Comment

From PROSpect to the bedside: contextual and pathophysiological considerations

Jeremy M Loberger, Robinder G Khemani

Lancet 2026; 408: 1788-1789

https://doi.org/10.1016/S0140-6736(26)02054-4

Paediatric acute respiratory distress syndrome (PARDS) is a common, severe manifestation of direct or indirect lung injury in children. Management focuses on the careful implementation of lung protective strategies with high-frequency oscillatory ventilation (HFOV) and prone positioning often deployed in the most severe cases. As such, we read the Prone positioning and high-frequency oscillatory ventilation in early paediatric acute respiratory distress syndrome (PROSpect) trial by Martha A Q Curley and colleagues published in The Lancet with great interest.¹ The study team should be commended for their work in conducting such a complex, multinational clinical trial (with an intervening pandemic) to answer a central question in PARDS.

PROSpect used a randomised, 2 × 2 factorial design with response-adaptive randomisation to compare the early use of HFOV versus conventional mechanical ventilation (CMV) and supine versus prone positioning in patients with moderate to severe PARDS.² The trial enrolled over 400 children with PARDS (54 of 75 participating intensive care units enrolled at least one patient), from countries in North America, South America, Europe, Oceania, and Asia. Patients were assigned to supine and CMV (n=126), prone and conventional mechanical ventilation (n=130), supine and HFOV (n=90), or prone and HFOV (n=78). The median age of patients was 3·3 years with equal representation of females and males. The authors concluded that use of prone positioning or HFOV as early first-line strategies did not improve ventilator-free days to day 28 (the primary outcome) in comparison to supine positioning or CMV.¹ Of note, the trial was stopped early for futility and slow enrolment. The natural question after reading the results of this trial focuses on how to apply these findings to the use of HFOV and prone positioning in our PARDS patients in the future.

We believe PROSpect has helped to answer a crucial question for the bedside paediatric intensivist. For the typical PARDS patient, early use of HFOV or prone positioning is unlikely to yield an appreciable benefit over CMV or supine positioning. Hence, routine application of these therapies when patients first meet moderate or severe PARDS criteria does not seem warranted, particularly when rates of adverse events such as pressure injuries and endotracheal tube obstruction were slightly higher with prone versus supine positioning, as was haemodynamic compromise with HFOV versus CMV.

However, as the authors also state, there might still be a role for these therapies in PARDS. Although PROSpect focused on the early use of prone positioning and HFOV, paediatric practice typically reserves these therapies for patients who do not respond to conventional management (ie, rescue therapy). Observational studies such as the international, cross-sectional PARDS Incidence and Epidemiology (PARDIE) study highlighted that prone positioning and HFOV in usual practice were typically initiated with worse oxygenation (median oxygenation index of 19·9 for prone positioning and 25·7 for HFOV in PARDIE compared with 20·8 in PROSpect) and later in the course.³ This practice might be naturally selecting for patients where the risk–benefit profile for prone positioning or HFOV is different than the global population of moderate to severe PARDS tested in PROSpect.

Specifically, PARDS clinical practice guidelines recommend considering alternative strategies (such as prone positioning or HFOV) when lung protective limits for plateau and driving pressure (like the CMV arm in PROSpect) yield insufficient gas exchange (such as those with the most severely impaired respiratory system compliance).⁴ This trial provides no clear evidence against these recommendations, as prone positioning or HFOV might still have a better risk–benefit profile than alternatives (such as extracorporeal membrane oxygenation [ECMO]) for patients not responding to CMV. Interestingly, ECMO was used in only 35 (9%) of the PROSpect patients, including the CMV group. This outcome highlights that for the majority of patients with moderate to severe PARDS, conventional lung protective ventilation strategies can provide adequate gas exchange within lung protective limits, although daily driving pressure was not reported in this trial so we cannot confirm whether these limits were maintained.

There has been an increasing focus over the past decade on trying to identify the populations of patients most likely to benefit from an intervention (ie, predictive enrichment), with notable heterogeneity in treatment effect for many critical care interventions (ie, differential benefit of a therapy based on patient-specific factors). However, the phenotype of patients who receive rescue use of HFOV or prone positioning in routine clinical practice might already represent an enriched population, as they are not responding to conventional therapy. There might also be biological, physiological, or clinical factors which drive this response.5, 6, 7, 8 There is a strong possibility that there is heterogeneity in the treatment effect of routine application of prone positioning or HFOV based on patient characteristics, and we look forward to secondary analyses of the PROSpect trial to potentially identify target populations who might have benefited. As the authors highlight in the discussion, they have preplanned analyses which will focus on biological pathways such as inflammation to identify subphenotypes that might drive heterogeneity in treatment effect.⁹

We also believe there are clinical and physiological factors which might drive the response to therapies such as prone positioning and HFOV. These include elements such as respiratory system compliance (as a marker of risk for failure to maintain lung protective targets), lung recruitability (as a marker of potential to respond to HFOV), PARDS trigger (as a potential source for more severe ventilation–perfusion mismatch), and lung versus chest wall mechanics (as a marker of potential response to prone positioning). Hopefully, some of this information can be gleaned from secondary analysis of PROSpect to help identify the populations of patients who might benefit from either routine or rescue use of these therapies, which can shape our future research priorities.

JML declares no competing interests. RGK is a medical advisor of Nihon Kohden; has research funding from the US National Institutes of Health; and is Chair of the Observational Study Monitoring Board for the POST-VENT Neonatal ICU.