PHYSIOLOGY
Osmolarity and partitioning
of fluids
Ben R Waterhouse
Andrew D Farmery
Abstract
This article first outlines definitions and descriptions of key terms that will
be used in the discussions that follow including osmolarity, osmolality,
osmotic and oncotic pressure. The physicochemical properties of water,
ions and organic molecules are discussed in terms of their biological
roles. Similarly, the interactions of amphipathic molecules and their
three-dimensional structures in aqueous and non-aqueous environments
are then explored. The movement of solutes and solvents across semiper-
meable membranes is considered and an assessment is made of the
contribution of such mechanisms to normal physiology. Firstly, simple
diffusion is described, followed by comparisons with facilitated diffusion
and energy-dependent active transport focusing on the differences in ki-
netics and rate-limiting factors arising as a result. The mechanisms under-
lying GibbseDonnan equilibria are discussed, with particular emphasis as
to how they arise across a selectively permeable membrane by way of a
worked example. The equilibrium of forces influencing fluid movements
across the capillary endothelium, known as Starling forces, is then
described. Comparisons are made between the classic model as first pro-
posed by Starling in 1896 and the modified glycocalyx model, which has
developed over the past 25 years. Finally, the impact of these differences
on our understanding of trans-capillary fluid flux is discussed.
Keywords Active transport; diffusion; GibbseDonnan equilibrium; gly-
cocalyx; ion channels; MichaeliseMenten; osmolality; osmolarity; Starling
forces
Royal College of Anaesthetists CPD Matrix: 1A01.
Properties of fluids
Key definitions
Osmolarity: the osmolarity, or osmotic concentration, of a so-
lution is dependent on the number of particles present per unit
volume regardless of their nature and is measured in osmoles/
litre of solution. Due to the low concentrations of solutes in
physiological fluids, quoting osmolarity in milliosmoles/litre
(mOsm/litre) is often more appropriate, for example plasma
osmolarity has a normal range of 285e295 mOsm/litre. The
relationship between the molarity of a solution and its osmolarity
depends on the number of particles derived from each molecule.
For example glucose remains intact in solution and so a 5 mmol/
Ben R Waterhouse BA BM BCh is an Academic Foundation Doctor in North
Bristol NHS Trust, UK. Conflicts of interest: none declared.
Andrew D Farmery BSc MA MD FRCA is a Fellow and Tutor in Medicine and
Physiology at Wadham College, Oxford, UK.
Conflicts of interest: none declared.
ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11
Learning objectives
After reading this article, you should be able to:
C define terms relating to fluid composition, compare and
contrast passive and active forms of solute transport, under-
stand the factors that affect and limit the rate of molecular
transport
C understand how a GibbseDonnan equilibrium is reached across
a selectively permeable membrane
C compare the classical and modified model of Starling forces
across capillary endothelia and understand the implications of
the differences between them
litre solution will have an osmolarity of 5 mOsm/litre.
Conversely, ionic compounds such as sodium chloride will
dissociate in aqueous solution though technically an ionic solid,
NaCl can be conceptually considered as molecules each con-
taining one Naþ and one Cl ion. Therefore a 100 mmol/litre
NaCl solution will produce an osmolarity of 200 mOsm/litre.
Conforming to this pattern, each conceptual molecule of calcium
chloride will dissociate to give three moieties in solution and so
100 mmol/litre CaCl2 will dissociate produce a 300 mOsm/litre
solution and so on. This of course becomes more complex when
multiple solutes are present or when solutes do not fully disso-
ciate, as is the case in blood plasma. The freezing point of water
containing solute is depressed proportionally to the number of
particles. Thus osmolarity can be measured by an osmometer
using this principle of freezing point depression.
Osmolality is defined as the number of osmoles per kilogramme
of solvent (Osm/kg). For biological solutions, which are pri-
marily water, this is not appreciably different from the
osmolarity.
Osmotic pressure is the pressure required to prevent the diffu-
sion of water from a weak solution to a strong one via a semi-
permeable membrane. A fluid of high osmolarity exerts
substantial pressure while the osmotic pressure of pure water is
zero. The osmotic pressure of a solution can be calculated using
Equation 1.
p  R$T$C ð1Þ
Equation 1: calculating osmotic pressure, where p is osmotic
pressure, R is the ideal gas constant (0.0826 litre atm/K/mol), T
is the absolute temperature (K), and C is the concentration of the
solution (mOsm/litre). For example, for a 1 osmolar solution at
0 C the osmotic pressure exerted is 22.4 atmospheres. From this
logic, we can define an ‘osmole’ is that quantity of solute which
when dissolved in 22.4 litres of water, generates an osmotic
pressure of 1 atmosphere.
Oncotic pressure: the oncotic, or colloid osmotic, pressure arises
when apertures between cells allow the passage of water and
small solutes but not macromolecular particles, for example
proteins. Large molecules such as gelatin, starch and proteins
584 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY

tend to form viscid, glue-like solutions. These solutions are

described as colloids. Different configurations of amphipathic molecules

Properties of water
About 70% of the mammalian cell comprises water. Although a
water molecules do not have true positive or negative charge
they do have a polarity. The two hydrogen atoms are bonded to
the oxygen by covalent bonds, that is a sharing of an electron
pair. However, the greater gross positive charge, or electroneg-
ativity, of the oxygen atomic nucleus favours these electrons
existing closer to it. As a result there is a higher electron density
around it resulting in a partial charge differential between it and b
the hydrogen atoms (Figure 1).
This uneven charge distribution gives rise to the characteristic
properties of water. It allows the formation of hydrogen bonds
between neighbouring molecules. This restriction on movement
of water molecules gives water a high boiling point relative to its
molecular size. As a solvent, water dissolves ions and polar
molecules rather than non-polar moieties due to its own polarity.
c
Properties of organic molecules
The polarity of water molecules contrasts with the non-polar
nature of organic molecules, which are made up mainly or
entirely of carbon and hydrogen atoms, for example the alkanes
and alkenes. In these molecules, the carbonehydrogen bonds Lipid
lead to an even distribution of electrons. As a result partial
charges do not arise and so the molecules do not interact with
each other by formation of hydrogen bonds. For this reason,
organic molecules larger than water have much lower boiling
points. However, while in their liquid state these act as non-polar
solvents, which are able to dissolve other non-polar molecules
but not ions or polar molecules. These substances are therefore Figure 2 Configurations of (a) bilayer, (b) micelle and (c) chylomicron
molecules.
known as hydrophobic or lipophilic.

Amphipathic molecules: it contain areas of polarity as well as
non-polar sections. This allows the polar portion to exist in an
aqueous phase while the non-polar portion is in an oil phase.
This is the principle that enables detergents to remove non-polar
oils and greases from clothing. Figure 2 shows how amphipathic
molecules, arrange themselves within the aqueous solutions of
the body.
Bi-layered sheets e (Figure 2a) in this configuration the
water-soluble heads of the molecules are in the aqueous envi-
ronment while their hydrophobic tails coalesce internally. This
structure is the basis of phospholipid cell membranes; the
aqueous intracellular and extracellular fluids are in contact with
only the polar regions at the phosphate head. Meanwhile, the
non-polar fatty acid tails interact shielded from the surrounding
water. An interesting consequence of this structure is that even
small ions are very poorly soluble in the non-polar inner portion
Structure and polarity of a water molecule
δ–
O
δ+H Hδ+
Figure 1
of the membrane and so are not able to diffuse directly through
it. Meanwhile, uncharged moieties such as urea can cross the
membrane freely.
Micelles e (Figure 2b) these are spherical droplets where the
molecules have their polar heads facing outwards and their hy-
drophobic tails pointing towards the centre of the structure.
These configurations can be observed to form spontaneously if
the membranes of cells suspended in an aqueous solution are
manually disrupted.
Chylomicra e (Figure 2c) hydrophobic lipids must be trans-
ported in the aqueous plasma in order to be utilized in the
appropriate areas of the body. This is achieved by amphipathic
bile acids enabling the formation of chylomicra. The non-polar
portions of amphipathic molecules are in the lipid phase of the
chylomicron and the polar portions are on the outside, forming a
protective envelope. In this way, the fat forms an emulsion that is
stable in an aqueous environment.
The positions of the polar and non-polar regions of any large
molecule determine its shape when placed in an aqueous, a lipid,
or a lipid and aqueous environment. Amino acids possessing
ionized side-chains are hydrophilic while those possessing aro-
matic side-chains are hydrophobic. The distribution of amino
acid types determines the three-dimensional structure of pro-
teins. The structure of proteins is fundamental to their functions
as enzymes or carrier molecules and the functional structure will

ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11 585 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY
be formed only when the molecule is in the appropriate envi-
ronment. A non-polar region is incapable of forming hydrogen
bonds with water, causing disruption of the hydrogen bond
network. This causes the bonds to be rearranged tangential to the
surface in order to minimize this disruption creating a water cage
around the non-polar surface. This state is energetically unfav-
ourable, prompting structural rearrangement to minimize such
contact. If a trans-membrane pore is considered, the non-polar
portions of the protein efface the lipid components of the cell
membrane whilst the polar portions help form the water-filled
pore or trans-membrane carrier for water-soluble molecules.
Properties of ions
Ion is a term that refers to any moiety possessing a net charge.
These can be small inorganic monoatomic ions such as sodium
(Naþ), small molecules such as bicarbonate (HCO3 ) or even large
macromolecules such as proteins. They can be divided into
positive ions, cations, which have lost one or more electrons
from their outer shell, or negative anions, which have gained one
or more electrons. At a physiological pH, proteins generally have
a net negative charge thus are considered anionic. Depending on
the net charge of the ion, it can be described as monovalent (Kþ,
Naþ, Cl), divalent (Ca2þ, S2), etc.
Many molecules are charged because they contain groups that
ionize in water. Common examples of this are the hydroxyl
group (OH), which tends to lose a hydrogen ion and become
negatively charged (O); the amino group (NH2), which tends to
gain a hydrogen ion and become positively charged (NH3 ); and
organic acids, in which the carboxyl (COOH) group tends to lose
a hydrogen ion and become negatively charged (COO). This
dissociation allows molecules such as glucose, which contains a
number of hydroxyl groups, and amino acids to be water-soluble.
At physiological pH, organic acids and bases exist in both
ionized and nonionized forms. As discussed above, the un-
charged moiety can permeate the phospholipid bilayer much
more freely than the ionized form. If the new microenvironment
favours ionization the molecule may become trapped on that side
of the membrane this is known as ‘ion-trapping’. For example,
local anaesthetics, such as lidocaine are weak bases that exist in
an equilibrium between the protonated and unprotonated form at
physiological proton concentrations. The protonated form does
not easily diffuse across cell membranes. Interstitial acidosis
secondary to inflammation can induce protonation, this prevents
diffusion into the cell thus inhibiting local anaesthetic function.
As the pH tends to be lower inside the cell, the ionized form is
favoured causing the molecule to become confined. Furthermore,
this mechanism is a reason for caution in intrapartum epidural
anaesthesia as ion-trapping in an acidotic foetus could theoreti-
cally lead to accumulation. Despite this, toxicity is generally
considered unlikely without exceptionally high concentrations of
the drug.
The function of ions in the body is vastly diverse. Ions provide
much of the osmotic pressure of body fluids, sodium and po-
tassium being the principle extracellular and intracellular cat-
ions, respectively. The movement of these ions is central to the
generation of action potentials required for neural signalling and
muscular contraction. Calcium ions are integral in any number of
intracellular signalling mechanisms. Some ions have specific
roles to play in the structure of the body such as calcium in bones
ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11
and teeth. Iron forms components of the mitochondrial electron
transport chain and other oxygen-carrying pigments such as
haemoglobin and myoglobin and thus is vital for aerobic respi-
ration. In some cases, only minute amounts of ions are present in
the body, including copper, zinc and chromium, these are known
as trace elements. These ions have important functions as co-
factors for enzymes and as carriers for the products of enzyme-
mediated reactions.
Movement of molecules
Passive transport of a substance occurs down a concentration
gradient by simple diffusion, facilitated diffusion, or osmosis in
the case of water, and does not involve the expenditure of
metabolic energy. Movement against a concentration gradient
requires energy to be used, this is known as active transport.
Depending on the relative environments either side of a mem-
brane a solute may move through in one direction by a passive
process and in the other direction by an active one.
Diffusion
Diffusion, or simple diffusion, is the process whereby substances
move down their concentration gradients. Even when solutes
exist as a mixture, they diffuse down their own concentration
gradients irrespective of other molecules present. This process
continues until an equilibrium is reached, and the solute is
equally distributed. Gas molecules also diffuse in proportion to
their own concentration gradients (partial pressures), as if they
were the only type of molecule present, to become evenly
distributed throughout the gas mixture, each contributing its own
partial pressure to the whole.
The rate at which a solute diffuses is described by Fick’s law
of diffusion (Equation 2). In which k is the Boltzmann constant,
T is the absolute temperature, r is the radius of the molecule
(with the somewhat dubious assumption of a spherical mole-
cule), h is the viscosity of the medium, A is the surface area
across which it may diffuse and dC/dx represents the concen-
tration gradient.
k$T dc
J¼ $A$ ð2Þ
6p$r$h dx
Equation 2, Fick’s law: All molecules are in motion at a speed
proportional to the thermal energy of the substance. In fluids
this movement is random, due to collisions with other mole-
cules, this is known as Brownian motion. Thus, a dynamic
equilibrium exists, the solute continues to move but the net
distribution remains constant such that the concentration is
equal in all parts of the containing medium. Increasing the
temperature would increase the thermal energy of the molecules
increasing their speed of movement; however, as physiological
temperature is tightly regulated it is largely redundant in a
biological context.
Smaller molecules are able to diffuse more rapidly than larger
molecules which will diffuse more slowly. Interestingly, Gra-
ham’s law states that rate of diffusion is inversely proportional to
the square root of molecular weight (Equation 3) while Fick’s law
refers only to molecular size. It is presumed the two are generally
proportional.
586 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY
1 proteins that span the bilayer, in a process known as facilitated
J a pffiffiffiffiffiffiffiffi ð3Þ
mw
Equation 3, Graham’s law: The first part of the Fick equation is
often referred to collectively as the diffusion coefficient, D and
the concentration gradient can be thought of as the difference
between the concentrations either side of the membrane. Thus
the Fick equation can be simplified as shown in Equation 3.
J ¼ D$A$DC ð4Þ
Equation 4, Fick’s law (simplified): Accordingly, for a given area
of membrane under physiological conditions the concentration
gradient is the primarily determinant of flux. Nonetheless, the
rate of solute movement through the bilayer may be slowed
compared with their movement across the aqueous phase on
either side. The two principal factors that affect movement
through the membrane are solubility in the bilayer and the size of
the molecule. Thus, small lipid-soluble molecules diffuse the
most easily and large polar ones with the greatest difficulty.
Ions are charged and therefore, in addition to following their
concentration gradients, their direction of movement is also
influenced by electrical gradients. The tendency of Naþ to move
to the inside of a cell is favoured both by the concentration
gradient but also by the fact the intracellular compartment has a
relative negative charge thus attracting the positive cation.
Conversely, Kþ moves out of the cell, down its concentration
gradient, but in opposition to its electrical gradient. Thus the
overall electrochemical gradient is the combination of these
chemical and electrical forces.
Osmosis
When two aqueous solutions of different concentrations are
separated by a semipermeable membrane (i.e. one permeable to
water but not the solute), water moves from the low osmolarity
solution to the high osmolarity solution by the process of
osmosis. In doing so, it is moving down its own concentration
gradient. Most cell membranes are permeable to water; despite
its polarity it manages to cross freely possibly via integral pore-
proteins known as aquaporins.
The amount of water that enters cells is regulated by changes in
their osmolarity (i.e. the amount of solute within the cell). The
principal intracellular solutes are Naþ, Kþ and Cl, but they also
include proteins, organic phosphates and other metabolic in-
termediates. If a mammalian cell is placed in a solution that is hy-
potonic, water enters the cell until the osmotic pressure inside the
cell is the same as that outside causing it to swell. Erythrocytes are
capable of the largest increase in volume, owing to their biconcave
shape, being able to accommodate up to 167% of their resting
volume. However, most other cells can withstand only modest
changes in volume. In extreme cases, holes develop in the cell
membrane and the cell loses its intracellular contents, or bursts. At
an osmolarity of 150 mOsm, about 50% of red blood cells lyse. By
contrast, when placed in hypertonic solutions, water passes from
the cell to the surrounding medium causing the cell to shrink.
Facilitated diffusion
Charged or large molecules will diffuse very slowly through cell
membranes and so tend to do so via specific carrier or channel
ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11
diffusion (Figure 3). This mechanism allows water-soluble mol-
ecules to cross the membrane at a much higher rate than would
be possible by simple diffusion alone.
Channel proteins: as discussed above, even small ions cannot
diffuse across the cell membrane owing to its hydrophobic
component. Channel proteins allow an aqueous channel through
which ions can diffuse from one side of the membrane to the
other and protect them from the surrounding lipid (Figure 3a).
These channels are important in the transport of simple inorganic
ions such as Naþ, Kþ, and Caþ. Most of these ion channels will
be specific to the passage of a single ion. The ions probably
interact with charged group in the channel hastening the transfer
across the membrane.
Many ion channels are gated in order to control the passage of
ions through the membrane. Therefore, despite being a passive
process, the flux through channel proteins can be tightly regu-
lated. For example, voltage-gated sodium channels only take on
an open state when the intracellular voltage increases by from
baseline, this regulation is responsible for the generating action
potentials only in response to appropriate stimulation. Alterna-
tively, many ion channels are ligand-gated responding to neu-
rotransmitters such as acetylcholine.
Carrier proteins: solutes bind to these trans-membrane proteins
inducing a conformational change; the rearrangement of the
protein’s tertiary structure leaves the solute facing the aqueous
environment on the opposite side of membrane (Figure 3b).
Assuming the concentration is low on this side of the membrane,
the solute dissociates from the binding site. This allows the
protein to revert to its previous conformation and the process to
be repeated.
This remains a passive process, following the electrochemical
gradient, but both forms of facilitated diffusion differ from simple
diffusion in a number of ways:
Rate e facilitated diffusion allows for faster transfer of solutes
across membranes than simple diffusion.
Saturation e if the solute concentration is high, and all the
binding sites on the proteins become occupied then these trans-
port proteins can become saturated. Thus the kinetics for trans-
port can be described similarly to enzyme kinetics using the
MichaeliseMenten equation (Equation 5).
VMAX $½S
V¼ ð5Þ
KM þ ½S
Equation 5, The MichaeliseMenten equation: where V repre-
sents the rate of transport across the membrane; VMAx is the
maximum possible rate of transport if all binding sites are filled;
KM is the Michaelis constant and [S] is the solute concentration.
If this equation is plotted on a graph, as shown in Figure 4, then
we can see that rate of transport increases with increased
concentration.
However, the relationship is hyperbolic rather than linear as
would be expected for simple diffusion as described by Fick’s
law. The rate is seen to plateau towards VMAx despite further
increases in substrate concentration thus the system is said to be
saturated.
587 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY

Facilitated diffusion and active transport across the cell membrane

a Channel protein b Carrier protein

c Active transport

Uniport Symport Antiport

ATP ATP
ADP +iP
ADP +iP

Figure 3 Facilitated diffusion (a, b) and active transport (c) across the cell membrane. ADP, adenosine diphosphate; ATP, adenosine triphosphate.

Specificity e carrier proteins are specific to a particular
molecule that can bind to them, likewise most ion channels will
only allow specific ions to pass through it.
Inhibition e both ion channels and carrier proteins can be
inhibited pharmacologically in order artificially to regulate their
function. For example, lidocaine as described above is able to
selectively block sodium channels.
Active transport
The movement of a solute across a membrane against a con-
centration gradient requires metabolic energy, usually in the
form of adenosine triphosphate (ATP). When movement of the
solute is coupled directly to an energy-yielding reaction, the
process is known as primary active transport. When indirectly
coupled, for example by use of energy stored in an electro-
chemical gradient established by primary active transport, this is
known as secondary active transport. To denote their active role
in the generation and maintenance of electrochemical gradient
active transporters are frequently referred to as ‘pumps’.
transport is converted from energy stored in readily oxidized,
reduced or hydrolysed chemical bonds. Typically this is the hy-
drolysis of ATP to form adenosine diphosphate (ADP) and
inorganic phosphate, PO4 , thus releasing energy previously
produced by oxidative phosphorylation. An example of this is the
sarco/endoplasmic reticulum Ca2þ ATPase (SERCA), which
pumps calcium from the cytosol into the sarcoplasmic reticulum,
at the expense of ATP, to facilitate muscle relaxation following
contraction.
Michaelis–Menten kinetics: effects of solute
concentration on rate of transport across the membrane
V

Uniports: a single type of ion can be transported across a

membrane against or down its concentration gradient; the term is
also applied to proteins involved in facilitated diffusion as dis-
cussed above. However, as depicted in Figure 3c the transport of
a single ion without linked movement of another can be driven
by active transport. As these mechanisms only involve one
[S]
moiety being transported across the membrane they can only be
a form of primary active transport. The energy required for Figure 4

ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11 588 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY

Antiporters: these co-transport two moieties across the mem-

brane in opposite directions (Figure 3c) and can be described as Generation of a Gibbs–Donnan equilibrium across a
primary or secondary active transport. An important example of selectively permeable membrane
primary active transport is the ubiquitous Naþ/Kþ-ATPase
pump. Hydrolysis of ATP drives three Naþ ions out of the cell Side A Side B

Selectively permeable membrane

and two Kþ ions in, in both cases this is against the concentration
gradient. This mechanism is vital to maintain the composition of
K+
intracellular and extracellular fluids as is discussed in more detail K+ CI –
A–
below. An example of a secondary active transport antiporter is
the sodium/calcium exchanger (NCX) utilizing the existing so- CI –
dium gradient to drive calcium transport. A–
K+ K+
Symporters: active transport of solutes in the same direction
occurs through symporters, typically this represents a form of
secondary active transport. (Figure 3c) The Naþ/Kþ-ATPase Side A Side B

Selectively permeable membrane

established sodium gradient driving the uptake of glucose by the
sodiumeglucose transport proteins (SGTP) in the intestinal mu-
A– K+ K+
cosa is just one example. CI –

(1)
Partitioning of fluids CI – CI –
A– K+
GibbseDonnan equilibria K+
If two compartments are separated by a selectively permeable
membrane, permeant to all ions but one the GibbseDonnan ef-
fect can be observed. It has been established above that proteins
tend to be large anionic molecules unable pass through cell Side A Side B

Selectively permeable membrane

membranes. Figure 5 demonstrates such a situation in which the
anionic protein is represented as A.
 Initially Side A contains Kþ and anionic protein, A, while
Side B contains Kþ and Cl.
 As Cl is not present on Side A, it diffuses down its con-
centration gradient from B to A.
 Side A is now net negative, causing Kþ to move from B
down its electrical gradient.
The result of this is an unequal distribution of the total ions
between the two sides with more ions present on Side A. As a
consequence Side A will be net negative relative to Side B thus
generating a potential difference across the membrane.
If we think of Side A as the protein containing intracellular
environment, this process acts to establish the resting membrane
potential. From the Nernst equation, the following equation
describing the equilibrium can be derived.
 þ    þ  
KA $ ClA ¼ KB $ ClB
Equation 6, example GibbseDonnan equilibrium: The equilib-
rium reached above assumes a fixed volume for Side A and Side
B. However; the model can then be extended further.
 The osmolarity is now higher on Side A, and so water will
move by osmosis from B to A.
 This, by dilution of Kþ will disrupt the GibbseDonnan
equilibrium causing more Kþ to diffuse from B to A until a
new GibbseDonnan equilibrium is established, which in
turn will cause more water to enter Side A.
Steps 4 and 5 would then continue to occur cyclically.
Returning to our cellular model this would lead to cell swelling
and eventually lysis. The Naþ/Kþ-ATPase, combined with
intrinsically low sodium permeability of the membrane, prevents
this from happening. In maintaining the gradient Naþ essentially
(2)
A– K+
K+
K+
CI – CI –
K+
A– K+
CI –
Figure 5
becomes an extracellular cation that cannot permeate the mem-
brane thus generating its own opposing GibbseDonnan effect
and stabilizing cell volume. This can be tested by inhibition of
the pump with ouabain resulting in cellular swelling being
observed.
ð6Þ
Starling forces
The classic model
The factors regulating the flux of fluid across the capillary
endothelium are often considered in terms of the Starling forces.
In the latter part of the 19th century Ernest Starling conducted
experiments in which he infused fluid into the interstitial
compartment of dogs’ hind limbs. He observed that when
isotonic saline was injected, the fluid was absorbed into the
vasculature and haemodilution of the venous blood followed as a
result. However, when serum was used no measurable absorp-
tion was seen. This led Starling to conclude the capillary endo-
thelium to be a semipermeable membrane permissible to small
ions but impermeable to colloids. Although, Starling himself
noted this selective permeability to be imperfect as plasma

ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11 589 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY

proteins such as albumin can be detected in the interstitium and The modified model
so he proposed that the barrier must leak over time. The proposed caveat to the original model is that the glycocalyx
This prompted the hypothesis that net fluid movement across the may provide the primary barrier to fluid and solute exchange
capillary bed is determined by the imbalance between the oncotic across the endothelium. This structure is a negatively charged
pressures arising from differential plasma protein distribution and proteoglycan fibre matrix that acts as a surface coat extending
the hydrostatic pressure gradient generated by cardiac output. between the cells on the luminal surface of the endothelium
These forces opposing each other can be considered as Equation 7 (Figure 7).
from which the net fluid flux across the membrane can be predicted. As seen in Figure 7, the filtration of fluid is restricted to narrow
spaces between cells known as paracellular clefts. As fluid is
J pumped through the capillary, the filtration at these narrow clefts
¼ LP ½ðPc  Pi Þ  sðpc  pi Þ ð7Þ
A will be at high speed. This is akin to placing a thumb over the nozzle
of a hosepipe, constricting the cross-sectional area of the channel
Equation 7, the classic model of Starling forces: Here flux, J, is
through which fluid is flowing and thus increasing its velocity. This
considered per unit surface area, A. The Starling forces are given
high-velocity flow will briskly dissipate as fluid passes beyond the
as Pc and Pi representing the hydrostatic pressures and pc and pi
mouth of the cleft, reducing 100-fold with micrometres.
representing the oncotic pressures, in the capillary and inter-
The freshly filtered fluid will be protein-free, as albumin is not
stitium respectively. Lp is the hydraulic permeability and s is the
filtered through glycocalyx. Due to its molecular size and
Staverman osmotic reflection coefficient, a measure of the
possibly also due to both possessing negative charge the barrier
membrane’s leakiness to plasma proteins. In this model change
effect is enhanced. Notwithstanding, the tissue albumin con-
in flux across a fixed size capillary bed is driven by differences in
centration is approximately 50e60% of that in plasma, theoret-
either hydrostatic or oncotic pressures.
ically transported by vesicular shuttling known as the large pore
The hydrostatic pressure of blood entering the capillary bed
pathway. However, despite the protein-rich interstitial beneath,
from the arterial system will be higher than as it exits the cap-
the rapid flow of filtrate prevents back-diffusion of interstitial
illaries into the venous circulation. From this a model developed
proteins into the cleft. This will mean that the sub-glycocalyx
of initial filtration down the hydrostatic gradient followed by
space has very little protein content and so the oncotic pres-
reabsorption along the oncotic gradient as shown in Figure 6.
sure, pg , will be much lower than the values ascertained from the
Based on the Starling equation above experiment measure-
interstitial fluid below, pi .
ments have been made for each parameter allowing the flux of
If this is considered in the context of the traditional Starling
fluid within a given capillary bed to be predicted. From this the
equation (Equation 7), a clear source of miscalculation can be
rate of formation of interstitial fluid can be inferred. Assuming a
seen. If we incorporate the proposal that the glycocalyx, not the
constant volume over time, the rate of uptake into the lymphatic
epithelium, provides the barrier for fluid exchange, then the
system should be equal to the calculated filtration. However,
oncotic gradient is derived from the difference between pc and
when lymphatic flow is measured it has been found to be much
pg . Hence this gradient is underestimated in the classic model. A
lower than Equation 7 would predict. This inconsistency, noted
new form of the equation was proposed which at high rates of
in the 1980s, was an early indicator that this model may not
flux can be simplified to Equation 8.
completely describe fluid balance. Experiments conducted in
which pi was artificially varied by interstitial albumin infusion, J  
demonstrated that the increase in filtration was only a fraction of ¼ Lp ðPc  Pi Þ  s2 $pc ð8Þ
A
that predicted by the equation.
Equation 8, the modified model of Starling forces: This form of the
equation does not incorporate the oncotic pressure of the inter-
stitium, as the rapid flow through the cleft makes the effective pg
Fluid movement across capillaries as described in
the classic model

The glycocalyx model of capillary fluid filtration

Lumen
Glycocalyx

50
Pressure (mmHg)

Pc
Endothelium
Net absorption
25 πc
Net filtration
Interstitium
0
Paracellular cleft

Figure 6 Figure 7

ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11 590 Ó 2015 Elsevier Ltd. All rights reserved.
 PHYSIOLOGY

near 0. However, at lower filtration rates there may be upward Conclusions

diffusion of albumin from the interstitium into the sub-glycocalyx
 The physicochemical properties of water, ions and organic
space. This means that the oncotic gradient is dependent on the
molecules determine their structures, distributions and
hydrostatic gradient.
interactions within the body.
Due to the negligible volume of the sub-glycolyx space, pg can
 Cell function is dependent on the maintenance of cell
fluctuate rapidly with changes in hydrostatic pressure. This can be
membrane potentials and the transport of molecules across
seen within the length of a single capillary. The arterial extreme of
those membranes.
the capillary will have a much higher Pc than the venous extreme.
 The mechanisms controlling transport of molecules can be
As a result the paracellular clefts in this part of the vessel will have
passive or active energy-consuming processes.
high-velocity filtration giving rise to much lower values for pg .
 In the absence of active ion pumping, the protein con-
This generates a large oncotic gradient, favouring reabsorption in
taining intracellular compartment would expand due to the
this region. Considering Equation 7, this reduces the net filtration
GibbseDonnan effect.
occurring at the beginning of the capillary. Similarly, the reduced
 The classical model of Starling forces across the capillary
hydrostatic drive at the venous end will be produce slower flow
underestimates reabsorption thus a modified model has
through the clefts and a decreased oncotic gradient, allowing low
been proposed.
volume hydrostatic filtration to go unopposed. This would seem
 Regulation of capillary fluid flux is key to the evolution and
to suggest a model of steady filtration across the length of the
resolution of oedema as well as in the generation of
capillary rather than one of intense filtration followed by reab-
compensatory haemodilution to maintain volume homeo-
sorption as shown in Figure 6.
stasis following haemorrhage. A
This new model explains previous experimental evidence
suggesting that reabsorption of fluid from tissue cannot be sus-
tained over time. The nature of these clefts allows for a rapid FURTHER READING
washout and reduction in oncotic pressure generating net ab- Barrett KE, Barman SM, Boitano S. Ganong’s Review of medical physi-
sorption and a protection against oedema formation. For example, ology. 23rd edn. Lange Medical, 2009.
during exercise muscle will receive increased flow theoretically Hall JE. Guyton and Hall textbook of medical physiology. 12th edn.
increasing the hydrostatic pressure and net filtration. However, Saunders, 2010.
this effect will produce washout of the clefts and a high oncotic Pocock G, Richards CD. Human physiology: the basis of medicine, (Oxford
gradient, which will partially counteract oedema formation. Core Texts). 3rd edn. Oxford University Press, 2006.

ANAESTHESIA AND INTENSIVE CARE MEDICINE 16:11 591 Ó 2015 Elsevier Ltd. All rights reserved.