Morphometric studies of the alveolar epithelium in dog lungs in vivo after increased rates of outward filtration.код для вставкиСкачать
THE ANATOMICAL RECORD 205:39-45 (1983) Morphometric Studies of the Alveolar Epithelium in Dog Lungs In Vivo After Increased Rates of Outward Filtration DAVID 0. DEFOUW AND FRANCIS P. CHINARD Departments of Anatomy, Medicine, and Physiology, UMDNJ-New Jersey Medical School, Newark, NJ 07103 ABSTRACT Increased pulmonary artery pressures and decreased albumin concentrations were established in dogs, in vivo, by circulatory overload and by plasmapheresis. Each change, occurring singly, would be expected to produce an increase in outward filtration rate. The ultrastructure of the alveolar epithelium after the production of this presumed increase in filtration was evaluated by established morphometric techniques. The extent of extravascular fluid accumulation was assessed by multiple indicator-dilution studies and by light microscopic analysis. Previous studies from this laboratory (DeFouw and Berendsen, 1979; DeFouw, 1980) define increased alveolar surface densities, increased type I cell vesiculation, and depleted type I1 cell lamellar body contents after increased filtration had induced septal edema and alveolar flooding in isolated perfused dog lungs. In the present study septal edema and alveolar flooding were not detected. Nor were changes observed in either alveolar epithelial cells or in alveolar configurations. However, there was significant accumulation of extravascular fluid in cuffs around extraalveolar vessels. These results are consistent with the interpretation that the initial stage of cardiodynamic edema formation, namely, perivascular cuffing, is not associated with structural aberrations of the alveoli or their lining epithelial cells, in vivo. Since such changes were detected previously in isolated lungs after development of the final stages of edema, namely, septal fluid accumulation and alveolar flooding, additional studies of septal edema and alveolar flooding in intact animals are necessary to establish functional correlates to ultrastructural aberrations in severely edematous lungs. Previous studies from this laboratory have established that severe edema formation in isolated perfused dog lungs is associated, in part, with depletion of type I1 cell lamellar body contents, elevations of alveolar surface densities, and increased vesiculation within type I epithelial cells and in capillary endothelial cells (DeFouw and Berendsen, 1979; DeFouw 1980). That the alveolar epithelium and the surfactant system should be affected in the production of acute pulmonary edema is suggested by two concepts: (1) that the lamellar bodies of type I1 cells represent intracellular storage depots of pulmonary surfactant (Chevalier and Collet, 1972;Massaro and Massaro, 1972; Smith and Kikkawa, 19781, and (2)that vesicles contribute to macromolecule transport across the alveolar epithelium (Dermer, 1970). Support for this view can be found in the report of Harlan et al. (1966) on the detection of 14C-labelled palmitic acid in tracheal foam from edematous dog lungs. However, the mechanism whereby the loss of surfactant to edematous fluid initiates a depletion of type I1 cell lamellar body contents remains uncertain. The present investigation describes morphometric analyses of dog lungs, in vivo, after circulatory overload and plasmapheresis had presumably produced increased rates of outward filtration from the microcirculation. Thus, comparisons with the previous studies of isolated dog lungs are made possible. Filtration rates from the pulmonary microcirculation are generally considered in relation to the differences between plasma hydrostatic and osmotic (oncotic) pressures. The importance of corresponding tissue pressures is rec- 0003-276W83/2051-0039$02.50 0 1983 ALAN R. LISS, INC Received July 2, 1981;accepted Oct. 12,'82. 40 D.O. DEFOUW AND F.P. CHINARD ognized but cannot be assessed. However, under the present conditions tissue hydrostatic pressure could be expected to increase and tissue osmotic pressure to decrease. In the first set of studies of this report hydrostatic pressures were elevated by circulatory overload, while oncotic pressure remained essentially unchanged; in the second set, oncotic pressure was reduced under conditions of relatively stable hydrostatic pressure. Previous studies by others have been shown that the respective procedures produce marked elevation in pulmonary lymph flow rates from lungs, in vivo, in response to a presumed but not directly measured increase in rates of outward filtration (Uhley et al., 1967; Zarins et al., 1978). METHODS Circulatory Overload Studies Four mongrel dogs, with body weights of approximately 20 kg were anesthetized with pentobarbital (30 mgkg, IV) and given heparin (25,000 IU) intravenously. The trachea was cannulated and the lungs artifically ventilated with humidified air containing 5% COz with end-expiratory pressures of 5 cm HzO. A Swan Ganz catheter was positioned in the pulmonary artery via the right external jugular vein for monitoring pulmonary arterial pressures. The left femoral artery and vein were cannulated for monitoring systemic arterial and venous pressures. Indicator-dilutionstudies were carred out as described elsewhere (Chinard, 1975; Per1 et al., 1976) in order to obtain measures of tissue water accumulation from the extraction of tritium oxide (THO) and from the extravascular lung water volume. A Wescor vapor pressure osmometer was used for determining osmolalities and a membrane osmometer was used for determining oncotic pressures on plasma samples obtained prior to and after the circulatory overload. After the initial indicator-dilution run, an infusion of 4% bovine serum albumin, containing red blood cells adjusted to a hematocrit of 25-30%, was pumped into the right femoral vein at a rate of 50 cm3 minute-'. ARer approximately 6 liters of the infusion solution had been administered and a second indicatordilution run had been recorded, the thorax was opened via midsternal incision and the lungs were fixed via tracheal instillation. The pulmonary trunk and aorta were ligated simultaneously with the onset of tracheal instillation of 2% glutaraldehye in 0.15 M bicarbonate buffer at 30 cm HzO pressure. This procedure was used to avoid artefacts of extravasation of blood that we have noted when fixation was initiated without blockage of the pulmonary circulation. Plasmapheresis Studies Four mongrel dogs, 20 kg approximate body weight, were anesthetized and artifically ventilated as described in the circulatory overload studies. Pulmonary arterial pressures and systemic oncotic pressures were also measured as above. After the first indicator-dilution run, the right femoral artery was connected to a roller pump which assisted in bleeding the dogs at 50 cm3 minute-l for 10-minute periods. The 500 cm3 of withdrawn blood was centrifuged and the plasma removed. The red cells were then reinfused as a suspension of 100-cm3cells in 1,050 cm3 saline into the right femoral vein a t 50 cm3 minute-'. Six bleedings and subsequent infusions were performed. The thorax was incised and the lungs fixed as in the circulatory overload studies after the second indicator-dilution run. Morphometric Analyses The fixed lungs from the two experimental groups were immersed in glutaraldehyde overnight and subsequently sampled via a stratified sampling technique which provided 24 tissue blocks from each of the lungs (Weibel, 1973). A portion of each block was processed for light microscopy and the remaining portions were postfixed in 1%osmium tetroxide in 0.15 M bicarbonate buffer, dehydrated in graded series of ethanols and propylene oxide, and embedded in Epon. From the primary sample of 24 blocks, ten blocks were randomly selected and sectioned on a Sorvall MT2B Ultramicrotome at 70-80-nm thickness. The thin sections were examined with a Philips 300 transmission electron microscope and micrographs of the alveolar septa with final magnifications of x 6,100, x 21,000, and x 81,500 were obtained for the morphometric analyses. Each of the micrographs was analyzed with a multipurpose stereologic test grid composed of 168 test points and 84 test lines (Weibel, 1973). Counts of test line intercepts with the alveolar epithelial surface profiles and counts of test points falling on type I epithelial cells were obtained from micrographs a t x 6,100 to provide estimates of alveolar surface densities and average thicknesses of the type I cells. Point counting volumetry was used to estimate type I1 cell organelle volume densities and type I 41 LUNG EPITHELIUM AND EARLY EDEMA cell vesicle volume densities from the micrographs at x 21,000 and x 81,500, respectively. A group of three normal mongrel dogs, approximately 20-kg body weights, were prepared for morphometric analyses in a manner identical to that described for the experimental groups. Statistical analyses of the normal and experimental morphometric parameters were performed by a one-way analysis of variance. Comparisonsof the physiological data from the experimental groups were performed with a paired t test. For all statistical tests P < 0.05 was accepted as significant. RESULTS The multiple indicator-dilution studies were undertaken in the overload and plasmapheresis series in order to obtain measures of the increases of extravascular lung water expected to result from the increased outward filtration. Two parameters are reported here. The first, the extraction of THO, indicates the extravascular volume of distribution of water relative to the vascular compartment volume. This measure is not affected by recruitment of vessels with extravascular volumes similar to those of initially perfused vessels. If it increases, there is fluid accumulation somewhere in the lungs. In both the overload and plasmapheresis series such an increase in the extraction of THO was recorded in each experimental animal; however, the variability of the increases in the overload series was such as to limit the statistical significance of the differences between the means of the normal and overload data (Table 5). The second parameter, the extravascular volume of distribution of water (calculated essentially from the product of the flow of blood and the difference of the mean transit times of tracer water and a vascular reference indicator) is expected to increase with recruitment as well as with the development of edema. In the overload group extravascular lung water volume was increased but again with a variability limiting the statistical significance of the differences between averages of the normal and experimental data. An increase in extravascular lung water was not recorded in the plasmapheresis series (Table 5). In association with changes in the extraction of THO and of the extravascular lung water volume, changes in plasma hydrostatic and oncotic pressures in the experimental groups also indicated increased rates of outward filtration from the microcirculation. After circulatory overload, mean pulmonary arterial pressure increased from an average of 13 mm Hg in control periods to an average of approximately 38 mm Hg. Colloid osmotic pressure, on the other hand, remained in the range of 20-23 mm Hg (Table 3). Thus, the substantial increase in differences between hydrostatic and osmotic pressures would insure an increase in outward filtration provided that corresponding moderating changes did not occur in the tissue pressures. In the plasmapheresis group, mean pulmonary aterial pressure was maintained a t approximately 14-15 mm Hg while colloid osmotic pressure was decreased from an average of approximately 22 mm Hg to a n average of 8 mm Hg (Table 4). Again, increased outward filtration rates would be expected to occur with these changes although the increases would be less in the plasmapheresis group than in the overload group because of the lesser increase in the difference between pulmonary artery and oncotic pressures. Table 1 presents the stereologic parameters which were measured on the type I alveolar TABLE 1 . Average thicknesses of type I epithelial cells, type I cell vesicular volume densities, and alveolar surface densities’ Normal 0.26 f .01 Type I thickness (pm) 8.2 f 1.7 Vesicle volume density (% cytoplasm) Relative vesicle distributions (% vesicle population) 16 f 5 Luminal Cytoplasmic 67 f 8 17 f 4 Ab1umina1 Alveolar surface density 595.6 (cm2/cm3lung) f 69.6 ‘Mean values 2 one SD Overload Plasmapheresis 0.25 * ,063 7.7 c 2.5 0.28 f .04 9.1 * 2.9 12 c 7 71 c 8 17 t 5 624.7 e 88.5 14 ? 2 69 % 3 17 f 2 554.7 f 45.5 42 D.O. DEFOUW AND F.P. CHINARD epithelial cells. Average thicknesses of the type I cells were unchanged after either circulatory overload or plasmapheresis. The volume of type I cell cytoplasma occupied by vesicles was also unchanged and was estimated at approximately 8-9% in each of the groups. Moreover, the intracellular distributions of vesicles were similar in the normal and experimental groups: 70% of the vesicles were observed to be unattached to the cell surfaces and the remaining 30% were directly attached to either the luminal or abluminal membrane surfaces in the single plane of tissue section. Table 1also presents the alveolar surface densities, defined as cm2 alveolar surface per cm3 lung volume, for the normal and experimental groups. The slight changes in alveolar surface densities in the experimental groups were not statistically significant. Table 2 presents the estimates of organelle volume densities from type I1 cells of the normal and experimental lungs. Organelle volume densities are defined as the proportion of type I1 cell cytoplasm occupied by the respective organelles. As indicated, mitochondria1 volume densities were approximately 7-9% in each case. Volume densities of the rough endoplasmic reticulum, including polyribosomes, also remained essentially constant at approximately 7-10%. Likewise, the smooth endoplasmic reticulum, including the Golgi complex, occupied approximately &lo% of the type I1 cell cytoplasm in the normal and experimental groups. The lamellar body volume densities were decreased slightly from approximately 20% to 17-18% after both the plasmapheresis and circulatory overload procedures. In addition to the preceding morphometric results, examination of the alveolar septa demonstrated no fluid accumulation in the alveolar TABLE 2. Organelle volume densities in the cytoplasm of type II cells' Normal 9.1 i 1.5 Mitochondria 7.1 lr 3.5 Rough ER2 8.6 -c 1.5 SmoothER Lamellar bodies 20.5 f 0.3 Overload Plasmapheresis 7.0 f 1.1 8.5 i 2.6 8.1 f 2.4 16.6 2 1.5 8.8 i 1.6 9.5 f 2.5 10.8 lr 2.8 18.4 i 2.8 * 'Mean values one SD. 2ER, endoplasmic reticulum TABLE 3 . Hydrostatic and oncotic pressures in the circulatory overload studies' Normal 13.0 _t 1.5 Pulmonary artery pressure 19.5 i 4.3 Colloid oncotic pressure Resultant pressure difference -6.4 i 2.9 Overload 37. C 6.1* 22.5 c 8.5 15.4 C 9.6* 'Mean values (mm Hg) f one SD. The pressure differences were calculated for each of the runs and these differences were averaged. 'P < 0.05. TABLE 4. Hydrostatic and oncotic pressures in the plasmapheresis studies' Pulmonary artery pressure Colloid oncotic pressure Resultant pressure difference Normal Plasmapheresis 13.9 i 3.6 14.9 i 0.9 21.9 i 5.1 8.4 f 5.5* -8.1 6.7 -c 6.0* f 4.7 'Mean values (mm Hg) t one SD. The pressure differences were calculated for each of the runs and these differences were averaged. *P < 0.05. TABLE 5 . Cuffingof extraalveolar vessels and extractions of THO' ~ _ _ _ _ Normal lungs Circulatory overload Plasmapheresis _____ ~ Number of vessels examined Percentage with cuffs 164 455 171 11.0 56.0 64.0 ~ Extraction of THO 0.49 i 0.08 0.68 t 0.14 0.65 _t 0.03* Extravascular lung water cm3 . (100 gm lung)-' 42.9 2 13.3 111.2 i 73.2 34.6 f 15.5 'Values for the extraction of THO (means and standard deviations) were obtained from the multiple-indicatordilution studies carried out in the animals subjected to circulatory overload and to plasmapheresis.We have pooled the control values in these studies and listed them for normal lungs in the table to simplify the presentation of the data. In the overload experiments, the control values were 0.47 2 0.09 and for the plasmapheresis experiments 0.52 f 0.07. Similarly, the control values for extravascular lung water were 50.2 ? 11.1 and 35.7 ? 12.3. *P < 0.05. LUNG EPITHELIUM AND EARLY EDEMA 43 DISCUSSION interstitium. Light micrographs, on the other hand, containing branches of the larger pulmonary blood vessels and airways, showed fluid accumulation within the perivascular connective tissue spaces in both the circulatory overload and the plasmapheresis groups. In the normal lungs 18 extraalveolar vessels displayed cuffs from a total of 164 observed vessels. After circulatory overload 255 vessels possessed cuffs from a total of 455 observed vessels. In the plasmapheresis group 110 of 171 observed vessels had fluid cuffs (Table 5 ) . Figure 1 illustrates a pulmonary artery and its surrounding fluid cuff in one of the circulatory overload lungs. Thus, detectable edema fluid accumulation was restricted to the extraalveolar interstitial spaces of the intact dog lungs a h r circulatory overload and after plasmapheresis. The results of the present morphometric analysis of alveolar epithelial cells of dog lungs, in vivo, differ from the results of similar analyses we have reported previously on isolated perfused dog lung preparations. This difference is consistent with the interpretation that increases in filtration from the pulmonary microcirculation are not associated with changes in ultrastructural characteristics of the alveolar epithelium unless septa1 edema and alveolar flooding also occur. The contention that filtration rates were increased in the present investigation is based upon the marked increase in differences between hydrostatic and osmotic (oncotic) pressures of the circulating blood after both the overload and plasmapheresis procedures. Although corresponding tissue pressures sur- Fig. 1. The light micrograph from a lung of the circulatory overload group illustrates prominent penarterial fluid cuffs surrounding the branches of the pulmonary artery. The peribronchiolarconnective tissue sheath, on the other hand, is not marked by interstitial fluid accumulation. The distention of the lymph vessel adjacent to the pulmonary artery is consistent with the suggestion that lymph flow from these lungs was increased. 44 D.O. DEFOUW AND F.P. CHINARD rounding the microvasculature cannot be assessed, tissue hydrostatic pressure could be expected to increase and tissue osmotic pressure to decrease under these conditions. Thus, the effect of the marked changes in intravascular forces on filtration could have been moderated or even abolished by the presumed compensatory tissue pressure changes. The increased extraction of THO in both the overload and the plasmapheresis series provides evidence of increased outward filtration in those experiments. That the extravascular lung water volume also increased in the overload experiment is interpreted as an indication of recruitment. Since this parameter did not increase in the plasmapheresis experiments, there was no recruitment in that series, only excess fluid accumulation. These findings and the findings of increased numbers of extraalveolar vessels with cuffs are mutually compatible. The interpretation that there was increased outward filtration in both the overload and in the plasmapheresis series is thus supported by both the functional and the anatomical studies. Since the procedures of circulatory overload and plasmapheresis were also reported to be associated with markedly increased pulmonary lymph flow in lungs, in vivo (Uhley et al., 1967; Zarins et al., 1978), it is apparent that pulmonary lymphatic capacity, in vivo, exceeds that in isolated lung preparations. That is, comparable methods of alveolar septal edema induction were effective in isolated perfused dog lungs but ineffective in vivo. Moreover, pulmonary lymphatic capacity, though not definitively known, is considerable, as tenfold increases in pulmonary lymph flow have been reported (Staub, 1974). A finite lymphatic capacity is indicated, however, by the presence of perivascular fluid cuffs surrounding the extraalveolar lung vessels after circulatory overload and plasmapheresis. This fluid accumulation is consistent with the anatomic evidence presented by Lauweryns (1971) that lymphatic vessels are present in the extraalveolar connective tissue spaces only. It is reasonable, if not established at this time, to consider that both the capacity of the extraalveolar interstitium to accumulate fluid and the capacity of the lymphatic system to provide runoff must be overcome before septal edema and subsequent alveolar flooding occur. Previous studies from this laboratory have described an association between alveolar interstitial edema and alveolar flooding with increased numbers of vesicles in type I epithelial cells as well as in capillary endothelial cells. Similarly, Gil and Magno (1980) observed increasedvesiculation in squamousepithelial cells in fluid-filled rabbit lungs. The functional significance of this increased vesiculation, however, remains uncertain. The present study clearly establishes that changes in the factors which can be expected to provide increased filtration from the pulmonary microcirculation, in vivo, are not associated with an increase in type I cell vesiculation. Furthermore, intact animals could possess adaptive responses to increased filtration which are absent in isolated perfused lungs. Moreover, isolated lung preparations, which are denewated, ventilated with positive pressure respiration, and perfused in a nonpulsatile manner, could themselves contribute to abnormal cellular responses under edematous conditions. In another study from this laboratory severely edematous isolated lungs, perfused at 15°C also failed to display increased vesiculation in type I cells (Chinard and DeFouw, 1981). Apparently, septal edema and alveolar flooding in isolated-perfused lungs do not influence type I cell structure unless normal membrane fluidity and cytoplasmic viscosity are also maintained. Additional studies, possibly with ultrastructural tracers, may help to determine whether type I vesiculation is a response to abnormal fluid accumulation in the lung parenchyma and serves to limit expansion of the alveolar interstitium. Our studies provide no evidence that the vesicles provide a principal mechanism by which fluid accumulates in the alveoli during the alveolar flooding stage. In previous studies from this laboratory increased alveolar surface densities after septal edema and alveolar flooding in isolated dog lungs were detected. Similar increases in alveolar surface areas in saline-filled lungs were reported by Gil et al. (1979). As suggested by these authors, after alveoli become fluid filled, their configurations are no longer influenced by interfacial tensions; therefore, tissue pressures would primarily determine alveolar shapes and hence, surface densities. In addition, alveolar interstitial fluid accumulation could affect tissue tensions normally created by the fibrous and cellular components of the septa. Thus, pressure differences between the septal tissues and intraalveolar fluid could also strongly influence alveolar configurations. The results of the present study are consistent with the purported role of these determinants of alveolar configurations. Specifically, changes in LUNG EPITHELIUM AND EARLY EDEMA alveolar surface densities would not be expected if alveolar interstitial edema or alveolar flooding did not occur. Our earlier studies also identified an association between increases in alveolar surface densities and depletion of type I1 cell lamellar bodies in isolated perfused lungs. The present results, in which cardiodynamic changes in the pulmonary microcirculation, in vivo, occurred without associated changes in alveolar configurations or depletion of type I1 cell lamellar bodies, are consistent with the interpretation that alterations in alveolar configurationscould represent one of several possible stimuli which trigger a diminution of lamellar bodies in the alveolar type I1 cells. Since the rate of surfactant secretion was not measured in our present or previous studies, further assessment of surfactant synthesis and secretion under normal and edematous conditions is required to test this interpretation. As pointed out by Mason and Williams (1977),cellular mechanisms regulating surfactant release from type I1 cell lamellar bodies are uncertain. ACKNOWLEDGMENTS The authors thank Drs. Parimal Chowdury and William Cua for their participation in the experimental aspects of these studies and Mrs. Perlita Hill and Ms. Gail Smith for their technical assistance. The late Mr. Gary Schroller also participated in these studies. This investigation was supported in part by Public Health Service grants HL 19571 and HL 12879 and by grant-in-Aid from the American Heart Association with funds contributed in part by the New Jersey affilate and the Bergen-Passaic and Ocean County chapters. LITERATURE CITED Chevalier, P., and A.J. Collet (1972)In vivo incorporation of ~holine-~H, Ie~cine-~H, and galacto~e-~H in alveolar type I1 pneumocytes in relation to surfactant synthesis. A quantitative radioautographic study in mouse by electron microscopy. Anat. Rec., 174289-310. Chinard, F.P. (1975)Estimation of extravascular lung water by indicator dilution techniques. Circ. Res., 37:137-145. 45 Chinard, F.P., and D.O. DeFouw (1981)Ultrastructural changes of the air-blood barrier aRer spontaneous development of edema in isolated dog lungs perfused a t 15°C. Microvasc. Res., 21:48-56. DeFouw, D.O., and P.B. Berendsen (1979)A morphometric analysis of isolated perfused dog lungs a h r acute oncotic edema. Microvasc. Res., 17:90-103. DeFouw, D.O. (1980)Ultrastructural changes in type I and type I1 alveolar epithelial cells in isolated perfused dog lungs aRer production of moderate and severe hydrostatic edema. Am. J. Anat., 158:355-364. Dermer, G.B. (1970)The fixation of pulmonary surfactant for electron micrumpy. II. Transport of surfactant through the air-blood barrier. U. Ultrastruct. Res., 31:229-264. Gil, J., H. Bachofen, P. Gehr, and E.R. Weibel (1979)Alveolar volume-surface area relation in air and saline-filled lungs fixed by vascular perfusion. J. Appl. Physiol., 47:990-1001. Gil, J., and M. Magno (1980)Fusion of pinocytotic vesicles in liquid filled lungs: A mechanism of cellular damage. Exp. Lung. Res., 1:43-56. Harlan, W.R., S.I. Said, and C.M. Banerjee (1966)Metabolism of pulmonary phospholipids in normal lung during acute pulmonary edema. Am. Rev. Respir. Dis. 94., 94:938-947. Lauweryns, J.M. (1971)The blood and lymphatic microcirculation ofthe lung. In: Pathology Annual 1971.S.C. Sommers, ed. Appleton-Century-CF, New York, pp. 365-415. Mason, R.J., and M.C. Williams (1977)Type I1 alveolar cell. Defender of the alveolus. Am. Rev. Respir. Dis. 115, (Suppl.):81-91. Massaro, G.D., and D. Massaro (1972)Granular pneumocytes. Electron microscopic radioautographic evidence of intracellular protein transport. Am. Rev. Respir. Dis., 105:927-93 1. Perl, W., R. Silverman, A.C. Delea and F.P. Chinard (1976) Permeability of dog lung endothelium to sodium, diols and amides. Am. J. Physiol., 230:1708-1721. Smith, F.B., and Y. Kikkawa (1978)The type I1 epithelial cells of the lungs 111. Lecithin synthesis: A comparison with pulmonary macrophages. Lab Invest., 38:45-51. Staub, N.C. (1974) Pulmonary edema. Physiol. Rev., 54:678-811. Uhley, H.N., S.E. Leeds, J.J. Sampson, and M. Friedman (1967)Right duct lymph flow in experimental heart failure following acute elevation of left artial pressure. Circ. Res., 20:306410. Weibel, E.R. (1973)Stereologic techniques for microscopic morphometry. In: Principles and Techniques of Electron Microscopy., Vol. 3.M.A. Hayatt, ed. Von Nostrand Reinhold, New York, pp. 237-296. Zarins, C.K., C.L. Rice, R.M. Peters, and R.W. Virgilio (1978) Lymph and pulmonary response to isobaric reduction in plasma oncotic pressure in baboons. Circ. Res., 43:925-930.