Oxygen's Second Face: The Lung at the Dose Limit of Hyperbaric Therapy

Editorial scientific diagram of an alveolus cutaway under pressure, with paired oxygen molecules pressing against the thin alveolar-capillary membrane and a fine teal shield arc protecting the endothelial barrier behind it

Hyperbaric oxygen therapy occupies a strange position in medicine: it is a regulated treatment that works by administering a gas which is, at sufficient dose, directly poisonous to the organ doing the breathing. The FDA has cleared chambers for decompression sickness, carbon monoxide poisoning, certain non-healing diabetic wounds, late radiation tissue injury, gas gangrene, crush injury, thermal burns and a handful of other conditions. Push the exposure far enough, though, and the same 100 percent oxygen that helps close a wound begins to break down the alveolar–capillary membrane and flood the lung.

That ceiling — pulmonary oxygen toxicity — is the least discussed and most consequential limit on oxygen therapy. Two papers published this October, both from the same naval-medicine group in Shanghai, have now traced how the lung tries to defend itself.

A gas that heals, and a gas that scars

The mechanism of hyperoxic lung injury has been studied for decades, but its molecular choreography has stayed murky. What is clear is that the damage is inflammatory and dose-dependent: sustained high oxygen partial pressures generate reactive oxygen species in the airways and alveoli, which draw in neutrophils, inflame the endothelium, and eventually thicken and leak the barrier that separates air from blood.

The Shanghai group approached the problem from an unexpected direction. Rather than asking what goes wrong, they asked what the lung switches on when it is exposed to hyperbaric oxygen — and found a family of detoxifying enzymes rising in step with the exposure.

Two papers, one laboratory

The first of the pair, in Redox Biology, reports that hyperbaric oxygen raises expression of CYP1A1 and CYP1A2 — two isoforms of the cytochrome P450 1A subfamily — in mouse lung tissue and in primary mouse lung microvascular endothelial cells. When both genes were deleted, the animals did markedly worse: greater pulmonary endothelial dysfunction, more mitochondrial oxidative stress, more endothelial cell death. When the researchers gave a CYP1A inducer, beta-naphthoflavone, the injury eased. The proposed route runs through heat shock factor 1 (HSF1), which is recruited to the promoter of a mitochondrial subunit gene, SDHC, preserving mitochondrial function and the integrity of the endothelial barrier.

The companion paper, in Chemico-Biological Interactions, follows a different thread of the same response. Here the damaging player is the neutrophil extracellular trap, or NET — a web of DNA and enzymes that neutrophils extrude when overstimulated, and which has been implicated in a wide range of inflammatory tissue injuries. Hyperbaric oxygen triggered NET formation in the lung; deleting CYP1A1 and CYP1A2 made it worse; inducing CYP1A reduced it. The protective effect, the authors argue, is indirect. CYP1A appears to act in the alveolar epithelial cells, lowering oxidative stress there through an NRF2-dependent pathway and so creating a microenvironment in which neutrophils are less likely to fire their traps.

Read together, the two papers describe hyperbaric oxygen provoking its own partial antidote — an adaptive brake that the lung applies when it is pushed too hard.

What the papers do not say

Two caveats belong next to that finding, and they matter more than the mechanism itself.

The first is that this is mouse work. There are no patients in either study, no clinical endpoints, no measurement of whether any of this translates to a human being in a chamber. The evidence level is preclinical mechanism — the kind of result that generates hypotheses and, occasionally, drug targets, but not clinical recommendations.

The second is that both papers come from a single programme. The senior authors (Zhu XY and Bao XC) and the underlying hyperbaric mouse model are shared, and the two studies are best understood as two arms of one investigation rather than two independent confirmations of the same hypothesis. That distinction matters: a mechanism described once, in one laboratory, in one species, is a lead, not a fact.

There is a third caution, and the authors are right to stop short of making it themselves. Beta-naphthoflavone is a laboratory tool compound, and CYP1A sits squarely inside the aryl hydrocarbon receptor pathway — the same system that dioxin and other environmental toxins activate. “Boost CYP1A” is not a benign instruction. Nothing here should be read as a reason to take a supplement, a drug, or a chamber session on the theory that it will harden the lungs. It is a map of one protective pathway, not a therapy.

The dose accountants

If the mouse papers explain what happens at the tissue level, a third paper this quarter shows how the operational side of the field has learned to count.

Writing in Undersea and Hyperbaric Medicine, researchers at China’s Naval Specialized Medical Center took the schedules of the country’s helium-oxygen decompression table and computed three conventional exposure metrics for each one: the unit pulmonary toxic dose (UPTD), the K index, and equivalent surface oxygen time (ESOT). Their analysis found that K and ESOT were more conservative predictors of hyperoxic lung injury than UPTD. After deeper dives, all three metrics commonly ran above recommended thresholds — and their proposed remedy is operational rather than clinical: lengthen the oxygen periods, add oxygen after surfacing, and most schedules’ predicted lung injury should resolve within about twelve hours of the diver reaching the surface.

This is a desk exercise on a dive table, not a controlled trial, and the schedules involved are saturation and commercial diving profiles rather than hospital protocols. Its importance is conceptual. The field now treats oxygen as a cumulative dose to be tracked the way a pharmacist tracks a drug — which is precisely the discipline that separates therapeutic oxygen from oxygen as a hazard.

What the divers show

Two human studies in Diving and Hyperbaric Medicine put a number on the practical side of that trade-off.

The larger one followed twelve technical divers through 173 closed-circuit rebreather dives over 31 days in Chuuk Lagoon, Micronesia, with a mean maximum depth of 38 metres and a mean dive time of 101 minutes. Oxygen exposure was calculated against multiple models and stayed within the daily limits for pulmonary and central nervous system toxicity. Weekly spirometry showed no statistically significant change across the expedition, and visual acuity fluctuated without any overall trend. Two individual findings were flagged — one diver lost ground across several lung function parameters, another developed mild obstruction by the end — which is exactly the sort of signal that a small expedition cannot distinguish from noise. With twelve self-selected divers and no control group, the study is reassuring rather than conclusive. Its takeaway is the conditional one its authors draw: expeditions can be conducted safely when oxygen exposure guidelines are respected.

The smaller study is a useful corrective to the assumption that “mild” hyperoxia is inert. Nineteen experienced divers had their critical flicker fusion frequency — a standard index of cortical processing speed — measured at poolside, at four metres’ depth breathing air, and at four metres breathing 100 percent oxygen, an oxygen partial pressure of 140 kPa. Performance barely moved, falling from 37.9 to 37.2 Hz, which the authors read as no meaningful impairment of alertness. But the cardiovascular picture shifted noticeably: heart rate and respiratory rate both fell, the low-frequency/high-frequency ratio halved, and a parasympathetic marker rose. Breathing oxygen at that pressure is not nothing. It measurably changes autonomic regulation.

That pressure matters for the wellness market. An oxygen partial pressure of 140 kPa is roughly what a person breathing pure oxygen at 1.4 atmospheres absolute would experience — the range marketed by so-called mild hyperbaric lounges. The variable that governs pulmonary toxicity is oxygen partial pressure multiplied by time, not the size of the chamber or the reassuring word “mild.”

The pressure injuries that have nothing to do with oxygen

A case report this month in International Maritime Health underlines that pressure, not only oxygen, is a hazard. Turkish clinicians describe a professional fish-farm diver of nearly three years’ experience who developed reverse sinus barotrauma — injury on ascent rather than descent — with air dissecting into the tissue around the eye. The injury is preventable, they argue, and prevention is unglamorous: understand the physics, know your fitness on the day, and abort the dive when equalisation fails.

For the inhale-injury end of the spectrum, a review in the Balkan Medical Journal on smoke and chemical inhalation injury is a useful check on enthusiasm. Surveying pathogenesis and management, the authors conclude that adjunctive therapies, “including hyperbaric oxygen,” remain of uncertain benefit — a reminder that even where the physiology is compelling, the clinical evidence for a given use is often thinner than the mechanism suggests.

Where the clinical line actually sits

None of this undermines hyperbaric oxygen treatment for the conditions it is cleared to treat. Pulmonary oxygen toxicity is the main reason clinical protocols are built the way they are: modest pressures of roughly 2.0 to 2.5 atmospheres absolute, sessions of 60 to 90 minutes, air breaks, and limits on cumulative exposure. The lung findings are an argument for disciplined dosing and monitoring, not for avoiding a medically indicated chamber session.

What they do not do is license the wellness framing. The FDA has not cleared hyperbaric oxygen for longevity, anti-ageing, general “detox,” or cognitive enhancement, and the agency has warned specifically about unproven uses. In an August 2025 letter to health care providers, it also reported fires involving hyperbaric oxygen devices that caused serious injuries and deaths, and set out recommendations on staff training, continuous patient monitoring, hyperbaric-compatible clothing such as cotton, and the avoidance of prohibited electrical or static-generating items. Cleared devices are Class II, identified through the 510(k) process under Product Code CBF. CMS, for its part, covers chamber therapy under National Coverage Determination 20.29 only for a defined list of conditions, and only within stated limits — including requiring that diabetic wounds fail a measurable course of standard care first.

Oxygen under pressure is a drug with a narrow therapeutic index, a real ceiling, and a lung that fights back a little every time the dose rises. That is the honest picture — and it is a far more interesting one than the tonic being sold in a lounge.

This article reports what published research and regulators say. It is not medical advice, and any decision about hyperbaric treatment belongs with a qualified clinician.

Sources

← Back to blog