Effects of High Flow Nasal Cannula on Deadspace Reduction and Regional Distribution of Ventilation

Background and Significance Historically, oxygen nasal cannulae have been used in hypoxemic patients as a therapy to increase arterial oxygenation. In the neonatal community, oxygen flows >2 L/min and > 4 L/min in the pediatric patient have seldom been used due to the potential for excessive drying of the nasal mucosae and risk for developing hypothermia. It has been speculated that larger pediatric and adult patients are able to better tolerate flow rates in excess of 4 L/min because the large nasal mucosal surface area can support hydration of dry medical grade gases. High flow nasal cannula (HFNC) therapy is a form of respiratory support that provides flows that are believed to be in excess of a patient's spontaneous inspiratory and expiratory flow rates. In the last decade, a proliferation of heated-humidified high flow nasal cannula (HFNC) devices has been introduced into the clinical setting. These devices are intended to provide optimal heating and humidification of medical gases and regardless of flow setting. In the case of neonates, which has been the predominant patient population receiving HFNC, there clearly is no consensus on which flows constitute "high" in this population. If "high" is intended to mean a flow rate that exceeds spontaneous inspiratory flow rate then there is no reasonable clinical measurement to ascertain this relationship. Thus, the true definition of HFNC remains an elusive term and our protocol hopes to shed some light on this subject.

Our prospective randomized trial of the effects of 3 different oxygen nasal cannula flow rates (low, medium, and high) on respiratory rate, SPO2, transcutaneous CO2 and regional distribution of ventilation as measured by EIT will help clinicians define a range of HFNC flow rates in which deadspace washout occurs without positive distending pressure (low range) and in which deadspace washout occurs with positive pressure that creates regional distribution of ventilation changes.

1. Preliminary Studies One neonatal source defines HFNC as flows > 1 L/min and another defines flows >3 L/min as HFNC.1 Classification is further complicated in larger pediatric and adult patients where flows during HFNC have been reported to be in excess of those traditionally used with a standard nasal cannula ~ 6 L/min) and as high as 30-40 L/min.2 With the technologic ability to provide better heat and humidity, clinicians have found that HFNC may be able to support a larger fraction of patients that would otherwise require continuous positive airway pressure (CPAP), noninvasive ventilation (NIV), or invasive mechanical ventilation. There are several proposed mechanisms by which HFNC may provide greater respiratory assistance than standard oxygen delivery devices.

Flows that exceed the expiratory flow rate may provide "back-pressure" at the nasal airway opening during exhalation that is similar to nasal CPAP. Additionally, gases may provide a physiologic purging of carbon dioxide from the anatomic dead space via anatomic leak (nasal/oral airway). These effects are likely to vary based on the flow, minute ventilation, patient size, leak, and nasal airway opening/prong size relationship. There are currently three FDA approved HFNC systems. Clinical acceptance is related to the fact that HFNC is less expensive, simpler to operate and requires a less complicated airway interface than a standard CPAP or NIV device. Another proposed benefit is that HFNC prongs are generally less occlusive and may cause less nasal airway injury than CPAP prongs or a BiPAP mask. The widespread acceptance and use of this approach has been implemented with very little experimental data to support HFNC flow settings as a safe and effective option in all patients with hypoxic respiratory failure.

Electrical Impedance Tomography (EIT) Electrical impedance tomography capitalizes on changes in impendence in air-filled versus tissue-filled spaces to characterize and quantify regional distribution of lung volume at the bedside. This technology has been validated in animal3 and human4, 5 studies performed over the past decade at Boston Children's Hospital. The technology utilizes a series of 16 electrodes placed across the patient's chest (Figure 1). As small currents are passed between the electrodes, impedance is measured between and amongst the series. Through a complex interrogation and manipulation of these impedance values, a two-dimensional image is formed (Figure 2), and has been shown to correlate with clinical and radiographic changes in patients4. The ability to estimate lung volume and regional distribution of gas non-invasively and in real time may give us insight as to what mode of ventilation is more effective.

Physiologic effects and safety of HFNC There are several proposed physiologic mechanisms by which HFNC is believed to be effective. These include: 1) flushing the upper airway deadspace of CO2, allowing for better alveolar gas exchange; 2) providing a flow adequate to support inspiration therefore reducing inspiratory work of breathing; 3) improving lung and airway mechanics by eliminating the effects of drying/cooling; 4) reducing or eliminating the metabolic cost of gas conditioning; and 5) providing positive expiratory distending pressure. While these variables have been measured in either animals or humans, short-term studies only enrolled small numbers of subjects and were not specifically designed to address safety of HFNC.

Hasan et al observed the effects of pressure generated in a static neonatal lung model using two commercially available HFNC devices with flows between 0-12 L/min and at different leak settings.6 They demonstrated in a nares model where leak was minimized (mouth closed) that a systematic increase in simulated tracheal pressures was proportional to increased flows. The measured airway pressures were similar to those reported with nasal CPAP (~5-6 cmH2O at flows of ~6-8 L/min).

Frizzola and colleagues measured tracheal pressures and gas exchange in 13 lung-injured neonatal piglets supported by N-CPAP and HFNC under high-leak and low-leak conditions.7 The major finding of this study was that HFNC tracheal pressures were comparable to CPAP pressures at the same flow range, and washout of nasopharyngeal dead space is associated with improved ventilation and oxygenation independent of tracheal pressure alone during HFNC.

A number of short-term studies have evaluated the magnitude of distending pressure in the lung in a small group of infants. Sreenan et al found that similar end expiratory pleural pressures could be maintained between a standard oxygen delivery nasal cannula (1-2.5 L/min) and N-CPAP in a group of 40 premature infants with no differences in desaturations, bradycardia, and apnea.8 However, this pressure is likely to be highly variable due to leak and airway/cannula size relationship. Lampland observed similar end-expiratory pleural pressures between HFNC (2-6 L/min) and N-CPAP 6 cm H2O in premature neonates.9 A recent Cochrane meta-analysis evaluated prospective, randomized controlled trials that were limited to the premature infant population.1 The primary objective of this meta-analysis was to determine safety and efficacy of HFNC. When used as primary respiratory support after birth, one trial found similar rates of treatment failure in infants treated with HFNC (5-6 L/min) and nasal CPAP.10 Following extubation, one trial found that infants treated with HFNC (1.8 L/min) had a significantly greater need for reintubation than those treated with nasal CPAP.11 Another trial found similar rates of reintubation for humidified and non-humidified HFNC (~ 2-3 L/min)12 and the fourth trial found no difference between two different models of equipment used to deliver humidified HFNC (6 L/min).13 There were few patients enrolled in these studies and in two studies (Woodhead, Miller) poor methodological approaches were used. In one study enrollment was stopped short due to infections related to the HFNC device being used.10 Based on these findings, there is insufficient evidence to establish the safety or effectiveness of HFNC within the range of commonly used flows, and as a form of respiratory support in preterm infants. Also, these data demonstrate that when used following extubation, HFNC may be associated with a higher rate of reintubation than nasal CPAP.14 There have been no studies evaluating safety and efficacy in larger infants and other pediatric patients. Despite the lack of supporting evidence, HFNC is still being implemented in many pediatric intensive care units (PICU) using flows ≥ 20 L/min and few adult studies have been conducted using these high flows during HFNC. Thus, it is difficult to extrapolate these findings to promote the use of similar flows in children. It is apparent that HFNC may provide some of the same clinical benefits as CPAP or even NIV with a less complicated nasal airway or naso-oral interface. While nasal CPAP serves as an intermediary form of support between oxygen therapy and invasive ventilation in neonates, it is more common for larger pediatrics and adults to be supported with bi-level NIV as an alternative to invasive ventilation. There is compelling data to also support the use of NIV in the neonatal population to augment the ventilation effects of nasal CPAP. HFNC not only provides a baseline pressure similar to CPAP but it also augments alveolar ventilation to a level that may be similar to NIV.

1. Design and Methods
2. Study Design
3. In a prospective, randomized trial of three different flow rates of HFNC we will evaluate the outcome of EIT regional distribution, transcutaneous CO2, and respiratory rate.
4. Patient Selection and Inclusion/Exclusion Criteria
5. Inclusion Criteria i. All patients who are receiving HFNC for hypoxia ii. Age: 1 day (> 38 weeks GA neonate or older) to 17 years. b. Exclusion Criteria i. Patients who have congenital heart defects. ii. Patient who the medical team feels may require urgent escalation of non-invasive therapy or imminent intubation. iii. Patients who are on FIO2 > 0.6 at the highest level of flow offered within the study. iv. Patients who are immunocompromised and/or status post bone marrow transplant v. Patients who are on vasoactive support to maintain blood pressure or heart rate vi. Patients with a known airway anomaly, e.g. Pierre-Robin, tracheomalacia. vii. Patients less than 38 weeks gestational age viii. Patients less than 3 kilograms ix. If the EIT band/electrodes are not able to be properly positioned on the chest due to size/weight limitations x. If the medical team feels that the patient is not appropriate to enroll in the study based on medical, social or emotional concerns
6. Description of Study Treatments or Exposures/Predictors Following informed consent, patients who are ordered to receive HFNC therapy will be randomized to escalating flows (low to high) or deescalating (high to low) each hour and then returned to their previously order flow setting. According to age they will be placed on three different flow settings for 1 hour as shown in table 1 with a nasal cannula outside diameter no larger than 50% of the inside diameter of the naris. Table 1 was developed based on three known tidal volumes of 4, 6, and 8 mL/Kg with a 33% inspiratory time and highest normal respiratory rate. Measurements of SPO2, EIT, TCM CO2, and respiratory rate will be recorded every 15 minutes. FIO2 will be adjusted to maintain SPO2 90-95%. Transcutaneous CO2 (TCM) (Sentec) will be placed 30 minutes before randomization in order to allow for equilibration with the surface of the skin. We will place the TCM on left upper chest on patients less than 15 Kg and earlobe for those > 15 Kg. The device is FDA approved and warms the skin to allow for diffusion of CO2 across the membrane of the skin and sensor. This device will allow us to develop a modified ventilation index. EIT measurements - EIT measurements will be taken prior to randomization and 1 hour after each of the three flow range changes. This involves the placement of a 16 electrode band around the patients' chest, just below the nipple line. Pulse oximetry and co-oximetry - SpO2, S/F ratio (SpO2 to FIO2 ratio) SpHb (non-invasive HGB), and SpOC (oxygen content) will be monitored continuously for the 3 hour period. The average SpO2, S/F ratio, and SpOC will be calculated from the streamed data collected by the Masimo RAD-7 via a lap-top computer. Desaturations will be defined as a SpO2 < 85% and reported. Excel will be utilized to perform these functions. An FDA approved disposable SPO2 probe will be applied to the patients' finger, thumb or toe.
7. Definition of Primary and Secondary Outcomes/Endpoints a. Primary i. Oxygenation 1. Higher SPO2 on lower FIO2 (S/F ratio) 2. SpOC, and frequency of desaturations ii. Ventilation 1. Lower respiratory rate 2. Lower TCM CO2 b. Secondary i. Regional distribution difference measured by EIT 1. Area and upper to lower ratio will be the primary data analyzed 2. Regional filling of the lung will be compared.
8. Data Collection Methods, Assessments, Interventions and Schedule (what assessments performed, how often) Data will be recorded continuously on each of the devices' (Draeger EIT, Sentec TCM, Masimo SPO2) computer. Data will be downloaded with either a USB drive or PCMC card and merged into one excel spreadsheet for analysis. Data will be manually recorded every 15 minutes for the entire 3 hours.
9. Study Timeline a. See figure three.
10. Adverse Event Criteria and Reporting Procedures Due to the fact that this is a pilot study the PI will review each major adverse event. The following complications will be monitored but only major complications will reported to the IRB. Minor events include: bradycardia (10% below baseline), increase in respiratory rate > 20%, increase in TCpCO2 by 10 mmHg, increase in FIO2 of > 0.3, hypoventilation (breath hold longer than 15 second) and desaturation (<88%).

Major events that will halt the study and be immediately reported to the IRB are:

• Desaturation < 80% (continuously monitored by pulse oximetry) for longer than 1 minute.
• Bradycardia < 60 BPM All major and minor events will be monitored and reported to the PI by the clinical research coordinator.

1. Data Management Methods Upon entry, each patient will be assigned a number unique and unlinked from their medical record for the purpose of patient tracking. This number will be entered into a privately owned BCH, password protected research drive accessible only by BCH study personnel.

A spreadsheet will be kept at the bedside during the data collection period (3 hours) for each data point to be entered manually.

1. Quality Control Method Quality of the transfer of data will be ensured by a second investigator, who will confirm the manual and electronic data. SPSS software will be used to help analyze the data and ensuring data integrity by establishing alerts for non-filled fields as well as unexpected or possibly misentered results.
2. Data Analysis Plan We will consider that >10% differences in respiratory rate between flow setting to be considered significant. We will consider a > 20% difference in TCM CO2 and SPO2 to be considered significant.

EIT data: The lung imaging system is the Dräger EIT Pulmovista 500 (Dräger Medical, Lübeck, Germany). Sixteen coplanar electrodes will be placed equidistantly around the thorax at the level of the parasternal sixth intercostal space. The reference electrode will be placed on the right side of the abdomen near the waistline. Electrodes #1 and #16 are symmetrically placed to the left and the right of the sternum, respectively, so that electrodes #8 and #9 straddled the spinal column. This configuration leads to transverse images in the radiological convention, caudal to cranial, similar to a cat scan. Lung image reconstruction will be done according to the Graz consensus for electrical impedance tomography (GREIT) (15) using the Electrical Impedance and Diffuse Optical Reconstruction Software (16). In simple terms, the idea is to observe any shifts in the center of ventilation by assessing the ratio of the ventral to dorsal impedance changes ( ) during each part of the study. Impedance changes indicate how open or closed the lung is. This approach has been previously described in detail by our research group (17).

1. Statistical Power and Sample Considerations Power analysis reveals based on repeated measures of three different flows per patient that 35 patients (7 per age category) are required for an effect size of 0.2, alpha of .05 and power of 0.8. Differences in mean values oxygenation (SPO2 and S/F ratio), ventilation (TCM CO2), and EIT (U/L ratio) between each treatment group will be compared every hour over the three hour testing period following randomization using ANOVA with Tukey post-hoc test.
2. Study Organization Single institution pilot study.