1
ElectronicSupplementary Material (ESM)
METHODS
Procedures and instrumentation
A total of 9 adult male New Zealand white rabbits (Charles River Labs, St Constant, Quebec, Canada) weighing 2.8-3.4 Kg were studied. All animals were anaesthetized by an intramuscular bolus of ketamine hydrocloride (35 mg/Kg) and xylazine (10 mg/Kg), followed by continuous intravenous infusion of ketamine hydrocloride (10 ml/Kg per hour), xylazine (2 mg/Kg per hour). Lactated Ringer’s solution (5 mL/Kg per hour) was continuously infused intravenously with a volumetric infusion pump. Arterial blood pressure (Datex-Ohmeda S/5) and blood for arterial blood gases (Ciba-Corning Model 248, Bayer, Leverkusen, Germany) were obtained from an ear artery. Transcutaneous oxygen saturation was measured with pulse oximetry (NONIN 8600 VTM, Nonim Medical Inc., Plymouth, MN) at the tail. Body temperature was measured with a rectal probe and was maintained between 37.5C and 39C with a heated surgical table. The animals were surgically tracheotomized and ventilated by a Servo-i (Maquet Critical Care, Solna Sweden).
Flow was measured via a pneumotach (Novametrix Series 3 Neonatal flow sensor; Cat. No.:6718-00) placed between the tracheostomy tube and the y-piece of the circuit. Tidal volume was obtained by integrating the flow. An 8-Fr catheter, with an array of 10 microelectrodes (5 mm inter-electrode distance) for recording of EAdi was positioned in the oesophagus at the level of the crural diaphragm [E1]. The catheter had two balloons (2.5 cm in length, 1 cm in diameter) to measure oesophageal pressure (Pes) and gastric pressure. Proper positioning of the catheter was allowed by online display of the electrocardiogram detected by the electrodes. Esophageal balloon positioning was confirmed by the occlusion method [E2]. At the end of the study procedure all animals were sacrificed by an overdose of anaesthesia.
EAdi, flow and volume signals were acquired at a rate of 62.5 Hz from the Servo-i ventilator (Maquet Critical Care, Solna, Sweden) via its RS232 serial communication port. Differential pressure signals from the pneumotach as well as Paw, Pes and Pga (4 channel pressure unit, NeuroVent Research Inc, Toronto, Canada) were acquired at 2 kHz, averaged and time synchronized with the RS232 data using a custom made acquisition system (16 channel Interface System, NeuroVent Research Inc, Toronto, Canada).
Method for NAVA
During NAVA, the EAdi is processed and used to control Paw delivered by the ventilator (Servo-i, Maquet, Solna, Sweden). Inspiratory assist is initiated when either the EAdi or the inspiratory flow exceeded a trigger threshold (first-come-first-served). During neural inspiration, the pressure delivered is proportional to the EAdi waveform. The assist is increased during NAVA, by increasing the NAVA level or “gain level”, proportionality constant with the unit cmH2O/μV. The breath is cycled-off at 70% of the peak EAdi to a user-defined PEEP.
Description of the protocol for ramp increases of the NAVA level with resistive loads and after ALI.
Resistive loading: the ramp increases of the NAVA level were tested separately for two resistive loads referred to as R1 (~0.15 cmH2O/ml/s) and R2 (~0.3 cmH2O/ml/s). The resistor was inserted between the endotracheal tube and the pneumotachograph 10 minutes prior to the ramp increase of the NAVA level started. A PEEP level of 5 cmH2O was applied at all times during the resistive breathing.
Acute lung injury:After induction of neuromuscular paralysis (pancuronium bromide 0.02 mg/kg), the mode of ventilation was switched to volume-controlled ventilation (tidal volume of 6 ml/kg, respiratory rate 30/min, PEEP 5 and FiO2 50%). An acute lung injury was induced during neuromuscular paralysis by intratracheal instillation of 1.5 mL/kg of hydrochloric acid (pH 1.5), followed by a recruitment maneuver (CPAP of 25 cmH2O for 5 sec). Compliance of the respiratory system (Crs) was measured by dividing inspiratory volume by the difference between plateau pressure and PEEPduring the respiratory paralysis and volume control ventilation before and after the lung injury. Arterial blood gas analysis was obtained at 1 and 5 minutes after HCl instillation. A PaO2/FiO2 lower than 200 was required and if this target was not reached, a second instillation of HCl (1 mL/kg) was performed.
After the recovery from paralysis, NAVA was resumed with a level of 0 and PEEP of 5 cmH2O and FiO2 of 50%. Once EAdi and Pes had stabilized, ramp increases in NAVA level were performed.
Two different rates of ramp increases of the NAVA level were used for each resistive load (R1 and R2) and during ALI: slow (steps of 0.2 cmH2O/V every 20 sec) and fast (steps of 0.4 cmH2O/V every 20 sec).
Additional description of PVBC and associated indices
To study the contribution of the respiratory muscles to the generated Vtinsp, the ratios of Vtinsp and ∆Pes for both non-assisted [(Vtinsp/∆Pes)no-assist] and assisted breaths [(Vtinsp/∆Pes)assist] were calculated as to calculate [(Vtinsp/∆Pes)no-assist/(Vtinsp/∆Pes)assist] and compared to the PVBC index values.
To evaluate the validity of using EAdi to normalize non-assistedand assisted breaths the ratios of ∆PL,dyn and ∆EAdi for bothnon-assisted,[(∆PL,dyn/∆EAdi)no-assist] and assisted breaths[(∆PL,dyn/∆EAdi)assist] were calculated. Their ratio [(∆PL,dyn/∆EAdi)no-assist/(∆PL,dyn/∆EAdi)assist] was then calculated and was assumed to be linear to ∆Pes/∆PL,dynif the ∆EAdi normalization was valid.
RESULTS & DISCUSSION
Figures E1a,b,c. Time plots of airway (Paw, blue), esophageal (Pes, red), and gastric (Pga, green) pressures as well as diaphragm electrical activity (EAdi, magenta) during 10 s of breathing with NAVA at high assist level. No data was available for ALI in experiments 3 and 7, Pga was not available during ALI in experiments 4, 8 and 9. The variables represent the three conditions of acute lung injury (ALI), high (R1) and very high resistance (R2). The figures demonstrates that Pga deflections were minute relative to Pes and they did not change much between non-assisted and assisted breaths whereas Pes deflections increased somewhat. This could suggest that the relative contribution between diaphragm and extradiaphragmatic muscles may have been altered somewhat between assisted an non-assisted breaths, probably due to increased intercostal and accessory muscle activation.
Figure E2: Relationship between inspiratory deflections in transpulmonary pressure (PL,dyn, x-axis) and tidal volume (Vtinsp, y-axis) for both assisted (filled symbols) and non-assisted (open symbols) breaths at each quintile of unloading during acute lung injury (ALI, circles) high (R1, squares) and very high (R2, triangles) resistive load. Blue and red symbols indicate slow and fast ramp increases of the NAVA level, respectively. *= datapoints indicate values obtained at zero NAVA level. ‡=datapoints indicate values obtained at highest NAVA level. VTinsp was related to PL,dyn for the non-assisted breaths (open symbols) and also for the assisted breaths (solid symbols). Both the assisted and non-assisted relationships were shifted towards the right (lower VTinsp for a givenPL,dyn) by the load imposed - ALI to R1 to R2. Progressive increases in the NAVA level increased PL,dyn and Vtinsp during assisted breaths, whereas PL,dyn and Vtinsp decreased during the non-assisted breaths.
Figure E3: Relationship between patient ventilator breath contribution (PVBC) index (x-axis) and the ratio of inspiratory volume (Vtinsp) and mean inspiratory deflection in esophageal pressure (Pes) during non-assisted and assisted breaths ([(VTinsp/∆Pes)no-assist/(VTinsp/∆Pes)assist], y-axis) during progressive increases of the neurally adjusted ventilatory assist (NAVA) level (y=1.1x-0.09, R2=0.96).
The linear relationship shows that Vtinsp normalized to either neural (EAdi) or mechanical (Pes) parameters during assisted and non-assisted breaths have a proportionally linearly increasing difference with increasing unloading. Thus, the curvilinearity for the relationship between Pes/PL,dyn and PVBC cannot be explained by neuromuscular uncoupling. The difference in the slopes and intercepts is likely mainly induced by inability to suppress EAdi at maximal unloading, thus resulting in pressure generation when load is removed during the non-assisted breath.
Values are depicted for all conditions and levels of unloading during acute lung injury (ALI, circles) high (R1, squares) and very high(R2, triangles) resistive load. Blue and red symbols indicate slow and fast ramp increases of the NAVA level, respectively. PVBC = patient ventilator breath contribution.
Figure E4: Relationship between indices determining the respiratory muscles inspiratory pressure contribution by either calculating the ratio of the mean inspiratory deflection in esophageal and transpulmonary pressure during assisted breaths only (Pes/PL,dyn) or by calculating the ratio of PL,dyn during non-assisted and assisted breaths while normalizing for the diaphragm electrical activity (EAdi) i.e. [(PL,dyn/EAdi)no-assist/(PL,dyn/EAdi)assist].
The near linear relationship shows that using EAdi to normalize for changes in neural drive corrects for changes in effort between assisted and non-assisted breaths for [(PL,dyn/EAdi)no-assist/(PL,dyn/EAdi)assist]. The difference in the slopes and intercepts is likely mainly induced by inability to suppress EAdi at maximal unloading, thus resulting in pressure generation when load is removed during the non-assisted breath.
Values are depicted for all conditions and levels of unloading during acute lung injury (ALI, circles) high (R1, squares) and very high (R2, triangles) resistive load.
Blue and red symbols indicate slow and fast ramp increases of the NAVA level, respectively. Figure E4depicts a strong and near linear relationship (R2=0.95) between (PL,dyn/EAdi)no-assist/(PL,dyn/EAdi)assist] and Pes/PL,dynwith a close fit (y=0.97x+0.11) (data represent all conditions and unloading levels).
Figure E5. Bland-Altman plots of data presented in Figure 4. Left panel shows the mean of Pes/PL,dyn and PVBC indices on the x-axis and their difference of the y-axis. Right panel displays the mean of Pes/PL,dyn and PVBC2 indices on the x-axis and their difference of the y-axis. The PVBC2 index reduced the relative zero-offset from -0.25 to -0.06 and the 95% confidence interval from 0.15 to 0.09 when compared to the PVBC index.
In the present study, Vtinspat highest NAVA level was relatively small (about 6-8 ml/kg) and did not change much between conditions (Table 1, Figures 3 and E2) and therefore its effects on the Pes generating capacity for a given neural drive must be considered relatively small and consistent between conditions. However, Pes could have been influenced by dynamic hyperinflation during resistive loading, as suggested by increasing intrinsic PEEP (Table 1). Also, one could speculate that increasing Vtexp relative to Vtinsp observed during non-assisted breaths (Figure 3) could have been suggestive of air-trapping due to instant increase of inspiratory load during the non-assisted breath (reduction in Vtinsp) with no alteration of the expiratory load. As NAVA is synchronized to neural efforts despite intrinsic PEEP, air-trapping could be anticipated during extreme respiratory challenges, both as a consequence of the very high breathing frequencies during ALI and the high airflow limitation during high resistive loading. However, the very strong and near linear relationship for Pes and EAdi (Figure 2), suggests that the functional impact of hyperinflation on Pes generating capacity was small.
The level of applied PEEP affects both EAdi and lung mechanics [E1, E3].Given that the applied PEEP was always set to 5 cmH2O during all conditions and rates of increasing NAVA level we do not expect that our results were influenced by the applied PEEP.
In the present study, we used the ratio ΔPes/ΔPL,dyn to calculate the “true” sharing of effort between patient and ventilator. Given that ΔPes/ΔPL,dynis calculated as ΔPes/(ΔPaw – ΔPes), ΔPes appears both in the numerator and in the denominator which conveniently sets the range of the ΔPes/ΔPL,dyn index between 0 and 1. However, it also opens the possibility that there is some degree of mathematical coupling. Given that the present study protocol consistently increased ΔPaw from initially being very low to finally being very high, causing ΔPes to decrease from its highest to its lowest, changes in their relative contribution should relatively accurately reflected. However, it is important to know that within the limits of physiological measurements lies the problem of obtaining exact data. It is admitted that the resistance of the endotracheal tube causes a decrease in the Paw reaching the lungs, as well as there are also inaccuracies with balloon measurements of gastric and esophageal in small animals. However, we do not believe that neither of these limitations had a major impact on the results in the present study.
A curious finding during NAVA is that the Pes can be abolished whereas EAdi is still present (Figure 2 and Table1). We have previously observed and discussed this phenomenon [E4].Although not entirely sure about all themechanisms it is clear that the application of positive pressure reduces the load to overcome by the respiratory muscles. This in turn reduces the respiratory drive in order not to hyperventilate, in turnwhich could cause hypocapnia and pH shifts. At very high assist levels the load is abolished and drive is low (however still present since hyperventilation does not take place during NAVA). Consequently, pressure can be generated by the inspiratory muscles, unless there is load.
There is still no consensus on the precise definition of the concept of neural inspiratory time or how we define what neural inspiratory time really is. In the NAVA mode, the inspiration to expiration cycling normally occurs when the EAdi drops to 70% of its maximum value (lower percentage if peak EAdi is less than 4 times the EAdi trigger level). This criteria is a compromise a) to maintain a sensitive off-cycling without interrupting inspirations due to miniscule fluctuations of the EAdi signal and b) to best accommodate the complex interplay between neural timing and flow generation (E4) and preventing premature opening of the exhalation valve. One could argue that a "true" neural inspiration is lasting from the onset of EAdi to its peak and that the relaxation phase is not a part of the true neural inspiration.
One could also arguethatthe inspiratory time (Ti) should be considered when comparing the Vtinsp/ΔEAdi between the assisted and the non-assisted breaths? Vtinsp/Ti (i.e. mean inspiratory flow) is a derivative of volume and thus introduces more non-linearity and noise. Changes in Tibetween the assisted and non-assisted breath would alter respiratory mechanics and thus cause changes in e.g. resistance due to changes in flow, the EAdi would change and correct for this (e.g. increased resistance would result in lower Vtinsp and/orincreased EAdi). Normalizing Vtinspto the time integral of the EAdi would introduce an error since a prolonged inspiration at an EAdi plateau would not increase Vtinsp. To conclude, introducing Ti would not likely improve the PVBC index.
REFERENCES
E1. Allo JC, Beck J, Brander L, Brunet F, Slutsky AS, Sinderby C. Influence of neurally adjusted ventilatory assist and positive end-expiratory pressure on breathing pattern in rabbits with acute lung injury. Crit Care Med 2006; 34: 2997-3004
E2. Baydur A, Behrakis PK, Zin WA, Jaeger M, Milic-Emili J. A simple method for assessing the validity of the oesophageal balloon technique. Am Rev Respir Dis 1982; 126: 788–791
E3. Petrof BJ, Legaré M, Goldberg P, Milic-Emili J, Gottfried SB (1990). Continuous positive airway pressure reduces work of breathing and dyspnea during weaning from mechanical ventilation in severe chronic obstructive pulmonary disease. Am Rev Respir Dis 141:281-289E4. Sinderby C, Beck J, Spahija J, de Marchie M, Lacroix J, Navalesi P, Slutsky AS (2007) Inspiratory muscle unloading by neurally adjusted ventilatory assist during maximal inspiratory efforts in healthy subjects. Chest 131:711-717.
E4. Yamada Y, Du HL (2000). Analysis of the mechanisms of expiratory asynchrony in pressure support ventilation: a mathematical approach.J Appl Physiol 88:2143-50.
E5. Laghi F. Effect of inspiratory time and flow settings during assist-control ventilation (2003).Curr Opin Crit Care. 9:39-44.
Table E1. Group mean values for inspiratory time (Ti), expiratory time (Te), and respiratory rate (RR), as a function of each quintile of unloading expressed as Pes in % of Pes at zero NAVA level (i.e. 100% = no unloading), for acute lung injury (ALI), high (R1), and very high (R2) resistive load as well as fast and slow increases of the NAVA level. It should be noted that, during all conditions, Ti decreased during the assisted breath with increasing unloading. This was likely due to a flow or volume inflation reflex (E5) terminating inspiration earlier as the desired Vt was reached faster with increasing levels of assist. During resistive breathing, Te increased during the assisted breath with increasing unloading and RR remained unchanged. This was probably due to the fact that the resistance affected both inspiration and expiration.