Properties and contents of cerebrospinal fluid

The precious bodily fluid which laps at the shore of your brain is sometimes accessible for analysis, and can be an endless source of physiological amusement. In the CICM exams, the CSF questions typically take the shape of "does this CSF look infected to you?" SAQs of this form include Question 11.1 and Question 11.3 from the second paper of 2008, as well as Question 3.2 from the second paper of 2013, and Question 28.4 from the first paper of 2022.

In summary:

  • Normal CSF volume and circulation:
    • you normally have 150ml or so, and produce 20ml/hr
  • The biochemical properties of CSF:
    • low in protein and high in chloride
  • Protein in the CSF:
    • With intracranial bleeding: for every 100 RBCs you get a 0.01g/L increase in protein
    • Apart from infection, there is a whole list of conditions which cause CSF protein to rise, among them malignancy sarcoidosis and MS.
  • Glucose in the CSF
    • Decreases with most things, most of all with bacterial meningitis
  • Cellular contents of the CSF
    • With intracranial bleeding, for every 1 leucocyte in the CSF you should have 500-1500 RBCs.
  • Unusual tests for CSF include lactate, CK, LDH, and β2 transferrin
  • CSF findings in common clinical scenarios
    • Subarachnoid haemorrhage: xanthochromia, RBCs and low glucose
    • Guillain-Barre Syndrome: lymphocytosis and raised protein
    • Status epilepticus: raised protein

Venkatesh, our former college head, has published a nice article about the various interesting qualities of CSF. In the article, one may find a nice timeline of the history of CSF research, which begins with Galen describing the ventricles, and Antonio Valsalva draining some dog's lumber sack.

Venkatesh is in good company, that of Hippocrates Galen and Cushing. The first opinion on the CSF is credited to Swedendorg, a 16th century theologian whose discussions of the "highly gifted juice" were published well over a century after his death.

Normal volume and circulation of the CSF


A normal-sized person has 125-150ml of CSF, and produces about 20-24ml of it per hour.

The choroid plexus secretes this clear acellular fluid by some mixture of ultrafiltration and active transport.

Probably ultrafiltration has little role to play, because the concentration of electrolytes in the CSF is under fairly tight control and this would not occur if the choroid plexus dumbly pushed plasma through a filter.

The active transport component of CSF production relies on the active transport of sodium and chloride, with water movement through aquaporins and leaky cellular junctions; the whole thing is rather complex and instead of getting carried away with ion channel diagrams I will merely reference an article which scrutinises this topic with exhausting attention to detail.

CSF flow and volume

It is said that the rate of CSF production is about 0.4ml per minute. Thus, one can expect one's EVD output to be 24ml/hr if all of the produced CSF issues forth though the EVD rather than circulating around in a normal manner. In this fashion, one can expect a person to produce about 576ml of of CSF every day, which seems like quite a lot.

Biochemical properties of the CSF

Essentially, this substance is plasma. There is one notable difference, however: it is the near-total absence of protein. This influences the concentration of the ions in the CSF. After all, who will contribute all that extra negative charge, in the absence of anionic protein?

Certanly not the bicarbonate. The pH of the CSF very closely resembles the pH of plasma because the CO2 diffuses freely in and out of the CSF space, and undergoes the same conversion into HCO3-; there is perhaps a subtle difference in CSF HCO3- incomaprison to plasma because the CSF is devoid of all those helpful buffering proteins.

But this is a digression. Early 1960s dog studies compared CSF electolytes to the electrolytes of plasma. Within the normal measurement error ranges, most of the cation concentrations were essentialy the same, with perhaps the tiniest bit less of potassium in the CSF.

CSF electrolyte concentration

Not so for the anions, obviously. Without albumin to donate anionic charge, the chloride concentration in the CSF is about 15-20mmol higher than in the plasma. The dog studies put it at 132mmol/L or so.

Protein content of the CSF

The blood-CSF barrier excludes protein almost completely.

Indeed, there is about 100-200 times less protein in the CSF under normal conditions; so where the total serum protein may be 80g/L, the CSF protein will be 0.4g/L.

Under unusual circumstances, one may find an excessive amount of protein in the CSF, and this tends to be a sign of terrible things happening.

Elevated CSF protein in intracranial bleeding

One can imagine that additional protein in the CSF may have come from the plasma, as a result of some sort of unhappy meeting between blood and CSF. This may be the result of a proper intracranial haemorrhage, or it may be the result of a traumatic tap. Generally speaking, for every 1000 RBCs per microlitre, there should be an extra 0.01g/L of protein, which is not very much.

Elevated CSF protein in infectious and non-infectious conditions

Contrary to popular belief, a slightly elevated CSF protein does not exactly scream about infection.

There are a variety of reasons why the protein might be elevated.

Increased CSF protein due to exudation of protein

  • Viral or bacterial meningitis, with inflamed meninges exuding the protein
  • Viral or bacterial encephalitis
  • Malignancy, leaking protein into the CSF through immature capillaries and disrupted blood-brain barriers
  • Mycobacterial infection, CNS tuberculosis

Increased CSF protein due to increased local synthesis

The specific protein we are talking about here is IgG. This is the major immunoglobulin in the CSF.

It can be elevated in a whole host of conditions:

  • Multiple sclerosis
  • Neurosyphilis
  • Sarcoidosis
  • Viral or bacterial meningitis or encephalitis

Increased CSF protein due to decreased protein resorption

This is typically the case in hydrocephalus, particularly in the communicating hydrocephalus one develops after a large-scale subarachnoid haemorrhage.

Conversely, in the context of a chronic CSF leak, there is a persistently decreased CSF protein.

Glucose content of the CSF

Glucose moves into the CSF by facilitate ransport, and is then sucked back out again by the walls of the CSF cisterns. All those cells eat it right up. Thus, the CSF glucose content is always somewhat lower than that of serum - the ratio is about 0.6. Thus, a patient with a BSL of 10mmol/L will be expected to have a CSF glucose of 6mmol/L.

As there are numerous cells, organisms and chemical processes which feed on glucose, so there are numerous reasons why the CSF glucose might be abnormally low.

Causes of low CSF glucose:

  • Bacterial meningitis
  • viral meningitis, to a far lesser extent
  • Intracranial malignancy
  • CNS sarcoidosis
  • Subarachnoid haemorrhage (the RBCs also consume glucose)

Of all these, the lowest glucose concentrations would be in bacterial meningitis. All those reproducing organisms tend to gobble up glucose at a rapid rate, and the CSF value is typically below 1.0 mmol/L

CSF glucose is usually normal in viral meningitis or encephalitis. Usually. Obviously, exceptions always exist. One would not hang their diagnosis purely on the CSF protein.

Cellular content of the CSF

Having referred to this fluid as acellular, I must grudgingly accept that even in normal CSF some errant cells may be sneaking around. 4-5 RBCs per millilitre is said to be normal; anything more than 3 WCCs is generally abnormal.

Classically, it is said that a neutrophil-dominant WCC around 500 is suggestive of bacterial meningitis, and a monocyte-dominant WCC around 100 is more suggestive of viral meningitis.

However, there are really no fixed rules. Lets say you collect CSF too early in the course of bacterial meningitis- the WCC count may actually be quite normal, seeing as the CSF has not yet had time to change into frank pus.

Additionally, there is time-associated degradation. Let us consider a scenario where the CSF is allowed to brew quietly in the corner of the peripheral emegency department, waiting to be collected and taken by taxi to the nearest laboratory. Over the first few hours, a fair poportion of the WCCs will lyse, and defy detection. Another proportion will settle to the bottle of the plastic tube or stick to the walls. For this reason, one really needs to get the cell count done within 60 minutes or so.

In order to confuse things further, there is a certain amount of WCCs which is expected if there has been an intracranial bleed or traumatic tap. Generally speaking, for every 500-1500 RBCs, 1 leucocyte is permitted in the CSF. The figure changes somewhat for patients with EVDs -the mere presence of an EVD tends to stimulate a CSF pleocytosis, and the pragmatic intensivist will permit a larger proportion of white cells. Of course, this is an evidence-free zone. How much does one relax their standards for RBC/WCC ratio? Depends on how many episodes of ventriculitis one is prepared to ignore.

So lets hypothetically consider a patient recovering from a subarachnoid haemorrhage, with a CSF RCC of oh, say 30,000. This patient would therefore be expected to have a WCC around 60-180. Anything more than that could then be viewed as "CSF pleocytosis". Of course, in order to call it ventriculitis one would actually have to demonstrate an organism, but a rising white cell count in the CSF would certainly make you want to pull out the infected EVD.

This, of course, is a very crude approximation; but it stems from the general belief that any bleeding into the CSF is essentially just blood being diluted by CSF, an act which keeps the proportions of white cells and red cells intact. Reasoning in this way, one can arrive at a more mathematically correct method, comparing the ratio of cells in the CSF with the ratio of cells in the bloodstream. Dr Beer from the ventriculitis article describes this "cell index" as follows:

CSF cell index equation

Under conditions of a totally sterile non-infectious CSF and blood mixture, the cell index should be 1.

In this fashion, one can predict what the CSF WCC should be no matter how deranged and bizarre the hematological abnormalities. Say the patient has a neutrophil count of 110,000 because of some sort of berzerk myeloproliferative disorder? No matter! Cell index should still be 1. Any increase in this index suggests that there are more than the expected amount of WCCs in the CSF, and so some sort of inflammatory process is taking place there.

So, what is causing this pleocytosis?

Causes of raised CSF granulocytes

  • Bacterial meningitis, ecephalitis or ruptured brain abscess
  • Early stages of any sort of meningitis
  • Haemorrhagic stroke (as it gradually resolves)
  • Recently injected intrathecal drugs

Causes of raised CSF lymphocytes and monocytes

  • Mycobacterial, fungal or viral meningitis
  • Guillain-Barre syndrome
  • CNS vasculitis

So, any damn thing, really.

Other tests performed on the CSF

No all of these are exactly routine, but its worth knowing about them.

CSF Lactate

A rises in CSF lactate can be the result of any damn thing, be it bacterial meningitis, acute brain injury, stroke... Its really not a specific marker of anything. Venkatesh's article examines some of its uses; seems as if it an be used to distinquish between viral and bacterial meningitis. This marker is not in routine use.


Lactate dehydrogenase rises in the CSF in response to non-specific cell injury. Generally speaking, CSF LDH levels are 10 times lower than those of serum. It is said that in bacterial meningitis the serum LDH levels are massively elevated, whereas in viral meningitis they are more modestly elevated or actually normal. How many more CSF tests do we need to diagnose bacterial meningitis? This one can be thrown into the same corner.


Creatine kinase in the brain is the result of neuronal breakdown, and there is some relationship between its level and the extent of brain damage. Naturally, the more brain tissue gets smooshed, the more CK it leaks, and the higher the CSF CK level. Now, the article I keep quoting actually presents CK levels which have been associated with certain degrees of injury, which might give one the impresson that CSF CK is somehow useful in prognostication. However, between the volume of smooshed brain and the quality of neurological outcome exists a very feeble relationship. I would be very reluctant to rest a family's hopes on this single biological marker, early in the course of a head injury.

CSF β2 transferrin

There are very few proteins which are specific to the CSF, and which are absent from the plasma. β2 transferrin is one of these. It is the product of the normal β-transferrin which seeps into the CSF, and is then worked over by CNS neuraminidase. It can be useful in detecting the leakage of CSF. And lets face it, there are many circumstance during which we gather around a spot of wetness on the patients' pillow, and wonder whether that spot is CSF or just snot.

Sure, there are the traditional investigations. One can dribble the fluid into some gauze or absorbent tissue and watch for the "target sign", where the blood remains central while the CSF spreads outwards. Or one can collect some of the fluid and send it to the lab for some sort of biochemical confirmation of it being CSF. This, naturally, brings up delicate questions regarding the biochemical composition of saliva and nasal mucus. How much protein should there be? Chloride? Lactate? And are these parameters constant? Could there, even in some alternate universe, be a scenario where the biochemical composition of one's mucus is very similar to the composition of one's CSF?

Fortunately the existence of the β2 transferrin assay spares one the need to closely examine these questions.

Changes in CSF in common clinical situations

Post neurosurgical patients, with EVD in situ

Apart from having had their heads cut open, these patients face the peculiar experience of having a thin silicone rubber tube in their brain. This is ok for 3 or so days; thereafter, the risk of infection begins to increase exponentially. Unfortunately, the actual presence of such a brain tube tends to irritate the delicate brain parenchyma, causing an increase in CSF WCC and protein, making it difficult to diagnose such an infection.

The presence of fever tends to be the decisive feature; and of course there is always the CSF culture.

Subarachnoid haemorrhage

The SAH patient will have RBCs(obviously), boring old xanthochromia, as well as a decrease in the CSF glucose (because the RBCs have been metabolising it).

Guillain-Barre Syndrome

Late in the game these people develope a bit of a CSF lymphocytosis, but the real money is in the protein. A pathognomonic feature of the disease is a rise in CSF protein. Weirdly, critical illness polyneuropathy tends to have an increased CSF protein also, but not quite to the same level as GBS.

Status epilepticus

Somehow, epilepsy results in a leaky blood-brain barrier. The CSF is thus enriched with plasma protein (though not usually cells, though sometimes the CSF WCC is mildly elevated). A slight rise in CSF protein is therefore to be expected after prolonged seziures.


Venkatesh, B., P. Scott, and M. Ziegenfuss. "Cerebrospinal fluid in critical illness." Critical care and resuscitation 2 (2000): 42-54.

SCHAIN, RICHARD J. "Cerebrospinal fluid and serum cation levels." Archives of Neurology 11.3 (1964): 330.

POSNER, JEROME B., AUGUST G. SWANSON, and FRED PLUM. "Acid-base balance in cerebrospinal fluid." Archives of neurology 12.5 (1965): 479.

Brown, P. D., et al. "Molecular mechanisms of cerebrospinal fluid production."Neuroscience 129.4 (2004): 955-968.

Steele, Russell W., et al. "Leukocyte survival in cerebrospinal fluid." Journal of clinical microbiology 23.5 (1986): 965-966.

Beer, R., P. Lackner, and B. Pfausler. "Nosocomial ventriculitis and meningitis in neurocritical care patients." Journal of neurology 255.11 (2008): 1617-1624.

Zaret, Deborah L., et al. "Immunofixation to quantify beta 2-transferrin in cerebrospinal fluid to detect leakage of cerebrospinal fluid from skull injury."Clinical chemistry 38.9 (1992): 1908-1912.

Swedenborg, E. "The Brain." Considered Anatomically, Physiologically and Phylosopically.(translated and edited by RL Tafel), London (1887).

Hajdu, Steven I. "Discovery of the cerebrospinal fluid." Annals of Clinical & Laboratory Science 33.3 (2003): 334-336.