When a critically ill patient suffers acute brain injury and acute respiratory distress syndrome at the same time, doctors face a physiological paradox: the treatment that protects one vital organ can endanger the other. A new narrative review in Intensive Care Medicine argues that this conflict cannot be solved with a single ventilation protocol. Instead, clinicians should continuously adjust respiratory support according to the patient’s changing lung mechanics, blood gases, circulation and cerebral physiology.

Acute respiratory distress syndrome, or ARDS, causes widespread inflammation and fluid accumulation in the lungs. As air sacs collapse or fill with oedema, the amount of functioning lung tissue shrinks, creating the “baby lung” effect. Blood continues flowing through poorly ventilated regions, producing severe ventilation–perfusion mismatch and low blood oxygen. Acute brain injury, meanwhile, includes traumatic brain injury and non-traumatic events such as intracranial haemorrhage. The review reports that ARDS affects up to one-third of critically ill patients with acute brain injury, with particularly high rates after severe trauma and intracranial bleeding. Its development is associated with higher mortality, longer ventilation and poorer neurological recovery.

The interaction runs in both directions. Brain injury can trigger lung dysfunction through aspiration, impaired consciousness, an intense sympathetic stress response, neuroinflammation and temporary immune suppression. A damaged lung can then worsen the brain’s condition through hypoxaemia, carbon dioxide disturbances, unstable blood pressure and impaired cerebral perfusion. This creates a dangerous feedback loop in which respiratory failure promotes secondary brain injury, while neurological instability makes conventional respiratory treatment more hazardous.

The most familiar example is low-tidal-volume ventilation, a central component of ARDS care. A ventilator’s tidal volume is the amount of air delivered with each breath; limiting it, generally to about 4–8 millilitres per kilogram of predicted body weight, reduces over-stretching of fragile alveoli and helps prevent ventilator-induced lung injury. But reducing the volume of each breath can cause carbon dioxide to accumulate, especially when only a small fraction of the lung remains available for gas exchange. Rising arterial carbon dioxide, or PaCO₂, rapidly relaxes cerebral blood vessels. That increases cerebral blood volume and can raise intracranial pressure, a potentially dangerous consequence in a swollen or poorly compliant brain.

The opposing problem appears with positive end-expiratory pressure, or PEEP. By maintaining pressure in the lungs at the end of exhalation, PEEP can prevent alveoli from collapsing, improve oxygenation and reduce repetitive opening and closing injury. Yet higher intrathoracic pressure may restrict venous blood returning to the heart, lower cardiac output and reduce mean arterial pressure. Because cerebral perfusion pressure depends largely on arterial pressure relative to intracranial pressure, this combination can reduce the blood supply available to injured brain tissue. High PEEP may also increase resistance in the lung’s blood vessels, strain the right ventricle and obstruct cerebral venous drainage. Moderate levels of roughly 5–10 centimetres of water are often tolerated, the authors write, but higher levels require close assessment of intracranial pressure, perfusion and cardiovascular stability. Elevating the head of the bed by about 30 degrees may help venous drainage.

Oxygen therapy presents a similarly narrow path. Too little oxygen deprives vulnerable neurons of the substrate needed to generate energy, while excessive oxygen may cause cerebral vasoconstriction and oxidative stress. The review highlights observational evidence suggesting a U-shaped relationship between arterial oxygen and outcomes after brain injury: risk appears to rise at both low and very high levels, although these studies cannot prove that oxygen exposure itself caused the harm. Consensus targets commonly place PaO₂ between 80 and 120 millimetres of mercury, but the appropriate range may differ from patient to patient. Measuring brain tissue oxygen tension, or PbtO₂, can reveal whether oxygen is actually reaching the injured region. A low brain-oxygen reading despite an apparently adequate arterial PaO₂ may indicate inadequate blood flow or microvascular dysfunction rather than a simple need for more inspired oxygen.

Carbon dioxide targets may need similar individualisation. In patients without intracranial hypertension, available cohort data support normocapnia, generally a PaCO₂ of 35–45 millimetres of mercury. When pressure inside the skull rises, clinicians may cautiously use low-normal values around 35–38 millimetres of mercury, with mild hypocapnia reserved for escalation and moderate hypocapnia used only briefly in severe situations. Values below 30 millimetres of mercury are discouraged because excessive cerebral vasoconstriction can reduce blood flow. Conversely, carefully controlled mild hypercapnia might improve cerebral blood flow in selected patients whose intracranial pressure is stable but whose brain tissue remains oxygen-starved. Such manoeuvres are unsafe without monitoring because a rise in PaCO₂ can abruptly worsen intracranial hypertension.

The review also reassesses treatments that are often viewed as too risky for patients with brain injury. Non-invasive support, including high-flow nasal oxygen, continuous positive airway pressure and non-invasive ventilation, may help carefully selected patients avoid intubation and sedation. But impaired airway protection, a weak cough, haemodynamic instability or severe gas-exchange failure should prompt caution. In ARDS patients with a PaO₂-to-FiO₂ ratio below 150, prior observational evidence has linked non-invasive ventilation with higher mortality than invasive ventilation, possibly because delayed intubation allows respiratory effort and lung injury to intensify. If neurological status deteriorates, secretions cannot be cleared or gas exchange worsens, delaying intubation may be more dangerous than the procedure itself.

Prone positioning, one of the most effective treatments for moderate-to-severe ARDS, has historically been avoided in brain-injured patients because turning patients face down can increase chest and abdominal pressures and interfere with cerebral venous outflow. Evidence reviewed by the authors now suggests that the procedure can be feasible in selected patients without significant uncontrolled intracranial hypertension. In reported studies, rises in intracranial pressure were usually modest, while oxygenation and sometimes brain tissue oxygenation improved. Safe implementation requires continuous monitoring, careful padding to limit abdominal pressure, attention to head and neck position and maintenance of adequate cerebral perfusion. Recruitment manoeuvres that briefly expose the lungs to very high pressures are less promising: they can increase intracranial pressure, lower cerebral perfusion and destabilise the circulation, and current ARDS guidelines generally advise against their routine use.

For the most severe cases, venovenous extracorporeal membrane oxygenation, or VV-ECMO, may provide a way to oxygenate and remove carbon dioxide from the blood while allowing the ventilator to operate at ultra-protective settings. Brain injury has traditionally been considered a major barrier because ECMO commonly requires anticoagulation, which can worsen intracranial bleeding. Growing observational experience suggests that carefully selected patients, particularly those without active bleeding or an absolute contraindication to anticoagulation, should not automatically be denied ECMO. Heparin-sparing approaches may reduce bleeding risk but increase the chance of circuit clotting. The transition onto ECMO also demands careful control of carbon dioxide: a sudden fall in PaCO₂ can constrict cerebral vessels and has been associated with intracranial haemorrhage. Extracorporeal carbon-dioxide removal remains experimental in this population.

The authors emphasise that the same integrated logic must continue during weaning and extubation. A patient may meet conventional respiratory criteria yet remain unable to protect the airway or tolerate the neurological stress of spontaneous breathing. Clinicians should combine a neurological wake-up assessment with a spontaneous breathing trial, evaluating brainstem reflexes, pupils, intracranial pressure, work of breathing and blood gases. Extubation failure occurs in roughly 20–25 percent of patients with acute brain injury, compared with about 10–20 percent in general intensive-care populations. Weak airway protection, poor secretion clearance, advanced age and prolonged ventilation increase the risk. Tracheostomy, performed in approximately 20–30 percent of patients after acute brain injury, can reduce sedation needs and facilitate pulmonary hygiene and rehabilitation, but its timing remains individual.

The review’s central message is that no intervention in this setting is truly “lung-specific” or “brain-neutral.” Corticosteroids may help selected forms of inflammatory ARDS but can cause hyperglycaemia, immune suppression, delirium and muscle weakness, while previous brain-injury trials have raised serious safety concerns. Inhaled nitric oxide can improve oxygenation temporarily without improving survival. Conservative fluid management benefits the lungs, but excessive fluid removal can cause hypotension and worsen cerebral injury. Vasopressors may be necessary to preserve cerebral perfusion, yet their effects on the right ventricle and pulmonary circulation must also be considered.

Rather than treating fixed targets as universal rules, the authors propose a physiology-driven strategy. Initial settings might combine low tidal volume, moderate PEEP, plateau pressure below 30 centimetres of water and driving pressure below 15, alongside targets for oxygen, carbon dioxide, intracranial pressure and brain tissue oxygenation. If oxygenation or intracranial stability deteriorates, clinicians could adjust PEEP, blood pressure, carbon dioxide or ventilator volume while observing the brain’s response. Driving pressure, calculated as tidal volume divided by respiratory-system compliance, and mechanical power, which estimates the total energy transferred from the ventilator to the lungs, may be more informative than tidal volume alone. The review calls for randomised trials that measure lung and brain outcomes together. Until such evidence exists, the safest approach is not a rigid protocol but continuous coordination between pulmonary and neurocritical care teams, using multimodal monitoring to protect two organs whose physiology is inseparably linked.