Name the compartment,
and the manoeuvre explains itself.

Venous outflow — the first thing to check, and the cheapest to fix

• Head-up posture, conventionally about 30°, improves the pressure gradient from the dural sinuses to the right atrium. A neutral head position matters as much as the tilt: extreme rotation or flexion, a tight cervical collar or tight endotracheal tube ties occlude the internal jugular veins.
• Airway and intrathoracic pressure. Airway obstruction, a kinked tube, bronchospasm, gas trapping, tension pneumothorax and excessive PEEP all raise central venous pressure and impair cerebral venous drainage. Use the minimum PEEP that achieves adequate oxygenation — but not so little that the patient becomes hypoxaemic, because hypoxaemia raises ICP by its own route.
• Coughing and straining take ICP transiently to around 50 mmHg even in a normal brain. Adequate depth, adequate analgesia and neuromuscular blockade at laryngoscopy and during craniotomy are physiological interventions.

Arterial blood volume — carbon dioxide, oxygen and metabolic rate

• Avoid hypoxaemia absolutely. Below a PaO₂ of about 8 kPa (60 mmHg) there is abrupt cerebral vasodilatation, so hypoxaemia is not only a delivery problem but directly an ICP problem.
• Control carbon dioxide. The relationship is roughly linear across the physiological range, so PaCO₂ is the fastest handle on cerebral blood volume that the anaesthetist has.
• Suppress cerebral metabolic rate where flow is the problem: treat seizures and fever, and use hypnotic agents. Flow–metabolism coupling means a fall in metabolic rate carries flow, and therefore volume, down with it.

Choice of anaesthetic agent

Open cranium against closed

With the cranium and dura open, the container is no longer rigid, so the Monro–Kellie doctrine no longer produces a rising pressure — added volume escapes through the craniotomy. The physiology does not stop mattering; it changes what it costs you. Instead of a pressure you get a tight, bulging surgical field, poor access, brain herniating through the bone defect and mechanical injury to the herniated tissue. The same manipulations apply for the same reasons, but the result is judged by looking at the field rather than by reading a number. The same logic explains decompressive craniectomy: it converts a closed rigid box into an open one, so the volume-pressure curve loses its steep limb.

A structured approach to an acutely rising ICP

Work through the compartments in the order you can act on them fastest. This is the physiology of §12 turned into a checklist.

1. Venous. Head neutral and up? Collar or tube ties compressing the jugulars? Airway pressure high — obstruction, bronchospasm, kinked tube, pneumothorax, excessive PEEP? Coughing or straining?
2. Arterial. PaCO₂ and PaO₂? Febrile or fitting? Any vasodilator running — nitrous oxide, a volatile above 1 MAC, nitroprusside, a calcium-channel blocker? Arterial pressure appropriate for the CPP target?
3. CSF. Is there a ventricular drain, and is it clamped or blocked? Drainage is the only practical way to manipulate this compartment acutely.
4. Cellular and fluid. Is this a new or expanding mass lesion? A sharp rise over an hour in a head-injured patient means re-imaging, because an expanding haematoma is a surgical problem, not a physiological one. Osmotherapy and steroids act on the fluid compartment.

Cerebral protection means preventing the secondary insult. The primary injury — the clip on the vessel, the moment of arrest, the impact — has already happened and cannot be undone. What can be changed is everything that happens to the brain afterwards, and the brain injured once is far less able to defend itself against a second insult than a normal brain is against a first.

Physiology first, and by a wide margin

The single most useful conclusion in this field is not about a drug. Anaesthetic neuroprotection is real in experimental models but modest, and it is achieved only when physiological homeostasis is rigorously maintained. The potential to worsencerebral injury through physiological mismanagement is substantially greater than the protection any drug provides. Perfusion pressure, oxygenation, carbon dioxide tension, temperature, glucose and seizure control are therefore not the background against which protection happens — they are the protection.

Temperature, the one technique that unequivocally works

Hypothermia is the principal protective technique for circulatory arrest, and it unequivocally increases the brain’s tolerance of ischaemia. The reason is the one distinction this module keeps returning to. An anaesthetic can only suppress the electrophysiological component of cerebral metabolic rate — about 60% of CMRO₂ in the awake state — and once the electroencephalogram is isoelectric there is nothing further to suppress. Hypothermia reduces both that component and the energy spent maintaining cellular integrity, and so keeps working past the point at which the trace goes flat. Mild hypothermia may act preferentially on the second.

In the laboratory, cooling by as little as 2 to 4 °C during ischaemia confers substantial protection, and cooling begun immediately after the insult still benefits. The clinical picture is more divided, and the divisions are worth holding separately rather than as one verdict on hypothermia.

What the brain is being protected from, and for how long

The cascade itself belongs to lesson 5 — the pump fails, the membrane depolarises, calcium floods in, glutamate excitotoxicity amplifies it, and proteases, lipases, free radicals and endonucleases dismantle the cell. What matters here is when the dying happens, because that is what decides whether any intervention can reach it.

Neuronal death after ischaemia is not confined to the insult and the early reperfusion period, which was the traditional view. It runs in two phases. Where ischaemia is severe, most neurons die rapidly. Where it is more moderate, neurons that survive the initial insult undergo delayed death — a dynamic process continuing long after the initiating event, first shown in global ischaemia and since in focal ischaemia too. This is what makes a cerebral infarct expand gradually rather than declaring its final size at onset, and experimentally there is still evidence of cerebral inflammation six to eight months after the primary insult.

The vascular consequence runs alongside. After brain injury or global ischaemia autoregulation is impaired or abolished, often regionally, and where it remains intact the curve shifts to the right — so flow follows pressure more directly than it did, and both hypotension and hypertensive surges become dangerous in the same patient. That is the mechanism underneath the perfusion target in the table above, and the reason it is a target rather than a minimum to be undercut when convenient.

The drugs, and why none of them is the answer

Barbiturates, propofol, the volatile agents, xenon and ketamine all reduce injury in experimental models, and the brain under anaesthesia is less vulnerable to ischaemic injury than the awake or lightly sedated brain. But no direct comparison has shown any one agent or combination superior to another, the protection is not sustained once the insult is more than mild, and the human evidence is largely negative.

One minute on CSF

CSF is a clear, acellular, low-protein transcellular fluid filling the ventricles and the subarachnoid space — about 150 mL, made at about 0.35 mL per minute, roughly 500 mL a day, so it turns over three to four times daily. It is made mostly by the choroid plexus, in two stages: ultrafiltration across a fenestrated choroidal capillary, then active secretion across the choroid plexus epithelium, whose apical tight junctions form the blood–CSF barrier. Active sodium transport into the ventricle creates an osmotic gradient and water follows, so the fluid is near-isosmotic with plasma but differs from it — lower potassium, lower glucose, higher chloride, almost no protein, no cells, pH about 7.32. It provides buoyancy, reducing effective brain weight from 1400 g to under 50 g; mechanical protection; a stable ionic environment; a volume buffer for intracranial pressure; acid–base signalling to the central chemoreceptors; and waste clearance. It flows from the lateral ventricles through the foramina of Monro, the third ventricle, the cerebral aqueduct and the fourth ventricle, out through the foramina of Magendie and Luschka into the cisterns and subarachnoid space, and is absorbed by pressure-dependent bulk flow through arachnoid villi into the dural venous sinuses.

One minute on ICP regulation

Intracranial pressure is the pressure generated within the rigid cranium by its contents, normally 5 to 15 mmHg in a supine adult. By the Monro–Kellie doctrine the cranium is a fixed volume containing brain, blood and CSF, so one compartment can only grow at another’s expense. Only CSF and venous blood are displaceable. As a mass expands, CSF is translocated to the spinal subarachnoid space and its absorption rises, then venous blood is displaced into the internal jugular veins — the flat limb of the volume-pressure curve. When that reserve is spent, compliance falls, elastance rises, and ICP climbs steeply. Cerebral perfusion pressure, which is mean arterial pressure minus intracranial pressure, falls with it; below a CPP of about 60-65 mmHg autoregulation fails, cerebral blood flow becomes pressure-passive and ischaemia begins, causing further vasodilatation and oedema in positive feedback. Late, brainstem ischaemia triggers the Cushing response — sympathetically driven hypertension with reflex bradycardia and an abnormal respiratory pattern — and finally herniation. The determinants I can change are arterial carbon dioxide and oxygen tension, cerebral metabolic rate, venous outflow, and the choice and depth of anaesthetic agent.

Draw-from-memory checklist

• The intracranial volume-pressure (elastance) curve — axes with units, flat limb labelled with what is compensating, knee marked as the point of decompensation, steep limb, focal ischaemia from about 20 mmHg and global ischaemia above about 45 mmHg.
• The normal ICP waveform — pressure in mmHg against time in seconds, 20 mmHg dotted, three peaks of decreasing amplitude labelled P1, P2, P3.
• The reduced-compliance waveform — raised baseline, rounded, P2 taller than P1.
• The CSF pathway — nine steps from lateral ventricles to dural venous sinuses, in order, with the named foramina.
• A Monro–Kellie box — three compartments, fixed total width, and what happens to each as a mass is added.

Common misconceptions

The two barriers, side by side

Take-home points. CSF is actively secreted, not filtered, and its composition proves it. Absorption is pressure-dependent and production is not, which is where all of CSF’s buffering capacity comes from. The cranium is a fixed volume, so only CSF and venous blood can give way, and both are small. Compliance and elastance are reciprocals, and the curve conventionally drawn plots elastance. ICP matters because of CPP, and CPP is a gradient, so a normal blood pressure guarantees nothing. The Cushing response is late. And every anaesthetic manoeuvre for a tight brain acts on one of four compartments — name the compartment, and the manoeuvre explains itself.