Comment

Effect of normobaric hyperoxia in ischaemic stroke

Diederik W J Dippel, Paula M Janssen

Lancet 2025; 405: 442-443

DOI: https://doi.org/10.1016/S0140-6736(25)00153-9

In ischaemic stroke, discontinuation of local arterial blood supply to the brain leads to neurological deficits. The assumption that increased delivery of oxygen in patients with acute ischaemic stroke might lead to amelioration of neurological deficit is interesting, but almost all large clinical trials were neutral, with no effect compared with usual care.1, 2 Similar studies in patients with other serious illnesses were also neutral.³

In hindsight, animal experiments might have warned us that administering high concentrations of oxygen to patients with ischaemic stroke without adequate reperfusion might not be effective. For instance, the few experimental rodent studies that yielded convincingly positive results all made use of temporary arterial occlusions.4, 5Additionally, one experimental study reported no beneficial effects in a randomised experiment with permanent occlusion, and positive effects in a model of temporary occlusion.⁶

In The Lancet, Weili Li and colleagues⁷ report the effect of delivery of normobaric hyperoxia in a multicentre, randomised, single-blind, sham-controlled trial (the OPENS-2 trial) in patients with acute ischaemic stroke who underwent endovascular treatment.⁷282 patients (median age 65 years [IQR 57–71], 75 [27%] of 282 were female and 207 [73%] were male, all 282 [100%] were of Chinese Han ethnicity) were randomly assigned to either normobaric hyperoxia (n=140) or sham normobaric hyperoxia (n=142), to be started before and continued until after completion of endovascular treatment. Normobaric hyperoxia treatment involved inhalation of 100% oxygen at a flow rate of 10 L/min through a non-rebreather mask for 4 h (which is similar to the methods of experimental animal studies4, 5, 6) or an inspiratory oxygen fraction (FiO₂) of 1·0 in participants who were intubated. The sham treatment consisted of 100% oxygen delivered at a flow rate of 1 L/min via a non-rebreather mask that had the bilateral side valves open for the same duration or at an FiO₂ of 0·3. The primary outcome was the modified Rankin Scale score at 90 days (±14), and the effect measure was the common odds ratio, representing the shift on this 7-point scale, adjusted for major prognostic factors.

The trial was positive, with a common odds ratio of 1·65 (95% CI 1·09–2·50) representing a shift in the direction of improved outcomes, with less disability, for patients who had normobaric hyperoxia versus the sham normobaric hyperoxia group.⁷ Similar and significant effects of the intervention were seen for the secondary outcomes of infarct volume at 24–48 h and National Institutes of Health Stroke Scale score at 24 h and 72 h, but not at 7 days. Quality of life and disability measures at 90 days showed differences that were generally favouring the intervention, but were not significant.

Notably, in clinical care globally, reducing costs and improving sustainability of health care is very important, especially concerning new treatments. As such, normobaric hyperoxia has an advantage over other methods such as new neuroprotective agents because of its relatively low costs and wide availability, even in resource-limited settings around the world. Hence it might become an efficient addition to acute stroke unit care.

In pragmatic trials, the management of patients in the control group should meet standard guidelines to avoid inflating the treatment effect. In the current trial, we are reassured that there were no differences in vital parameters and the occurrence of adverse events between the two treatment groups. The difference between the two groups appeared predominantly in the arterial oxygen pressure at the end of therapy, which was increased by more than 60% in the intervention group compared with the control group. This was a potential driver of the treatment effect, along with prolonging the delivery of the treatment until several hours after the reperfusion treatment.

The OPENS-2 trial was a well designed, well prepared, and well conducted trial.7, 8 It was preceded by a smaller pilot and a dose-escalation study.9, 10 The trial has only a few limitations. The sham alternative when blinding study participants is challenging and is a good solution. Although the primary outcome was assessed by study staff who were supposedly not aware of the actual treatment allocation, treatment allocation might have become known to local assessors. Knowledge of the treatment allocation might push the assessment of functional outcome in favour of the new and promising intervention that is being investigated. Therefore, it should be common practice in single-blind trials to test the effectiveness of the blinding with a forced binary choice experiment.¹¹

The OPENS-2 trial was rather small, and consequently, the significant effect was quite large, with an average shift of more than 10% in the direction of more favourable outcome on the modified Rankin Scale. Although the generally consistent effect on other outcome parameters is reassuring, nevertheless, we can expect the true effect of normobaric hyperoxia to be smaller than reported here.¹² Therefore, these promising results require confirmation in larger trials and, encouragingly, at least one trial has started recruitment (NCT06666764).