Fluid Physiology
4.5 Water & the Blood-Brain Barrier


When considering the Starling hypothesis it is usual to consider the important special cases of the glomerulus and the lung. However, the situation with the cerebral capillaries is very different and this seems to be rarely appreciated. The capillary membranes in most of the body are permeable to the low molecular solutes present in blood but are more or less impermeable to the large molecular weight proteins. The only solutes present that can exert an osmotic force across the capillary wall in most of the capillaries are the proteins so the oncotic pressures of plasma and the interstitium are two important Starling’s forces. The low molecular weight solutes can easily cross most capillary membranes so they are not effective at exerting an osmotic force across the capillary endothelial cells.

How are the brain capillaries different?

The differences are due to the blood-brain barrier:

The capillary membrane in the cerebral capillaries is relatively impermeable to most of the low molecular weight solutes present in blood (as well as to the plasma proteins).

The ions Na+ and Cl- make up most of these solutes. These solutes are effective at exerting an osmotic force across the cerebral capillary membrane (the site of the blood-brain barrier). As a consequence, the Starling’s forces in the cerebral capillaries are:

the hydrostatic pressure in the cerebral capillaries

the hydrostatic pressure in the brain ECF (ICP)

the osmotic pressure of the plasma

the osmotic pressure of the brain ECF

Note that it is total osmotic pressure rather than oncotic pressure. The oncotic pressure is extremely small in comparison to the huge osmotic pressure exerted by the effective small solutes in the cerebral capillaries. The small leak of these low molecular weight solutes can be accounted for by a reflection coefficient as with the plasma proteins in other capillary beds. A one milliOsmole /kg increase in osmotic gradient between blood and brain interstitial fluid will exert a force of 17 to 20 mmHg. At an osmolality of 287 mOsm/kg then the total osmotic pressure is about 5400mmHg as can be calculated with the van't Hoff equation. In comparison, the plasma oncotic pressure of 25 mmHg is tiny. 

Therefore even small changes in plasma tonicity can have a marked effect on the total fluid volume of the intracranial compartment. It is not just the intracellular volume of the brain cells but also the volume of the brain ECF that are decreased by an increase in plasma osmolality. In other tissues of the body, an increase in plasma osmolality would increase ISF volume but decrease ICF volume in that tissue.

 Effect of Increase in Plasma Osmolality on Tissue Fluid Volumes


 ISF volume

 ICF volume

 Total fluid volume




 ALWAYS decreased

 Other tissues



 Dept on balance between the increased ISF 
 & the decreased ICF

Infusion of hypertonic solutions of any effective small molecular weight solute (eg hypertonic saline, mannitol or urea) will dehydrate the brain. In the peripheral capillaries, these solutes are not effective at exerting an osmotic force because they can easily cross these capillary membranes. Hypertonic solutions of sodium (as saline) or mannitol are however effective at the cell membrane and will cause cellular dehydration in all body cells. Urea can cross most cell membranes relatively easily and is a much less effective solute at this membrane.

A final comment should also be made about the water permeability of the blood brain barrier. The fluid flux across the capillary membrane is proportional the the net pressure gradient (as stated in the Starling equation). The constant of proportionality in this equation is the filtration coefficient and the value of this is a measure of how easily water crosses the membrane. As discussed earlier, this filtration coefficient is the product of the total area of the capillary walls and the hydraulic conductivity. This hydraulic conductivity is a measure of the water permeability of the membrane. The point to make is that in comparison to other body capillaries the hydraulic conductivity (ie water permeability) of the cerebral capillaries is very much lower. This greatly minimises the amount of water that is lost from the brain in response to changes in plasma tonicity and this is fortunate in view of the huge changes in osmotic forces that occur with tonicity changes of only a few millOsmoles/kg. This very low filtration coefficient is necessary for maintaining a constant intracranial volume. 

Note the difference between the reflection coefficient and the filtration coefficient
The reflection coefficient gives a measure of how well solutes cross a membrane and the filtration coefficient (or more accurately the hydraulic conductivity) gives a measure of how well the solvent (water) crosses a membrane. This distinction is important to consider in the brain because cerebral damage does not necessarily result in equal changes in each coefficient in the area of damage. For example it is often said that hypertonic mannitol solutions are less effective at dehydrating abnormal or damaged areas of the brain (as compared to normal areas) but this is not necessarily correct. A damaged area may have a lower reflection coefficient for low molecular weight solutes so an increase in osmotic gradient due to mannitol will be less effective in this area. However, the damaged area may also have a higher hydraulic conductivity and water is more able to leave the brain in this area. The net effect is that the damaged brain may be dehydrated as much as (or more) than undamaged areas.


The blood brain barrier is impermeable to low molecular weight solutes so the plasma osmotic pressure (rather than plasma oncotic pressure) is the Starling force to be considered here. For the same reason, the brain interstitial osmotic pressure is also a Starling force (rather then the oncotic pressure of the interstitial fluid). 

The reflection coefficient due to these solutes is used rather than the reflection coefficient for the proteins. This reflection is very high for most of these water-soluble solutes.

The Starling equation is also altered for another reason: the hydraulic conductivity of the cerebral capillaries is very much lower than in other capillaries. The filtration coefficient is low. This minimises the amount of dehydration that occurs in response to changes in plasma tonicity. The application of the Starling equation to the brain is different from that anywhere else in the body and it is surprising this is so little appreciated especially in view of the important clinical relevance (eg use of hypertonic mannitol solutions).

Finally, because of Pascal's principle, the interstitial fluid pressure in the brain is equal to the CSF pressure (ie intracranial pressure).

The cerebral capillaries are indeed an important 'special case' as regards the application of Starling's hypothesis.


Last updated
All material © Copyright - Kerry Brandis, 2001