp. St. Louis, MO, USA) on ice, homogenized immediately, centrifuged at 4°C for 10 minutes, and the supernatant was stored in aliquots at -80°C until analyzed. The left lung was filled with paraformaldhyde 4% at a hydrostatic pressure of 15 cmH2O, submerged with the other organs in paraformaldhyde 4% for 24 hours, rinsed in phosphate buffered saline 1% (Sigma-Aldrich Inc, St. Louis, MO) for another 24h, and then stored in alcohol 70% at 4°C until embedded in paraffin.

Measurements after the experiments

Complete sets of material for the assessment of lung histology; of inflammation and coagulation parameters; and of apoptosis in non-pulmonary organs were available in 5 VC 15-ml/kg animals, in 7 VC 6-ml/kg animals, and in 7 NAVA animals. For comparison, inflammation and coagulation parameters were also measured in 4 healthy, anesthetised, non-ventilated control animals that were sacrificed immediately after tracheostomy. Biochemical markers of organ dysfunction were measured in all ventilated animals.

Lung and non-pulmonary organ histology

A pathologist blinded to the study groups assessed eight high power fields (x 400) in paraffin embedded, hematoxylin and eosin (HE) stained sections of the dependent (dorsal) and non-dependent (ventral) portions of the left lower and upper lung separately (4 sections per animal x 8 fields each) and scored histological lung injury qualities per section (see Table 2) as follows: none=0, mild=1, moderate=2, severe=3 (for perivascular / peribronchial hemorrhage and edema, and for bronchial epithelial lesions); none=0, only few=1, in <50% lung volume=2, and in >50% lung volume=3 (for vascular congestion / distension, and for alveolar membranes and alveolar edema); none=0, <20%=1, >20%=2, >30%=3 (for alveolar collapse); none=0, up to 6 red blood cells per alveolus=1, 7-30 red blood cells per alveolus=2, >30 red blood cells per alveolus=3 (for intraalveolar hemorrhage), and none=0, only one found=1, more than one found=2, four or more found=3 (for intravascular thrombi). Alveolar polymorph-nuclear neutrophils and macrophages were calculated as average number per alveolus.

Sections of the heart, kidney, liver, and small intestines of all animals were examined by the same pathologist for histological organ injury, specifically for intravascular thromboemboli and for micro-infarctions.

Inflammatory, coagulation and fibrinolysis mediators

We measured IL-8 concentrations as a surrogate marker of a pro-inflammatory response because IL-8 was increased in animal [7-9] and human studies on VILI [10], and because plasma IL-8 levels were associated with both, mortality as well as ventilator and organ failure free days in patients with established ALI/ARDS [11]. In patients at risk for ARDS, Donelly et al. demonstrated that BAL IL-8 concentration was higher in those patients that subsequently progressed to ARDS suggesting that IL-8 might be an early marker of ARDS. Measurement of inflammatory mediators in rabbit models is generally limited by the availability of rabbit specific assays.

Plasma, BAL fluid, and tissue levels of interleukin 8 (IL-8) were measured using a human IL-8 ELISA kit (BioSource International Inc, Camarillo, CA, USA). It has been shown that the antibody against human IL-8 is cross-reactive with rabbit IL-8 [13]. Tissue IL-8 levels were normalized to lung tissue total protein concentrations (BIO-RAD Protein Assay, Hercules, CA, USA).

There is extensive cross–talk between inflammatory mediators and coagulation products, causing reciprocal modulation. We measured tissue factor (TF) and plasminogen activator inhibitor type 1 (PAI-1) because levels of these proteins reflect the extent of coagulopathy in various forms of lung injury, including VILI [16]. TF levels are low in the normal lung and elevated in disease [17,18]. In patients with ARDS an increase in soluble TF in BAL fluid has been demonstrated [19]. In association with enhanced fibrin production, fibrinolytic activity is depressed in BALF of patients with ALI/ARDS [20], related to high pulmonary concentrations of PAI–1. PAI–1 is increased in ALI/ARDS [21,22]. Alveolar PAI–1 levels have also been found to be associated with higher mortality in patients with ALI/ARDS [23].