The immunology of fibrogenesis in alcoholic liver disease

Original Articles The Immunology of Fibrogenesis in Alcoholic Liver Disease Antonio Chedid, MD; Selma Arain, MD; Ann Snyder, PhD; Philippe Mathurin, MD; Frédérique Capron, MD, PhD; Sylvie Naveau, MD ● Context.—Alcoholic liver disease in humans frequently leads to cirrhosis. Experimental models of hepatic fibrogenesis are available, but extrapolation of those findings to human ethanol-induced liver injury is difficult. Hepatic ethanol-induced fibrosis in humans has often been studied in relatively small patient populations. During the past decade, several animal models and human studies have attributed fibrogenesis in the liver to the role played by hepatocytes, Kupffer cells, endothelial cells, and especially stellate cells. Objective.—To determine the contribution of the main liver cell types to ethanol-induced fibrogenesis. For that purpose, we studied the expression of the following immunologic parameters: smooth muscle–specific a actin (SMSA), CD68, CD34, transforming growth factor b1, intercellular adhesion molecule 1, and collagen types 1 and 3. The Dako LSAB1 kit (peroxidase method) was used. Design.—We recently studied a large cohort of patients with alcoholic liver disease in France. In this cohort, we found 87 cases in which liver biopsies revealed only pericentral injury with nonpathologic portal areas. We compared cases in which the portal areas were nonpathologic with 324 patients in whom staging ranged from F0 to F3. Patients with cirrhosis (F4) were excluded from evaluation. To stage fibrosis, we used the METAVIR system. Furthermore, we selected 40 cases in which the biopsies measured at least 25 mm in length for further histochemical evaluation. Ten additional normal cases from our archives were used as controls. We divided this patient population into the following 5 groups of 10 patients each: group 1A, F0 with steatosis; group 1B, F0 without steatosis; group 2, F0 to F1, central injury; group 3, F3, fibrosis with multiple septa; and group 4, nonpathologic livers (controls). Results.—Smooth muscle–specific a actin was expressed by stellate cells, pericentrally, with increasing severity and intensity in the advanced stage of fibrosis of group 3, less intense expression was noted in group 2, and expression was practically absent in group 1 and in nonpathologic controls. CD68 was the best marker for Kupffer cells and was expressed diffusely within the lobules in all groups. Its expression correlated directly with the degree of disease severity, progressing from stage I through stage III, but was absent in nonpathologic livers. CD34 was consistently expressed by endothelial cells in the periportal areas in all groups. The expression of collagen type 1 was intense in the bands of fibrosis or bridging, while type 3 expression was poor. Transforming growth factor b1 and intercellular adhesion molecule 1 were not expressed in any group. Conclusions.—In this study, stellate cell activation (SMSA) was most intense pericentrally in the early stages and diffusely with progression to fibrosis and maximum intensity in stage III. Kupffer cell activation, as determined by CD68 expression, was intense and diffuse, while endothelial cells expressed CD34 periportally in a similar manner in all stages. Fibrogenesis in human ethanol injury is due to the activity of stellate cells, Kupffer cells, and to a lesser extent, to endothelial cells. (Arch Pathol Lab Med. 2004;128:1230–1238) C cirrhosis using names reflecting the opinion of the investigators concerning the etiology of any given case. Notable among these works were the reports of Mallory1 and Gall.2 Concerning the problem of alcoholic liver disease (ALD), the end stage of this condition, initially known as Laennec cirrhosis, was later renamed nutritional, micronodular, and portal cirrhosis. Mallory had already noted that sev- irrhosis of the liver for many years was classified based on the amount of connective tissue present and the variable size of its nodularity. Based on these parameters, the classic writers differentiated several forms of Accepted for publication July 1, 2004. From the Departments of Pathology (Drs Chedid and Arain), Medicine (Drs Chedid and Arain), Microbiology and Immunology (Dr Chedid), and Cellular and Molecular Pharmacology (Dr Snyder), Rosalind Franklin University/Chicago Medical School, North Chicago, Ill; the Department of Medicine, Antoine Béclère Hospital, Clamart, France (Drs Mathurin and Naveau); and the Department of Pathology, PitieSalpetriere Hospital/University of Paris, Paris, France (Dr Capron). The authors have no relevant financial interest in the products or companies described in this article. Reprints: Antonio Chedid, MD, Rosalind Franklin University/Chicago Medical School, 3333 Green Bay Rd, North Chicago, IL 60064 (e-mail: [email protected]). 1230 Arch Pathol Lab Med—Vol 128, November 2004 For editorial comment, see p 1212. eral stages characterized alcoholic liver injury, but the identification of acute alcoholic hepatitis as a particularly severe form of the disease preceding the stage of cirrhosis was, surprisingly enough, not sufficiently recognized until 1961–1962 by Beckett et al.3,4 Credit for the identification Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al of the perivenular nature of the early injury caused by ethanol should be given to Edmondson et al.5 Finally, an International Group in 1981 categorized the clinicopathologic manifestations of ALD,6 allowing sequential studies that led to the realization of the dismal prognosis of ALD in its various stages.7 However, a thorough, detailed evaluation of the various forms of early perivenular ethanol injury and fibrogenesis has not been undertaken. Recently, we had the opportunity to study a large population of alcoholic patients in France. These patients had undergone needle liver biopsies shortly after admission and before any form of therapy was given. Clinicopathologic evaluation and the nature of the early perivenular lesions of necrosis and fibrosis were determined by histologic and immunologic means. The results of this evaluation are reported here. MATERIALS AND METHODS The patient population was obtained from Antoine Béclère Hospital in Clamart, France. All patients had consumed ethanol daily for at least 1 year and many of them for many years. Mean alcohol consumption in this group consisted of 50 g of ethanol daily for 1 to several years, in contrast with the US criteria (80 g/d for 1 or more years) for inclusion in the Veterans Affairs Cooperative Studies.7 The following clinical and laboratory parameters were evaluated in the present study: age; sex; amount of ethanol consumed daily; number of years of ethanol consumption; prothrombin time (PT); serum bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and gglutamyltransferase (GGT) levels on admission or at the time of the biopsy; liver span in cm; splenomegaly; ascites; and esophageal varices. Only patients who were negative for hepatitis B virus, hepatitis C virus, and human immunodeficiency virus were entered into the study. The liver biopsies were evaluated for fibrosis following the METAVIR system,8 with minor modifications to allow for the separation of those patients in whom the tissue revealed only perivenular (central) injury (87 cases). Staging identified the following 5 general groups: group F0, 111 patients; group F1, 122 patients; group F2, 57 patients; group F3, 34 patients; and group F4, 179 patients. Of this total number of biopsies, in 37 cases the biopsies were considered inadequate for evaluation. A biopsy was considered inadequate if no liver tissue was found in the amount minimally necessary to establish a diagnosis with morphologic certainty. Patients with cirrhosis (F4) were excluded from this study. Thus, from this group, a population of 324 alcoholic patients with early lesions (F0 to F3) and without cirrhosis was studied. An additional 87 patients whose biopsies were characterized only by pericentral injury with normal or minimal involvement of portal areas were included. Thus, the 324 patients included in this study were divided for the purposes of final evaluation into 3 categories: category 1 (F0), category 2 (central injury only), and category 3 (F3). For the immunologic evaluation of fibrogenesis, we selected 40 cases. Ten nonpathologic livers from our archives were also investigated as normal controls. For the evaluation of fibrogenesis (after a thorough search of commercial sources), the following markers were found to best serve our purpose and were selected for investigation: Kupffer cells, CD68; Stellate (Ito) cells, smooth muscle–specific a actin (SMSA); endothelial cells, CD34; collagen, type 1 and type 3; cytokines, transforming growth factor b (TGFb1), tumor necrosis factor a (TNF-a), and intercellular adhesion molecule 1 (ICAM-1). These markers were selected because preliminary results in our laboratory showed they were best expressed in our control tissues and worked well in paraffin-embedded tissues, as advertised by the manufacturer. Thus, a total of 50 samples were divided into 5 groups, as follows: group 1A (F0), steatosis present in mild to moderate degree; group 1B (F0), steatosis absent; group 2 (F0 to F1), central injury; group 3 (F3), Arch Pathol Lab Med—Vol 128, November 2004 fibrosis with multiple septa; and group 4, normal liver biopsies (controls). Morphology Histologic examination was performed on liver biopsies fixed in formalin and embedded in paraffin. All cases were routinely studied with hematoxylin-eosin, Masson trichrome stain for collagen type 1, reticulin stain in some instances, and iron stain. Immunoperoxidase for collagen type 3 was performed with a monoclonal antibody. Furthermore, to characterize the other immunologic markers, only monoclonal antibodies were used with the single exception of TGF-b, which was investigated with a polyclonal antibody from Promega Corporation (Madison, Wis). The monoclonal antibodies for CD34, CD68, SMSA, and collagen type 3 were obtained from Dako Corporation (Carpinteria, Calif). The selection of the best marker for each cell type or parameter investigated was made after preliminary studies showed the markers chosen to give the best results in our paraffin-embedded biopsies compared to the positive controls used. Skin was used for CD34. Lymph nodes and tonsils were used for CD68, coronary branches within the heart for SMSA, and cirrhotic livers for collagen types 1 and 3. Furthermore, positive controls were selected from patients with cirrhosis to test for the expression of markers such as TGF-b, SMSA, and collagen types 1 and 3. Normal livers were used as negative controls for markers such as CD68, CD34, and collagens. Immunoperoxidase The method in use at our laboratory was the Dako LSAB1 kit, peroxidase method. This kit contains labeled streptavidin biotin. Cut sections of liver biopsy specimens (4 mm thick) were mounted on glued slides, air-dried for 10 to 15 minutes, and fixed in cold acetone. The initial step consists of 3% hydrogen peroxide treatment for 5 minutes, followed by application of the primary antibody for 30 minutes. This step was followed by the application of the link antibody for 15 minutes, and after that, application of the streptavidin peroxidase for 15 minutes. After application of the substrate chromogen diaminobenzidine for 5 minutes, the slides were counterstained with hematoxylin for 2 to 5 minutes. The diaminobenzidine stock solution was stored in 1mL aliquots at 2808C. The reagents were delivered to the sections with Dispenstirs (Becton Dickinson, Mountain View, Calif). Rinsing and washing with phosphate-buffered saline (33) were done routinely for total periods of 10 minutes following each step after the cold acetone fixation. Statistical Analyses Descriptive statistics are expressed as mean values 6 standard errors. The x2 or Fisher exact tests were used to compare qualitative variables. Comparisons between the different groups of individuals were performed with 2-way analysis of variance (ANOVA), and multiple comparisons used the Student-NewmanKeuls test. Relationships between parameters were evaluated with the Pearson correlation coefficient. Significance was defined as P , .05. RESULTS Morphologically, the following perivenular lesions were found in group 2: pericentral ballooning degeneration, pericentral steatosis, pericentral necrosis, pericentral fibrosis, and sclerosing hyaline necrosis (Figures 1 through 5). Pericentral ballooning was usually present alone, while frequently the other lesions appeared simultaneously. Mean liver span was 11.6 6 0.3 cm for group 1, 12.5 6 0.4 cm for group 2, and 13.4 6 0.5 cm for group 3. For this parameter, a significant difference was observed between group 1 and group 3, while group 2 did not differ significantly from either group 1 or group 3. Clinical parameters are depicted in Tables 1 through 3. Analysis of clinical parameters, such as alcohol drinking Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al 1231 Figure 1. Ballooning degeneration. The hepatocytes around the terminal hepatic veins (central veins) are swollen. Minimal collagen deposition is present (Masson trichrome, original magnification 320). Collagen is normally absent in this area. Even by electron microscopy, the central veins exhibit minimal collagen, and they appear to be more like dilated sinusoids. Figure 2. Pericentral steatosis. Mild collagen deposition (Masson trichrome, original magnification 340). Figure 3. Pericentral fibrosis. Central vein with collagen bands and steatosis (Masson trichrome, original magnification 3100). 1232 Arch Pathol Lab Med—Vol 128, November 2004 Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al Table 1. Alcohol Drinking History* Group 1 Men Women Total Group 2 Men Women Total Group 3 Men Women Total Age, y Ethanol Consumption, g/d Years of Ethanol Intake Ratio, g/d/y 44 6 1 (77) 47 6 3 (27) 45 6 1 (104) 145 6 12 (74) 111 6 17 (26) 136 6 10 (100) 16 6 1 (73) 15 6 3 (25) 16 6 1 (98) 15 6 2 (72) 16 6 4 (25) 16 6 2 (97) 49 6 2 (40) 53 6 2 (20) 51 6 1 (60) 136 6 14 (39) 80 6 14 (18) 118 6 11 (57) 21 6 2 (37) 14 6 3 (18) 19 6 2 (55) 11 6 2 (37) 22 6 6 (16) 14 6 3 (53) 49 6 2 (21) 47 6 4 (12) 48 6 2 (33) 109 6 9 (19) 124 6 26 (9) 110 6 11 (28) 23 6 3 (19) 13 6 3 (9) 20 6 2 (28) 963 (19) 14 6 4 (8) 11 6 3 (27) P Values 2-Way analysis of variance Group Gender Group 3 gender Multiple Group Group Group .03† .36 .54 comparisons for significant differences 1 vs group 2 .02† 1 vs group 3 .38 2 vs group 3 .21 .39 .06 .39 .57 .006† .13 .54 .13 .35 . . .‡ ... ... ... ... ... ... ... ... * Values are presented as mean 6 SEM of (n) subjects. † Indicates P values of differences that are statistically significant. ‡ Ellipses indicate data not computed. history, revealed statistical significance in terms of age between groups 1 and 2 and in the amount of ethanol consumption between men and women (Table 1). Other clinical data were examined by 2-way ANOVA (Table 2) and revealed statistically significant differences in PT between groups 1 and 2, groups 1 and 3, and groups 2 and 3. Total bilirubin was significantly different between groups 1 and 2 and groups 1 and 3. The difference in AST levels was statistically significant only between groups 1 and 3, while ALT did not exhibit any significant differences. Finally, GGT was significantly different between groups 1 and 2 and between groups 1 and 3. The 2-way ANOVA was performed to detect any sex differences. A significant difference was observed between sexes; the more severe the degree of fibrosis, the lesser the duration in years of ethanol consumption by women in relation to men, that is, women ingested much less ethanol to achieve the same degree of fibrosis as the men. The Pearson correlation coefficients (Table 3) revealed statistically significant positive correlations between age, grams per day of ethanol consumption, number of years, and the ratio of grams/day/number of years of alcohol consumption (P , .001). Prothrombin time correlated with total serum bilirubin (P , .001) and GGT levels (P 5 .005). Furthermore, serum bilirubin correlated with AST (P 5 .01) and GGT levels (P , .001). Finally, GGT levels correlated with PT (P 5 .005), total bilirubin (P , .001), AST (P , .001), and ALT (P 5 .002). The pairs of values with positive correlation coefficients and P values less than .05 tended to increase together. For the pairs with negative correlation coefficients and P values less than .05, one variable tended to decrease while the other increased. ← Figure 4. Pericentral necrosis. Ballooning of hepatocytes with moderate fibrosis (‘‘chicken-wire’’) and some inflammatory cells (Masson trichrome, original magnification 3400). Figure 5. Sclerosing hyaline necrosis. Fibrosis, steatosis, and acute inflammation (Masson trichrome, original magnification 3400). Figure 6. CD68 expression by Kupffer cells. A central area extending into the sinusoids (CD68 immunoperoxidase, original magnification 3100). Figure 7. Endothelial periportal expression. CD34 is expressed in the endothelial cells (CD34 immunoperoxidase, original magnification 3100). Figure 8. Pericentral ballooning, stage F1. A liver lobule expressing smooth muscle–specific a actin (SMSA) during an early stage of ballooning with F1 (SMSA immunoperoxidase, original magnification 3100). Arch Pathol Lab Med—Vol 128, November 2004 Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al 1233 Table 2. Clinical Data* Group 1 Men Women Total Group 2 Men Women Total Group 3 Men Women Total PT Total Bilirubin AST ALT GGT 97 6 1 (77) 95 6 3 (27) 97 6 1 (104) 14 6 1 (77) 10 6 1 (27) 13 6 1 (104) 36 6 5 (77) 27 6 5 (27) 33 6 4 (104) 35 6 5 (77) 25 6 4 (27) 33 6 4 (104) 125 6 14 (77) 112 6 24 (27) 122 6 12 (104) 95 6 2 (40) 89 6 3 (20) 93 6 2 (60) 19 6 2 (40) 32 6 12 (20) 23 6 4 (60) 44 6 5 (40) 42 6 7 (20) 43 6 4 (60) 36 6 4 (40) 29 6 4 (20) 34 6 3 (60) 303 6 53 (40) 338 6 90 (20) 315 6 46 (60) 78 6 3 (21) 85 6 5 (12) 81 6 3 (33) 36 6 9 (21) 33 6 12 (12) 35 6 7 (33) 42 6 9 (21) 64 6 12 (12) 50 6 7 (33) 28 6 2 (21) 26 6 4 (12) 27 6 2 (33) 371 6 85 (21) 313 6 82 (12) 350 6 61 (33) .02† .54 .18 .70 .20 .83 ,.001† .78 .73 P Values 2-Way analysis of variance Group Gender Group 3 gender ,.001† .72 .04† ,.001† .65 .13 Multiple comparisons for significant differences Group 1 vs group 2 .03† .004† .09 . . .‡ ,.001† Group 1 vs group 3 ,.001† ,.001† .02† ... ,.001† Group 2 vs group 3 ,.001† .10 .24 ... .72 * Promthrombin time (PT), total bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and g-glutamyltransferase (GGT) levels were determined on admission or at the time of biopsy. Values are reported as mean 6 SEM of (n) subjects. † Indicates P values of differences that are statistically significant. ‡ Ellipses indicate data not computed. Immunologic Findings Our immunologic evaluation shows that, irrespective of the degree of fat present, the expression of markers of liver cell types is not modified in a relevant manner by the presence of steatosis. The best markers were CD68 for Kupffer cells, CD34 for endothelial cells, and SMSA for stellate cells. CD68 was the best marker for Kupffer cells and was expressed diffusely (Figure 6) within the lobules in all groups. Its expression was intense throughout, especially if severe necrosis (ballooning) was present. Expression was intense and correlated directly with degree of disease severity. It increased from stage F1 through F3, but was absent from control livers. CD34 was best expressed (in mild to moderate degree) within periportal endothelial cells in all groups (Figure 7). Expression of SMSA increased in intensity and correlated with degree of disease severity. Furthermore, SMSA expression was practically absent in controls and in group 1, was minimal in group 2, and was most intense in stellate cells in the advanced stage of fibrosis within group 3 (Figures 8 through 10). The SMSA enhancement with disease progression reached a peak in stage F3 and in the cirrhotic bands of the positive controls (Figure 11). Collagen type 1 was mildly expressed pericentrally in group 1, more markedly and pericentrally expressed in group 2, and severely expressed in group 3. Collagen type 3 was absent in group 1, minimal in group 2, and intense- ly expressed in group 3. Collagen type 1 was intensely expressed in positive cirrhotic controls, while the expression of type 3 was less marked, although enhanced as well. Transforming growth factor b1, TNF-a, and ICAM-1 were not expressed. In conclusion, based on the intensity and periportal location of CD34 expression, the endothelial cells appear to play a role, as yet unknown, in the fibrogenesis of ALD. Kupffer cells are activated early, diffusely, and intensely and precede the activation of stellate cells. The intensity of their activation, based on CD68 expression, suggests that they play an initial role, perhaps by secreting cytokines. This role could not be confirmed in this study because some cytokines could not be tested with monoclonal antibodies, which are not available commercially. Stellate cells are the most intensely activated, and this activation progressively increases in severity from stage F1 to stage F3. The expression of SMSA within the collagen bands of F3 and positive cirrhotic controls suggests the existence of a continuum between early- and late-stage fibrogenesis. If this is the case, SMSA expression may be used to determine the transition between severe fibrosis and cirrhosis. 1234 Arch Pathol Lab Med—Vol 128, November 2004 Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al COMMENT To our knowledge, this study examined the largest population of patients with ALD in the early stages to date. Table 3. Clinical Correlations* Ethanol Consumption g/d Age, y R P n 20.291 ,.001† 186 Alcohol consumption, g/d R P n y 0.584 ,.001† 181 20.082 .28 177 Alcohol consumption, y R P n Alcohol consumption, g/d/y R P n Total Bilirubin g/d/y PT AST ALT GGT 20.386 ,.001† 180 20.121 .09 197 0.110 .12 197 20.099 .16 197 20.179 .01† 197 20.072 .31 197 0.450 ,.001† 176 0.045 .54 186 20.074 .32 186 20.002 .98 186 0.121 .10 186 20.092 .21 186 20.586 ,.001† 180 20.182 .01† 181 0.179 .02† 181 20.088 .24 181 20.134 .07 181 20.039 .60 181 0.032 .67 180 20.149 .046† 180 0.021 .78 180 20.083 .27 180 20.031 .68 180 20.092 .20 197 0.063 .38 197 20.201 .005† 197 0.180 .01† 197 0.068 .34 197 0.345 ,.001† 197 .560 ,.001† 197 .453 ,.001† 197 PT R P n 20.569 ,.001† 197 Total bilirubin R P n AST R P n ALT R 0.215 P .002† n 197 * PT indicates prothrombin time; AST, aspartate aminotransferase; ALT, alanine aminotransferase; and GGT, g-glutamyltransferase. R is the Pearson product moment correlation coefficient, and n represents the number of subjects. † Indicates P values of differences that are statistically significant. Concerning the clinical parameters, significant differences were found between the various groups in age, PT, serum bilirubin, AST, GGT, and liver span. Abnormalities of these parameters allow the clinician to infer the early presence of liver dysfunction and to identify ethanol-induced injury. Clinically, serum bilirubin, PT, and GGT levels appear to be the best indicators to differentiate the 3 groups we identified, that is, group 1 (portal fibrosis, F0) from group 2 (central injury, F0 or F1) and group 3 (fibrosis with multiple septa, F3). The evaluation with 2-way ANOVA and Pearson correlation coefficients revealed that other factors, such as age, grams per day of ethanol consumption, number of years, and the ratio of grams/day/number of years of alcohol consumption have significant positive correlations. Morphologically, there seems to be a sequential set of lesions, starting with pericentral swelling of hepatocytes followed by pericentral steatosis, pericentral necrosis, pericentral fibrosis, and finally sclerosing hyaline necrosis. A combination of steatosis with other lesions, such as fibrosis or necrosis, was relatively common. The expression of these parameters in such a large population of patients allows us to conclude that central injury (F1) seems to characterize a group of patients intermediate between Arch Pathol Lab Med—Vol 128, November 2004 those with normal portal areas (F0) and those with bridging fibrosis (F3). The lesions of early pericentral ethanolinduced injury were observed in this study quite often. They are sequentially characterized here. Pericentral swelling (ballooning) is usually present alone. On the other hand, the other lesions frequently may appear simultaneously, that is, lesions such as steatosis with fibrosis, or fibrosis with necrosis. Thus, we believe that the perivenular lesions begin sequentially with ballooning degeneration followed by steatosis, necrosis, fibrosis, and finally sclerosing hyaline necrosis. The second objective of this study deals with the evolution of fibrosis from the early stage of ALD. We found central vein fibrosis, although minimally expressed, at the swelling stage (early ballooning degeneration). Liver fibrogenesis has been a topic of considerable interest for many years. Experimental models have been developed preferentially in rats by administration of carbon tetrachloride,8 dimethylnitrosamine,9 and ethanol.10 The objective of these studies was to gain insight into the complex process of human liver fibrogenesis, which represents one of the most common causes of human morbidity and mortality. The development of liver fibrosis in humans has been Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al 1235 Figure 9. Central injury, group 2, stage F2. Expression of smooth muscle–specific a actin (SMSA) pericentrally with ballooning and collapse around a central vein, stage F2 (SMSA immunoperoxidase, original magnification 3400). Figure 10. Smooth muscle–specific a actin (SMSA), stage F3. A liver needle biopsy, stage F3. Porto-portal bridging. Some morphologists may be tempted to call this stage cirrhosis; however, there are no regenerative nodules and the central veins are patent (SMSA immunoperoxidase, original magnification 320). Figure 11. Positive controls in cirrhosis showing expression of 4 markers: transforming growth factor b (TGF-b1) (A), smooth muscle–specific a actin (SMSA) (B), collagen type 3 (C), and collagen type 1 (Masson trichrome) (D) (immunoperoxidase for TGF-b1, SMSA, and collagen type 3, original magnifications 3100 [A], 340 [B], 320 [C], and 340 [D]). The higher magnification of panel A reveals TGF-b1 expression within a regenerative nodule by the sinusoid-lining cells. attributed over the years to production of hepatic substances by the various cellular components of the liver under the influence of multiple noxious agents, including ethanol. Emphasis currently has been placed either on the role of central hypoxia11 or on the release of cytokines by sinusoid-lining cells.12 During endotoxemia, cytokines are known to undergo alterations by increasing intestinal absorption of lipopolysaccharide and lipopolysaccharidebinding protein. This is associated, especially in Kupffer cells, with increased secretion of TNF-a, TGF-b1, interleukin (IL)-1, IL- 8, reactive oxygen species, and nitric oxide. Historically, the first cellular element suspected of playing a role in fibrogenesis was the hepatocyte,13 followed by the Kupffer cell.12 More recently, studies of liver fibrogenesis have shifted interest to the role played by the stellate cell.14–16 The interest in the role of Kupffer cells in endotoxemia and related states is understandable and significant with respect to liver fibrogenesis. In the present study on human ALD, the expression of CD68 by Kupffer cells parallels the expression of SMSA by stellate cells in intensity. We could not test for the presence of TNF-a, IL-1, and IL- 1236 Arch Pathol Lab Med—Vol 128, November 2004 Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al 8 because our material had been treated with formalin and embedded in paraffin for some time. Under these circumstances (ie, formalin-fixed and paraffin-embedded tissues), monoclonal antibodies for these cytokines are not available. Activation of fibrogenesis has been attributed to several factors, including increased intestinal permeability by ethanol, increased absorption of intestinal endotoxin, activation of Kupffer cells, transformation of stellate cells into myofibroblasts, and finally stimulation of endothelial cells. Stellate cell activation appears to be preceded by Kupffer cell production of stimulatory factors, such as TGF-b,12 or unknown substances produced by hepatocytes after injury.14 Factors involved in Kupffer cell activation include, among others,11 TNF-a and TGF-b. Northern blot studies have shown production of TGF-b1 messenger RNA (mRNA) by Kupffer cells isolated from ethanol-treated rats. Furthermore, monocytes activated by intestinal lipopolysaccharide have a similar effect. This results in stimulation of stellate cells (lipocytes) with expression of collagen genes, especially collagen type 1.17 The bile duct ligation model yields unusually elevated levels of collagen type 1 expression by stellate cells, while the CCl4 model shows enhanced expression (14-fold) of collagen type 1 by endothelial cells and a 43-fold increase by lipocytes (Ito cells). The mRNA for type 3 collagen in normal rat liver has been reported to be equally increased in endothelial cells and lipocytes. This mRNA elevated in these experimental models is most dramatically expressed by stellate fat-storing Ito cells. In our study, SMSA, the marker for stellate cells, appears to be the most reliable indicator of progression of human fibrosis in ALD. Expression of SMSA is enhanced with disease progression, reaching a peak in F3, and is seen very prominently in the positive control cirrhotic bands. CD68 is the best marker for Kupffer cells, is expressed diffusely within the lobules in all groups, increases from stage F1 through stage F3, and is absent in normal livers. This pattern correlates directly with the clinical degree of disease severity. CD34 was expressed in this study in mild to moderate degree within periportal endothelial cells in all groups. The bile duct ligation model17 showed that in normal animals the endothelial cell is an important producer of collagen types 1 and 4, either higher than or similar to that of the stellate cell. We tested only for collagen types 1 and 3, and we used human livers. We also used a different immunologic marker. Thus, it is difficult to explain this difference. However, based on our information using CD34 expression by endothelial cells, the results do support a lesser role for these cells in the fibrogenesis of human ALD. The expression of mRNA for a given collagen type experimentally does not justify a direct extrapolation to the human condition. Several studies in human alcoholics18–20 showed no distinctive clinical features. However, in one of these studies,20 11 patients exhibited perivenular fibrosis very prominently, and the most characteristic features were myofibroblast proliferation and collagen deposition around the terminal hepatic venules. Of the 21 venules examined in the study, myofibroblasts represented almost half of the cellular population combined with fibroblasts, while fat-storing stellate cells as such were not recognized in the connective tissue surrounding the terminal hepatic Arch Pathol Lab Med—Vol 128, November 2004 venules. Furthermore, it was postulated that fat-storing cells, fibroblasts, and myofibroblasts belonged to the same family. In addition, it was stated that a transition occurred from Ito fat-storing cells to myofibroblasts. Three of the studies18,20,21 concluded that perivenular fibrosis was a precursor of more advanced forms of ALD, especially cirrhosis, if the patients continued their alcohol intake. Only one study19 concluded the opposite. Perivenular fibrosis in the alcohol-fed baboon model,20 as well as in the French model,22 confirmed the presence of myofibroblasts and stellate cells in the perivenular area and strongly suggested that they are the cellular elements responsible for the generation of collagen. Our study was based on a much larger number of patients, and numerous terminal hepatic veins were available for examination. Our findings support the hypothesis that perivenular injury (swelling, steatosis, fibrosis, and sclerosis) represents earlier stages of ALD from which, unless a patient stops alcohol consumption, they most likely progress to cirrhosis. In this study, SMSA expression increased progressively in intensity and correlated with the degree of severity of fibrosis, reaching the highest level in cirrhotic bands. Expression of SMSA was absent in normal controls but was very intensely expressed in stage F3. Thus, the increased expression of SMSA within the collagen bands of F3 and cirrhotic patients suggests the existence of a continuum between severe fibrosis and end-stage cirrhosis (F4). Finally, SMSA expression may be useful to determine the transition between severe fibrosis and cirrhosis. The earlier literature cited and our findings in this large population of alcoholics support the idea that a sequence of lesions, starting with pericentral ballooning and steatosis, progresses eventually in many patients to end-stage liver disease. The expression of collagen type 1 was absent in normal controls, but was very intense in F3 and in positive controls with cirrhosis. Collagen type 3 was progressively expressed as disease severity increased, but with a lesser intensity in the various groups. Expression of TGF-b1 has been reported to increase following liver injury in several animal models. Furthermore, TGF-b suppresses the proliferation of hepatocytes and other epithelial elements and induces active proliferation of hepatic stellate cells. Because previous studies with the French-Tsukamoto11,22 rat model revealed concentrations of Ito cells in the centrilobular areas and enhanced collagen production, it has been easy to assume that this phenomenon applies to most human liver diseases, including ethanol-induced fibrogenesis. Thus, it has been stated that TGF-b1 is a major factor stimulating stellate cell fibrogenic activity (enhanced collagen production), predominantly expressed in centrilobular areas, and finally that it correlates with enhanced expression of SMSA. We were unable to confirm this correlation in our patients. The reason for this discrepancy may be related to 3 different factors: (a) we studied formalin-fixed, paraffinembedded human livers from ALD patients, while in most experimental studies dealing with rat models sampling took place shortly after sacrifice; (b) we studied tissues already processed and stored for quite some time, while animal studies usually used fresh tissues; and (c) monoclonal antibodies were not available for all markers of our human studies. However, it must be noted that TGF-b1 was strikingly expressed in our positive controls Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al 1237 1. Mallory FB. Cirrhosis of the liver: five different types of lesions from which it may arise. Bull Johns Hopkins Hosp. 1911;22:69–75. 2. Gall EA. Posthepatitic, postnecrotic and nutritional cirrhosis: a pathologic analysis. Am J Pathol. 1960;36:241–271. 3. Beckett AG, Livingston AV, Hill KR. Acute alcoholic hepatitis. Br Med J. 1961;2:1113–1119. 4. Beckett AG, Livingston AV, Hill KR. Acute alcoholic hepatitis without jaundice. Br Med J. 1962;2:580–582. 5. Edmondson HA, Peters RL, Reynolds TB, et al. Sclerosing hyaline necrosis of the liver in the chronic alcoholic: a recognizable clinical syndrome. Ann Intern Med. 1963;59:646–673. 6. Alcoholic liver disease: morphological manifestations: review by an international group. Lancet. 1981;1:707–711. 7. Chedid A, Mendenhall CL, Gartside P, et al. Prognostic factors in alcoholic liver disease. Am J Gastroenterol. 1991;86:210–216. 8. Poynard T, Bedossa P. Opolon P, for the OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups. Natural history of liver fibrosis progression in patients with chronic hepatitis C. Lancet. 1997;349:825–832. 9. Martinez-Hernandez A. The hepatic extracellular matrix II: electron immunohistochemical studies in rats with CCl4-induced cirrhosis. Lab Invest. 1985;53: 166–186. 10. ZheQi, Nobuhiko A, Akira O, Akira T, Ueno H. Blockade of type b transforming growth factor signaling prevents liver fibrosis and dysfunction in the rat. Proc Natl Acad Sci U S A. 1999;96:2345–2349. 11. Tsukamoto H, Gaal K, French SW. Insights into the pathogenesis of alcoholic liver necrosis and fibrosis. Hepatology. 1990;12:599–608. 12. Matsuoka M, Tsukamoto H. Stimulation of hepatic lipocyte collagen production by Kupffer cell-derived transforming growth factor b: implications for a pathogenetic role in alcoholic liver fibrogenesis. Hepatology. 1990;11:599–605. 13. Sakakibara K, Saito M, Umeda M, Enaka K, Tsukada Y. Native collagen formation by liver parenchymal cells in culture. Nature. 1976;262:316–318. 14. Gressner AM, Lotfi S, Gressner G, Lahme B. Identification and partial characterization of a hepatocyte-derived factor promoting proliferation of cultured fatstoring cells (parasinusoidal lipocytes). Hepatology. 1992;16:1250–1266. 15. Nakano M, Lieber CS. Ultrastructure of initial stages of perivenular fibrosis in alcohol-fed baboons. Am J Pathol. 1982;106:145–155. 16. Mak KM, Lieber CS. Lipocytes and transitional cells in alcoholic liver disease: a morphometric study. Hepatology. 1988;8:1027–1033. 17. Maher JJ, McGuire RF. Extracellular matrix gene expression increases preferentially in rat lipocytes and sinusoidal endothelial cells during hepatic fibrosis in vivo. J Clin Invest. 1990;86:1641–1648. 18. Waes LV, Lieber CS. Early perivenular sclerosis in alcoholic fatty liver: an index of progressive liver injury. Gastroenterology. 1977;73:646–650. 19. Nasrallah SM, Nassar VH, Galambos JT. Importance of terminal hepatic venule thickening. Arch Pathol Lab Med. 1980;104:84–86. 20. Nakano M, Worner TM, Lieber CS. Perivenular fibrosis in alcoholic liver injury: ultrastructure and histologic progression. Gastroenterology. 1982;83:777– 785. 21. Worner TM, Lieber CS. Perivenular fibrosis as precursor lesion of cirrhosis. JAMA. 1985;254:627–630. 22. French SW, Miyamoto K, Wong K, Jui L, Briere L. Role of the Ito cell in liver parenchymal fibrosis in rats fed alcohol and a high fat–low protein diet. Am J Pathol. 1988;132:73–85. 23. Friedman SL. Molecular regulation of hepatic fibrosis, an integrated cellular response to tissue injury. J Biol Chem. 2000;275:2247–2250. 24. Wrana JL. Transforming growth factor-b signaling and cirrhosis [editorial]. Hepatology. 1999;29:1909–1910. 25. Roulot D, Sevcsik AM, Coste T, Strosberg AD, Marullo S. Role of transforming growth factor b type II receptor in hepatic fibrosis: studies of human chronic hepatitis C and experimental fibrosis in rats. Hepatology. 1999;29:1730– 1738. 26. Rockey DC, Fouassier L, Chung JJ, et al. Cellular localization of endothelin-1 and increased production in liver injury in the rat: potential for autocrine and paracrine effects on stellate cells. Hepatology. 1998;27:472–480. 27. Bataller R, Gines P, Nicolas JM, et al. Angiotensin II induces contraction and proliferation of human hepatic stellate cells. Gastroenterology. 2000;118: 1149–1156. 28. Ueki T, Kaneda Y, Tsutsui H, et al. Hepatocyte growth factor gene therapy of liver cirrhosis in rats. Nature Med. 1999;5:226–230. 1238 Arch Pathol Lab Med—Vol 128, November 2004 Immunology of Fibrogenesis in Alcoholic Liver Disease—Chedid et al (human cirrhotic livers), which were processed in a fashion similar to the precirrhotic cases under investigation. We are unable to explain this discrepancy. This matter has been examined recently by Friedman23 at the molecular level. According to relatively recent publications,24–26 it must be recognized that TGF-b does not act alone to cause liver fibrogenesis. Other factors seem to participate, and an important one seems to be hepatocyte growth factor, which appears to have effects just opposite to those of TGF-b, such as potent proliferative activity on hepatocytes. In experimental models, hepatocyte growth factor has been shown to suppress the fibrotic response and stimulate hepatocyte proliferation.24 In vitro studies suggest that hepatocyte growth factor functions predominantly to block the effects of TGF-b and that fibrogenesis in the liver may be the result of a delicate equilibrium between these 2 factors. Finally, another body of literature27,28 supports the idea that the major effector cell involved in fibrogenesis is the stellate cell, following its activation by factors released in response to injury. This response is associated with increased expression of smooth muscle proteins and increased production of extracellular matrix proteins. Stellate cells exposed in culture to vasoactive substances, such as endothelin-1 and angiotensin II, respond with enhanced production of SMSA and extracellular matrix proteins. In this study, we confirmed the enhanced production of SMSA by the stellate cells. The authors thank Mariann Eichorst, HT(ASCP), for her superb technical help with immunohistochemistry and Gail Hoppe for her invaluable secretarial assistance. References