COVID-19: Oxygen therapy harms lung microbiome contributing to lung injury


One of the hallmarks of severe COVID-19 is shortness of breath and significantly reduced levels of oxygen in the blood, called hypoxemia.

Upon hospitalization, these patients are administered oxygen in an attempt to bring their levels back up to normal.

However, a new study hints that this universal therapy may have unintended consequences via an unexpected source – the microbiome.

“It had been assumed that the lungs were relatively clean and free of bacteria,” says Shanna Ashley, Ph.D., a former Post-Doctorate Fellow with the Division of Pulmonary and Critical Care Medicine at U-M Medical School.

“We now know that the balance of bacteria inside the lungs matters much like it does in the gut.” Ashley worked with a team led by Robert Dickson, M.D., Assistant Professor of Pulmonary &

Critical Care Medicine and Microbiology and Immunology, whose lab has spent years exploring the role of the lung microbiome in health and disease. Their work has found that oxygen disrupts this balance, contributing to lung injury.

Scientists have long known that oxygen can damage the lungs. “Oxygen is actually a potent lung toxin,” says Dickson.

“If I put healthy mice in 100% oxygen, they will die in five days, and they’ll have the same kind of severe lung injury that patients with COVID-19 or other lung damage have.”

Patients in intensive care are often treated with high concentrations of oxygen for long periods of time.

Their team began to explore how therapeutic oxygen was affecting the lung microbiome. They looked at critically ill patients who were on a ventilator for more than 24 hours and studied bacteria detected in specimens from their lungs.

They found marked differences in the bacteria species present in samples from patients depending on whether they received low, intermediate, or high concentrations of oxygen.

Specifically, patients who received high oxygen concentrations were much more likely to grow Staphylococcus aureus, bacteria that are very oxygen-tolerant and a common cause of lung infections in the ICU.

“Different types of bacteria vary quite a bit from each other in how well they can handle oxygen,” Dickson says, “So we wondered if the oxygen we give our patients might be influencing the bacterial communities in their respiratory tract.”

To better understand the relationship between oxygen and lung bacteria, the team designed a series of experiments in mice.

They first exposed healthy mice to high concentrations of oxygen to determine the effects of oxygen on the lung bacteria of healthy mice.

“When we gave high concentrations of oxygen to healthy mice, their lung communities changed quickly, and just like we predicted,” said Ashley.

“The oxygen-intolerant bacteria went down, and the oxygen-tolerant bacteria went up.” After three days of oxygen therapy, oxygen-tolerant Staphylococcus was by far the most commonly detected bacteria in mouse lungs.

The team next designed experiments to answer a key “chicken or the egg” question: do these altered bacterial communities contribute to lung injury?

Or are bacterial communities altered because the lung is injured?

They first addressed this by comparing the relative timing of changes in lung bacteria as compared to the onset of lung injury.

Using mice, they were able to demonstrate that while the lung microbiome was changed by high oxygen concentrations after only a day, lung injury wasn’t detectable until after 3 days, proving that damage to the lung followed the disruption of the microbiome, and not the other way around.

Furthermore, they showed that natural variation in lung bacteria was strongly correlated with variation in the severity of inflammation in oxygen-exposed mice.

To further strengthen the causal link, they turned to germ-free mice, which completely lack a microbiome.

“We wanted to see whether there was a selective advantage or disadvantage to having bacteria-free lungs when exposed to therapeutic oxygen,” says Ashley. When comparing two groups of genetically identical mice—one with bacteria and one without—the mice without bacteria were protected from oxygen-induced lung injury.

“That was an extraordinary finding for us,” said Dickson. “Compared to conventional mice, these germ-free mice have the same genetics and receive the same oxygen dosing, but their lungs are protected from injury.

Nothing in our current understanding of oxygen-induced lung injury can explain that finding.”

“It really makes the case that the microbiome is somehow playing a role in lung injury,” said Ashley.

Targeting the microbiome

Critically ill patients receiving oxygen are typically administered antibiotics as well. The team wondered: Could antibiotics alter the severity of oxygen-induced lung injury in mice?

“The short answer is yes, we can affect the severity, but it wasn’t in the direction we predicted,” says Dickson. Vancomycin, an antibiotic that targets gram-positive bacteria like Staphylococcus, had no effect on lung injury, while ceftriaxone, a gram-negative antibiotic, made things worse.

“The microbiome is not all good and not all bad,” comments Dickson.

“That’s why it’s important for us to figure out the mechanisms here. We’re currently using very non-specific interventions, when what we need is targeted manipulation of the microbiome.”

Ashley agrees. “We need to think about using the microbiome as a therapeutic target to prevent doing further damage to patients’ lungs while they are on a ventilator or receiving oxygen.”

Dickson cautions against changing clinical practice prematurely based on these findings. “The question of how much oxygen to give critically ill patients is a complex one, and a topic of intense study,” says Dickson.

“Our findings are exciting, but I still look to randomized controlled trials to inform my decisions about how to dose oxygen in sick patients.”

James Kiley, director of the Division of Lung Diseases at the National Heart, Lung, and Blood Institute, part of the National Institutes of Health, agrees.

“This study provides important insights into the contributions of the microbiome toward inflammation and damage in lungs exposed to varying levels of oxygen, and supports the continued importance of understanding how the microbiome and related factors impact lung disease and clinical outcomes.”

Supplemental oxygen.

In patients with moderate severe COVID-19, supplemental oxygen can be provided using simple nose prongs or face masks with an oxygen flow up to around 5–6 L O2/minute.

Flow rates can be titrated using pulse oximetry monitoring, targeting an arterial oxygen content (SpO2) greater than 88%, which is a much more liberal target than in other causes of pneumonia.

If the patient shows desaturation less than 88% for prolonged periods of time, oxygen delivery can be increased by using a non-rebreathing mask. These masks contain an additional reservoir bag where oxygen flows in, which is inhaled through a valve during inspiration.

This can provide a fraction of inspired oxygen (FiO2) of 0.6–0.8 but will require an oxygen flow from the oxygen cylinder or piped oxygen of minimal 10 –15 L/minute. The use of non-rebreathing masks can be an important additional modality to increase oxygen delivery to patients with severe COVID-19, both in resource-rich and in resource-limited settings.

In addition, patients can be nursed in the prone position or encouraged to lay on their front, which may give a remarkable improvement in oxygenation in COVID-19 patients (Schultz, personal communication). Sitting straight up can be an alternative, especially in patients for whom prone positioning is not feasible, for example, in severe obesity.

Prone positioning facilitates ventilation of posterior lung field, improving the ventilation-perfusion mismatch, and thus oxygenation. High-flow nasal oxygen (HFNO) can also be used to importantly increase FiO2.

Experience in the use of HFNO in coronavirus pneumonia is limited, and an important disadvantage for the resource-poor setting is the very high oxygen flow of up to 60 L/minute needed.

Adult HFNO can either be delivered by the mechanical ventilator or by stand-alone systems, such as OptiflowR, which require a permanent power source, because they are often not battery-operated. This can be very dangerous in low-resourced settings when a permanent power supply cannot be guaranteed.

In patients treated with only supplemental oxygen, it is important to monitor fatigue or exhaustion because of increased work of breathing, in addition to monitoring oxygen saturation. A proportion of patients will not be sufficiently supported by just increasing FiO2.

Using a non-rebreathing mask will not build up any positive end-expiratory pressure (PEEP), which is important to help preventing collapse of small airways and alveoli in the diseased lung at the end of expiration.

As an intervention before invasive mechanical ventilation, PEEP can be generated by using continuous positive airway pressure (CPAP) or noninvasive ventilation (NIV) using biphasic positive airway pressure.

Continuous positive airway pressure can be delivered with specific devices containing a PEEP valve providing resistance to exhalation, linked to a tight-fitting oral, nose, or full-face mask or a specific CPAP hood or helmet.

Noninvasive ventilation requires a mechanical ventilator attached to a tight-fitting mask, and will in addition to PEEP also deliver additional inspiratory pressure to assist inspiration.

Its use in patients with Middle East respiratory syndrome-related coronavirus showed a high failure rate of NIV, where it did not prevent intubation for ventilation. [Arabi YM, Arifi AA, Balkhy HH, Najm H, Aldawood AS, Ghabashi A, Hawa H, Alothman A, Khaldi A, Al Raiy B, 2014. Clinical course and outcomes of critically ill patients with middle east respiratory syndrome coronavirus infection. Ann Intern Med 160: 389–397.]

Currently, the recommendation is that HFNO, CPAP, or NIV in severe COVID-19 should only be used in selected patients with hypoxemic respiratory failure and that these patients are closely observed for early detection of further deterioration (

In the current practice, these modalities are often used in patients where it is decided to forgo intubation for mechanical ventilation, for instance, because mechanical ventilation is not available, or in patients with a “do not resuscitate” directive.

There is uncertainty around the potential for aerosolization when using HFNO, CPAP, or NIV, and these modalities should be used with airborne precautions until further evaluation of the safety is completed.

An additional issue for resource-limited settings is that all these modalities require specific equipment, which can be expensive and difficult to procure in these times of a rapidly spreading pandemic. An exception could be CPAP.

There are several initiatives for producing cheap CPAP helmets, which can be directly attached to an oxygen and compressed air cylinder.

COVID-19: acute respiratory distress syndrome (ARDS) and hyperbaric oxygen therapy (HBOT)

hyperbaric oxygen therapy (HBOT) that consists of exposure to 100% oxygen under increased atmospheric pressure up to 2.4 atm could be a great resource to improve the outcome from the infection when it is administered at early stages as soon as a reduction of arterial oxygen concentration is detected.

Indeed, experimental animal studies have shown that an initial HBOT improved dramatically the outcome from sepsis, which was correlated with a reduction of the inflammatory response triggered by the initial insult (Halbach et al. 2019).

The great advantage of HBOT is that it delivers oxygen at elevated partial pressure allowing this gas to penetrate tissues very rapidly and in higher concentration, which is more effective than hemoglobin oxygen delivery.

Both mechanical ventilators, the current treatment for severely ill hypoxic COVID-19 patients in critical care, and HBOT are able to elevate the levels of arterial O2, but in addition, HBOT provides a crucial function that ventilators lack.

The increased concentration of O2 delivered to cells in tissues by HBOT at 2.4 atm provides a signal for cells to induce two powerful transcription factors, Nrf-2 which stimulates the production of literally hundreds of cell defense proteins most of which participate in oxidative stress responses and heat shock transcription factor 1 which induces cells to produce additional defense proteins that are also anti-inflammatory (Godman et al. 2010).

The elevated supply of oxygen is likely to preserve cellular metabolism and organ function. Indeed, HBOT has been reported to improve mitochondrial function (Tezgin et al. 2020).

Moreover, HBOT alters the balance between glycolysis and mitochondrial respiration, possibly countering an effect of viral infection on cellular caloristasis networks (Tezgin et al. 2020) and improving hypoxia in COVID-19 patients.

An additional advantage of HBOT is its capacity to reduce the inflammatory response (Buras et al. 2006; Halbach et al. 2019). Several studies have shown that HBOT improves kidney function after infection (Edremitlioglu et al. 2005) and reduces kidney damage in diabetic patients (Harrison et al. 2018).

It has also been reported to protect from ischemia/reperfusion injury (Buras and Reenstra 2007; Yu et al. 2005) and diminish UV skin damage (Fuller et al. 2013).

HBOT has been used extensively with great safety in the treatment of patients for a variety of maladies. HBOT is the treatment of choice for carbon monoxide poisoning and gas embolism (Goodman 1964; Tibbles and Edelsberg 1996).

It has been very effective in the treatment of diabetic ulcers (Gill and Bell 2004; Stoekenbroek et al. 2014) and radiation injury (Kirby 2019a) and in the improvement of wound healing (Kirby 2019b).

Therefore, HBOT could be a potential intervention to improve the outcome of COVID-19 patients. It has been shown to be safe during the use of mechanical ventilation (Bessereau et al. 2017).

A small study from China has shown excellent potential for its use in the treatment of COVID-19 patients (

In this study, five critically ill patients with COVID-19 and signs of hypoxia were subjected to HBOT. After two treatments, a dramatic improvement in the clinical condition of the patients was observed with an increase in blood oxygen saturation level and reduced lung inflammation, as observed by CT scans.

There were no concerns about viral contamination and the spread of the disease to medical attendants.

All of these observations pointed out that HBOT could be a useful tool for improving the conditions of COVID-19 patients, particularly if the intervention occurs at early stages, although it could also be positive during the intubation period.

Obviously, there are some logistics in the use of HBOT in the ICU setting. Hyperbaric chambers occupy significant space, and they may not be available continuously to the ICU units.

Thus, patients need to be transported to the HBOT facility. In addition, the number of bed settings per chamber is limited. It would be of great utility to have portable chambers that could be easily installed within the patient ICU bed.

Although we appear highly enthusiastic about the potential role of HBOT in the treatment of COVID-19 patients, sound clinical trials are needed to test whether or not this intervention could save lives during the current pandemic.

More information: S.L. Ashley el al., “Lung and gut microbiota are altered by hyperoxia and contribute to oxygen-induced lung injury in mice,” Science Translational Medicine (2020). … scitranslmed.aau9959


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.