The Ventricles and Cerebrospinal Fluid

The four ventricles and the doors between them

The brain is not solid. Buried inside it is a connected system of cavities, the ventricles, which are the remnants of the hollow neural tube from which the whole central nervous system develops. Their shape is the shape of that tube after the brain has grown around and over it, which is why they are so irregular: they are what was left when the tissue expanded.

There are four, and they are arranged in series.

The channels connecting them are as important as the chambers, because it is the channels that block.

1. Interventricular foramina (of Monro)

   Two small openings, one on each side, connecting each lateral ventricle to the single third ventricle. This is the first narrowing in the system, and the first place a lesion can dam the flow.

2. Cerebral aqueduct (of Sylvius)

   A slender canal running through the midbrain from the third ventricle to the fourth. It is the narrowest passage in the whole ventricular system, and it is therefore the classic site of obstruction: aqueductal stenosis is a leading cause of congenital hydrocephalus.

3. Foramina of Luschka and Magendie

   Three exits from the fourth ventricle, the paired lateral foramina of Luschka and the single midline foramen of Magendie. These are the only routes by which fluid leaves the interior of the brain and reaches the subarachnoid space on its surface. Block all three and the entire ventricular system dams up behind them.

A useful way to hold the layout: lateral, lateral, third, fourth, in series, with the narrowest point in the middle of the chain. Everything upstream of a blockage swells; everything downstream stays normal or shrinks. That single rule lets you read most obstructive hydrocephalus off a scan.

The choroid plexus makes fluid; it does not filter it

Cerebrospinal fluid is produced by the choroid plexus, a frilled, highly vascular tissue that hangs into the ventricles: into the bodies and temporal horns of the lateral ventricles, into the roof of the third, and into the roof of the fourth. Structurally it is a tuft of capillaries wrapped in a single layer of specialised epithelial cells, joined to each other by tight junctions.

Those tight junctions are the whole point, and they are the reason the standard shorthand, that CSF is "a filtrate of blood", is wrong.

If CSF were a filtrate, it would be produced by pressure: blood pushed through a sieve, with the small molecules passing and the large ones held back, and the resulting fluid would be plasma minus its proteins. Some accounts still describe it that way. But the choroid plexus epithelium is not a sieve. It is sealed by tight junctions, so nothing simply leaks across it. Instead the epithelial cells pump. They use sodium-potassium ATPase and a set of associated transporters and channels, including carbonic anhydrase and aquaporins, to move ions from the cell into the ventricle, and water follows osmotically. Production is therefore an active, energy-consuming, and above all regulated process.

Two consequences follow, and both matter. The first is compositional. CSF is not plasma with the protein removed: it is a chemically distinct fluid, with a much lower protein concentration, lower potassium, lower calcium, and a slightly higher chloride and magnesium concentration than plasma, and these differences are maintained. The second is pharmacological and clinical. Because production is an active transport process, it can be inhibited: acetazolamide, a carbonic anhydrase inhibitor, reduces CSF production and is used for exactly that reason in some conditions of raised intracranial pressure. You cannot pharmacologically inhibit a passive filter. You can inhibit a pump.

The circulation: 500 mL a day into a 150 mL space

Trace a molecule of fluid from the moment it is secreted. It is released by the choroid plexus of a lateral ventricle. It moves through the interventricular foramen into the third ventricle, down the cerebral aqueduct into the fourth, and then out through the foramina of Luschka and Magendie into the subarachnoid space, the layer between the arachnoid and pia mater that wraps the outer surface of the brain and spinal cord. There it circulates over the convexities of the cortex and down around the cord, before being reabsorbed.

Reabsorption happens principally at the arachnoid granulations (also called arachnoid villi), small projections of arachnoid membrane that push up into the dural venous sinuses, above all the superior sagittal sinus running along the top of the brain. Fluid passes from the subarachnoid space, through these granulations, and into venous blood, which carries it away. The process is largely pressure-driven: when CSF pressure exceeds venous pressure, fluid crosses; when it does not, it does not. That is a simple and rather elegant self-regulating valve. Additional drainage along cranial and spinal nerve roots into lymphatics is now well recognised and is an active area of research; the older picture in which the granulations were the sole exit has been complicated.

Now the numbers, which are the most instructive part of this section.

Put those first two figures side by side and something important falls out. The standing volume is about 150 mL. Production is about 500 mL a day. The system therefore replaces its entire contents roughly three to four times over in the course of a day. This is not a sealed reservoir topped up occasionally. It is a river.

And that has an immediate corollary that explains why hydrocephalus is dangerous. Production is not strongly regulated by pressure: the choroid plexus goes on pumping, at very nearly the same rate, whether the fluid can get out or not. If drainage is obstructed and secretion continues, roughly half a litre a day of new fluid must be accommodated inside a rigid skull. It cannot be. The pressure rises, and it rises quickly.

Buoyancy: why a 1.4 kg brain weighs 50 g

Of all the functions of cerebrospinal fluid, this is the one to lead with, because it is the most physically striking and the most easily verified from first principles.

The problem is real. The adult human brain has a mass of roughly 1.4 kilograms and the mechanical properties of a very soft gel. Living brain tissue is soft enough that it deforms visibly under its own weight when removed from the skull, which is precisely why anatomists fix a brain in formalin before attempting to slice it: fresh, it will not hold its shape. Yet inside the head this fragile mass sits above a floor of bone, pierced by holes and crossed by ridges, and it does not tear, compress, or occlude its own blood supply. Why not?

Because of Archimedes. His principle states that an object immersed in a fluid experiences an upward buoyant force equal to the weight of the fluid it displaces. The consequences depend entirely on how the density of the object compares with the density of the fluid, and here the numbers are remarkable.

Brain tissue has a density of roughly 1.04 to 1.05 grams per cubic centimetre. Cerebrospinal fluid has a density of roughly 1.00 to 1.01. These are almost, but not quite, the same. The brain is very slightly denser than the fluid it floats in, which means that the buoyant force very nearly, but not quite, cancels its weight. What is left over, the small residual, is the brain's effective weight, and it is on the order of 50 grams rather than 1400.

Once you see this, several other facts stop being separate facts and become consequences of one.

It explains why the brain does not crush its own base. Fifty grams distributed across the skull base is nothing. Fourteen hundred grams, concentrated on the sharp ridges of the sphenoid bone and on the vessels and nerves passing through the base, would be catastrophic within hours: the tissue at the bottom would be compressed, its blood supply choked, and it would die.

It explains, too, one of the more disconcerting sights in a dissection room, the way a fresh brain lifted out of the skull immediately begins to sag and deform under gravity. Nothing has been damaged. The brain has simply been taken out of the water. Its true weight has been restored to it, and its true weight is more than its structure can support.

And it explains a subtlety that is often missed: the effect is one of neutral suspension, not of resting on a fluid mattress. The brain is supported from every direction at once, because pressure in a fluid acts on all surfaces. It is not standing on the CSF beneath it; it is held, evenly and everywhere, by the fluid around it. This is why the delicate vessels and cranial nerves that leave the brain are not sheared off by the weight of the organ they are attached to. There is almost no weight to shear them.

Cushioning, and the limits of the cushion

The second mechanical function is protection against impact, and it works by the same physics as the first but on a different timescale.

When the head is struck, the skull accelerates. If the brain were rigidly fixed to it, the brain would accelerate identically and instantly, which would deliver the full impulse of the impact to the tissue. Instead the brain is floating, mechanically coupled to the skull only through the fluid and the meninges. The fluid layer means that the brain's acceleration is spread out over a slightly longer time and distributed over its whole surface rather than concentrated at the point of contact. A gentler acceleration, applied everywhere at once, is a smaller force on any given piece of tissue.

The same suspension has a further use. The brain is not only buffered against blows; it is buffered against everyday movement. Every step you take jolts your skull, and every jolt is absorbed by the fluid rather than transmitted as a shock to the tissue.

But the cushion has limits, and it is worth being honest about them, because CSF is often described as though it made the brain safe.

Buoyant suspension protects well against static load and against modest, evenly distributed acceleration. It protects far less well against rapid rotational acceleration. If the head is spun sharply, the skull turns, and the brain, being suspended in a low-viscosity fluid and having some inertia of its own, lags behind and then twists. The shearing that results is concentrated in the long white matter tracts and at the boundaries between tissues of different density, and it is the mechanism behind diffuse axonal injury and behind a good deal of concussion. Similarly, in a decelerating impact the brain can strike the inside of the skull on the side of impact and then rebound against the opposite side, the coup and contrecoup pattern. The fluid layer softens all this. It does not abolish it.

There is a design lesson buried here. The very property that makes CSF a superb static support, its near-perfect density match, is what allows the brain to move relative to the skull under rapid acceleration. You cannot have neutral buoyancy and rigid restraint at the same time. The brain's suspension is an engineering compromise, and traumatic brain injury is the price of the compromise.

Chemical stability: the third job, and the quietest

The mechanical functions are the ones that get illustrated. The chemical one may matter more.

Neurons signal by moving ions across their membranes. An action potential is, at bottom, a rapid and precisely timed movement of sodium and potassium down electrochemical gradients, and the size and shape of that signal depend on the concentrations of those ions in the fluid outside the cell. A neuron sitting in an extracellular fluid whose potassium concentration wandered up and down with the contents of the last meal would be an unreliable neuron. Raise extracellular potassium and neurons depolarise and become hyperexcitable; the ionic environment is not a background condition but a determinant of whether the brain works at all.

Blood plasma is not stable in this way. Its composition shifts with diet, exercise, hydration, and hormonal state, within limits that the body tolerates perfectly well elsewhere. The brain cannot tolerate them. So it does not share the blood's fluid.

CSF, together with the brain's own interstitial fluid with which it freely exchanges, forms a separately regulated compartment, walled off by the blood-brain barrier at the capillaries and by the blood-CSF barrier at the choroid plexus, and actively maintained at a composition of the brain's choosing. Potassium in CSF is held at a lower concentration than in plasma and, more importantly, is held steady: when plasma potassium swings, CSF potassium barely moves. The choroid plexus is not merely a tap. It is a chemical regulator, and the continuous 500 mL a day of production is the flow that lets it keep the compartment where it wants it.

The same barrier arrangement means that CSF has an immunological character of its own, and a very low protein content, which is why the sudden appearance of white cells, protein, or bacteria in the fluid is such a powerful diagnostic signal. A normal CSF sample is a sample of a jealously guarded compartment. An abnormal one means the guard has been breached.

Waste clearance and the glymphatic hypothesis

Every tissue in the body produces metabolic waste, and the brain, which consumes around a fifth of the body's oxygen at rest, produces a great deal of it. In most organs, waste is collected by the lymphatic system. The brain parenchyma has no conventional lymphatic vessels running through it. So how is it cleaned?

Since 2012 the leading answer has been the glymphatic system, so named because the pathway is glial-dependent and performs a lymphatic-like function. The proposal is this: cerebrospinal fluid enters the brain along the outsides of arteries, in the perivascular spaces that surround them, is driven from there into the brain tissue proper through aquaporin-4 water channels concentrated on the endfeet of astrocytes, sweeps through the interstitial space collecting solutes as it goes, and exits along the perivascular spaces of veins, carrying the waste with it. In effect, the fluid is proposed to flush the tissue.

The finding that made this famous concerns sleep. In 2013, Xie and colleagues reported in Science that in mice this exchange of cerebrospinal fluid with interstitial fluid increased markedly during sleep and under anaesthesia compared with wakefulness, that the interstitial space itself expanded, and that clearance of injected amyloid-beta was faster in the sleeping animal (Xie L, et al. Science. 2013;342(6156):373-377). The implication drawn was arresting: that one of the functions of sleep is to wash the brain, and that the reason the brain must go offline is that this cleaning cannot easily be done while it is running.

Why does it matter enough to argue about? Because of what the fluid may be carrying. Amyloid-beta, the peptide that aggregates into the plaques of Alzheimer's disease, is one of the solutes whose clearance has been studied, and if clearance is genuinely impaired by chronic sleep loss, then a lifetime of poor sleep would plausibly contribute to the accumulation of the very protein that characterises the disease. That is a hypothesis with enormous public health implications, which is exactly why it deserves the scepticism rather than the enthusiasm it has often received in popular coverage. For the wider picture of what sleep is for, see sleep and the brain.

Hydrocephalus: when the plumbing fails

Return to the two numbers: production about 500 mL a day, standing volume about 150 mL, and production largely indifferent to pressure. Now consider that the skull is a closed rigid box containing three things, brain, blood, and CSF, whose combined volume is fixed. Any increase in one must be paid for by a decrease in another, or by a rise in pressure. This is the Monro-Kellie doctrine, and it is the reason a fluid problem becomes a pressure problem, and a pressure problem becomes a brain problem.

Obstructive (non-communicating) hydrocephalus occurs when the flow is blocked somewhere within the ventricular system: at the interventricular foramen, in the cerebral aqueduct, or at the outlets of the fourth ventricle. Fluid continues to be made upstream of the obstruction and cannot get past it. The ventricles above the block dilate; those below do not. Aqueductal stenosis is the classic example, and a tumour of the posterior fossa pressing on the fourth ventricle is another common one. The pattern of ventricular enlargement on imaging localises the blockage, because the swelling stops exactly where the obstruction is.

Communicating hydrocephalus occurs when the ventricles are all in free communication with each other and with the subarachnoid space, but reabsorption fails. The commonest causes clog the arachnoid granulations: blood, after a subarachnoid haemorrhage, or inflammatory debris after meningitis. Fluid circulates normally but cannot get out into the venous blood, and so the whole system, ventricles and subarachnoid space together, distends.

The danger in both is raised intracranial pressure. Because the box cannot expand (in adults; in an infant, whose skull sutures have not fused, it can, which is why untreated infantile hydrocephalus enlarges the head rather than immediately killing the child), the rising pressure compresses the brain and, critically, opposes the arterial pressure driving blood into it. Cerebral perfusion pressure is arterial pressure minus intracranial pressure. Raise intracranial pressure high enough and perfusion falls to nothing, whatever the blood pressure. Before that point is reached, the pressure gradient may force brain tissue through the openings in the skull and the folds of dura, herniation, which compresses the brainstem and is rapidly fatal.

The standard treatment is mechanical and slightly brutal in its simplicity: a shunt, a tube with a one-way valve, running from a ventricle to somewhere the fluid can be absorbed, usually the peritoneal cavity. A ventriculoperitoneal shunt does not correct the underlying failure. It simply provides an alternative drain.

A separate and puzzling condition, normal pressure hydrocephalus, presents in older adults with the triad of gait disturbance, cognitive decline, and urinary incontinence, together with enlarged ventricles but CSF pressure that measures within the normal range on a single reading. It is diagnostically contentious, it is over-diagnosed and under-diagnosed by turns, and it is one of the reasons the physiology of CSF remains a live subject rather than a closed one.

The lumbar puncture: reading the fluid

Because the subarachnoid space is continuous from the brain all the way down around the spinal cord, and because the spinal cord itself ends at around the first or second lumbar vertebra while the dural sac containing the fluid continues to about the second sacral segment, there is a region of the lower back where a needle can enter the fluid without any risk of striking the cord. It will pass among the loose nerve roots of the cauda equina, which drift aside. This is the anatomical basis of the lumbar puncture, and it is a rare gift: direct, safe, repeatable physical access to the internal environment of the central nervous system.

What it yields is a window. The opening pressure is measured, which is a direct reading of intracranial pressure. The fluid is inspected: normal CSF is as clear as water, and turbidity means cells. It is then analysed.

One caution completes the picture, and it follows directly from the physics of the previous section. If intracranial pressure is dangerously raised by a mass lesion, withdrawing fluid from the lumbar sac can create a pressure gradient down the neuraxis and precipitate herniation of the brain downward through the skull's openings. This is why a patient with signs of a mass lesion is imaged before a needle is put in. The window is a real one, and it can be opened at the wrong moment.

What CSF is not

Related topics