Highlight
– A multicenter pilot randomized trial compared conservative (post‑oxygenator PPOSTO2 100–150 mm Hg) vs liberal (FDO2 100%) extracorporeal oxygen strategies during VA‑ECMO for cardiogenic shock.
– Target separation was achieved (mean PPOSTO2 ~139 vs ~420 mm Hg), but conservative targets were maintained only 33% of the time.
– No between‑group differences were observed in the primary biomarker (plasma IFABP at day 2) or secondary markers of hepatic, renal dysfunction and inflammation.
Background: oxygen management during VA‑ECMO and the clinical question
Venoarterial extracorporeal membrane oxygenation (VA‑ECMO) is a life‑saving circulatory and respiratory support modality used in refractory cardiogenic shock. The extracorporeal circuit oxygenator allows clinicians to fully control gas exchange delivered to extracorporeal blood. Practice varies widely: many centers default to a liberal strategy (sweep gas fraction of inspired oxygen [FDO2] 100%), in part for safety and simplicity; others worry that sustained hyperoxia promotes oxidative stress, vasoconstriction and organ injury. Observational data in critical illness and postcardiac arrest settings link hyperoxia to worse outcomes, but randomized data are scarce, especially for patients on VA‑ECMO where native and extracorporeal oxygenation interact unpredictably.
The trial by Winiszewski and colleagues asked whether titrating extracorporeal oxygen to a conservative post‑oxygenator partial pressure (PPOSTO2 100–150 mm Hg) reduces early organ damage compared with maintaining FDO2 at 100% (liberal strategy), using intestinal fatty acid binding protein (IFABP) at day 2 as a marker of gut epithelial injury.
Study design and methods
This was a pilot, multicenter randomized controlled trial conducted in four ICUs across three teaching hospitals in eastern France. Adults with cardiogenic shock supported with VA‑ECMO were randomized to:
– Conservative arm: titration of sweep gas FDO2 to target post‑oxygenator PaO2 (PPOSTO2) 100–150 mm Hg.
– Liberal arm: sweep gas FDO2 maintained at 100%.
The primary endpoint was mean plasma IFABP at day 2. Secondary endpoints included feasibility metrics (proportion of time within target), biochemical markers of hepatic (AST, prothrombin time) and renal injury (serum creatinine, need for renal replacement therapy), lactate and vasoactive inotropic score, and inflammatory cytokines (TNF‑α, IL‑6, IL‑8) and antioxidant system markers. Fifty‑five patients were analyzed (29 conservative, 26 liberal).
Key results
Separation and feasibility
– The two groups were meaningfully separated in extracorporeal oxygenation over the first 7 days: mean FDO2 61 (±7)% in the conservative arm versus 98 (±5)% in the liberal arm; corresponding PPOSTO2 139 (±40) mm Hg vs 420 (±50) mm Hg.
– However, the conservative strategy was difficult to implement: patients in that arm spent a mean of only 33% (±20%) of monitored time within the predefined PPOSTO2 window (100–150 mm Hg).
Primary biomarker
– Plasma IFABP at day 2 (primary outcome) did not differ significantly: median 407 pg/mL (IQR 206–549) in the conservative arm versus 569 pg/mL (IQR 247–708) in the liberal arm (p = 0.25).
Secondary biochemical and clinical measures
– No significant differences were observed between groups in lactate, vasoactive inotropic score, AST, prothrombin time, serum creatinine, need for renal replacement therapy, or the inflammatory cytokines measured (TNF‑α, IL‑6, IL‑8) on days 0, 2, and 6 after randomization.
– The trial did not report major safety signals attributable to the conservative strategy (no excess hypoxia‑related adverse events reported), though event counts were small.
Interpretation and mechanistic considerations
The trial demonstrates two important, interrelated points. First, intentional reduction of extracorporeal oxygenation to a conservative post‑oxygenator target is technically feasible and produces a clear separation in post‑oxygenator PaO2 versus routine liberal practice. Second, implementation is operationally challenging — clinical teams only kept patients within the conservative window one‑third of the time — and no early signal of reduced organ injury (as measured by IFABP and other biomarkers) was observed.
Physiologic nuance is crucial when interpreting extracorporeal oxygen targets. Post‑oxygenator PaO2 reflects oxygen tension in blood exiting the membrane lung, but systemic arterial oxygenation (and oxygen delivery to organs) depends on complex interactions including native pulmonary gas exchange, cardiac output, ECMO flow, and the ‘mixing’ between native and extracorporeal blood streams (the so‑called Harlequin or north‑south syndrome). A lower post‑oxygenator PaO2 reduces the oxygen content delivered by ECMO, but systemic PaO2 — and tissue oxygen delivery — may remain adequate through native lung oxygenation or higher flows. Conversely, very high post‑oxygenator PaO2 may contribute to hyperoxia in arterial territories perfused principally by ECMO output, potentially exacerbating oxidative stress locally and systemically.
That the trial found no early biomarker benefit may reflect several possibilities: the intervention dose (time within target window) was modest; the pathophysiologic link between post‑oxygenator hyperoxia and early gut epithelial injury may be weaker than hypothesized; biomarkers selected may be insensitive to the relevant mechanisms; or a true effect may require larger sample size or longer exposure to manifest. Furthermore, patients on VA‑ECMO are hemodynamically heterogeneous and subject to many injury pathways (ischemia–reperfusion, inflammatory activation, multi‑organ hypoperfusion) that could obscure an oxygen effect.
Strengths and limitations of the trial
Strengths
– Randomized, multicenter design improves internal validity compared with observational work.
– Pragmatic intervention with measurable physiologic separation in extracorporeal oxygenation.
– Use of prespecified organ‑specific biomarkers (IFABP) provides biologic plausibility for detecting gut injury.
Limitations
– Pilot sample size (n=55) limits power to detect modest or clinically important differences in biomarkers or clinical endpoints.
– Poor adherence to the conservative window (33% time in target) attenuates the “dose” of the intervention and biases toward the null.
– Post‑oxygenator PaO2 may be an imperfect surrogate for tissue oxygen exposure across organs; systemic arterial oxygenation and regional perfusion were not the primary titration targets.
– Short horizon for biomarker assessment (primary at day 2) may miss delayed effects of oxygen exposure.
– Heterogeneity in ECMO cannulation, flows and patient pathophysiology could dilute effects.
– The study focused on biomarker endpoints rather than patient‑centered outcomes (organ‑failure free days, mortality), which are the ultimate clinical priorities.
Clinical implications and practice recommendations
For clinicians caring for VA‑ECMO patients, this pilot RCT provides useful but limited guidance. The findings do not support routine adoption of an aggressive conservative post‑oxygenator PaO2 target in routine practice solely to reduce early organ injury, because the trial detected no biomarker benefit and achieving the conservative target proved operationally difficult.
However, the trial also shows that lower extracorporeal FDO2 can be implemented safely in many patients, producing a large difference in post‑oxygenator PaO2 without overt adverse events in this sample. Pragmatically, units interested in reducing potential hyperoxia exposure should consider:
– Titration strategies focused on systemic arterial oxygenation (PaO2/SpO2) alongside monitoring of post‑oxygenator PaO2 to ensure tissue delivery.
– Protocolized adjustment schedules and education to improve adherence, or automated/closed‑loop control of sweep gas FDO2 where available.
– Caution in patients with low native pulmonary oxygenation or risk of differential hypoxia; individualized assessment of flows, cannulation sites, and mixing zones is essential.
– Monitoring for hypoxia and organ perfusion using lactate, mixed/central venous oxygen saturation, and clinical parameters rather than relying solely on post‑oxygenator PaO2.
Research implications and next steps
This pilot trial informs the design of definitive studies. Key considerations for future trials include:
– Larger sample sizes with sufficient power to detect clinically meaningful differences in organ failure or mortality.
– Improving intervention fidelity through more intensive protocolization, real‑time decision support, automated sweep gas control, or continuous post‑oxygenator monitoring with alerts.
– Targeting systemic arterial PaO2 (or SpO2) as the primary titration variable, or integrating both extracorporeal and systemic measures to better reflect organ oxygen exposure.
– Selecting patient subgroups most likely to benefit (e.g., those with preserved native lung function vs those with little native oxygenation) and standardizing ECMO flow strategies.
– Inclusion of patient‑centered endpoints (organ‑failure free days, ICU mortality, neurologic outcomes) along with mechanistic biomarkers of oxidative stress, endothelial injury, and microcirculatory function.
– Preclinical mechanistic work to clarify how hyperoxia delivered via the ECMO circuit affects different vascular beds and organ systems.
Conclusion
In this pilot randomized trial of patients on VA‑ECMO for cardiogenic shock, a conservative extracorporeal oxygen strategy (targeting post‑oxygenator PaO2 100–150 mm Hg) produced a clear separation from a liberal strategy but was difficult to maintain in practice and did not reduce early biomarkers of gut, hepatic, renal injury or inflammation. The data do not currently justify routine adoption of a conservative post‑oxygenator target to prevent organ dysfunction, but they support the feasibility and relative safety of lower extracorporeal FDO2 in selected settings. Definitive trials with improved adherence, clinically important endpoints and mechanistic measures are needed to establish whether extracorporeal oxygen titration can improve outcomes in VA‑ECMO patients.
Funding and trial registration
As reported: Winiszewski H et al., Crit Care Med. 2025. Details of funding and clinical trial registration are reported in the original manuscript (Winiszewski et al. 2025).
References
1. Winiszewski H, Puyraveau M, Kimmoun A, et al. Impact of Conservative Vs. Liberal Extracorporeal Oxygen Target During Venoarterial Extracorporeal Membrane Oxygenation Support for Cardiogenic Shock: A Pilot Randomized Control Trial. Crit Care Med. 2025;53(12):e2552‑e2561.
2. Extracorporeal Life Support Organization (ELSO) guidelines and resources (general reference for ECMO practice and monitoring).