Abstract
BACKGROUND: Inhaled nitric oxide (INO) has been used with heated and humidified high-flow nasal cannula (HFNC), nasal CPAP and several forms of noninvasive ventilation (NIV). This study was designed to evaluate the delivered dose of INO, level of NO2 generation, and effect of net gas delivery (addition of INO to the ventilator circuit − gas removed for sampling) on lung pressure at different NO doses during noninvasive respiratory support.
METHODS: An infant lung model was supported with the different noninvasive modes during INO therapy. NO and NO2 were measured from within the patient circuit of the noninvasive devices and simulated neonatal trachea at several NO levels. Lung pressures were compared with and without INO and at several INO settings.
RESULTS: Accuracy of NO delivery was determined to be within the stated accuracy by the manufacturer with nasal CPAP and NIV, but accuracy was compromised during HFNC. INO and NO2 measured by the INOmax DSIR (Ikaria, Hampton, New Jersey) did not consistently reflect the delivered dose of NO or formation of NO2 across all types of neonatal noninvasive respiratory support. Tracheal NO2 levels were < 1.5 ppm with all forms of noninvasive support, except nasal intermittent mandatory ventilation at 40 ppm INO. Lung model mean airway pressures were mildly affected by gas sampling/delivery during combined INO therapy/HFNC at certain flows but remained stable with all other forms of noninvasive support.
CONCLUSIONS: Clinicians cannot always assume that the set INO level results in a similar lung dose when using all forms of neonatal noninvasive support. Clinical decisions regarding ways to improve INO delivery may need to include changing settings or placing patients on a different form of noninvasive support. The NO2 level delivered to the patient could be greater than the value recorded by the INO delivery system.
Introduction
Inhaled nitric oxide (INO) is a potent pulmonary vasodilator drug that is approved for use in term and near-term infants (> 34 weeks of gestation) with acute hypoxic respiratory failure associated with pulmonary hypertension.1 The vast majority of INO research in the neonatal population has been done in subjects receiving conventional or high-frequency mechanical ventilation. Due to the improved understanding of the role of invasive ventilation in the initiation of lung injury2,3 and potential for increasing pulmonary vascular resistance,4 there has been increased interest in combining noninvasive ventilation (NIV) with INO therapy.
Currently, there is only one FDA-cleared delivery device available that maintains a consistent delivered INO dose regardless of the patient gas flow. The device has been validated by the manufacturer to be used with several neonatal noninvasive respiratory support devices, including heated and humidified high-flow nasal cannula (HFNC), nasal CPAP (nasal CPAP), and several ventilators that have been used for NIV. There have been sporadic reports about the combined clinical use of INO therapy delivered through noninvasive devices. Kinsella et al5 first described noninvasive INO delivery through a nasal cannula following surgical repair in a select group of subjects with congenital diaphragmatic hernia and late pulmonary hypertension. There have also been a series of clinical studies describing INO/CPAP in premature infants as part of a treatment strategy, and there were no reported safety concerns in these trials.6–8
Factors that are known to affect the concentration and fluctuation in the inspired concentration of INO include the INO delivery system, respiratory support device type, settings and mode, circuit design, site of delivery of INO, size of mixing chamber (if any) in the patient circuit, concentration of inspired oxygen, and site and method of INO monitoring.9 When INO is delivered during conventional or high-frequency oscillatory ventilation via an endotracheal tube, it can be assumed that the preset INO dose is similar to what the patient receives in the lung.10,11 However, little is known about the compatibility, accuracy, and safety of INO delivery using the array of available noninvasive strategies.
Nitrogen dioxide (NO2) is a toxic by-product that forms when NO and O2 gases are allowed to mix. This chemical reaction can take place in the gas delivery system or noninvasive delivery device, the airway interface, and the lungs.1 Factors influencing the generation of NO2 include FIO2, dose of INO, and dwell time for INO and oxygen.12 NO2 accumulation is more likely to form when using high FIO2 in combination with high INO concentrations. It is unclear whether the complicated physical gas flow pathways of neonatal noninvasive support or the upper airway fosters an environment wherein toxic levels of NO2 can accumulate and be inhaled into the lungs. Additionally, INO delivery systems continuously sample gas, and it is unknown whether the subtraction of gas sampling from the patient circuit or addition of INO gas can affect the patient's mean airway pressure levels. As such, clinicians may be hesitant to use INO therapy in conjunction with noninvasive delivery devices.
This study was designed to evaluate the delivered dose of INO, level of NO2 generation, and effect of sampling on lung parameters at different NO doses in a realistic, spontaneously breathing, neonatal lung model using noninvasive respiratory support. We hypothesized that there would be no differences between: (1) tracheal INO and the preset INO level on the nitric oxide delivery system, (2) NO2 measured on the INO delivery device and tracheal NO2; and (3) mean airway pressure in the lung model at baseline (no INO therapy) and at multiple INO doses.
QUICK LOOK
Current knowledge
Inhaled nitric oxide (INO) is a selective pulmonary vasodilator used to treat acute hypoxemic respiratory failure associated with pulmonary hypertension during mechanical ventilation. Delivery of nitric oxide during noninvasive ventilation (NIV), including high-flow nasal oxygen, has increased with the recognition of the importance of avoiding intubation in infants.
What this paper contributes to our knowledge
In an infant lung model, INO delivery with a commercially available device was accurate with both nasal CPAP and NIV. During high-flow nasal oxygen via cannula, set and delivered nitric oxide concentrations were significantly different. Nitrogen dioxide formation was < 1.5 ppm in all conditions.
Methods
Ikaria (Hampton, New Jersey) reviewed the study protocol to provide technical support where required. The research was conducted by RMD, DD, and CK in the laboratory of RMD at Seattle Children's Hospital in Washington. All data collection, data analysis, and interpretation of data were done by RMD, DD, and CK. EC and JLG, both employees of Ikaria, were present for the study and reviewed the manuscript.
Nasal Airway/Lung Model
The neonatal test lung (ASL 5000, IngMar Medical, Pittsburgh, Pennsylvania) has a screw drive-controlled piston and uses a mathematical model of the equation of motion of the human respiratory system to simulate disease-specific pulmonary mechanics. The lung model was configured to represent the pulmonary mechanics of a term newborn infant (compliance of 2 mL/cm H2O and resistance of 25 cm H2O/L/s).13 The lung model was made to breathe spontaneously by setting a simulated pleural pressure profile of 12 cm H2O, yielding a tidal volume of ∼24 mL with a rate of 40 breaths/min. An anatomically accurate neonatal nasal/upper airway model, scaled to a 3.0-kg infant, was attached to the lung model via a simulated trachea (4.0-mm inner diameter endotracheal tube, 6 cm in length). A similar neonatal nasal airway model has been described in detail elsewhere.14 A potassium permanganate (KMnO4)/charcoal filter was placed at the inlet of the lung model to scrub INO and remove any exhaled NO/NO2 in the gas exiting the lung model to eliminate rebreathing of NO so that the simulated tracheal gas concentrations would consist of inhaled NO and NO2 only.15,16 Figure 1 shows a schematic of the experimental setup. Before and following testing, the efficacy of the KMnO4 filter was evaluated by measuring the NO and NO2 distal to the filter at a flow of 6 L/min with FIO2 of 1.0 and INO of 5, 20, and 40 ppm.
Noninvasive Delivery Devices
We chose to test an array of noninvasive delivery devices that are currently being used in neonatal ICUs within the United States. These included 3 different constant-flow, constant-pressure nasal CPAP devices; one constant-flow, variable-pressure nasal CPAP device; one HFNC (Fisher & Paykel Healthcare, Auckland, New Zealand); and 2 NIV devices (sigh positive airway pressure [SiPAP[, CareFusion, San Diego, CA; and nasal intermittent mandatory ventilation [N-IMV]). Table 1 provides a summary of the different delivery devices, mechanism of pressure generation, and settings. Each device uses inherently different mechanisms to generate pressure. These have been described in greater detail elsewhere.17 Many of these devices use a monophasic flow through a common circuit configuration and/or provide pressure through complex physical pathways. The HFNC provides a continuous monophasic flow to the nasal airway during the respiratory cycle, and patient exhalation occurs through a leak in the naris or oropharynx. The Infant Flow nasal CPAP (CareFusion), SiPAP, and Arabella (Hamilton Medical, Reno Nevada) devices use a fluidic flip-type control, whereas the AirLife (CareFusion) uses vortices to maintain pressure and provide exhalation.17 The bubble CPAP system (Fisher & Paykel Healthcare) uses a piece of tubing placed in a water column that provides pressure proportional to the depth below the water surface, and the gas exiting into the water column results in noisy, small-amplitude, high-frequency pressure oscillations. During N-IMV, variable flow from the ventilator (Avea, CareFusion) provides inspiratory flow and pressure through a dual-limb circuit, and an active exhalation valve located back at the ventilator maintains PEEP. All noninvasive devices were set up according to the manufacturer's recommendations18 with settings that are commonly used in current clinical practice (Table 1). Not all devices tested have been validated by the manufacturer; clinicians should refer to the Ikaria INOmax DSIR operation manual for a list of validated devices. In all testing conditions except HFNC, occlusive short bi-nasal prongs were used. The prong size was selected to ensure a snug fit into the nasal airway opening to minimize leak. The non-occlusive prongs used with the HFNC device were chosen so that the external diameter of the prongs occupied approximately one half of the internal diameter of the naris. Customary heated and humidified gases were used with a Fisher & Paykel MR730 heated humidifier. All devices were set to FIO2 of 1.0 to maximize NO2 generation during testing.
INO Delivery System and Chemiluminescence Analyzer
The INOmax DSIR was used to deliver the INO for all test conditions. The delivery system is capable of delivering a dose of 0.1–80 ppm and uses an injector module containing a mass hot film sensor to continuously monitor inspired gas flow, injecting proportional amounts of INOmax (800 ppm NO with balanced N2) source gas to maintain a constant delivered dose. The INOmax DSIR adds NO/N2 gas to the breathing circuit in proportion to the NO setting and gas flow output of each device. For example, at an 80 ppm NO setting (the maximum NO setting with an 800 ppm NO cylinder concentration), the INOmax DSIR adds 10% more gas to that delivered by the ventilator and proportionally less for lower NO settings.18
The INOmax DSIR has a built-in gas analysis system capable of continuously monitoring INO, NO2, and FIO2 using electrochemical cells for each gas. The monitoring sample system removes gas at a rate of 230 mL/min. Monitored values are updated every 100 ms using a 10-sample rolling average, and the on-screen display is updated every second. The INOmax DSIR was calibrated, and performance was verified according to the manufacturer's recommendations.18 Additional NO gas analysis was performed within the simulated neonatal trachea using a fast-response chemiluminescence analyzer (Sievers NOA 280i, GE Healthcare, Madison, Wisconsin). Chemiluminescent NO measurement is more sensitive and reproducible than electrochemical cells and is considered the accepted standard for measuring NO.19,20 The monitor is capable of analyzing NO samples of < 1.0–500,000 ppb (500 ppm) with a 67-ms response time. Gas samples are withdrawn at ∼200 mL/min. Data from the monitor was continuously uploaded to a computer and recorded at 10 samples/s. The INOmax DSIR, sample system tubing, and injector module were affixed to the respective gas outlets or inspiratory side of each device. The sampling port was configured following manufacturers' recommendations.18 NO2 levels were monitored using the INOmax DSIR sample system and a 3-way stopcock to switch between the sampling locations on the inspiratory circuit of the noninvasive devices and simulated trachea (see Fig. 1). Tracheal NO2 measurements were necessary to determine whether additional mixing of gases in the patient circuit, nasal airway interface, nasal airway model, and trachea created elevated levels of the by-product.
Experimental Conditions
All noninvasive respiratory support devices were attached to the spontaneously breathing lung model via the nasal airway, and following a 5-min stabilization period, breath-by-breath tracheal NO levels were measured using the Sievers chemiluminescence analyzer while NO and NO2 from the patient circuit were recorded from the INOmax DSIR monitor onto a spreadsheet for 1 min at 0 (baseline), 5, 20, and 40 ppm. The baseline measurements were obtained without the INOmax DSIR sampling adapter in-line. Mean airway pressure measurements were made from within the ASL 5000 (version 3.2) for 1 min at each of the testing conditions. Data were reduced from the ASL software to a spreadsheet (Excel, Microsoft, Redmond, Washington). Following this testing, the stopcock was turned (see Fig. 1), and tracheal NO2 was measured in the simulated trachea (before the KMnO4/charcoal filter) using the INOmax DSIR for 1 min, and breath-by-breath data were recorded on a spreadsheet.
Data Analysis
Differences in the preset INO level and mean tracheal NO measurements (n = 20) at each INO setting for each device were compared using a one-sample t test. Accuracy between the preset INO level and tracheal NO measurements were determined by calculating accuracy: % error = (tracheal NO − preset INO level)/preset INO level × 100.
We determined differences a priori between the set INO level and mean delivered tracheal NO as P < .05, but considered anything > 20% or ±2 ppm (whichever is greater) to be outside the manufacturer's specified accuracy for the INOmax DSIR. The stated accuracy of the chemiluminescence analyzer is ±5%. Differences in mean values of measured INO concentration on the INOmax DSIR and the tracheal NO were compared using a paired t test, whereas differences in mean values of NO2 measurements from the patient circuit and simulated trachea were compared using an independent t test. We also determined testing conditions that rendered tracheal NO2 levels of > 1.5 ppm as potentially toxic.
Differences in mean airway pressure between 0 (baseline), 5, 20 and 40 ppm were compared using one-way analysis of variance with Tukey post hoc testing. We determined differences a priori in mean values for mean airway pressure in the lung model between 0 (baseline), 5, 20 and 40 ppm as P < .05. These differences were also calculated: % change from baseline = (measurement − baseline)/baseline × 100, where baseline is the pressure measurement without NO or sampling from the INOmax DSIR, and measurement is the pressure at 5, 20, and 40 ppm NO.
Results
Efficacy of the KMnO4 Filter
NO and NO2 levels were < 0.1 ppm during the KMnO4 testing when NO was set at 5, 20, and 40 ppm before and following the experiment, suggesting that KMnO4 effectively eliminated NO and NO2.
Nitric Oxide Delivery
Measured tracheal NO levels were lower than the preset NO level on the INOmax DSIR for all test conditions during HFNC (P < .05) (Fig. 2). The disparity between the measured tracheal NO levels and the preset NO value were outside the manufacturer's specified accuracy (> 20%) for the INOmax DSIR at all testing conditions during HFNC (Table 2). The NO levels measured by the INOmax DSIR were greater than the tracheal NO levels under all testing conditions during HFNC (P < .05). Overall, increasing the flow above 4 L/min not only resulted in less disparity between the set INO and tracheal NO levels during HFNC, but the tracheal NO levels more closely reflected the set NO level on the INOmax DSIR.
Measured tracheal NO levels were lower than the preset INO level on the INOmax DSIR under every testing conditions for all forms of nasal CPAP and NIV (P < .05) (Fig. 3). The measured tracheal NO levels and the preset NO value were all within 20% under all testing conditions during nasal CPAP and NIV, which is within the specified accuracy for the INOmax DSIR. The percent errors between the preset INO level and tracheal NO for nasal CPAP and NIV devices are reported in Table 2. The NO levels measured by the INOmax DSIR were different from the tracheal NO levels under all testing conditions during nasal CPAP and NIV (P < .05), except for bubble CPAP at 5 ppm (P = .62), Infant Flow nasal CPAP at 5 ppm (P = .08), and AirLife nasal CPAP at 5 ppm (P = .08).
Nitrogen Dioxide Formation
Tracheal NO2 levels were greater than those measured in the patient circuit under all HFNC testing conditions (P < .05), except for HFNC (2 L/min) at 5 ppm NO (P = 1.0) (Fig. 4). There were no NO2 levels > 1.5 ppm measured during HFNC, regardless of flow setting or NO dose being used.
Tracheal NO2 levels were greater than measured NO2 levels made from within the patient circuit during nasal CPAP and NIV (P < .05), except for bubble CPAP at 5 ppm NO (P = 1.0) (Fig. 5). Tracheal NO2 levels were well within a clinically acceptable range (< 1.5 ppm) with all devices and at all NO levels, except for N-IMV at 40 ppm, where the tracheal NO2 level was > 2 ppm.
Lung Model Pressures
There were significant differences in mean airway pressure between baseline and the different NO levels for all HFNC testing conditions (P < .05) (Fig. 6). Differences between mean airway pressure at baseline and at 5, 20, and 40 ppm during HFNC were within 10% for all testing, except for HFNC: 2 L/min at 5 ppm (−13.04%), 2 L/min at 20 ppm (−10.43%), 4 L/min at 5 ppm (−21.36%), 4 L/min at 20 ppm (−19.75%), 6 L/min at 5 ppm (−41.48%), and 8 L/min at 5 ppm (−11.0%).
There were differences in mean airway pressure between baseline and the different NO levels for all nasal CPAP and NIV testing conditions (P < .05) (Fig. 7), except for bubble CPAP (P = .65). Pressures during INO therapy were within 5% of the baseline measurements for each of the noninvasive devices tested.
Discussion
The major findings of this study include the following. (1) The accuracy of NO delivery was within the stated accuracy range for all forms of support that used occlusive nasal airway interfaces (nasal CPAP and NIV), but due to the inherent leak during HFNC, the accuracy was outside of the manufacturer's range (> 20%). The clinically available measurement of INO and NO2, as measured by the INOmax DSIR, did not consistently reflect the delivered dose of NO or formation of NO2 in a tracheal airway model in most types of neonatal noninvasive respiratory support. (2) Tracheal NO2 levels were < 1.5 ppm for all forms of noninvasive support, except N-IMV. (3) Lung model mean airway pressures were mildly affected by gas sampling/delivery during combined INO therapy/HFNC at certain flows but remained stable with all other forms of noninvasive support.
Lindwall et al reported in 2 separate bench studies,21,22 followed by a clinical study,7 that INO can be delivered safely and effectively using a modified continuous-flow titration INO delivery system with the Infant Flow nasal CPAP system. Trevisanuto et al evaluated the feasibility of INO therapy using a neonatal CPAP helmet in a bench study,23 followed by a case report24 describing the successful application in the long-term treatment of an infant with pulmonary hypertension. INO therapy appeared feasible with that system and was found to be an effective treatment option in one subject. However, INO therapy was provided with antiquated continuous-flow NO titration systems, and these bench studies did not use a realistic lung model or fast-response NO analyzer. Since the time of these bench studies, FDA-approved INO delivery systems have continuously improved. Likewise, a number of noninvasive devices, other than nasal CPAP, have been introduced to the market. The current study was designed to provide a realistic clinical/physiologic environment by using a current FDA-approved INO delivery device, a nitric oxide-consuming filter, fast-response and highly accurate NO sensors, and a realistic nasal airway attached to a spontaneously breathing lung model using multiple noninvasive devices.
NO delivery was within the suggested accuracy range using all nasal CPAP/NIV devices. We believe that the major influence on accuracy with these devices was due to the occlusive nature of the nasal prongs. Although we cannot guarantee that these systems were air-tight, measures were used to ensure that similarly sized outer diameter prongs were used with each of the noninvasive devices and that they were fitted snuggly into the nasal airway opening. We postulate that the small differences that were observed between preset and measured NO concentrations are best explained by differences in the accuracy of the electrochemical and chemiluminescence analyzers. Tracheal NO2 accumulation and consequent dilution of tracheal NO may also help to describe why there were differences between NO concentrations measured in the 2 locations, regardless of the device being used. For example, differences in the INOmax DSIR and tracheal NO was negligible when 5 ppm was used because NO2 was nearly absent at the tracheal level. As the dose increased from 20 to 40 ppm, not only did the NO2 levels increase, but in most cases, the disparity between the 2 NO measurements increased. Although we did not test the hypothesis that there were differences in NO delivery between the different nasal CPAP/NIV devices, the respective values of the preset INO dose and tracheal NO concentrations were similar at the respective NO settings between devices. As such, the specialized gas flow pathways during each form of noninvasive support do not appear to be a contributing factor for the accuracy of INO delivery in this lung model.
We speculate that the major reason for the observed disparities between preset and measured NO concentrations during HFNC is due to entrainment of ambient air around the non-occlusive prongs. This leak accounted for nearly a 50% reduction in tracheal NO at flows ≤ 4 L/min at the different NO settings. Differences in NO accuracy can also be explained, with all devices and INO settings, by the dilution effects of NO2 mixing at the tracheal level, especially when higher INO concentrations were used. Our findings with HFNC are consistent with those reported by Kinsella et al,5 who obtained posterior pharyngeal INO measurements using a chemiluminescence analyzer in post-surgical congenital diaphragmatic hernia subjects receiving INO via a nasal cannula. The authors concluded that nasopharyngeal NO concentrations were 5.4 ± 0.5 and 2.4 ± 0.4 ppm with inhaled NO doses of 10 and 5 ppm, respectively. Although these authors used much lower flows with a standard (non-heated/humidified) oxygen cannula during INO administration, we observed a mean value for all HFNC testing that was also ∼40% lower than the preset NO level. Of particular interest, we found that increasing flow by > 2 L/min during HFNC reduced ambient gas entrainment and thus dilutional effects of NO, resulting in less disparity between the preset and delivered doses. On the basis of these data, patients who exhibit poor clinical response to INO therapy may benefit from using higher flows during HFNC. This may serve as an alternative to titration of the INO or FIO2 levels for improved clinical response. Additionally, transitioning to a noninvasive device that uses more occlusive nasal prongs might be another feasible option, but only if this benefit is weighed against the potential risk associated with reduced pulmonary blood flow secondary to increased alveolar pressures.
The NO2 levels measured from within the patient circuit poorly reflected the actual delivered tracheal NO2 dose during all forms of noninvasive support. In fact, we observed that in most devices, tracheal NO2 may be 2–4 times greater than the measured value on the INOmax DSIR. There is potential for NO2 to form in the patient circuit distal to the sampling site and nasal airway interface, but we suspect that these differences are best explained by NO2 accumulation within the simulated nasal airway and trachea. The relatively large dead space of the infant's nasal airway and trachea may serve as a reservoir for NO and O2 to react to form NO2. Although this becomes less of an issue during invasive mechanical ventilation (because the upper airway and a portion of the trachea are bypassed with an endotracheal tube), it is an important consideration when using any noninvasive device with INO therapy. Using FIO2 of 1.0 most likely contributed to the measured NO2 levels as well, and it would be reasonable to assume that lower levels will be seen at a lower FIO2.
As mention previously, NO2 is a toxic gas, and the Occupational Safety and Health Administration recommends that human peak exposure levels remain < 5 ppm.25 In animal studies, inhaled NO2 at ∼2 ppm affected alveolar development and surfactant production, altered the epithelial lining of the terminal bronchioles, and induced loss of cilia.21,26,27 In human studies, inhaled NO2 at ∼2 ppm affected alveolar permeability28 and increased airway responsiveness.29 In another study in humans, alterations of airway reactivity have been reported at exposures as low as 1.5 ppm NO2.30
Potentially toxic NO2 levels (> 2 ppm) were observed only during N-IMV using 40 ppm NO. It is very unlikely that many infants would be supported clinically with 40 ppm NO and FIO2 of 1.0, but nonetheless, this represents a worst case scenario for NO2 to form during noninvasive support. We believe that the observed high NO2 levels during N-IMV are related to the low bias flow (2 L/min) used by the Avea ventilator. All other nasal CPAP and noninvasive devices used higher flows (∼8–10 L/min), and NO2 concentrations never exceeded 1 ppm. As such, lower flows may result in a longer dwell time for gases to chemically react and allow NO2 accumulation within the nasal airway model, simulated trachea, and inspiratory limb of the ventilator.
Sokol et al9 measured NO2 during simulated neonatal ventilation and found NO2 values of < 1 ppm using 80 ppm with FIO2 of ∼1.0. They used a pressure-limited time-cycled continuous-flow ventilator, and flows are generally set between 8 and 10 L/min in these ventilators. The nasal CPAP/NIV devices other than N-IMV used in the current study were preset with flows of ∼8–10 L/min. Although we used 40 ppm as the maximum NO setting, NO2 was < 1 ppm with all testing conditions.
Lindberg and Rydgren12 demonstrated for one ventilator that uses bias flows of < 2 L/min that even small changes in system dead space can result in considerable changes in NO2 formation during NO delivery. They found that insertion of an empty humidifier in the inspiratory limb increased the concentration of NO2 from 4.9 to 8.1 ppm (using 80 ppm NO). When the humidifier was filled with water, NO2 decreased to 6.6 ppm. They also observed greater NO2 concentrations as the ventilatory flows were decreased. This supports the finding from the current study that dead space from the nasal airway model, combined with low bias flows, increases residence time for NO2 and O2 during exhalation, resulting in a bolus of inhaled NO2 during simulated N-IMV. It was interesting to note that NO2 concentration was < 1 ppm with HFNC at 2 L/min and 40 ppm. Because this device uses a monophasic flow, it is likely that fresh gas penetrated the anatomic dead space of the nasal airway model, and NO2 was eliminated through the nasal airway leak.
We report for the first time pressures obtained from within a spontaneously breathing neonatal lung model with combined noninvasive/INO therapy. As mentioned previously, the INOmax DSIR adds gas proportional to the INO setting and the measured flows being generated by the noninvasive device. Conversely, the INO-monitoring system subtracts volume from inspired gases at a rate of 230 mL/min. According to the INOmax DSIR operator's manual, the approximate amount of INO flow added to the system when using 20 ppm at 10 L/min is 250 mL/min. This would account for a net gain of gas of ∼20 mL/min being added to the system. At the lower NO settings, less NO gas is added to the system, but the gas sampling rate is fixed at 230 mL/min. Conversely, NO settings greater than 20 ppm at 10 L/min add > 250 mL/min NO to the system. As such, sampling at lower NO settings could affect lung recruitment by reducing the mean airway pressure. Adding additional gases to the system at higher NO settings could increase mean airway pressure and place infants at greater risk for lung injury. We found that gas sampling affects mean airway pressure levels during HFNC, especially at flows of > 4 L/min, but there did not appear to be a large effect when using other forms of noninvasive support.
Although we did not observe major differences in pressure related to INO therapy, it is still necessary to continuously monitor the ventilation system for disparities between the set and measured parameters before and after implementation of INO therapy and with changes in the INO settings.
Limitations
In this in vitro study, we took careful measures to simulate a realistic neonatal clinical environment. We used modern noninvasive respiratory support devices, a heated humidifier, a realistic nasal airway model, a filter that consumed NO and NO2 similar to the human lung, and an actively breathing test lung configured with mechanics similar to those measured in neonates. However, like all data obtained from studies in vitro, the results must be approached with some trepidation. We used only one lung model configuration and settings that are commonly used with noninvasive devices. We did not test the range of settings that are commonly used. It is possible that different support settings and lung mechanics configurations could render different results. Furthermore, we used a chemiluminescence analyzer that sampled gas from the simulated neonatal trachea. It is quite possible that this sampling may have entrained room air around the non-occlusive prongs with HFNC, resulting in potentially lower tracheal NO levels than those that would be delivered under ideal physiologic conditions (without an analyzer). Because this testing included only a simulated nasal airway, all delivered gas concentrations could be grossly overestimated because an oral leak was not present in this model. As such, it should be assumed that in the presence of an oral leak during noninvasive/INO therapy, entrainment/dilution by ambient air will reduce the NO/NO2 levels to the patient. Finally, we tested only FDA-approved neonatal noninvasive devices because INO is FDA-approved only in neonates at > 34 weeks of gestation. Clinicians should not extrapolate from these data and expect similar results in a larger pediatric or adult lung model using INO with approved noninvasive devices used in this population. Future studies enrolling live subjects need to be conducted to determine whether these findings are similar.
Conclusions
On the basis of these findings, clinicians cannot assume that the NO level set on the INO delivery system will always result in a similar dose being applied to patients when using all forms of neonatal noninvasive support. Leak at the nasal airway interface may be a major limiting factor for neonates receiving the desired INO dose with any noninvasive device. HFNC is intended to be used as an inherently leaky device, and disparities between the preset and delivered INO levels are always likely to exist. Clinical decisions regarding ways to improve INO delivery may need to include manipulating noninvasive settings, minimizing leak, or placing patients on a different form of support. The NO2 level delivered to the patient could be greater than the value recorded by the INO delivery system because the nasal airway interface and nasal airway serve as additional reservoirs for oxygen to combine with NO, particularly at FIO2 of 1.0. Pressures are relatively unaffected with most forms of noninvasive support combined with INO therapy. Although pressures are typically not monitored during HFNC, INO therapy sampling at higher flows may affect delivered pressures.
Footnotes
- Correspondence: Robert M DiBlasi RRT-NPS FAARC, Center for Developmental Therapeutics, Seattle Children's Hospital Research Institute, 1900 Ninth Avenue North, Seattle, WA 98101. E-mail: robert.diblasi{at}seattlechildrens.org.
This study was funded in part by Ikaria through donation of the drug and delivery system. The authors have disclosed no other conflicts of interest.
- Copyright © 2015 by Daedalus Enterprises