Effects of Graded Increases in Pulmonary Vascular Pressures on Lung Fluid Balance in Unanesthetized Sheep

Transcription

1 1164 Effects of Graded Increases in Pulmonary Vascular Pressures on Lung Fluid Balance in Unanesthetized Sheep RICHARD E. PARKER, ROBERT J. ROSELLI, THOMAS R. HARRIS, AND KENNETH L. BRIGHAM SUMMARY Fourteen experiments were conducted on 12 chronically instrumented unanesthetized sheep in which we monitored pulmonary arterial, left a trial, and aortic pressures, lung lymph flow, and lymph-to-plasma ratios (L/P) of total proteins and four endogenous protein fractions during baseline and progressive elevations of left atrial pressure. We found that L/P for total proteins decreased as lung lymph flow increased until lymph flow exceeded four to five times baseline and thereafter remained nearly constant (filtration independent) at The four protein fractions exhibited a filtration independent L/P that was related to the effective molecular radius of the protein fraction. The minimal L/P were 0.36, 0.28, 0.18, and 0.09 for the protein fractions that had effective molecular radii of 36, 38.6, 59, and 101 A, respectively. In addition, we found no evidence supporting a stretched pore phenomenon over the pressure range utilized in this investigation. Ore Res 49: , 1981 MANY investigators have shown that an increase in left atrial pressure causes lung lymph flow to increase and lung lymph protein concentration to decrease (Warren and Drinker, 1942; Erdmann et al, 1975; Vreim et al., 1976; Woolverton et al., 1978; McNamee and Staub, 1979). Relationships among pressures, lymph flow, and protein concentrations are expressed by the equation: = QL = K fc [(Pmv - Ppmv) where Q v is the fluid volume crossing the walls of the lung microvessels per unit time (equal to lung lymph flow, QL, in the steady state); K fc (filtration coefficient) is an index of the hydraulic conductance of lung microvessel walls; Pmv and Ppmv are the hydrostatic pressures within the lung microvessels and interstitial fluid surrounding the microvessels, respectively; o<j is the average osmotic reflection coefficient for plasma proteins; and n mv and H pio y are the respective total oncotic pressures of the plasma and interstitial fluid (lymph). It is apparent from Equation 1 that decreases in From the Pulmonary Circulation Center, Vanderbllt University School of Medicine, Nashville, Tennessee TTua work was supported by National Heart, Lung, and Blood Institute Grant HL (SCOR in Pulmonary Vascular Diseases), a Parker B. Francis Foundation Research Grant, an Upjohn Company Grant, the H. J. Morgan Fund for Cardiology, M.W. Straus-H.H. Straus Foundation, Inc, and the John W. Cooke, Jr., ajid Laura W. Cooke Fund for Lung Research. Dr. Parker is supported by National Institutes of Health Training Grant 5 T Dr. Roselli is supported by National Institutes of Health NHLBI Young Investigator Award HL Dr. Bi-igham was an Established Investigator for the American Heart Association. Address for reprints: Richard E Parker, Pulmonary Circulation Center, Room B-1308, Vandertrilt University Medical Canter, Nashville, T»ns Received May 22, 1980; accepted for publication July 13, (1) interstitial fluid protein concentration (and thus ripmv) oppose the filtration caused by increased microvascular hydrostatic pressure. However, the degree to which filtration is reduced depends on the value of ff d - Our purpose in this investigation was to determine quantitatively the influences of increased lung lymph flow and decreased interstitial protein concentration on lung fluid balance in response to a broad range of increased pulmonary vascular pressures. To determine the importance of the oncotic pressure gradients, we estimated a d from lymph-toplasma protein concentration ratios (L/P) at very high pressures where L/P is independent of lymph flow (Granger et al., 1979; Granger and Taylor, 1980). We found that o a for total proteins averaged 0.74 for the lung microvessels. We also found that decreased interstitial fluid oncotic pressure was an important protective mechanism against pulmonary edema for moderate microvascular pressure elevations but that this mechanism was exhausted at high pressures. Conversely, increased lymph flow per se protected the lung against edema formation at high microvascular pressures after the other safety factors against pulmonary edema seemed to be exhausted. Even at left atrial pressures as high as 35 to 53 cm H 2 O, we saw no evidence of increased vascular permeability. Methods Experimental Methods We chronically instrumented 12 sheep as described in detail in previous publications (Brigham et al., 1974). In brief, we put catheters in the main pulmonary artery, left atrium, right carotid artery, right jugular vein, and the efferent duct of the caudal mediastinal lymph node. In addition, we put

2 PRESSURE EFFECTS ON FLUID BALANCE/PorAer et al a 16 French Foley balloon catheter into the left atrium to mechanically elevate pulmonary microvascular pressure. To minimise contamination of the lung lymph from systemic sources, we ligated the tail portion of the caudal mediastinal lymph node. The sheep recovered from surgery for 4-6 days before experiments were conducted, at which time there was a stable flow of lymph, free of any red blood cells. We continuously monitored pulmonary arterial (Ppa), left atrial (Pla), and aortic pressures (all pressures were referenced to the chest midpoint), recorded lung lymph flow rate by measuring the volume collected each 15 minutes in a graduated tube, and took blood samples each half hour. After monitoring all of the above variables for 1-3 hours during stable baseline conditions, we elevated left atrial pressure to an arbitrary value by inflating the Foley balloon catheter and allowed the lymph flow and vascular pressures to become stable for at least 1 hour. In four experiments we elevated left atrial pressure once, and in ten other experiments we elevated left atrial pressure to progressively higher levels, waiting for a steady state at each level. After each experiment, the total protein concentrations in plasma and lymph were measured by the biuret technique (Failing et al., 1960) on an automated device (AutoAnalyzer, Technicon Instruments). Steady state lymph and plasma samples were electrophoresed on polyacrylamide gradient gels (4-30%) as described in a previous publication (Brigham and Owen, 1975). We were able to estimate the effective molecular radii of the different protein fractions by comparing their relative migration distances to a standard curve composed of six proteins of known molecular weights and Einstein- Stokes radii (bovine serum albumin, mol wt = 67,000; beef heart lactate dehydrogenase, mol wt = 140,000; beef liver catalase, mol wt = 232,000; horse spleen ferritin, mol wt = 440,000; hog thyroid thyroglobulin, mol wt = 669,000; and human serum /?- lipoprotein, mol wt = 3,200,000). We estimated lung microvascular pressure (Pmv) from the empirical equation suggested by several investigators (Kuramoto and Rodbard, 1962; Gaar et al., 1967): Pmv - Pla (Ppa - Pla) (2) where Ppa and Pla are pulmonary arterial and left atrial pressures, respectively. Oncotic pressures of the plasma and lymph were calculated from the four measured protein fractions and their calculated molecular weights and corrected to sheep body temperature of 38 C (Harris and Roselli, 1981): Cy 1 X * Use of Equation 3 implies that protein-protein interaction* are similar for different die proteins. See Appendix A of Harris and Roselli, J Appl Phyriol (1981) 50: where IIT is the total protein oncotic pressure (cm H 2 0), IT, is the oncotic of protein fraction i (cm H 2 0); R is the universal gas constant, T is the absolute temperature, Cj is the protein concentration of protein fraction i (g/dl), MW, is the molecular weight of protein fraction i, and Cr is the total protein concentration (g/dl). Statistical analyses were performed by the conventional methods (arithmetic mean ± standard error, Student's -test, and linear regression), and we assumed significance for P values of less than 0.05 (Snedecor and Cochran, 1967). Theoretical Methods Recently, Granger and Taylor (1980) have proposed a method to estimate a minimal value of the average osmotic reflection coefficient (aa) from the lymph-to-plasma total protein concentration ratio (L/P) at high lymph flows where L/P is independent of the filtration rate. The calculation is based on a modification of the Patlak equation (Patlak et al., 1963) derived for a single sized protein moving through a single cylindrical pore. Thus, if the microvascular barrier were composed of pores of uniform size: L/P = (1 - ff f )/(l - are"*) (4) where /? is a modified Peclet number $ = (1 - ar)qi7ps. (5) Here at is the solvent drag reflection coefficient, QL is lymph flow rate, and PS is the permeabilitysurface area product. As lymph flow becomes large (»oo + ) the limiting lymph to plasma protein concentration ratio, (L/P)miD, becomes (L/P) =2 or (6) Brace et al. (1977) have shown that, for a single solute in a heteroporous system, at approaches Od when the L/P is filtration independent. Therefore, the osmotic reflection coefficient can be estimated by the equation: ad = 1 - (L/P). (7) Results We conducted 14 experiments in 12 sheep. Figure 1 is a representative experiment which demonstrates that for each elevation of left atrial pressure there are concomitant increases in pulmonary arterial pressure and lung lymph flow while the lymph-to-plasma ratio for total proteins decreases. The measured variables and calculated microvascular and osmotic pressures obtained during steady state conditions for each experiment are listed in Table 1. To estimate the osmotic reflection coefficient for total proteins, we plotted the steady state lymphto-plasma ratio of total proteins as a function of lung lymph flow shown for one experiment in Figure 2. Although the fraction of total lung lymph drained through the caudal mediastinal lymph node is

3 1166 CIRCULATION RESEARCH VOL. 49, No. 5, NOVEMBER i MEAN PRESSURES 40- (CmH,O) FLOW (ml/hr) PLASMA J 20 " 0 J 0.6- TOTAL " 0.0 J TlME (HOURS) FIGURE 1 Representative experiment depicting the effects of increased left atrial pressure (Pud on pulmonary arterial pressure (PpJ, lung lymph flow, and the lymphto-plasma concentration ratio for total proteins. The solid bars indicate the period of time assumed to be steady state. thought to be high (Staub, 1975), it may vary among sheep. Therefore, lymph flow for all experiments was normalized to baseline for each experiment as shown in Figure 3. The solid curve in Figure 3 is a line of best fit through the data points and is similar to the multiple data points in one experiment shown in Figure 2. Figures 2 and 3 show that L/P for total proteins initially decreased rapidly as lung lymph flow increased but slowed considerably at higher relative lymph flows, becoming constant at approximately four times control lymph flow. The dashed line in Figure 3 is the extrapolation of that portion of the curve which was considered to be filtration independent. The extrapolated line intercepts the ordinate at 0.26, giving a minimal value for a A, as determined from Equation 7, of 0.74 for total proteins. If the 0d value obtained from Figure 3 were used to determine the effective osmotic pressure gradient (AIIE), the equation for this pressure gradient would be: MIE = 0.74 (FLnv - lip ). (8) Figure 4 is a plot of the net effective osmotic pressure gradient and normalized lung lymph flow as a function of changes in microvascular pressure with the lines drawn being the lines of best fit. The net effective osmotic pressure gradient was calculated by subtracting the effective osmotic gradient obtained during steady state baseline conditions from the steady state effective osmotic pressure gradient obtained during experimental conditions. When pulmonary microvascular pressure was elevated by approximately 10 cm H2O, the average lung lymph flow increased by only 70%. However, when microvascular pressure was elevated by cm H2O, lung lymph flow increased dramatically (3.2 to 4.3-fold). The average net effective osmotic pressure gradient increased for moderate elevations of pulmonary microvascular pressure (up to 30 cm H 2 O), but did not continue to increase significantly for Pmv values above 30 cm H2O. Figure 5 is an electrophoretogram of lung lymph and plasma from a sheep. In all of our baseline lymph and plasma samples which were electrophoresed, we could identify nine fractions. At high lymph flow rates, however, the concentrations of some of the fractions (especially fractions 5, 6, and 7 as shown in Figure 5) were so low that they could not be delineated accurately. Because of this, and to simplify our analysis, we grouped the fractions into four major groups (I, II, HI, IV) as shown at the bottom of Figure 5. Table 2 lists the cumulative data for the four protein fractions and the calculated oncotic pressures of both lymph and plasma. The average migration distance relative to albumin for each of the four protein groups was calculated from the concentration average of each group. The four protein fraction groups had effective molecular radii which averaged 36, 38.5, 59, and 101 A and molecular weights of 67,000, 106,000, 376,000, and 1,300,000 for groups I, II, III, and IV, respectively. Figure 6 shows relationships between the lymphto-plasma concentration ratio of a small protein fraction (36 A) and a larger protein fraction group (101 A) and lung lymph flow. The L/P values for the larger proteins were initially much lower than those of the smaller proteins and decreased to a very low value at high lung lymph flow rates. When the L/P values for all four protein fraction groups are plotted as a function of their effective molecular radii for both baseline and higher lymph flow rates (>4 times control), it is apparent that molecular size plays an important role in the permeability of the proteins under both conditions, as shown in Figure 7. The o<j values for these four protein groups (from Eq. 5) were 0.64,0.72,0.82, and 0.91 for groups I, II, III, and IV, respectively. Discussion The amount of fluid crossing the walls of exchange vessels depends on the balance of intra- and extravascular hydrostatic and oncotic pressures (Pmv, Ppmv, ITmv, and ITprnv) and the characteristics of the exchange vessel walls (kf C and o<i). The relationship of these forces and coefficients is shown in Equation 1. Many investigators (Guyton and

4 PRESSURE EFFECTS ON FLUID BALANCE/Parker et al 1167 TABLE 1 Cumulative Data for the Variables and Calculated Microvascular Pressures during Steady State Sheep no. VS-3-79 #1 (ml/hr) C L (g/dl) Cp (g/dl) PLA (cmh,o) PPA (crnh,o) PMV (cm H.O) VS-3-79 # VS-4-79 # VS-4-79 # VS VS VS-2O VS VS VS VS VS VS VS Abbreviations: Qi. = lung lymphflow;ci. >= lung lymph total protein concentration; Cp = plasma total protein concentration; PLA left atrial pressure; PPA main pulmonary arterial pressure; PMV calculated pulmonary microvascular pressure. Dashes indicate data not taken or calculated J , Lindsey, 1959; Erdmann et al., 1975; Vreim et al., safety factors, which include: (1) increasing inter- 1976; Woolverton et al, 1978) have shown that stitial fluid pressure, (2) decreasing interstitial fluid pulmonary microvascular hydrostatic pressure oncotic pressure, and (3) increasing lung lymph must exceed some critical level before significant flow. Our data permit us to evaluate the relative amounts of fluid accumulate in the lung. Those contributions of the latter two safety factors over a observations have led to the concept of lung edema broad range of pulmonary vascular pressures.

5 1168 CIRCULATION RESEARCH VOL. 49, No. 5, NOVEMBER PLASMA FLOW EXPERIMENTAL BASELINE ) ' * * 0.8-0, *y* LrMPH FLOW (ml. / hour ) FIGURE 2 The relationship of steady state lymph-toplasma concentration for total proteins to steady state lung lymph flow for one sheep. Decreased Interstitial Oncotic Pressure In order to estimate the oncotic pressure of the lung interstitial fluid we assumed that the lymph was exclusively derived from the pulmonary circulation and that during steady state conditions the oncotic pressure of the interstitial fluid was equal to that of the lymph collected. This latter assumption is based on evidence that proteins are not concentrated within lymphatics (Vreim et al., 1976; Taylor and Gibson, 1975). Although we have not attempted to distinguish the relative contributions of the various mechanisms for decreasing Ilpmv, there are two major ones which should be mentioned. The first is removal (or "washing" out) of interstitial proteins PLASMA ,4-0,9-0.B ,6- O.3-0,2 - O.I FLOW / EXPERIMENTAL \ 9 10 FIGURE 3 The cumulative data depicting the relationship of steady state lymph-to-plasma concentration ratios for total proteins to steady state normalized lung lymph flow. NET (cm Hfi) 12 -, CHANGE IN MICROVASCULAR PRESSURE (cm H20) FIGURE 4 Relationship of normalized lung lymph flow (top panel) and net effective oncotic pressure gradient (bottom panel) to increased pulmonary microvascular pressure. The solid lines are lines of best fit. (Taylor et al., 1973), and the second, dilution of proteins in the interstitium. Dilution may be accomplished by physically increasing the fluid volume of the interstitial space (Fishman, 1972) and by decreasing the volume of the interstitium which excludes proteins (Comper and Laurent, 1978; Parker et al., 1979). However, it should be emphasized that the dilution mechanisms are only transient means by which the protein concentration within the interstitium is decreased. The steady state protein concentration in the available interstitial fluid is determined by (and equal to) the protein concentration of the microvascular nitrate (Granger and Taylor, 1980). The effective oncotic pressure gradient across pulmonary exchange vessels depends on the value of a d (i.e., if a d equals 1.0, the total oncotic pressure gradient would be exerted across the microvascular membrane; whereas, if oa equals 0.0, no oncotic pressure gradient would exist because the proteins would be as permeable as water). Granger and Taylor (1980) have shown that a minimal value for CTd can be estimated from the lymph-to-plasma total protein concentration ratio (L/P) at lymph flows high enough that L/P is filtration independent. Utilizing this technique, we calculated a rninirnal ad for total proteins of 0.74 and rninirnalod values for protein fraction groups I, II, III, and IV of 0.64, 0.72, 0.82, and 0.91, respectively. This increasing Od

6 PRESSURE EFFECTS ON FLUID BALANCE/Parker et al FIGURE 5 An electrophoretogram of sheep lung lymph and plasma. The nine fractions are numbered by the Arabic numerals and the four protein fraction groups are numbered by Roman numerals. HE iz: with increasing molecular size is as expected and lends some credence to the calculation. If the averaged L/P values of each major fraction obtained at high lymph flow rates (>4 times control) are divided by their baseline L/P values, Figure 8 is obtained. From Figure 8 it can be seen that the greater the size of the protein the greater the relative decrease in its L/P value. However, this is not consistent with results found by McNamee and Staub (1979), who found that lymph/plasma concentrations of their protein fractions all decreased proportionally as microvascular pressure increased. Our data for estimating the net effective osmotic pressure gradient is consistent with that of other investigators (Erdmann et al., 1975; Bowers et al., 1977). Erdmann et al. (1975) found that, for the Pmv range they studied in sheep, decreased interstitial oncotic pressure provided a safety factor which was 50% of the increment in pulmonary microvascular pressure. However, they assumed oa to be equal to one. If aa is taken to be 0.74, their data are similar to ours. There are possible errors inherent in our investigation. We assumed that membrane or protein electrical charge had no effect on protein sieving by the pulmonary microvascular membrane. There are data that suggest that this assumption may be invalid in the kidney (Chang, 1975) and small intestine (McElearney and Granger, 1979), and that could also be true in the lung. Also, the method of calculating microvascular pressure from an empirical equation (Eq. 2) is a possible source of error, since the pre- to postcapillary resistance ratio may change as left atrial pressure is elevated. If the phenomenon of "pore stretching" occurs, the minimal value we obtained for a as estimated by the L/P technique would be an underestimation of the Od at normal left atrial pressures (Taylor et al., 1973). We found no evidence suggesting such a phenomenon in this investigation; in fact, we found some evidence against it. The L/P for total proteins never increased over that of the previous value for any elevation of pulmonary microvascular pressure in any sheep. In addition, the lymph-to-plasma ratio for each of the various protein fraction groups decreased from the values obtained at moderate Pmv elevations when Pmv was elevated to high values (Figs. 6 and 7). However, it is possible that pore stretching occurs at microvascular pressures higher than those of this investigation. Increased Lung Lymph Flow Obviously, if all the fluid filtered into the lung interstitium is immediately removed by the lung PLASMA I.O-i 0, ,0 I- O o 8 0 o FLOW / EXPEPHCHTAL \ \ BASELWE I FIGURE 6 Steady state lymph-to-plasma ratios for a small (fraction I denoted by the solid circles) and large protein (fraction IV denoted by the open circles) as a function of normalized lung lymph flow.

7 1170 CIRCULATION RESEARCH VOL. 49, No. 5, NOVEMBER 1981 TABLE 2 Cumulative Data for Total Proteins and Four Protein Fractions in Lymph and Plasma CT C, c,, Cm C,v Sheep no. L P L P L P L P L P Vs-3-79 # L n T P VS-3-79 # & VS-4-79 # VS-4-79 # VS VS VS Vs VS VS VS VS VS VS S6 l.u Abbreviations: CT " total protein concentrations; Ci, Cn. Cm, Crv ~ protein concentrations of protein fractions; OT ~ calculated oncotic prenurog; L lymph; P» plasma. Dashes indicate data not taken or calculated lymphatics, edemafluidwould not accumulate. The top panel of Figure 4 shows that lung lymph flow increased much less when pulmonary microvascular pressure was elevated by 10 cm H2O than at higher elevations, probably because interstitial fluid pressure increased and/or the oncotic pressure of the interstitial fluid decreased. However, lung lymph flow becomes more important in preventing edema at Pmv elevations above 10 cm H 2 O, where the relationship between lymphflow and Pmv becomes steep. A comparison of our data showing relationships

8 PRESSURE EFFECTS ON FLUID BALANCE/Parker et al 1171 I.O-i LlWG FLOW EXTRAVASCULAR PLASM4 ('EXPEF^IMEr^TAL NTALN 5 V BASELII INE I /EXPERIMENTAL^ \ BASELINE / MOLECULAR RADIUS (1) FIGURE 7 Average ± SEM steady state lymphtoplasma ratios of the four protein fraction groups during baseline conditions (solid circles) and at high lung lymph flows (open circles) caused by increased left atrial pressure as a function of their effective molecular radii. between lung lymph flow and microvascular pressure and the data of Erdmann et al (1975) showing relationships between extravascular lung water and microvascular pressure in sheep is illustrated in Figure 9. That figure shows that the highest pressures at which we were able to obtain steady state data were near the pressures at which the relationship between lung water and pressure becomes very steep. Since, as we have shown, the protection against edema resulting from falling interstitial oncotic pressure is exhausted at much lower pressures, it is reasonable to infer that high lymph flow is a major protective mechanism at high pressures. In conclusion, the results of this investigation indicate that the safety factor against pulmonary L/ P (EXPERIMENTAL) L/P (BASELINE) 0.6 T , A PMV FIGURE 9 Comparison of normalized lung lymph flow for this study (solid line) to lung extravascular water to bloodless dry lung weight ratios of sheep lungs (dashed line) as reported by Erdmann et al. (1975), as a function of increased lung microvascular pressure. edema provided by decreasing interstitial oncotic pressure depends on both the value of o d, which is at least 0.74 for total protein, and the rate of transmicrovascular filtration (i.e., 70% of this safety factor is exhausted when lung lymph flow is three times control). It also appears that lung lymph flow per se plays an important role in opposing the development of pulmonary edema, due to elevated pulmonary micro vascular pressure, for pressure elevations greater than 15 cm H2O. For microvascular pressure elevations below 15 cm H2O, lymph flow appears to be important mainly for the removal of interstitial fluid proteins, thereby permitting Ilpmv to decrease. We found no evidence for increased microvascular permeability in response to increased pressure over the pulmonary microvascular pressure range studied in this investigation MOLECULAR RADIUS (A) FIGURE 8 Relationship of the relative decrease of lymph-toplasma protein concentration ratios of the four protein fraction groups as a function of their molecular size.

9 1172 CIRCULATION RESEARCH VOL. 49, No. 5, NOVEMBER 1981 References Bowers R, Brigham K, Owen P (1977) Salicylate pulmonary edema: Mechanism in sheep and review of the clinical literature. Am Rev Reap Dia 115: Brace RA, Taylor AE, Granger DN (1977) Analysis of lymphatic protein flux data. II. Effect of capillary heteroporosity on estimate* of reflection coefficient»nd PS product. Microvaac Res 14: Brigham KL, Owen PJ (1975) Increased sheep lung vascular permeability caused by hutamine. Circ Res 37: Brigham KL, Woolverton WC, Blake LH, Staub NC (1974) Increased sheep lung vascular permeability caused by Paeudomonas bacteremia. J Clin Invest 54: Chang RLS, Den WM, Robertson CR, Brenner BM (1975) Permselectivity of the glomerular capillary wall III. Restricted transport of polyanions. Kidney Int 8; Comper WD, Laurent TC (1978) Physiologic function of connective tissue polysaccharides. Physiol Rev 58: Erdmann J, Vaughan T, Brigham K, Woolverton W, Staub N (1975) Effect of increased vascular pressure on lung fluid balance in unanesthetized sheep. Circ Res 37: Failing J, Buckley M, Zak D (1960) Automatic determination of 3erum protein. Am J Pathol 33: Fishman AP (1972) Pulmonary edema. The water-exchanging function of the lung. Circulation 46: Gaar KA, Taylor AE, Owens LJ, Guyton AC (1967) Pulmonary capillar pressure and filtration coefficient in the isolated perfused lung. Am J PhysioL 213: Granger DN, Taylor AE (1980) Permeability of intestinal capillaries to endogenous macromolecules. Am J Phyaiol 238: H457-H464 Granger DN, Richardson PDI, Taylor AE (1979) The effects of iaoprenaline and bradykinin on capillary filtration in the cat small intestine. Br J Pharmacol 67: Guyton A, Lindsey A (1959) Effect of elevated left atrial pressure and decreased plasma protein concentration on the development of pulmonary edema. Circ Res 7: Harris TR, Roselli RJ (1981) A theoretical model of protein, fluid, and small molecule transort in the lung. J Apply Physiol 50: 1-14 Kuramoto K, Rodbard S (1962) Effects of blood flow and left atrial pressure on pulmonary venous resistance. Circ Res 11: McEleamey PM, Granger DN (1979) Intestinal capillary walls as a charge-selective filter (abstr). Physiologist 22: 85 McNamee JE, Staub NC (1979) Pore models of sheep lung macrovascular barrier using new data on protein tracers. Microvasc Res 18: Parker JC, Falgout HJ, Parker RE, Granger DN, Taylor AE (1979) The effect of fluid volume loading on exclusion of interestitial albumin and lymph flow in the dog lung. Circ Res 45: 44CM50 Patlak CS, Goldstein DA, Hoffman JF (1963) The flow of solute and solvent across a two membrane system. J Theor Biol S: Snedecor G, Cochran W (1967) Statistical Methods, ed 6. Ames, Iowa, Iowa State University Press Staub N, Bland R, Brigham K, Demkng R, Erdmann J, Woolverton W (1975) Preparation of chronic lung lymph fistulas in sheep. J Surg Res 19: Taylor AE, Gibson WH (1975) Concentrating ability of lymphatic vessels. Lymphology 8: Taylor AE, Gibson WH, Granger HJ, Guyton AC (1973) The interaction between intracapillary and tissue forces in the overall regulation of interstitial fluid volume. Lymphology 6: Vrieim CE, Snashall PD, Demling RH, Staub NC (1976) Lung lymph and free interstitial fluid protein composition in sheep with edema. Am J Physiol 230: Warren M, Drinker C (1942) The flow of lymph from the lungs of the dog. Am J Physiol 136: Woolverton WC, Brigham KL, Staub NC (1978) Effect of positive pressure breathing on lung lymph flow and water content in sheep Circ Res 42: