High-flow nasal oxygen versus noninvasive ventilation in acute respiratory failure: reflections on the RENOVATE trial

We read the recent RENOVATE trial in the December 10th, 2024 edition of the Journal of the American Medical Association with great interest (1). This study adds to a growing body of literature exploring the use of high-flow nasal oxygen (HFNO) compared with noninvasive ventilation (NIV) in acute respiratory diseases (2-4). Specifically, RENOVATE was a randomized controlled trial of 1,766 patients comparing the use of HFNO versus NIV for those with hypoxemia and five different clinical conditions: immunocompromised state, non-immunocompromised state, acute cardiogenic pulmonary oedema (ACPO), chronic obstructive pulmonary disease (COPD) exacerbation with respiratory acidosis, and coronavirus disease 2019 (COVID-19). Many of these diseases would traditionally be managed with NIV (particularly ACPO and COPD with respiratory acidosis), rather than HFNO, owing to a wealth of robust literature and experience in support of this approach in these clinical conditions (5,6). While this study demonstrated noninferiority of HFNO with respect to the primary outcome of endotracheal intubation and 7-day mortality in all but the immunocompromised cohorts, discussion regarding the limitations of this study and its place within the context of existing literature on the topic is necessary to understand how it may influence patient care.

This study supports the superiority of HFNO for patient comfort in instances of acute respiratory failure (ARF) as compared with NIV. Patients receiving HFNO reported significantly higher comfort scores at every interval, and sedative use was more frequent in all NIV groups, particularly among patients with COVID-19 (7). This finding is clinically meaningful, as intolerance of NIV is a common driver of treatment failure, escalation to intubation, increased sedative exposure, and prolonged intensive care unit length of stay. Importantly, sedative use is not benign and carries increased risk of adverse outcomes, including delirium, hypotension, and bradycardia (8). These risks may be mitigated through use of a more tolerable respiratory support modality when clinically appropriate.

Despite these advantages, the generalizability of the RENOVATE trial is constrained by the mild-to-moderate severity of illness across cohorts. In the ACPO group, patients had partial pressure of oxygen in arterial blood (PaO₂) to fraction of inspired oxygen (FiO₂) ratios of 247 in the HFNO arm and 235 in the NIV arm. Moreover, patients requiring NIV prior to randomization were excluded, potentially biasing this cohort toward less severe disease. While prior studies have demonstrated benefit of HFNO in more severe presentations of ACPO, the RENOVATE data cannot be extrapolated to such patients (9-11).

Similar concerns regarding illness severity apply to the COPD cohort. Patients with COPD and hypercapnia had median baseline arterial pH values of 7.30 and 7.32 and partial pressure of carbon dioxide in arterial blood (PaCO₂) levels of 64 and 55 mmHg in the HFNO and NIV groups, respectively. The finding of non-inferiority of HFNO in this population contrasts with prior literature evaluating patients with mild-to-moderate hypercapnic respiratory failure. For example, Tan et al. did not find non-inferiority of HFNO compared with NIV in a similar cohort (3). However, an important methodological distinction is that crossover to NIV was considered treatment failure in the Tan study, whereas it was permitted in RENOVATE. Taken together, these findings may suggest that HFNO may be reasonable as an initial respiratory support strategy for mild-to-moderate ACPO and COPD with acute hypercapnic respiratory failure, provided there is a low threshold for escalation to NIV or intubation.

Extending these concerns about disease severity and external validity, the non-immunocompromised cohort also warrants careful interpretation. While these patients had some of the lowest PaO₂/FiO₂ ratios across the trial, they were still not as severely ill as patients in prior work demonstrating a mortality benefit of HFNO over NIV or conventional oxygen support (2). Notably, a trend toward reduced mortality in favor of HFNO did not emerge until the 28-day analysis in RENOVATE, which is the same timeframe where benefit was observed in the trial by Frat et al. These findings further support the feasibility of HFNO as a primary respiratory support modality in non-immunocompromised patients with hypoxemia, particularly in less severe disease states.

While these findings are clinically meaningful, they must be considered in the context of several important limitations, some of which the authors acknowledge. First, the composite outcome of intubation or death within 7 days is both patient-centric but also of unequal significance. Indeed, death is a more severe outcome than intubation, yet both are consequential in the context of ARF. Importantly, death was not included in the original primary outcome and was added nearly two years into the study after the power calculations had been performed. Simultaneously, the use of a 7-day composite outcome may better capture the acute phase of illness than the more commonly used 28-day endpoint. These methodological choices complicate interpretation and temper the strength of the conclusions that can be drawn from this study.

A second methodological issue relates to the criteria for endotracheal intubation, which were notably conservative. As outlined in Supplement 1 of the RENOVATE trial, indications included hemodynamic instability, Glasgow Coma Scale less than 11, arrhythmia with hemodynamic compromise, refractory hypoxemia, and others. Of these, refractory hypoxemia, defined as “an inability to maintain an oxygen saturation of 92% despite an FiO₂ of 60%” is particularly noteworthy, especially since it represented the most common reason for intubation. This threshold may not reflect contemporary clinical practice and does not account for the role of expiratory positive airway pressure (EPAP), a major determinant of oxygenation. Notably, EPAP was not administered at pressures higher than 10 cmH₂O in any cohort despite an allowance of up to 12 cmH₂O by protocol. These conservative intubation criteria likely influenced the number of intubations observed during this study, further limiting its generalizability.

Another consideration is the overall trial design, which introduces additional complexity. The RENOVATE trial may be conceptualized as one manuscript with five studies, rather than one study with five subgroups. Their primary outcome was analyzed using dynamic borrowing, which is a statistical method in which data is shared across groups, and they modeled which groups were most similar. They utilized four prespecified clusters to guide dynamic borrowing, a Bayesian approach in which outcome data are shared across groups deemed most similar. This method assumes that the clustered populations are sufficiently alike in terms of underlying pathophysiology, response to respiratory support, and clinically meaningful outcomes. However, given the heterogeneity among the included disease states, the appropriateness of borrowing across these groups is uncertain. Differences in disease mechanisms and expected treatment effects may limit the validity of this approach and complicate interpretation of the primary outcome (12).

The timing of the trial during the height of the COVID-19 pandemic presents an additional challenge. Although the authors appropriately designated COVID-19 as a separate cohort, this group was larger than the other four cohorts combined and likely exerted a disproportionate influence on the borrowing analysis. Therefore, these data from the non-COVID groups need to be taken in context as they are much smaller populations. Moreover, extrapolating findings from the COVID-19 cohort to contemporary practice is difficult, given rapid and substantial changes in management during and after the trial period, including the adoption of systemic corticosteroids (13), evolving strategies regarding timing of intubation (14), use of adjunctive therapies (15), emergence of new viral variants, and widespread vaccination. As a result, the true effect of oxygen delivery modality on intubation and mortality in COVID-19-related respiratory failure remains uncertain.

ARF is a complex and heterogeneous syndrome and regardless of the modality chosen, these patients require close monitoring and thoughtful care. The RENOVATE study represents an ambitious effort to explore a common and important question of whether HFNO can be used in place of NIV across a range of ARF scenarios. As with many large trials, methodological limitations necessitate cautious interpretation and further validation. With these considerations in mind, HFNO appears to be non-inferior to NIV for several conditions, including COVID-19, mild-to-moderate ACPO, and mild-to-moderate COPD exacerbation with hypercapnic respiratory failure, for which HFNO has traditionally not been used. Additionally, HFNO is almost certainly superior to NIV with respect to patient comfort. Future research should focus on identifying which subsets of patients can safely and effectively be managed with HFNO as a primary modality, while incorporating patient-centered outcomes such as functional recovery and quality of life in addition to mortality.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.