Pathophysiology of Edema

Transvascular Fluid Filtration

Fluid movement across the endothelium is driven by forces that regulate fluid flux across a semipermeable membrane. These forces are the hydrostatic and protein osmotic, or oncotic, pressures of the intravascular and interstitial spaces. The ostensibly quantitative expression of how these forces influence fluid filtration is shown in the equation.

where J _v represents net transvascular fluid flow, K is a coefficient that accounts for the permeability of the barrier and surface area for filtration, σ is the reflection coefficient of the barrier to protein, P _mv and P _i are the hydrostatic pressures of the microvascular compartment and interstitium, respectively, and π _mv and π _i are the osmotic pressures generated by the protein in the microvascular and interstitial spaces, respectively. For a barrier that is completely impermeable to protein, σ would assume the value of 1, and for a barrier across which protein moves without restriction, σ would assume the value of 0. The most complete analysis of fluid filtration across the endothelium would include values for σ that are specific for each plasma protein.

Countless attempts have been made to quantify each of the variables in Eq. 165.1 for different organs in animals and in man. All experimental approaches involve assumptions about the validity of the values measured, and thus all values so measured lack a sense of finality, particularly in the fetus and newborn. ^,

Under usual circumstances J _v is greater than 0 in most tissue spaces. That is, the sum of forces regulating fluid filtration result in a net movement of fluid out of the microcirculation. However, it is important to recognize that fluid filtration processes are dynamic and at any one time, in any specific section of the microcirculation, filtration forces may not result in fluid moving into the interstitium, even though the net J _v of the tissue or organ is greater than 0. For instance, the return of lung liquid into the circulation that occurs after birth takes place predominantly across the microcirculation and not by lymphatic channels.

Intravascular pressure in the microcirculation, P _mv , can be measured directly using micropuncture or indirectly techniques using isogravimetric methods. It is unlikely that intravascular pressure remains constant across the entire fluid-exchanging surface because resistance to flow along the vessel results in a drop in pressure over the length of the vessel. Thus P _mv is understood to be the net hydrostatic force for the surface area involved in fluid exchange, even if the hydrostatic pressure at the arterial end of the vessel results in fluid filtration, and hydrostatic pressure at the venous end results in fluid reabsorption.

P _mv is influenced by the relative vascular resistances in the circulation before and after the fluid-exchanging regions. An increase in upstream resistance or a decrease in downstream resistance reduces P _mv and fluid filtration, and a decrease in upstream resistance or an increase in downstream resistance has the opposite effect. The importance of considering the profile of vascular resistance distribution is that the effect of a change in arterial pressure on fluid filtration cannot be predicted with certainty. The redistribution of vascular resistance in response to an intervention or a change in condition ultimately determines whether transvascular fluid filtration is affected, and because resistance cannot be assessed clinically, the ability to predict whether a change in transvascular filtration will occur is difficult. For instance, alveolar hypoxia increases pulmonary arterial pressure in both adults and neonates, but alveolar hypoxia affects transvascular fluid filtration only in the newborn.

Interstitial pressure, P _i , has been assessed by several different techniques, including porous capsule embedment, direct micropuncture, and cotton wick insertion into the adventitial space. It is not possible to assign a general value to P _i because it assumes different values depending on the tissue bed under study. Values that exceed atmospheric pressure, and those that are subatmospheric have been measured. Although P _i often assumes a value near 0 mm Hg, changes in interstitial pressure in response to physiologic disturbances can be dramatic. For example, in tissue that has suffered thermal trauma, P _i may decrease many folds over baseline, thus contributing, in part, to the rapid accumulation of interstitial fluid seen with burn injuries. The molecular mechanisms underlying control of interstitial pressure are not clear, but the binding of collagen molecules on the surface of adventitial cells may play a role, because antibodies to the β ₁ -integrin lower interstitial pressure and prompt edema formation.

Microvascular protein osmotic pressure, π _mv , has been the focus of investigation and clinical study to understand better whether changing this value can improve total body fluid balance. ^, In the strictest analysis, π _mv should be measured for each plasma protein constituent, along with the corresponding σ , but like the aforementioned P _mv , a net value of osmotic pressure generated by the sum of plasma proteins suffices to account qualitatively for the relative importance of plasma oncotic pressure to transvascular fluid filtration. The osmotic force of all components of the extracellular fluid is on the order of 5000 to 6000 mm Hg. However, because these components pass unimpeded across the endothelial barrier of the microcirculation (i.e., they have a reflection coefficient of 0), they do not affect fluid flux. In contrast, plasma proteins do not move with complete freedom across the circulation and have a significant effect on fluid filtration despite an osmotic pressure in plasma being in the range of 10 to 20 mm Hg.

Interstitial protein osmotic pressure, π _i , has been measured by direct micropuncture, implanted tissue capsules, and absorbent materials placed within the tissue space. ^, In vivo measurements of π _i have relied on collection of lymph from organs of interest, with the assumption that the protein concentration of lymph is equivalent to that of interstitial fluid (this is likely to be a safe assumption under steady-state conditions when lymph is collected from afferent lymphatics).

The membrane parameter, σ , represents the sieving ability of a semipermeable membrane for protein. High values of σ imply that osmotic pressure differences across the vascular barrier will exert a greater effect on fluid flux than will lower values of σ . One can measure σ in vitro as the ratio between the measured and expected osmotic pressure generated by the protein of interest. Furthermore, the value of this coefficient can be estimated in vivo from experiments in which transvascular fluid filtration is maximized (i.e., under conditions in which protein flow is nearly all convective).

Eq. 165.1 describes the driving forces for transvascular fluid movement, but a different mathematical relationship exists for describing transvascular protein movement. This relationship contains two components, one describing flow of protein as a result of convective movement:

where J _s is transvascular protein flow, σ is the protein reflection coefficient, P is the concentration of protein in the vasculature, and J _v represents net transvascular fluid movement. The second relationship describes the flow of protein as a result of diffusion:

where K is the product of the permeability and surface area of the microcirculation, and P and L represent the concentration of protein in the vasculature and lymph (interstitium), respectively. Combining these two equations yields the equation that describes net total transvascular protein movement:

Under steady-state conditions, lymph contains all the transvascular protein filtered if no metabolism occurs in the interstitium, and thus J _s is reduced to LJ _v . In this analysis, J _v is equal to lymph flow and L is equal to lymph protein concentration. Rearranging the above equation, simplifying it, and arranging experimental conditions in which lymph flow is maximized yields the equation σ = 1 − L / P . When measured experimentally, σ is well over 0.5 in most tissue beds and usually in the range of 0.75 to 0.90, even in the neonate. Capillary beds that contain fenestrations, as in the liver, represent little restriction to transvascular protein movement, and under such conditions σ would be low. Capillaries in other vascular beds contain few if any discontinuous regions, and σ under these circumstances would be closer to 1. The closer σ is to unity, the greater the influence of plasma proteins on fluid filtration.

The lower the value of σ , the less able proteins are to generate a force counteracting fluid filtration. This arises for two reasons. First, with a less restrictive barrier, π _i increases numerically toward π _mv because the sieving quality of the membrane is diminished. Second, any difference between π _mv and π _i is minimized as σ decreases. Thus, under conditions in which the vascular barrier is injured, allowing a greater degree of protein leak, administration of protein intravenously to augment vascular protein osmotic pressure and to reduce edema formation theoretically should have little, if any, effect. The administration of protein to affect fluid balance in some other fashion (e.g., on barrier function per se or to provide other favorable effects unrelated to fluid balance [e.g., to maintain plasma drug binding]) may subserve a clinical benefit, but such a rationale would be independent of the presumed advantage of an increase in microvascular protein osmotic pressure on transvascular fluid filtration.

Finally, the coefficient K represents the product of barrier hydraulic conductivity and the surface area available for fluid filtration. These two components of K are difficult to separate experimentally. Hydraulic conductivity itself is predominantly a function of the density of the pathways for liquid and solute movement, and is not necessarily a measure of protein permeability. That is, more pathways, or pores, for solute and liquid exchange might exist under different conditions without the individual pathways being more permeable. Thus changes in K may occur with true alterations in liquid and solute permeability (i.e., pathways that allow passage of larger proteins), but changes in K may occur simply with an alteration in the density of pathways for liquid and solute movement or with changes in surface area for filtration, as would occur when unperfused capillaries are filled.

Consideration of Eq. 165.1 allows several general statements about fluid filtration under normal conditions during steady-state. First, because some degree of sieving always occurs across the microvascular barrier, the protein concentration in the interstitium is less than that in the vascular space. This difference in protein concentration yields a difference in protein osmotic pressures that is subtracted from the hydrostatic pressure difference term. In this sense, protein osmotic pressure attempts to balance the “edema-promoting” effect of microvascular hydrostatic pressure. Second, because a net exit of fluid from the circulation occurs, the hydrostatic pressure difference term in Eq. 165.1 must exceed the difference in osmotic pressure term when total body fluid balance is under consideration. No experimental evidence exists that indicates the presence of net “active” transport of water and solute across the endothelium that would influence transvascular fluid flux.

Although the concepts expressed in Eq. 165.1 serve physiologists and clinicians well when considering fluid balance in a general sense, the equation does not always allow a precise calculation of the change in fluid filtration ( J _v ), even when only one variable in the equation changes. This arises because a change in one of the variables in Eq. 165.1 usually results in a change in one of the other variables, even if such a change is unexpected. For instance, under conditions in which microvascular hydrostatic pressure is increased, one might assume that transvascular flow will increase by an amount that can be arrived at arithmetically from Eq. 165.1 . When P _mv increases in the presence of a stable vascular barrier, J _v increases, but as it does, the interstitial protein concentration decreases. This occurs because the driving force for liquid exceeds the bulk flow of protein across the barrier (because σ is greater than 0 for protein); there is no sieving of water. When interstitial protein concentration is reduced, the protein osmotic pressure difference between the vascular and interstitial spaces increases. Thus, from Eq. 165.1 , J _v will increase in response to an increase in P _mv , but as ( π _mv − π _i ) becomes larger, J _v assumes a new steady-state value that is less than that predicted by the increase in P _mv alone. Additionally, if excess fluid expands the interstitium, interstitial pressure increases to some extent, although the magnitude of this change is not predictable because tissue space compliance is not linear in the presence of increasing interstitial edema. This increase in tissue hydrostatic pressure also acts to slow transvascular fluid movement. Changes that occur in driving forces for filtration that counteract the change in the “edema-promoting” variable are referred to as the edema safety factors. The implication of the edema safety factors is that increases in transvascular filtration are blunted because of the countervailing changes seen in other variables in Eq. 165.1 . ^{, , ,}