Four Patients, One Chamber: A First-in-Human Test of Hyperbaric Oxygen in Septic Shock

Stylised editorial diagram of three stacked strata — a stalled capillary mesh, a cutaway mitochondrion and drifting signalling molecules — in cool teal and amber, with a small outlined hyperbaric chamber at the lower right

Septic shock is one of the few conditions in medicine where the mortality is high enough, and the therapeutic pipeline thin enough, that almost any plausible mechanism attracts a trial. The World Health Organization counts roughly 48.9 million sepsis cases and 11 million sepsis-related deaths a year, about one death in five worldwide, at an average hospital cost above US$32,000 per patient in high-income countries.

Against that backdrop, a line in a new paper from the University of Iowa lands hard: septic shock, the authors write, has seen “no successful, novel therapies emerge in the past two decades.”

Their response was not a new drug. It was a chamber. In the third-quarter issue of Undersea and Hyperbaric Medicine, Dr. Lingyao Sun and colleagues at the University of Iowa Carver College of Medicine, with a co-author from McMaster University, report what they describe as a first-in-human pilot of hyperbaric oxygen treatment in septic shock — four patients treated, four matched controls, and a result that is negative, clean, and quietly informative.

The case for pressure in a disease of perfusion

Sepsis kills through a sequence that resuscitation only partly touches. Fluids and vasopressors can restore a normal blood pressure while the microcirculation stays sluggish, and organ dysfunction in septic shock is driven by three overlapping failures: microvascular, mitochondrial and immune. Cells starved of usable oxygen at the tissue level do not necessarily respond to a better number on the monitor.

Hyperbaric oxygen attacks that gap directly. Breathing 100 percent oxygen at two to three times atmospheric pressure raises the amount of oxygen dissolved in plasma and pushes tissue oxygen tension far above what any normobaric circuit can achieve. Animal work, the Iowa authors note, supports the idea that this improves mitochondrial function and dampens the inflammatory cascade — enough of a rationale to justify treating human beings.

A parallel review published in September in Frontiers in Cellular Neuroscience maps the most extreme version of the problem, sepsis-associated encephalopathy. A team at the Fifth Clinical Medical College of Henan University of Chinese Medicine describes blood-brain barrier disruption, neuroinflammation, microglial and astrocytic activation, oxidative stress, mitochondrial injury, ferroptosis and pyroptosis as interlocking processes, and argues that hyperbaric oxygen could plausibly interrupt several of them. Animal models of sepsis-associated encephalopathy suggest the treatment can attenuate barrier injury and hippocampal neuroinflammation and improve neurobehavioural outcomes. The review’s own verdict on the clinical question is blunt: “Current evidence remains predominantly preclinical.”

Sixty minutes at 2.5 ATA

The Iowa trial was a phase I, open-label, matched design — the smallest honest step from that preclinical base to a patient. Four experimental subjects received 60 minutes of hyperbaric oxygen at 2.5 atmospheres absolute, with a 25-minute descent and ascent, alongside standard sepsis management. Four controls were matched on age, which ranged from 29 to 75, and on vasopressor dose at enrolment, with norepinephrine equivalents between 0.05 and 0.35 micrograms per kilogram per minute.

The primary outcome was microvascular function, measured as the rate of reoxygenation after a period of upper-limb ischaemia using real-time near-infrared spectroscopy — a bedside way of asking whether the smallest vessels actually recovered their capacity to deliver oxygen. Secondary outcomes looked one level deeper: mitochondrial function in neutrophils and monocytes sampled ex vivo, and serum concentrations of the inflammatory cytokines IL-6, IL-10 and TNF-alpha. Analysis was graphical, comparing matched pairs, rather than a formal statistical test. With four subjects per arm, that was the only defensible option.

What the data showed

Hyperbaric oxygen produced transient improvements in microvascular function in two of the four treated patients and no change in the other two. All four controls deteriorated.

That asymmetry is the finding most likely to be quoted out of context, so it is worth stating what it is not. There was no difference between groups in mitochondrial function, and no difference in the cytokine profiles — the two secondary outcomes intended to show that the treatment was engaging the biology the animal literature points to. The authors also recorded no adverse events in the intervention arm, which in a patient population this fragile is a result in itself.

Their conclusion is unambiguous: “Our study did not find a statistically or clinically significant benefit of hyperbaric oxygen in septic shock.” What they propose instead is what the field actually needs — adequately powered trials, and enough participants to identify the subgroups, if any, most likely to respond.

Why the “half improved, all controls declined” pattern cannot carry weight

Septic shock is not one disease. Source of infection, organism, host response and time from onset all vary enormously between patients, and much of the modern trial literature has failed precisely because a heterogeneous population was treated as one. The Iowa team matched on vasopressor dose, a pragmatic proxy for haemodynamic severity, but matching cannot compensate for unmeasured differences in a sample of eight.

Add the absence of blinding — patients in a chamber know they are in a chamber — and a graphical analysis, and the honest reading of the microvascular signal is narrow: reoxygenation kinetics may be modifiable in some patients. It is a hypothesis for a larger study, not a treatment effect. The result that deserves the headline is the null one, and it is a credit to the authors that they published it that way.

The regulatory floor under all of this

Septic shock is not a cleared indication for hyperbaric oxygen therapy. In the United States, hyperbaric devices are Class II medical devices cleared through the 510(k) pathway under Product Code CBF, and the cleared uses are a defined list: carbon monoxide poisoning, decompression sickness, gas gangrene, crush injury, compromised skin grafts and flaps, selected non-healing wounds, late radiation tissue injury, severe anaemia, idiopathic sudden sensorineural hearing loss and intracranial abscess. Medicare’s national coverage determination 20.29 tracks that list.

The FDA’s letter to health care providers of 25 August 2025 remains the agency’s current device-safety guidance, and it is worth paraphrasing in a critical-care context: the agency is “aware of recent reports of fires that occurred with HBOT devices that resulted in serious injuries and deaths,” and it asks facilities to follow manufacturer instructions, maintain fire prevention and training, monitor patients continuously in the chamber, and control prohibited electrical and static-generating items.

Now place a patient in septic shock inside that frame. Vasopressors, a ventilator, arterial and central lines, continuous renal replacement in some cases — this is the least transportable patient in the hospital, and the chamber is not where they are. An 11-year single-centre review from the Medical University of Gdańsk, published in Journal of Clinical Medicine in September, quantifies how rare the combination is: of 6,410 hyperbaric patients treated between 2013 and 2023, 176 were mechanically ventilated intensive care patients, 2.7 percent, accounting for 1,381 sessions — 56.6 percent of them at night. Running a chamber for the sickest patients is a staffing and infrastructure problem before it is a biological one.

Where hyperbaric oxygen does have traction in severe infection

The conditions in which hyperbaric oxygen is genuinely used in life-threatening infection are local tissue diseases, not shock states. Gas gangrene — clostridial myonecrosis — is a cleared indication, and necrotising soft-tissue infections more broadly remain the leading intensive-care reason patients reach a chamber. A 2026 systematic review and meta-analysis in Medical Gas Research pooled 13 retrospective studies of Fournier’s gangrene, 322 patients treated with hyperbaric oxygen against 366 controls, and found significantly lower mortality in the treated groups, with length of stay and number of debridements inconsistent between studies. Retrospective, non-randomised, and about dead tissue in the perineum — not about reversing shock.

At the outer edge sit the case reports. In December 2025, Diving and Hyperbaric Medicine published an account from Alfred Health in Melbourne of a 20-year-old trauma patient who developed disseminated Saksenaea and Fusarium infection with multiorgan dysfunction and sepsis, and remained on venovenous ECMO throughout 13 hyperbaric sessions at 2.4 ATA for 95 minutes each. Maximal surgery and antifungal therapy had failed; the authors describe the chamber as an experimental addition, after which disease control was achieved. One patient, no control group, and a mechanism of angioinvasive fungal disease rather than septic shock — the kind of result that justifies the next study and not the next claim.

What would actually move the needle

The Iowa trial does not close the question of whether hyperbaric oxygen can help in septic shock. It closes a smaller one: at 2.5 ATA for 60 minutes, in four patients already on vasopressors, it appears tolerable and produces no measurable cytokine or mitochondrial signal. Whether the microvascular flicker seen in half of them is real, and whether any subgroup would benefit, requires a blinded, adequately powered trial with microcirculatory endpoints — expensive, slow, and unglamorous.

Meanwhile the same physiology is being sold in far less rigorous packaging. Soft-sided and mild chambers marketed for “cellular oxygenation” and immune support produce pressures that do not reach the range used here, for indications the FDA has not cleared, in settings the agency’s 2025 letter specifically warns about.

The Iowa result is a modest one. Nothing helped, nothing hurt, and the authors said so. In a field where the loudest claims come from the least controlled settings, that is worth more than it looks.

Sources

← Back to blog