Non-invasive Respiratory Support in Preterm Infants

BACKGROUND:

Bronchopulmonary dysplasia (BPD) is the most common morbidity of prematurity, and affects up to 40% of all preterm infants. It is independently associated with long term pulmonary and neurodevelopmental impairment, and confers a significant cost to society. Although development of BPD is multifactorial in nature, endotracheal mechanical ventilation (EMV) is known to be a significant contributor and strategies aimed at avoiding endotracheal mechanical ventilation may indeed reduce the incidence of bronchopulmonary dysplasia.

Use of non-invasive respiratory support (NRS) has increased over the last decades in an effort to minimize dependence on endotracheal ventilation and reduce BPD in preterm neonates. Nasal continuous positive airway pressure (NCPAP) remains the prototypical form of NRS and is most commonly utilized. Another commonly used modality is non-invasive positive pressure ventilation (NIPPV) though it is often relegated as a "rescue" NRS mode following failure of NCPAP. NIPPV mimics tidal breathing by delivering "breaths" through a nasal interface at regular intervals, and results in higher mean airway pressures compared to NCPAP. Despite the increase in use of NRS and evidence from clinical studies demonstrating effectiveness in reducing BPD,results at a population level have been rather disappointing, with a recent report from the United States actually showing an increasing trend in BPD.

In recent years, in an effort to further minimize dependence on endotracheal ventilation and BPD, some new strategies in NRS use are emerging. One of these approaches is the early use of NIPPV in place of NCPAP, rather than as a rescue mode after failure of NCPAP. Two recent Cochrane reviews have evaluated NIPPV as an alternative to NCPAP, and suggested superiority over NCPAP Based on these results, a number of NICUs have started utilizing NIPPV in place of NCPAP as the default NRS mode. However, there are a number of methodological limitations of studies that were included in these reviews. First, most studies did not allow for rescue use of NIPPV in the NCPAP arms, which is not representative of current clinical practice at most NICUs. Secondly, the mean airway pressures in NIPPV vs. NCPAP were vastly different, and the question remains whether it is truly the mechanism of NIPPV or simply higher mean airway pressure that leads to the superior outcomes reported. These limitations warrant consideration because most NIPPV use is not synchronized to patients' respiratory efforts, the optimal and safe peak pressures remain unknown.

An alternative approach that has emerged in recent years is the use of high end-expiratory pressures (defined as pressures > 8 cm H2O) on NCPAP. As of 2015, use of high NCPAP pressures had only been adopted at 5 out of 28 Canadian neonatal intensive care units (NICU), and McMaster Children's Hospital NICU is currently one of only few Canadian centres employing pressures > 12 cm H2O on NCPAP. The reasons for the lack of widespread use of high end-expiratory pressures on NCPAP are unknown, but may relate to the theoretical concerns of altered systemic hemodynamics including impaired cardiac venous return, decreased venous drainage of the cerebral circulation as well as the potential for air leak syndromes such as pneumothoraces. On the other hand, such high end-expiratory pressures are very commonly and safely used during endotracheal mechanical ventilation, as well as during NIPPV. As such, the selective use of high end-expiratory pressures on NCPAP non-invasively may be quite appropriate. However, the use of high end-expiratory pressures on NCPAP is currently based on very limited evidence, and whether adopting this strategy leads to a meaningful decline in the dependence on endotracheal ventilation and/or bronchopulmonary dysplasia remains unknown. In this study, we aim to comparatively evaluate use of NIPPV vs. high end-expiratory pressures (> 8 cmH2O) on NCPAP after failure of NCPAP use at traditional pressures (≤ 8 cmH2O) among preterm infants. This will be the first prospective clinical study to evaluate high end-expiratory pressures on NCPAP, and also the first to compare NIPPV and NCPAP at equivalent mean airway pressures. If high NCPAP is shown to be as effective and safe as NIPPV, it will lead to a significant advancement in the understanding of NRS and the respiratory management of extremely preterm neonates.

OBJECTIVES AND HYPOTHESIS:

To determine the feasibility of the conduct of a larger and definitive trial comparing two NRS strategies (NIPPV vs. high NCPAP pressures) in preterm infants. Therefore to prepare for this a pilot trial is proposed. We hypothesize that the conduct of a definitive trial comparing these two NRS modes to assess clinical outcomes will be feasible.

METHODS:

Parents of all patients with gestational age < 29 weeks who do not meet any of the exclusion criteria will be approached for consent by the research assistant at the earliest opportunity. Consent will be sought within the first 72 hours. Subjects whose parents provide consent AFTER 72 days will still be eligible, as long as there is no administration of high NCPAP or NIPPV outside of randomization for greater than 4 continuous hours. A subject for whom consent is obtained from the parents will be considered an "enrolled subject". The research assistant will be responsible for indicating this by placing a laminated card at the bedside, and informing both the medical and respiratory therapy teams.

Enrolled subjects who meet NRS failure criteria despite NCPAP 8 cmH2O, or are being extubated from invasive mechanical ventilation with a mean airway pressure >/= 10 cmH2O will be eligible for randomization to either high CPAP > 8 cmH2O (experimental arm) or NIPPV (control arm). Once NRS failure criteria (indicated on bedside laminated card) and randomization eligibility is confirmed, the subject will be randomized using an electronic web-based secure platform (REDCap). An enrolled subject who gets randomized to one of the two arms will be considered a "randomized subject". A study initiation form will be completed by the research team within 24 hours of randomization, and a new laminated card indicating arm of randomization and extubation criteria will replace the enrollment card. This form will include demographic data about the subject, the NRS failure criteria that was met which led to randomization, and the suspected underlying pathophysiological mechanism for NRS failure.

Guidelines for initiation and recommended increments of both high CPAP and NIPPV will be provided, with a view towards maintaining similar mean airway pressures between the two modes (which will also be placed at patient bedside in randomized subjects). Given the existing variability of practice in NRS application and lack of evidence-based guidelines, the strategies will NOT provide guidance with regards to day-to-day changes in settings or modalities, but rather provide only ceiling limits and maximum allowed pressures for each NRS mode. Interfaces that will be allowed include either nasal masks or short bi-nasal prongs, as well as non-occlusive prongs (e.g. RAM cannula). However, ceiling limits and maximum allowed pressures for each NRS mode will be provided, as delineated below:

• High CPAP - Maximum PEEP: 15 cmH2O
• NIPPV - Maximum (set) PIP: 25 cmH2O; AND/OR Maximum (set) PEEP: 10 cmH2O; AND/OR Maximum (calculated) MAP: 15 cmH2O

If NRS failure criteria are met despite escalating settings (with a strongly suggested - but not mandated - minimum MAP of 12 cmH2O) whether either randomized arm of high NCPAP or NIPPV, the clinician will have one of following three options:

• Intubation and endotracheal mechanical ventilation
• Escalation of settings within the randomized arm beyond the aforementioned ceiling limit - will still be considered failure of assigned mode
• Use of an alternate mode of NRS (NIV-NAVA or NIHFV) - but no crossover

Of note, the maximum recommended settings for each NRS mode need not be reached before decision to intubate a subject, at the medical team's discretion. Furthermore, the NRS strategy corresponding to the randomization arm reflects use of NRS at any and all times after randomization until a randomized subject is discharged or transferred. The decision to wean down to traditional NCPAP levels from either arm will be at discretion of the medical team. Finally, subjects randomized to a particular strategy will NOT be allowed to cross-over to the other arm until the first intubation. Patients may be pulled out from the study protocol only at the discretion of the attending medical physician following the first intubation, and the reason must be discussed in advance with the PI (A.M.). A customized protocol violation form will then be completed by the attending physician, facilitated by the research assistant.

Two major forms of protocol violations will be tracked prospectively (by research coordinator):

1. Use of NIPPV or high CPAP >4 hours in an enrolled patient outside of randomization. Such patients will not be included in any further analysis of clinical outcomes, but will count towards feasibility outcome.
2. A patient with protocol violation related to cross-over; such a patient will be treated as intention-to-treat for clinical outcomes, but a set of sensitivity analyses excluding such patients (i.e. those that underwent protocol violation) will be conducted (see section below on Sensitivity Analyses).

Assessment of cerebral and renal regional perfusion will be performed using near-infrared spectroscopy (NIRS) for all subjects (at the primary study site only) at time of randomization. At the time an enrolled subject meets NRS failure criteria, the NIRS leads will be placed and recording will be initiated. Perfusion will be assessed immediately prior to and continuously for up to 7 days post-randomization (or intubation if sooner that 7 days). This will be conducted by the research assistant during office hours and the respiratory therapist at night.

At the end of the study period, relevant demographic data and outcome variables for all randomized subjects will be extracted from the local Canadian Neonatal Network (CNN) database (or other equivalent local database) and from patient charts/electronic records by the research coordinator using a customized data collection form.

DATA ANALYSIS, STATISTICS AND SAMPLE SIZE CALCULATIONS

All randomized patients will be analyzed using intention-to-treat principle. Mean (SD) or Median (IQR) will be reported for continuous variables depending on normality of data, and percentages will be reported for categorical variables. T-test (or Wilcoxon rank sum test for non-normally distributed variables) and Chi-square test (or Fisher's exact test as appropriate) will be used to compare continuous and categorical variables, respectively. All analyses will be performed using SAS version 9.4.

Sensitivity analyses (for only the main secondary outcome) based on the following variables:

The following 4 sets of sensitivity analyses will be conducted for secondary outcome #1 only by exclusion of:

1. Patients with at least one protocol violation due to crossover to alternate study arm within 1st week post-randomization
2. Patients for whom a minimum MAP of 12 cmH2O was NOT utilized prior to non-invasive support failure on at least one occasion within 1st week post-randomization
3. Patients who failed post-randomization non-invasive support due to a non-respiratory pathology on at least one occasion within 1st week post-randomization
4. Patients placed on study intervention using RAM cannula at time of randomization

We will only perform sensitivity analyses if any one or more of the above results in exclusion of ≥20% of the randomized cohort.

Regression analyses: No regression analyses are planned a-priori for this pilot trial.

Regional perfusion data using NIRS (from participant data at primary study site only) will be compared within each randomization arm (immediately prior to and 2 hours post-randomization) as well as between the two randomization arms 2 hours-post randomization using paired t-test and unpaired t-tests, respectively.

A convenience sample of 100 subjects (40 at the lead institution and 20 per site at the other 3 centres) will be chosen for the purposes of this pilot trial. Analysis of secondary outcomes from this pilot study will help inform the design and sample size for a definitive trial comparing high CPAP and NIPPV. The feasibility outcome goals were used to determine the sample size, based on methods described by Thabane et al.

DATA SAFETY AND MONITORING BOARD (DSMB): A DSMB will perform two reviews - one following the transition of the first phase of the study (30 randomized patients) and prior to commencement of the second phase, with a second review occurring after 40 patients have been randomized (70 total) in the second phase.

Appendix: NRS Failure Criteria

• FiO2 >50% or rise in FiO2 >20% in ≤12 hours
• High CO2 with pH <7.20 (respiratory acidosis) on arterial or capillary blood gas
• Increased work of breathing (with RR >80 bpm) for at least 10 minutes
• Modified DOWNES' score to be completed in real time to objectively describe components of increased work of breathing
• Apnea/Desaturation/Bradycardia spells (>1 requiring bagging in a 4 hour period or >4/hour requiring moderate stimulation x 4 hours)
• Need for intubation related to non-respiratory pathology including but not limited to necrotizing enterocolitis, septic deterioration and hemodynamically significant ductus arteriosus where none of the above failure criteria are applicable [this option only applicable to post-randomization NRS failure]