Water exchange across the blood–CSF barrier: A systematic review, 2026, Hjørnevik and Eide

[SNT Gatchaman]

Water exchange across the blood–CSF barrier: A systematic review
Trine Hjørnevik; Per Kristian Eide

Interest in cerebrospinal fluid (CSF) physiology and its relevance to neurological disease has increased markedly in recent years. Classical descriptions portray CSF as a unidirectional flow from the choroid plexus to the dural venous sinuses and rarely distinguish between its solutes and the solvent (water) component, which constitutes ~99% of CSF.

We conducted a systematic literature review to evaluate current evidence for water exchange between CSF and blood across the blood–CSF barriers (BCSFB). Eighteen studies met the inclusion criteria: 15 in experimental animals and six with humans, spanning more than 70 years and employing diverse methodologies.

This literature review shows that CSF water moves freely and bidirectionally between CSF and blood across multiple BCSFB sites along the craniospinal axis, including the choroid plexus, ependymal surfaces, pial vessels, and perivascular spaces. The net direction of movement varies locally with hydrostatic, osmotic, and molecular gradients that transiently favor either inflow or outflow. Blood–CSF water exchange occurs predominantly by diffusion and is modulated by aquaporins and local vascular forces.

These findings challenge the classical concept of unidirectional CSF production and absorption, supporting instead a dynamic equilibrium where distributed, gradient-driven water flux maintains brain water homeostasis.

Web | DOI | PDF | Journal of Cerebral Blood Flow & Metabolism | Open Access

 

[SNT Gatchaman]

Water exchanges continuously between blood, interstitial fluid, and CSF along the entire craniospinal axis, driven by local gradients in hydrostatic pressure, surface area, permeability, and osmolality. This exchange can occur independently of solute transport, underscoring the need to distinguish CSF water dynamics from solute circulation.

Recognizing CSF as a heterogeneous fluid, whose water and solute components follow distinct anatomical and molecular pathways, fundamentally reframes classical notions of “production” and “absorption” as manifestations of transient, region specific biases in a broader bidirectional equilibrium. AQP4 provides the principal low-resistance pathway linking vascular, glial, and ependymal interfaces, while disease states such as hydrocephalus, hypertension, and neurodegeneration likely disrupt this finely regulated hydraulic network.

Despite these insights, contemporary evidence remains limited, and the molecular and pharmacological determinants of blood–CSF water exchange are poorly understood.

 

[Jonathan Edwards]

Thanks for finding that. This is very much consistent with what I concluded on the other glymphatic thread. What is intriguing is that they do not seem to have identified the overarching reason why CSF has to exist in the context of the unique pressure gradients across brain blood vessels - the Levick/Michel principle. At least not in the abstract. I will read through (the PDF link didn't work for me).

 

[Jonathan Edwards]

I had a further scan through the PDF, which I got through the DOI. They talk of an equilibrium rather than directional flow. This suggests a misunderstanding. The key point from Levick is that there is a far-from-equilibrium oncotic state the needs maintaining. That might look like 'flow equilibrium' if you don't consider the oncotic problem though. They note that there can be net water flux without net ion flux - which is what an oncotic gradient does, but maybe not connect the dots?

 

[SNT Gatchaman]

the PDF link didn't work for me

I think I've corrected it. Sometimes the journal gives a slightly or temporarily incorrect URL. Otherwise it should be accessible via the Web link.

 

[Jonathan Edwards]

I have had a read through this. It provides some citations and findings about water flux but there seems to be a serious lack of understanding of what the results imply.

Water can move in tissues in three ways:
1. Convective flow. This is where water molecules all move together as in coming out of a tap or passing through large pores.
2. Diffusion under random molecular motion. This is what happens when you inject deuterated water into one pace and see how fast it appears in another compartment.
3. Diffusion under hydrostatic or oncotic gradient. This is subtle. It is like what happens when you place rice in warm water and the water swells the rice up. In that case the gradient is created by fixed water excluding macromolecules but in tissues the same effect occurs due to oncotic solutes reducing the chemical potential of water. This effect operates across pores that are too small for water to go through in a convective bulk way.

Number 2 isn't of any great interest because for plain water it is just molecules changing places with no physiological change. But it is what investigators have mostly measured with tracers.

Number 3 is the interesting one, but I think you can maybe only measure it indirectly by placing pipettes into each compartment.

The review basically argues that people thought CSF flow was all type 1. They then cite type 2 movement to argue against this but type 2 doesn't have any function. They seem to miss the potential importance of type 3. At least that is how it comes across to me.

The one observation that seems to stand out is that water passes very easily between compartments in the skull, including vessels, so the idea that brain vessels do not let water across seems to be misguided.

I hope to have a chat with a local neurosurgeon soon, who works on tracer studies.

 

[Utsikt]

Possible nonsensical question:

Could OI be caused by reduced blood pressure (not blood flow per se) in the brain, disrupting the fluid dynamics?

Or would the reduced brain BP be detectable with normal BP measurements? Now that I think about it, I seem to remember that the pressure inside a container is evenly distributed, which would ruin my hypothesis.

 

[Jonathan Edwards]

Now that I think about it, I seem to remember that the pressure inside a container is evenly distributed, which would ruin my hypothesis.

Only where there is no flow.

Low systemic blood pressure would reduce the rate of nutrient/oxygen bearing blood flow through brain capillaries and that is why people feel faint when they stand up too quick normally. The blood volume in brain need not change, just how fast it is moving through.

The hydrostatic pressure in small vessels falls as you go along so there isn't any one measurement you can use. Constriction of brain arteries could make the poor flow worse but there doesnt seem much reason to invoke that.