- Email: [email protected]
Effects of experimental cirrhosis on splanchnic microvascular fluid and solute exchange in the rat
GASTROENTEROLOGY
1984:87:165-72
Effects of Experimental Cirrhosis on Splanchnic Microvascular Fluid and Solute Exchange in the Rat JAMES A. BARROWMAN
and D. NEIL GRANGER
Faculty of Medicine, Memorial University of Newfoundland. St. John’s, Newfoundland and Department of Physiology, College of Medicine, University of South Alabama, Mobile, Alabama
In human cirrhosis, there is evidence that there are considerable alterations in fluid and solute exchange in the hepatic and intestinal microcirculations. Experimental cirrhosis was induced in rats b) using oral phenobarbitone and carbon tetrachloride inhalation over an 8-wk period. Portal venous pressure, hepatic and intestinal lymph flows, and lymph and plasma protein concentrations were measured. Liver samples were obtained for histologic examination. Portal venous pressure increased from a normal value (control animals) of9.0 (6.3-23.1) cmHzO to 17.9 (9.0-29.0) cmH,O in cirrhotic rats. There was a strong correlation between the degree of fibrotic change on histology and portal venous pressure. Lymph flows from the intestine and liver in cirrhotic animals were increased threefold and 30-fold, respectively, over values obtained from control animals. There was a good correlation between intestinal and liver lymph flows and portal venous pressure. Analysis of lymph/plasma protein concentration ratios at various lymph flows suggests that capillary permeability in the small intestine during sustained portal hypertension is comparable to that in normal animals. However, the highly permeable blood-lymph barrier of the normal liver becomes markedly restrictive in cirrhotic animals. There is considerable evidence in the literature that the exchange vessels (sinusoids) in normal liver are highly permeable to macromolecules (1,~). The permeability characteristics of these microvessels can be attributed to their discontinuous nature, i.e., they Received September 26, 1983. Accepted February 13, 1984. Address requests for reprints to: Dr. D. Neil Granger, University of South Alabama. Department of Physiology, College of Medicine. MSB 3024, Mobile, Alabama 36688. This study was supported by funds from the Faculty of Medicine. Memorial University of Newfoundland. Dr. Granger is a recipient of a Research Career Development Award from the National Heart, Lung and Blood Institute (HL00816). 0 1984 by the American Gastroenterological Association 0016-5085/84:$3.00
lack a basement membrane and have wide interendothelial apertures (1000-5000 A in diameter) (3). Cirrhosis, by contrast, is characterized by striking structural alterations in the sinusoid, namely there is deposition of a basement membrane and the vessels take on the appearance of continuous capillaries (4). The functional correlate of these vascular alterations in human cirrhosis is evidence of reduced macromolecular permeability from hepatic lymph protein measurements (5) and indicator diffusion studies (6). A major aim of the present study was to quantify the permeability characteristics of the hepatic microcirculation in cirrhosis in terms of “pore” dimensions. To study this, hepatic lymph protein composition was examined in normal rats and animals with carbon tetrachloride-induced cirrhosis. Cirrhosis, through portal hypertension, also has important effects on intestinal microvascular fluid and solute exchange. It has been suggested that the slow development of chronic portal hypertension leads to compensatory changes that oppose interstitial edema formation and filtration secretion, which are observed in acute portal hypertension (7). Two of these changes, namely increased lymph flow and decreased capillary permeability, were examined in the present study using the carbon tetrachloride model.
Methods The studies were performed using 37 male SpagueDawley rats fed a standard rat diet. Ten animals served as controls and in 27 rats, cirrhosis was induced by a modification of the method of McLean et al. (8). This procedure consisted of the addition of 0.5 glL of sodium phenobarbitone to all drinking water and regular periods of inhalation of carbon tetrachloride in a 150-L capacity chamber. The
Abbreviations used in this paper: C,, lymph concentration; CI+ plasma concentration; ad, capillary osmotic reflection coefficient.
166 BARROWMAN AND GRANGER
GASTROENTEROLOGY Vol. 87, No. I
Figure 1. Histologic appearance of control (A] and cirrhotic livers (B). Masson trichrome stain. rats were given phenobarbitone 1 wk before commencement of gassing (9). This program was carried out over a period of 8 wk. Cirrhosis could be demonstrated in all treated animals both macroscopically and microscopically (Figure 1). Ascites was present in 50% of the cirrhotic animals.
Experimental
Procedures
The rats were anesthetized with ether and the abdomen opened with a midline incision. The main hepatic and intestinal lymph trunks were identified, dissected, and cannulated with clear vinyl tubing (OD 0.8 mm, Dural Plastic Engineering, Dural, Australia] as previously described (10,ll).Lymph flows were measured by following the movement of lymph in calibrated micropipettes. Lymph and plasma samples were taken for protein concentration measurement by refractometry. The plasma sample was obtained from the portal vein at the end of each study. Lymph and plasma samples were stored frozen for subsequent electrophoresis on polyacrylamide gradient gels (12). Portal venous pressure was measured by inserting a ZOgauge needle, connected to a water manometer, into the vein and securing it with a drop of cyanoacrylate glue.
At the end of each experiment, body weight was recorded and the liver and small intestine were excised and weighed. A sample of liver was stored in Bouin’s fixative, embedded in paraffin, sectioned, and stained with Masson’s trichrome stain. Using the histologic criteria described by Orrego et al. (13), specimens were scored blindly for necrosis, fibrosis, and fat on their scale of O-3. Protocols Controls. After laparotomy, basal lymph flows from the intestine and liver were measured and samples were taken for protein concentration in lymph and plasma. Portal pressure was then measured. Acute portal hypertension. In 7 of the 10 control animals, after obtaining basal data, a snare was placed around the portal vein above the needle used to measure portal pressure. The portal vein was partially occluded to achieve a portal venous pressure between 20 and 30 cmH20, values comparable to the upper range of pressures recorded in cirrhotic animals. Intestinal lymph flow and protein concentration were allowed to reach a new steady state before measurements and samples were taken. Cirrhotics. In this group of animals the measurements were the same as those made in the control group.
July
MICROVASCULAR
1984
Table
I. Carbon
Tetrachloride-Induced Body
Cirrhosis
weight
Liver
(g)
Controls
Cirrhotics (n = 27) p value ” Mean
+ SE. ” Values
weight
11.9 20.46 (9.2-13.3) 15.9 20.76 (9.6-25.9) co.01
in parentheses
represent
values for ud were obtained for nine plasma protein fractions identified on gradient gel electrophoresis. The molecular radius of each protein fraction was determined by extrapolation from curves obtained with standard propore radii for the hepatic microcircuteins (12). Equivalent graphical analysis of 1 - CT~was plotted
against solute radius. The larger molecules were fit to a “large pore” line generated using the equation of Drake and Davis (15). When possible, a small pore curve was derived by subtracting the 1 - 0;~ values of the large pore curve from points lying above this curve. For animals, the gd values for the various plasma fractions were derived from basal values of C&p, ing maximal sieving at normal filtration rates.
control protein assum-
Statistics Standard statistical analyses were applied to the data [mean * SEM, unpaired Student’s t-test). Data were also analyzed using a linear regression analysis that mini-
25
I
i i
r=12
t
Figure
IL It=7 2
Plasma protein concen-
Portal pressure (cmH,O)
tration (g (5))
9.0 fO.SU (6.3-13.1) 17.9 21.0 (9-29) co.01
6.07 to.06 (5.3-6.5) 5.68 20.15 (5.0-6.5) 0.01
range.
In all experiments, lymph to plasma protein concentration ratios (C&p) were calculated. From the relationship between CL/Cp and lymph flow, the capillary osmotic reflection coefficient (ud) was assumed to equal 1 - CL/C,J at high rates of lymph flow (12). In the liver,
through the nine proteins,
Intestine weight (g) 8.6 kO.57 (7.0-12.4) 9.6 “0.44 (6.8-13.9) 0.10
Calculations
lation were obtained Renkin (14). For the
167
in Rats
(81
365.8” 28.7 (321-414)b 365.7 56.9 (287-417) 0.50
(n = 10)
EXCHANGE
n=, 3
2. Relationship between extent of fibrosis and portal venous pressure (mean 2 SE] in control and cirrhotic rats. Number of animals in each group is indicated.
mizes errors variables.
in both
the
dependent
and
independent
Results Table 1 lists basic data from control and cirrhotic rats. Liver weight was significantly increased (34%) and portal pressure was doubled in cirrhotic animals. Portal pressures as high as 29 cmH,O were recorded in cirrhotic animals. Histologic examination of liver specimens clearly distinguished controls from experimental animals. Within the experimental group, however, there was a wide range in the degree of pathological change in terms of fibrosis, necrosis, and fatty infiltration. Of these three parameters, fibrosis showed the best correlation (r = 0.78, p < 0.01) with portal venous pressure [Figure 2). The correlation coefficient for the relationships between portal venous pressure and necrosis was 0.71 (p < O.Ol), and for fatty infiltration it was 0.55 (p < 0.01). Figure 3 illustrates the effects of cirrhosis on hepatic and intestinal lymph flows and lymph to plasma total protein concentration ratios (C&). Lymph flows in the liver and intestine of control animals were 0.012 -4 0.002 and 0.127 * 0.016 ml/ min x 100 g, respectively. In cirrhotic animals these values increased by 2%fold and threefold, respectively. Lymph/plasma protein concentration ratios for liver and intestinal lymph were 0.87 + 0.03 and 0.50 t 0.03, respectively; these values fell to 0.70 c 0.03 and 0.37 * 0.05, respectively, in cirrhotic animals. The relationships between intestinal and liver lymph flows and portal venous pressure in control and cirrhotic animals are shown in Figure 4. In addition, the effects of acute portal hypertension on intestinal lymph flow in control animals are illustrated. In all instances there was a significant (p < 0.01) linear correlation (r = 0.72-0.92) between lymph flows and portal pressure. There was no significant difference in the lymph flow-portal pres-
168 BARROWMAN AND GRANGER
[I[7
Controls
0
Cirrhotics
GASTROENTEROLOGY
*
I
*
-L T
_m 0 ._
5
a : ._
0.60 -
2 L
za c? :
0:60 -
.5 Q 5 t
0.40 -
: : rr i
0.20 -
E” 5 OLiver
Intestine
Figure 3. Hepatic and intestinal lymph flows sod lymph/plasma protein concentration ratios in control and cirrhotic livers (mean ? SE). An asterisk (*) indicates a significant difference between control and cirrhotic animals, p < 0.05.
sure relationship in acute and chronic (cirrhosis] portal hypertension. Figure 5 illustrates the relationships between lymph to plasma total protein concentration ratio (CJCp) and lymph flow in the intestine of cirrhotic and control animals, with and without acute portal hypertension. In both groups, CL/CP decreased with increasing lymph flow to low flow-independent values ranging between 0.10 and 0.23. The relationship between C&p and hepatic lymph flow in control and cirrhotic animals is shown in Figure 6. As observed in the small bowel (Figure 5), C&P decreases with increasing lymph flow with C&p decreasing to as low as 0.43.
Vol. 87, No. 1
Figure 7 illustrates the graphical analysis for pore sizes based on C&r data for nine plasma protein fractions in the control and cirrhotic liver. In control animals, the sieving data would only fit a single pore size, i.e., 360 A radius. Whereas, in cirrhosis, sieving across the hepatic blood-lymph barrier could be described by two pore populations, a small pore system df 70 A radius and a system of large pores with 230 A radius.
Discussion Since the first report in 1936 (16)that repeated administration of carbon tetrachloride to rats results in cirrhosis, this method has been used extensively to create a model of human cirrhosis. Certain differences between carbon tetrachlorideinduced cirrhosis and the human disease are recognized, including the topography of the fibrosis and the collagen types involved. This model, nevertheless, has many similarities to human alcoholic cirrhosis, the type for which most histologic and biochemical information is available (17). Among these similarities is the occurrence of portal hypertension, and, in a proportion of animals, ascites. In the present study, the degree of portal hypertension that developed correlated well with the extent of histologic change in the liver, notably the amount of fibrosis. Structural changes in liver architecture and its vascular tree, by increasing resistance to blood flow at the sinusoidal or postsinusoidal level, are considered an important factor in the pathogenesis of portal hypertension. The other important contribution is an increase in splanchnic arterial inflow (17). At the microvascular level in the liver, cirrhosis is associated with collagenization of the space of Disse and the development of a basement membrane beneath the sinusoidal endothelium, i.e., “capillarization” of the sinusoid (4). This collagenization of the space of Disse may be partly responsible for the development of portal hypertension inasmuch as an association has been demonstrated between Disse space collagenization and portal hypertension in alcoholics (13). Hemodynamic alterations in the microcirculation of the liver and splanchnic bed of cirrhotic animals are demonstrated by the greatly enhanced liver and intestinal lymph flows in the present study. These flows imply enhanced capillary filtration, presumably due to raised sinusoidal and splanchnic capillary pressures. Acute elevations of hepatic venous or portal venous pressure in several species have been shown to enhance filtration from the liver and (18-201. splanchnic microcirculations, respectively In the present study, an acute increase in portal venous pressure in control animals caused a marked
July
MICROVASCULAR
1984
INTESTINE
EXCHANGE
169
LIVER
I OOr
IOO-
0
Control
?? Acute
Portal
Hypertension
??Cirrhosis
.
75-
.
50 -
25-
. . /
??? ?
? ?
? ?
@LlO,-
0 Portal
Pressure
I
IO
I
20
Portal
fcmli20)
30
Pressure
fcmti,O)
Relationship between intestinal and liver lymph flow and portal pressure in control and cirrhotic animals. In the left-hand panel the broken line defines the relationship between intestinal lymph flow and portal pressure in control animals with and without acute elevation of portal pressure, whereas the solid line gives the relationship between intestinal lymph flow and portal pressure in chronic portal hypertension (cirrhotic rats).
INTESTINE
0
Control
??
Acute Portal
nyPertenSlOn
?? Cirrhosis
60-
60-
. 40-
.
.
zo-
.
Lymph (mllmin
Figure
5
Flow x
1OOg)
Relationship between lymph/plasma protein concentration ratio panel) and control rats with and without acute portal hypertension not been included in the analysis.
I
I
.30
60 Lymph (mllmin
Flow x
and flow of intestinal lymph (right-hand panel). The ringed
loop)
in cirrhotic rats (left-hand point (ieft-hand panel] has
170
GASTROENTEROLOGY
BARROWMAN AND GRANGER
LIVER
the washdown of intestinal lymph protein is shown as a function of lymph flow in cirrhotic animals and a similar curve relates C&p and lymph flow in control animals with acute elevation of portal venous pressure. Although there are insufficient data at high lymph flow to calculate an accurate ad in these studies, the results obtained under the two conditions provide a range of Ud values between 0.77 and 0.90. These values can be compared with the mean value of 0.88 obtained in an in situ autoperfused segment of rat small intestine (23). Because Ud provides an estimate of capillary permeability to macromolecules and inasmuch as the values for the rat intestine in chronic portal hypertension and in the normal intestine with acute portal hypertension appear to be similar, the present study suggests that no substantial alteration in intestinal capillary permeability occurs as a result of sustained portal hypertension. In contrast to the intestine, the response of liver lymph total protein concentration in normal animals to acute hepatic venous pressure elevation is a rise
0 Control ??
Cirrhotics
0 .
.
40-
20 -
0
I
I
I
.30
60 Lymph
(mllmin
Vol. 87. No. 1
J
.90 LIVER
Flow x 1OOg)
Figure 6. Relationship between lymph/plasma protein concentration ratio and flow of liver lymph in control and cirrhotic rats. The ringed point has not been included in the analysis.
rise in intestinal lymph flow. It was not possible, however, to achieve partial suprahepatic venous occlusion in the animals we studied without seriously compromising the entire circulation. In the absence of compensatory changes in chronic portal hypertension, one should expect a high capillary filtration rate in the intestine with the possibility of interstitial edema and filtration secretion. It has been proposed that these effects of sustained portal hypertension may be offset by enhanced lymphatic fluid clearance or by a change in capillary permeability, or both (7). Although our data are limited in number we could not distinguish any difference in the capillary filtration (lymph flow) response to acute portal hypertension in normal animals as compared with that in animals with chronic portal hypertension. In the intestine and many other tissues, elevation of venous pressure increases capillary filtration rate (lymph flow) and leads to a concomitant “washdown” in CJCp until this value reaches a filtrationindependent minimum (12,21,22). The osmotic reflection coefficient (odd) can be calculated from this minimum CLICP value as 1 - CL/CP (12). In Figure 5,
l.OO.60-
.06-
,011
, 20
I
I
I
I
40
60
60
100
SOLUTE
Figure
7.
RADIUS
1
120
140
(a)
A plot of 1 - CT~ versus solute radius for the nine plasma protein fractions studied for liver lymph in control and cirrhotic rats. The equivalent pore curves were predicted using the equation of Drake and Davis (1978). For control animals only one curve could be fitted. For cirrhotic rats subtraction of the large pore curve from 1 - CT~values for smaller values represented by open squares fitted the small pore curve.
solutes gives the to which can be
MICROVASCULAR
July 1964
toward a value of unity at high lymph flows from basal values of -0.8 (18,21). There are two possible explanations for this phenomenon. One is that the interstitial matrix lying between the sinusoid and initial lymphatic is the limiting restrictive barrier to blood-lymph movement of macromolecules at normal hydration states. With hepatic venous pressure elevation, the interstitial matrix is expanded due to enhanced sinusoidal filtration and the selectivity normally afforded by the matrix is lost (24).An alternative explanation is that there are two contributions to hepatic lymph, namely, the filtrates from the highly permeable sinusoids and the peribiliary plexus of continuous and consequently less permeable capillaries. With elevation of hepatic venous pressure, a relatively greater contribution of hepatic lymph may come from the sinusoidal vessels thereby causing a rise in CJCp. In contrast to studies in normal liver, C&e fell with high liver lymph flow (Figure 6) in cirrhotic animals, with a minimum value of 0.43 being recorded. Thus, the permeability characteristics of the liver microcirculation have dramatically changed to resemble those of continuous or fenestrated capillary beds. Such an alteration in filtration characteristics has also been observed in human cirrhosis (5) and the structural counterpart is presumably the capillarization referred to previously. Multiple indicator dilution studies in human cirrhosis have also yielded data consistent with such a change in the permeability characteristics of the liver microcirculation (6). The pore-stripping analysis illustrated in Figure 7 shows that the data obtained from control livers can only be described in terms of a large pore system of 360 A radius. The cirrhotic liver data can be fitted to a two pore system of 70 A and 230 A radius. The control liver data presumably describe the porosity of the principal restrictive barrier between blood and lymph in the normal liver, which may reside in the interstitium, i.e., the space of Disse, as morphologic studies fail to demonstrate any potential sieve at the level of the endothelium. The pore sizes obtained for the cirrhotic liver, however, are in the range of values normally reported for continuous capillaries (22). The structural equivalents of the two pore populations predicted for the cirrhotic liver are not readily apparent from our study; however, either a proliferation of continuous capillaries (6)or deposition of a basement membrane (a), or both, could provide the morphologic correlate for the two pore dimensions. The results of the present study provide further functional evidence in support of an important change in the microvascular permeability of the liver in cirrhosis. These changes, by providing a barrier to diffusion, will alter the exchange of solutes between
EXCHANGE
171
sinusoidal plasma and the hepatocytes. In this connection, the demonstration of a correlation between the extent of collagenization in the space of Disse and biochemical and clinical parameters of hepatocyte dysfunction in alcoholic liver disease is of interest (13). It has also been proposed that collagenization of the space of Disse, by reducing sinusoidal diameter, can lead to portal hypertension through an increase in resistance to sinusoidal blood flow (13). Diminished sinusoidal permeability to fluid and solutes would also contribute in that the decompression of the hypertensive portal venous system by transcapsular fluid filtration from the liver might be compromised by creating a greater degree of hypertension in the splanchnic bed.
References Barrowman JA, Granger DN. Hepatic lymph. In: WW Lautt, ed. Hepatic circulation in health and disease. New York: Raven, 1981;137-52. 2. Goresky CA. A linear method for determining liver sinusoidal and extravascular volumes. Am J Physiol 1963;204:626-40. 3. Wisse E. An electron microscopic study of the fenestrated lining of rat liver sinusoids. J Ultrastruct Res 1970;31:125--50. 4. Schaffner F, Popper H. Capillarization of hepatic sinusoids in man. Gastroenterology 1963;44:239-42. 5. Witte MH, Witte CL, Dumont AE. Estimated net transcapillary water and protein flux in the liver and intestine in patients with portal hypertension from hepatic cirrhosis. Gastroenter010gy i9ai;ao:265-72. 6. Huet P-M, Goresky DA, Villeneuve J-P. et al. Assessment of liver microcirculation in human cirrhosis. J Clin Invest i982;70:1234-44, 7. Lifson N. Fluid secretion and hydrostatic pressure relationships in the small intestine. In: Binder HJ, ed. Mechanisms of intestinal secretion. New York: Alan R Liss, 1979:249-61. a. McLean EK, McLean AEM, Sutton PM. Instant cirrhosis: an improved method for producing cirrhosis of the liver in rats by simultaneous administration of carbon tetrachloride and phenobarbitone. Br J Exp Path01 1969;50:502-6. 9. Chatambra K, Proctor E. Optimal timing of phenobarbitonei carbon tetrachloride in production of cirrhosis of the liver in rats. J Physiol (Lond) 1980;312:22P. 10. Tso P, Ragland JB, Sabesin SM. Isolation and characterization of lipoprotein of density ~1.006 g/ml from rat hepatic lymph. J Lipid Res 1983;24:810-20. 11. Bollman JL. Cain JC, Grindlay JH. Techniques for collection of lymph from liver, small intestine or thoracic duct of the rat. J Lab Clin Med 1948;33:1340-52, 12. Granger DN, Taylor AE. Permeability of intestinal capillaries to endogenous macromolecules. Am J Physiol 1980;238: 1.
H457-64. 13. Orrego H, Medline A, Blendis LM. et al. Collaginisation
of the Disse space in alcoholic liver disease. Gut 1979;20:673-9. 14. Renkin EM, Watson PD, Sloop CH. et al. Transport pathways for fluid and large molecules in microvascular endothelium of the dog paw. Microvasc Res 1977;14:205-14. 15. Drake R, Davis E. The corrected equation for the calculation of reflection coefficients. Microvasc Res 1978;15:259-60. 16. Cameron GR, Karunaratne WAE. Carbon tetrachloride cirrhosis in relation to liver regeneration. J Path01 Bacterial 1936; 42:1-21.
172
BARROWMAN AND GRANGER
17. Tamayo RP. Is cirrhosis of the liver experimentally produced by Ccl, an adequate model of human cirrhosis? Hepatology 1983;3:112-20. 18. Granger DN, Miller T, Allen R, et al. Permselectivity of the liver blood-lymph barrier to endogenous macromolecules. Gastroenterology 1979;77:103-9. 19. Laine G, Hall JT, Lame SH, et al. Transsinusoidal fluid dynamics in canine liver during venous hypertension. Circ Res 1979;45:317-23. 20. Mortillaro NA, Taylor AE. Interaction of capillary and tissue forces in the cat intestine. Circ Res 1976;39:348-58. 21. Witte CL, Witte MH, Kintner K, et al. Colloid osmotic pres-
GASTROENTEROLOGY
Vol. 87, No. 1
sure in hepatic cirrhosis and experimental ascites. Surg Gynecol Obstet 1971;133:65-71. 22. Granger DN, Perry MA. Permeability characteristics of the microcirculation. in: Mortillaro NA, ed. The physiology and pharmacology of the microcirculation, Vol. 1. New York: Academic, 1983:157-208. 23. Anzueto L, Benoit JN, Granger DN. A rat model for studying the intestinal circulation. Am J Physiol 1984;246:G56-61. 24. Barrowman JA, Perry MA, Kvietys PR, et al. Exclusion phenomenon in the liver interstitium. Am J Physiol 1982: 243:G410-4.
1984:87:165-72
Effects of Experimental Cirrhosis on Splanchnic Microvascular Fluid and Solute Exchange in the Rat JAMES A. BARROWMAN
and D. NEIL GRANGER
Faculty of Medicine, Memorial University of Newfoundland. St. John’s, Newfoundland and Department of Physiology, College of Medicine, University of South Alabama, Mobile, Alabama
In human cirrhosis, there is evidence that there are considerable alterations in fluid and solute exchange in the hepatic and intestinal microcirculations. Experimental cirrhosis was induced in rats b) using oral phenobarbitone and carbon tetrachloride inhalation over an 8-wk period. Portal venous pressure, hepatic and intestinal lymph flows, and lymph and plasma protein concentrations were measured. Liver samples were obtained for histologic examination. Portal venous pressure increased from a normal value (control animals) of9.0 (6.3-23.1) cmHzO to 17.9 (9.0-29.0) cmH,O in cirrhotic rats. There was a strong correlation between the degree of fibrotic change on histology and portal venous pressure. Lymph flows from the intestine and liver in cirrhotic animals were increased threefold and 30-fold, respectively, over values obtained from control animals. There was a good correlation between intestinal and liver lymph flows and portal venous pressure. Analysis of lymph/plasma protein concentration ratios at various lymph flows suggests that capillary permeability in the small intestine during sustained portal hypertension is comparable to that in normal animals. However, the highly permeable blood-lymph barrier of the normal liver becomes markedly restrictive in cirrhotic animals. There is considerable evidence in the literature that the exchange vessels (sinusoids) in normal liver are highly permeable to macromolecules (1,~). The permeability characteristics of these microvessels can be attributed to their discontinuous nature, i.e., they Received September 26, 1983. Accepted February 13, 1984. Address requests for reprints to: Dr. D. Neil Granger, University of South Alabama. Department of Physiology, College of Medicine. MSB 3024, Mobile, Alabama 36688. This study was supported by funds from the Faculty of Medicine. Memorial University of Newfoundland. Dr. Granger is a recipient of a Research Career Development Award from the National Heart, Lung and Blood Institute (HL00816). 0 1984 by the American Gastroenterological Association 0016-5085/84:$3.00
lack a basement membrane and have wide interendothelial apertures (1000-5000 A in diameter) (3). Cirrhosis, by contrast, is characterized by striking structural alterations in the sinusoid, namely there is deposition of a basement membrane and the vessels take on the appearance of continuous capillaries (4). The functional correlate of these vascular alterations in human cirrhosis is evidence of reduced macromolecular permeability from hepatic lymph protein measurements (5) and indicator diffusion studies (6). A major aim of the present study was to quantify the permeability characteristics of the hepatic microcirculation in cirrhosis in terms of “pore” dimensions. To study this, hepatic lymph protein composition was examined in normal rats and animals with carbon tetrachloride-induced cirrhosis. Cirrhosis, through portal hypertension, also has important effects on intestinal microvascular fluid and solute exchange. It has been suggested that the slow development of chronic portal hypertension leads to compensatory changes that oppose interstitial edema formation and filtration secretion, which are observed in acute portal hypertension (7). Two of these changes, namely increased lymph flow and decreased capillary permeability, were examined in the present study using the carbon tetrachloride model.
Methods The studies were performed using 37 male SpagueDawley rats fed a standard rat diet. Ten animals served as controls and in 27 rats, cirrhosis was induced by a modification of the method of McLean et al. (8). This procedure consisted of the addition of 0.5 glL of sodium phenobarbitone to all drinking water and regular periods of inhalation of carbon tetrachloride in a 150-L capacity chamber. The
Abbreviations used in this paper: C,, lymph concentration; CI+ plasma concentration; ad, capillary osmotic reflection coefficient.
166 BARROWMAN AND GRANGER
GASTROENTEROLOGY Vol. 87, No. I
Figure 1. Histologic appearance of control (A] and cirrhotic livers (B). Masson trichrome stain. rats were given phenobarbitone 1 wk before commencement of gassing (9). This program was carried out over a period of 8 wk. Cirrhosis could be demonstrated in all treated animals both macroscopically and microscopically (Figure 1). Ascites was present in 50% of the cirrhotic animals.
Experimental
Procedures
The rats were anesthetized with ether and the abdomen opened with a midline incision. The main hepatic and intestinal lymph trunks were identified, dissected, and cannulated with clear vinyl tubing (OD 0.8 mm, Dural Plastic Engineering, Dural, Australia] as previously described (10,ll).Lymph flows were measured by following the movement of lymph in calibrated micropipettes. Lymph and plasma samples were taken for protein concentration measurement by refractometry. The plasma sample was obtained from the portal vein at the end of each study. Lymph and plasma samples were stored frozen for subsequent electrophoresis on polyacrylamide gradient gels (12). Portal venous pressure was measured by inserting a ZOgauge needle, connected to a water manometer, into the vein and securing it with a drop of cyanoacrylate glue.
At the end of each experiment, body weight was recorded and the liver and small intestine were excised and weighed. A sample of liver was stored in Bouin’s fixative, embedded in paraffin, sectioned, and stained with Masson’s trichrome stain. Using the histologic criteria described by Orrego et al. (13), specimens were scored blindly for necrosis, fibrosis, and fat on their scale of O-3. Protocols Controls. After laparotomy, basal lymph flows from the intestine and liver were measured and samples were taken for protein concentration in lymph and plasma. Portal pressure was then measured. Acute portal hypertension. In 7 of the 10 control animals, after obtaining basal data, a snare was placed around the portal vein above the needle used to measure portal pressure. The portal vein was partially occluded to achieve a portal venous pressure between 20 and 30 cmH20, values comparable to the upper range of pressures recorded in cirrhotic animals. Intestinal lymph flow and protein concentration were allowed to reach a new steady state before measurements and samples were taken. Cirrhotics. In this group of animals the measurements were the same as those made in the control group.
July
MICROVASCULAR
1984
Table
I. Carbon
Tetrachloride-Induced Body
Cirrhosis
weight
Liver
(g)
Controls
Cirrhotics (n = 27) p value ” Mean
+ SE. ” Values
weight
11.9 20.46 (9.2-13.3) 15.9 20.76 (9.6-25.9) co.01
in parentheses
represent
values for ud were obtained for nine plasma protein fractions identified on gradient gel electrophoresis. The molecular radius of each protein fraction was determined by extrapolation from curves obtained with standard propore radii for the hepatic microcircuteins (12). Equivalent graphical analysis of 1 - CT~was plotted
against solute radius. The larger molecules were fit to a “large pore” line generated using the equation of Drake and Davis (15). When possible, a small pore curve was derived by subtracting the 1 - 0;~ values of the large pore curve from points lying above this curve. For animals, the gd values for the various plasma fractions were derived from basal values of C&p, ing maximal sieving at normal filtration rates.
control protein assum-
Statistics Standard statistical analyses were applied to the data [mean * SEM, unpaired Student’s t-test). Data were also analyzed using a linear regression analysis that mini-
25
I
i i
r=12
t
Figure
IL It=7 2
Plasma protein concen-
Portal pressure (cmH,O)
tration (g (5))
9.0 fO.SU (6.3-13.1) 17.9 21.0 (9-29) co.01
6.07 to.06 (5.3-6.5) 5.68 20.15 (5.0-6.5) 0.01
range.
In all experiments, lymph to plasma protein concentration ratios (C&p) were calculated. From the relationship between CL/Cp and lymph flow, the capillary osmotic reflection coefficient (ud) was assumed to equal 1 - CL/C,J at high rates of lymph flow (12). In the liver,
through the nine proteins,
Intestine weight (g) 8.6 kO.57 (7.0-12.4) 9.6 “0.44 (6.8-13.9) 0.10
Calculations
lation were obtained Renkin (14). For the
167
in Rats
(81
365.8” 28.7 (321-414)b 365.7 56.9 (287-417) 0.50
(n = 10)
EXCHANGE
n=, 3
2. Relationship between extent of fibrosis and portal venous pressure (mean 2 SE] in control and cirrhotic rats. Number of animals in each group is indicated.
mizes errors variables.
in both
the
dependent
and
independent
Results Table 1 lists basic data from control and cirrhotic rats. Liver weight was significantly increased (34%) and portal pressure was doubled in cirrhotic animals. Portal pressures as high as 29 cmH,O were recorded in cirrhotic animals. Histologic examination of liver specimens clearly distinguished controls from experimental animals. Within the experimental group, however, there was a wide range in the degree of pathological change in terms of fibrosis, necrosis, and fatty infiltration. Of these three parameters, fibrosis showed the best correlation (r = 0.78, p < 0.01) with portal venous pressure [Figure 2). The correlation coefficient for the relationships between portal venous pressure and necrosis was 0.71 (p < O.Ol), and for fatty infiltration it was 0.55 (p < 0.01). Figure 3 illustrates the effects of cirrhosis on hepatic and intestinal lymph flows and lymph to plasma total protein concentration ratios (C&). Lymph flows in the liver and intestine of control animals were 0.012 -4 0.002 and 0.127 * 0.016 ml/ min x 100 g, respectively. In cirrhotic animals these values increased by 2%fold and threefold, respectively. Lymph/plasma protein concentration ratios for liver and intestinal lymph were 0.87 + 0.03 and 0.50 t 0.03, respectively; these values fell to 0.70 c 0.03 and 0.37 * 0.05, respectively, in cirrhotic animals. The relationships between intestinal and liver lymph flows and portal venous pressure in control and cirrhotic animals are shown in Figure 4. In addition, the effects of acute portal hypertension on intestinal lymph flow in control animals are illustrated. In all instances there was a significant (p < 0.01) linear correlation (r = 0.72-0.92) between lymph flows and portal pressure. There was no significant difference in the lymph flow-portal pres-
168 BARROWMAN AND GRANGER
[I[7
Controls
0
Cirrhotics
GASTROENTEROLOGY
*
I
*
-L T
_m 0 ._
5
a : ._
0.60 -
2 L
za c? :
0:60 -
.5 Q 5 t
0.40 -
: : rr i
0.20 -
E” 5 OLiver
Intestine
Figure 3. Hepatic and intestinal lymph flows sod lymph/plasma protein concentration ratios in control and cirrhotic livers (mean ? SE). An asterisk (*) indicates a significant difference between control and cirrhotic animals, p < 0.05.
sure relationship in acute and chronic (cirrhosis] portal hypertension. Figure 5 illustrates the relationships between lymph to plasma total protein concentration ratio (CJCp) and lymph flow in the intestine of cirrhotic and control animals, with and without acute portal hypertension. In both groups, CL/CP decreased with increasing lymph flow to low flow-independent values ranging between 0.10 and 0.23. The relationship between C&p and hepatic lymph flow in control and cirrhotic animals is shown in Figure 6. As observed in the small bowel (Figure 5), C&P decreases with increasing lymph flow with C&p decreasing to as low as 0.43.
Vol. 87, No. 1
Figure 7 illustrates the graphical analysis for pore sizes based on C&r data for nine plasma protein fractions in the control and cirrhotic liver. In control animals, the sieving data would only fit a single pore size, i.e., 360 A radius. Whereas, in cirrhosis, sieving across the hepatic blood-lymph barrier could be described by two pore populations, a small pore system df 70 A radius and a system of large pores with 230 A radius.
Discussion Since the first report in 1936 (16)that repeated administration of carbon tetrachloride to rats results in cirrhosis, this method has been used extensively to create a model of human cirrhosis. Certain differences between carbon tetrachlorideinduced cirrhosis and the human disease are recognized, including the topography of the fibrosis and the collagen types involved. This model, nevertheless, has many similarities to human alcoholic cirrhosis, the type for which most histologic and biochemical information is available (17). Among these similarities is the occurrence of portal hypertension, and, in a proportion of animals, ascites. In the present study, the degree of portal hypertension that developed correlated well with the extent of histologic change in the liver, notably the amount of fibrosis. Structural changes in liver architecture and its vascular tree, by increasing resistance to blood flow at the sinusoidal or postsinusoidal level, are considered an important factor in the pathogenesis of portal hypertension. The other important contribution is an increase in splanchnic arterial inflow (17). At the microvascular level in the liver, cirrhosis is associated with collagenization of the space of Disse and the development of a basement membrane beneath the sinusoidal endothelium, i.e., “capillarization” of the sinusoid (4). This collagenization of the space of Disse may be partly responsible for the development of portal hypertension inasmuch as an association has been demonstrated between Disse space collagenization and portal hypertension in alcoholics (13). Hemodynamic alterations in the microcirculation of the liver and splanchnic bed of cirrhotic animals are demonstrated by the greatly enhanced liver and intestinal lymph flows in the present study. These flows imply enhanced capillary filtration, presumably due to raised sinusoidal and splanchnic capillary pressures. Acute elevations of hepatic venous or portal venous pressure in several species have been shown to enhance filtration from the liver and (18-201. splanchnic microcirculations, respectively In the present study, an acute increase in portal venous pressure in control animals caused a marked
July
MICROVASCULAR
1984
INTESTINE
EXCHANGE
169
LIVER
I OOr
IOO-
0
Control
?? Acute
Portal
Hypertension
??Cirrhosis
.
75-
.
50 -
25-
. . /
??? ?
? ?
? ?
@LlO,-
0 Portal
Pressure
I
IO
I
20
Portal
fcmli20)
30
Pressure
fcmti,O)
Relationship between intestinal and liver lymph flow and portal pressure in control and cirrhotic animals. In the left-hand panel the broken line defines the relationship between intestinal lymph flow and portal pressure in control animals with and without acute elevation of portal pressure, whereas the solid line gives the relationship between intestinal lymph flow and portal pressure in chronic portal hypertension (cirrhotic rats).
INTESTINE
0
Control
??
Acute Portal
nyPertenSlOn
?? Cirrhosis
60-
60-
. 40-
.
.
zo-
.
Lymph (mllmin
Figure
5
Flow x
1OOg)
Relationship between lymph/plasma protein concentration ratio panel) and control rats with and without acute portal hypertension not been included in the analysis.
I
I
.30
60 Lymph (mllmin
Flow x
and flow of intestinal lymph (right-hand panel). The ringed
loop)
in cirrhotic rats (left-hand point (ieft-hand panel] has
170
GASTROENTEROLOGY
BARROWMAN AND GRANGER
LIVER
the washdown of intestinal lymph protein is shown as a function of lymph flow in cirrhotic animals and a similar curve relates C&p and lymph flow in control animals with acute elevation of portal venous pressure. Although there are insufficient data at high lymph flow to calculate an accurate ad in these studies, the results obtained under the two conditions provide a range of Ud values between 0.77 and 0.90. These values can be compared with the mean value of 0.88 obtained in an in situ autoperfused segment of rat small intestine (23). Because Ud provides an estimate of capillary permeability to macromolecules and inasmuch as the values for the rat intestine in chronic portal hypertension and in the normal intestine with acute portal hypertension appear to be similar, the present study suggests that no substantial alteration in intestinal capillary permeability occurs as a result of sustained portal hypertension. In contrast to the intestine, the response of liver lymph total protein concentration in normal animals to acute hepatic venous pressure elevation is a rise
0 Control ??
Cirrhotics
0 .
.
40-
20 -
0
I
I
I
.30
60 Lymph
(mllmin
Vol. 87. No. 1
J
.90 LIVER
Flow x 1OOg)
Figure 6. Relationship between lymph/plasma protein concentration ratio and flow of liver lymph in control and cirrhotic rats. The ringed point has not been included in the analysis.
rise in intestinal lymph flow. It was not possible, however, to achieve partial suprahepatic venous occlusion in the animals we studied without seriously compromising the entire circulation. In the absence of compensatory changes in chronic portal hypertension, one should expect a high capillary filtration rate in the intestine with the possibility of interstitial edema and filtration secretion. It has been proposed that these effects of sustained portal hypertension may be offset by enhanced lymphatic fluid clearance or by a change in capillary permeability, or both (7). Although our data are limited in number we could not distinguish any difference in the capillary filtration (lymph flow) response to acute portal hypertension in normal animals as compared with that in animals with chronic portal hypertension. In the intestine and many other tissues, elevation of venous pressure increases capillary filtration rate (lymph flow) and leads to a concomitant “washdown” in CJCp until this value reaches a filtrationindependent minimum (12,21,22). The osmotic reflection coefficient (odd) can be calculated from this minimum CLICP value as 1 - CL/CP (12). In Figure 5,
l.OO.60-
.06-
,011
, 20
I
I
I
I
40
60
60
100
SOLUTE
Figure
7.
RADIUS
1
120
140
(a)
A plot of 1 - CT~ versus solute radius for the nine plasma protein fractions studied for liver lymph in control and cirrhotic rats. The equivalent pore curves were predicted using the equation of Drake and Davis (1978). For control animals only one curve could be fitted. For cirrhotic rats subtraction of the large pore curve from 1 - CT~values for smaller values represented by open squares fitted the small pore curve.
solutes gives the to which can be
MICROVASCULAR
July 1964
toward a value of unity at high lymph flows from basal values of -0.8 (18,21). There are two possible explanations for this phenomenon. One is that the interstitial matrix lying between the sinusoid and initial lymphatic is the limiting restrictive barrier to blood-lymph movement of macromolecules at normal hydration states. With hepatic venous pressure elevation, the interstitial matrix is expanded due to enhanced sinusoidal filtration and the selectivity normally afforded by the matrix is lost (24).An alternative explanation is that there are two contributions to hepatic lymph, namely, the filtrates from the highly permeable sinusoids and the peribiliary plexus of continuous and consequently less permeable capillaries. With elevation of hepatic venous pressure, a relatively greater contribution of hepatic lymph may come from the sinusoidal vessels thereby causing a rise in CJCp. In contrast to studies in normal liver, C&e fell with high liver lymph flow (Figure 6) in cirrhotic animals, with a minimum value of 0.43 being recorded. Thus, the permeability characteristics of the liver microcirculation have dramatically changed to resemble those of continuous or fenestrated capillary beds. Such an alteration in filtration characteristics has also been observed in human cirrhosis (5) and the structural counterpart is presumably the capillarization referred to previously. Multiple indicator dilution studies in human cirrhosis have also yielded data consistent with such a change in the permeability characteristics of the liver microcirculation (6). The pore-stripping analysis illustrated in Figure 7 shows that the data obtained from control livers can only be described in terms of a large pore system of 360 A radius. The cirrhotic liver data can be fitted to a two pore system of 70 A and 230 A radius. The control liver data presumably describe the porosity of the principal restrictive barrier between blood and lymph in the normal liver, which may reside in the interstitium, i.e., the space of Disse, as morphologic studies fail to demonstrate any potential sieve at the level of the endothelium. The pore sizes obtained for the cirrhotic liver, however, are in the range of values normally reported for continuous capillaries (22). The structural equivalents of the two pore populations predicted for the cirrhotic liver are not readily apparent from our study; however, either a proliferation of continuous capillaries (6)or deposition of a basement membrane (a), or both, could provide the morphologic correlate for the two pore dimensions. The results of the present study provide further functional evidence in support of an important change in the microvascular permeability of the liver in cirrhosis. These changes, by providing a barrier to diffusion, will alter the exchange of solutes between
EXCHANGE
171
sinusoidal plasma and the hepatocytes. In this connection, the demonstration of a correlation between the extent of collagenization in the space of Disse and biochemical and clinical parameters of hepatocyte dysfunction in alcoholic liver disease is of interest (13). It has also been proposed that collagenization of the space of Disse, by reducing sinusoidal diameter, can lead to portal hypertension through an increase in resistance to sinusoidal blood flow (13). Diminished sinusoidal permeability to fluid and solutes would also contribute in that the decompression of the hypertensive portal venous system by transcapsular fluid filtration from the liver might be compromised by creating a greater degree of hypertension in the splanchnic bed.
References Barrowman JA, Granger DN. Hepatic lymph. In: WW Lautt, ed. Hepatic circulation in health and disease. New York: Raven, 1981;137-52. 2. Goresky CA. A linear method for determining liver sinusoidal and extravascular volumes. Am J Physiol 1963;204:626-40. 3. Wisse E. An electron microscopic study of the fenestrated lining of rat liver sinusoids. J Ultrastruct Res 1970;31:125--50. 4. Schaffner F, Popper H. Capillarization of hepatic sinusoids in man. Gastroenterology 1963;44:239-42. 5. Witte MH, Witte CL, Dumont AE. Estimated net transcapillary water and protein flux in the liver and intestine in patients with portal hypertension from hepatic cirrhosis. Gastroenter010gy i9ai;ao:265-72. 6. Huet P-M, Goresky DA, Villeneuve J-P. et al. Assessment of liver microcirculation in human cirrhosis. J Clin Invest i982;70:1234-44, 7. Lifson N. Fluid secretion and hydrostatic pressure relationships in the small intestine. In: Binder HJ, ed. Mechanisms of intestinal secretion. New York: Alan R Liss, 1979:249-61. a. McLean EK, McLean AEM, Sutton PM. Instant cirrhosis: an improved method for producing cirrhosis of the liver in rats by simultaneous administration of carbon tetrachloride and phenobarbitone. Br J Exp Path01 1969;50:502-6. 9. Chatambra K, Proctor E. Optimal timing of phenobarbitonei carbon tetrachloride in production of cirrhosis of the liver in rats. J Physiol (Lond) 1980;312:22P. 10. Tso P, Ragland JB, Sabesin SM. Isolation and characterization of lipoprotein of density ~1.006 g/ml from rat hepatic lymph. J Lipid Res 1983;24:810-20. 11. Bollman JL. Cain JC, Grindlay JH. Techniques for collection of lymph from liver, small intestine or thoracic duct of the rat. J Lab Clin Med 1948;33:1340-52, 12. Granger DN, Taylor AE. Permeability of intestinal capillaries to endogenous macromolecules. Am J Physiol 1980;238: 1.
H457-64. 13. Orrego H, Medline A, Blendis LM. et al. Collaginisation
of the Disse space in alcoholic liver disease. Gut 1979;20:673-9. 14. Renkin EM, Watson PD, Sloop CH. et al. Transport pathways for fluid and large molecules in microvascular endothelium of the dog paw. Microvasc Res 1977;14:205-14. 15. Drake R, Davis E. The corrected equation for the calculation of reflection coefficients. Microvasc Res 1978;15:259-60. 16. Cameron GR, Karunaratne WAE. Carbon tetrachloride cirrhosis in relation to liver regeneration. J Path01 Bacterial 1936; 42:1-21.
172
BARROWMAN AND GRANGER
17. Tamayo RP. Is cirrhosis of the liver experimentally produced by Ccl, an adequate model of human cirrhosis? Hepatology 1983;3:112-20. 18. Granger DN, Miller T, Allen R, et al. Permselectivity of the liver blood-lymph barrier to endogenous macromolecules. Gastroenterology 1979;77:103-9. 19. Laine G, Hall JT, Lame SH, et al. Transsinusoidal fluid dynamics in canine liver during venous hypertension. Circ Res 1979;45:317-23. 20. Mortillaro NA, Taylor AE. Interaction of capillary and tissue forces in the cat intestine. Circ Res 1976;39:348-58. 21. Witte CL, Witte MH, Kintner K, et al. Colloid osmotic pres-
GASTROENTEROLOGY
Vol. 87, No. 1
sure in hepatic cirrhosis and experimental ascites. Surg Gynecol Obstet 1971;133:65-71. 22. Granger DN, Perry MA. Permeability characteristics of the microcirculation. in: Mortillaro NA, ed. The physiology and pharmacology of the microcirculation, Vol. 1. New York: Academic, 1983:157-208. 23. Anzueto L, Benoit JN, Granger DN. A rat model for studying the intestinal circulation. Am J Physiol 1984;246:G56-61. 24. Barrowman JA, Perry MA, Kvietys PR, et al. Exclusion phenomenon in the liver interstitium. Am J Physiol 1982: 243:G410-4.