Occlusive Venous Lesions in Alcoholic Liver Disease

Transcription

1 GASTROENTEROLOGY 1982;83: Occlusive Venous Lesions in Alcoholic Liver Disease A Study of 200 Cases ZACHARY D. GOODMAN and KAMAL G. ISHAK From the Department of Hepatic Pathology, Armed Forces Institute of Pathology, Washington, D.C. The nature and significance of vascular lesions in alcoholic liver disease were studied in 200 autopsies. Three principal types of lesions were recognized: [a) Lymphocytic phlebitis, consisting of a chronic inflammatory cell infiltrate of the wall of terminal hepatic venules [central veins) or intercalated [sublobular) veins, was noted in 16.7% of patients with precirrhotic alcoholic hepatitis and 4.3% of patients with cirrhosis. [b) Phlebosclerosis, consisting of perivenular scarring with gradual obliteration of the lumen of terminal hepatic venules and sometimes intercalated veins was found to some degree in all patients with alcoholic hepatitis or cirrhosis. [c) Veno-occlusive lesions, consisting of intimal proliferation, fibrosis, and narrowing of the lumen of terminal hepatic venules, intercalated veins, and occasionally portal veins were found in 52.1 % of cases of precirrhotic alcoholic hepatitis with total occlusion of some terminal hepatic venules or intercalated veins, or both, in 14.6%. In alcoholic cirrhosis, veno-occlusive lesions were present to some degree in 74.1 % with totally occluded vessels found in 46.8%. Evidence of portal hypertension was present in 47.9% of patients with precirrhotic alcoholic hepatitis and was significantly associated with the degree of both veno-occlusive Received April 9, Accepted May 21, Address requests for reprints to: Major Zachary D. Goodman, Department of Hepatic Pathology, Armed Forces Institute of Pathology, Washington, D.C The authors thank Dr. Walter Foster for performing statistical analyses, Mr. Peter Emanuele for technical assistance, Miss Mary C. James and Ms. Karen L. Page for typing the manuscript. They also thank Dr. Hyman J. Zimmerman and Dr. Frank B. Johnson for review of the manuscript and helpful suggestions. The opinions and assertions contained herein are the views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense. change and phlebosclerosis, which tend to occur together. It is concluded that both veno-occlusive lesions and phlebosclerosis contribute to the development of portal hypertension in alcoholic liver disease. Veno-occlusive lesions in the cirrhotic liver may contribute to atrophy, with loss of functioning parenchyma. The etiopathogenesis of the vascular lesions in alcoholic liver disease requires further investigation. Evidence has accumulated in recent years that the structural alterations responsible for the development of cirrhosis and portal hypertension in alcoholic liver disease result from obliteration of terminal hepatic venules (central veins) by the process called sclerosing hyaline necrosis or alcoholic hepatitis (1-5). In this disease, acinar zone 3 (centrilobular) hepatocytes are destroyed and replaced by collagen deposition that surrounds and appears to compromise the terminal hepatic venules by external compression, a process which may be regarded as a type of "phlebosclerosis." Cirrhosis develops as the fibrosis extends to link terminal hepatic venules and portal tracts with nodule formation. However, the early obliteration of terminal hepatic venules may cause portal hypertension before cirrhosis has developed (1,6-8). In some cases of alcoholic liver disease, we have observed another type of vascular lesion. In contrast to the usual obliteration of terminal hepatic venules by external compression, the lesion appears to be a primary vasculopathy. It occurs most commonly in terminal hepatic venules and intercalated (sublobular) veins, (the first- and second-order hepatic outflow vessels), but is occasionally seen in small portal veins. The lesion closely resembles that seen in veno-occlusive disease (VOD) of the liver produced by pyrrolizidine alkaloids (9), radiation injury (10), and drugs (11-16), with intimal proliferation of

2 October 1982 VENO-OCCLUSIVE LESIONS IN THE ALCOHOLIC 787 connective tissue producing a subtotal or sometimes total occlusion of the lumen. Veno-occlusive lesions can be readily overlooked in routine hematoxylin and eosin stained sections and are best appreciated with special stains for elastic tissue that outline the vessel. Veno-occlusive lesions in alcoholic liver disease have only rarely been mentioned in the literature and are generally thought to be of little significance (3-5). In contrast to this, our experience is that these lesions can be found frequently in sections of liver taken at autopsy from patients with alcoholic liver disease. The lesions can also be seen in biopsy specimens, but less frequently, due, at least in part, to smaller sample size. The present study was undertaken to characterize veno-occlusive (VO) lesions in the alcoholic and to determine their incidence and possible contribution to the development of portal hypertension. Livers from 200 autopsies of alcoholic patients were studied and the pathologic findings were correlated with the stigmata of portal hypertension. Material and Methods The cases that were studied were selected from autopsies submitted to the Armed Forces Institute of Pathology between January 1976 and December The Institute receives autopsies from military hospitals on a routine basis and from Veterans Administration hospitals as part of a quality control program. Only a few of these were diagnostic problems submitted for consultation. Cases were selected if they met the following criteria: (a) a diagnosis of alcoholic liver disease or a history of alcoholism; (b) autopsy protocol, clinical summary, and microscopic slides available for review; and (c) paraffin blocks available for recuts and special stains. The case was excluded if the liver contained a primary or metastatic carcinoma. Two hundred consecutive cases were thus chosen. One case was contributed from a civilian hospital, 57 cases from Veterans Administration hospitals, and 142 cases from United States military hospitals. The autopsy protocol and clinical summary usually provided reliable information regarding the patient's terminal illness, mode of death, liver weight, findings indicating portal hypertension (ascites, esophageal varices, splenomegaly), and nonhepatic diseases. Portal hypertension was considered to be present if the autopsy revealed esophageal varices, > 300 ml of ascites, or a > 400 g spleen. In most cases, two or three of the findings were present, and splenomegaly was not found in the absence of varices or ascites. Histologic sections of liver from each case were stained with hematoxylin and eosin, Masson's trichrome, Movat's pentachrome, and Mallory's iron stain. The sections were evaluated independently by both authors, and any differences in interpretation were discussed and resolved in conference. Each case was evaluated for the presence of fat, Mallory bodies, neutrophils, and hemosiderosis. These were graded subjectively on a scale of 0 to 3+ with 0 = absent, 1 + = minimal or few, 2+ = moderate, and 3+ = marked or many. In precirrhotic cases, the degree of intralobular fibrosis was graded similarly as 0 to 3 +. Cirrhosis was considered to be present if there was fibrosis with loss of lobular architecture and parenchymal nodularity. Vascular lesions were characterized as phlebosclerosis when there was compression by perivenular fibrosis or veno-occlusive when there was intimal proliferation. These were likewise graded on a scale of 0 to 3 + with the grade being given for the most severe lesion in the section. Veno-occlusive lesions were considered to be 1 + if the intimal proliferation had occluded the lumen up to onethird of its diameter, 2+ if the lumen was one-third to twothirds occluded, and 3+ if there was greater than twothirds or total occlusion, with or without recanalization. Phlebosclerosis was 1 + when there was mild peri venular fibrosis, 2 + when there was more extensive fibrosis of acinar zone 3 with most terminal hepatic venules still identifiable, and 3+ when there was severe fibrosis with obliteration of many terminal hepatic venules. Associations between variables were tested for independence using the X 2 statistic corrected for continuity. For comparison with the vascular lesions in alcoholic cirrhosis, a group of cases was selected from the Armed Forces Institute of Pathology files, consisting of autopsies of patients with cirrhosis due to nonalcoholic liver disease. These included 9 patients with cirrhosis due to chronic hepatitis B infection and 15 patients with endstage primary biliary cirrhosis. Results Patient Characteristics The clinical features of the patients in this study reflected the type of alcoholic likely to die and have a postmortem examination in a military or Veterans Administration hospital (Table 1). The majority were men (77%), white (84.5%), and yr old (65.5%). Although all 200 patients were thought to be alcoholic by the physicians caring for them, detailed information on the duration, amount, and type of alcoholic beverages consumed was seldom available. In many cases death was from a combination of factors with multi organ failure, but the principal cause was gastrointestinal (usually variceal) hemorrhage in 36.5%, hepatocellular failure with hepatic coma in 19%, infection in 18%, and miscellaneous causes (pancreatitis, cardiovascular disease, trauma, suicide, or unexplained) in the remainder. At autopsy, complications of portal hypertension (esophageal varices, ascites, or splenomegaly) were found in 78%. Cirrhosis was present in the majority of these, but 14.7% of patients with signs of portal hypertension had pre cirrhotic alcoholic hepatitis. Only 1 patient with neither alcoholic hepatitis nor cirrhosis had findings suggesting portal hyperten-

3 788 GOODMAN AND ISHAK GASTROENTEROLOGY Vol. 83. No. 4 Table 1. Clinical Features of Patients Precirrhotic alcoholic Cirrhosis hepatitis Others Number Sex Male Female Age Mean Range Race White Black American Indian Unknown Portal hypertension Mode of death Gastrointestinal hemorrhage Hepatic coma Infection Other sion. This was an alcoholic with a massive fatty liver (5220 g), ascites, esophageal varices, and splenomegaly; he died of a coronary thrombosis. Alcoholic Liver Disease Only 1 of the 200 patients had a normal liver. Twelve (6%) had only steatosis, 48 (24%) had precirrhotic alcoholic hepatitis (with Mallory bodies and neutrophilic infiltrate), and 139 (69.5%) had cirrhosis. Among those with cirrhosis, 75.5% had continuing hepatocellular injury with Mallory bodies; another 19 (13.7%) showed some degree of fat accumulation. Polymorphonuclear leukocyte infiltration was variable, and neither this nor the degree of fat accumulation correlated well with the numbers of Mallory bodies. Some degree of iron accumulation was present in 52 cirrhotic livers (37.4%), but was marked in only 11 (7.9%) (Table 2). Among the 48 patients with precirrhotic alcoholic hepatitis, the severity of the various histologic findings varied from case to case. Most cases were judged to have moderate or severe alcoholic liver disease, but the findings of fat, Mallory bodies, neutrophilic infiltration, and fibrosis often did not correlate well. vith one another. Vascular Lesions Three principal types of lesions were encountered. Phlebitis. A lymphocytic phlebitis was present in 8 cases (16.7%) with precirrhotic alcoholic Table 2. Pathologic Findings Related to Alcoholic Liver DiseaseO Pre cirrhotic alcoholic Cirrhosis hepatitis Others Total number Fat % 2.1% 7.7% % 16.7% 30.8% % 22.9% 15.4% % 58.3% 46.2% Mallory bodies % 0 100% % 18.8% % 50.0% % 31.2% 0 Neutrophils % % % 41.7% 7.7% % 43.8% % 14.6% 0 Hemosiderosis % 52.1% 53.8% % 25.0% 23.1% % 22.9% 23.1% % 0 0 Fibrosis (precirrhotic) % % 53.8% % % 0 Weight (n = 136) (n = 47) (n = 13) Mean 1847 g 2602 g 2448 g Range g g g a Abbreviations: 0 = absent; 1+ = minimal or few; 2+ = moderate; 3+ = marked or many. hepatitis and 6 cases (4.3%) with cirrhosis (Table 3). Affected vessels were infiltrated by inflammatory cells, predominantly lymphocytes with occasional neutrophils (Figure 1). Both terminal hepatic venules and intercalated veins were involved. In most Table 3. Incidence of Hepatic Vascular Lesions in Two Hundred Autopsies on Alcoholics Pre cirrhotic alcoholic Cirrhosis hepatitis Others Total number Phlebitis 4.3% 16.7% 0 Phlebosclerosis % % 53.8% % % 25.0% 0 Vena-occlusive change (most severe lesion found) % 47.9% 100% % 20.8% % 16.7% B% 14.6% 0

4 October 1982 VENO-OCCLUSIVE LESIONS IN THE ALCOHOLIC 789 Figure 1. Lymphocytic phlebitis. A. An intercalated vein showing diffuse lymphocytic infiltrate involving most of its wall and extending into the perivascular parenchyma. (Hematoxylin and eosin, x 100; AFIP negative ). B. A terminal hepatic venule with concentrically thickened wall infiltrated by lymphocytes. Arrows indicate thickness of the wall. (Hematoxylin and eosin, x 160; AFIP negative ). cases there were several affected vessels in each section, and in a few cases almost every vessel was affected. There was usually mild to moderate thickening of the vessel wall. The lymphocytic infiltration was often patchy and segmental, with 10%-50% of the circumference involved in milder cases and 90%-100% in severe cases. Occasional cases also revealed perivenular lymphocytic infiltrates. Intimal proliferation (VO-like) was present in 5 of the 8 precirrhotic cases with total occlusion of at least one vein per section in 2 of these. All 8 precirrhotic cases had moderate to severe alcoholic hepatitis. Five of the 8 had evidence of portal hypertension including 1 case without intimal proliferation and only mild phlebosclerosis. All 6 cases with cirrhosis and phlebitis showed VO changes, with 3 cases having some totally occluded vessels. All 6 cases had signs of portal hypertension. The amount of superimposed alcoholic hepatitis in these varied from none (1 case) to severe (2 cases). Phlebosclerosis. This lesion has been extensively described in the literature (1-5), although not under this name. In its mild (1 +) form there was perivenular and acinar zone 3 fibrosis with minimal compression of the vein lumen (Figure 2A). Cases with moderate (2 +) phlebosclerosis showed more severe fibrosis with recognizable compromise of the lumen (Figure 2B). Severe (3+) phlebosclerosis was characterized by extensive fibrosis of acinar zone 3 and obliteration of most terminal hepatic venules (Figure 2C). Sometimes the Movat stain would show a remnant of elastic fibers where the venule had been obliterated. All livers with cirrhosis had 3+ phlebosclerosis (Table 3). The 48 cases with precirrhotic alcoholic hepatitis all had some degree of phlebosclerosis, mild (1 +) in 21 cases, moderate (2 +) in 15 cases, and marked (3+) in 12 cases. Mild zone 3 fibrosis with mild (1 +) phlebosclerosis was found in 7 cases with steatosis but no active alcoholic hepatitis. Vena-Occlusive Lesions. These lesions varied in quantity and severity from case to case. In the most severe cases virtually every terminal hepatic venule, intercalated vein, and portal vein branch was affected to some degree. In many cases only a few, or sometimes only one lesion was found per section. Affected vessels showed a subendothelial intimal proliferation of connective tissue. This was usually best appreciated with a Movat stain in which the elastic lamina marl<ed the location of the original

5 790 GOODMAN AND ISHAK GASTROENTEROLOGY Vol. 83, No. 4 Figure. 2. Alcoholic phlebosclerosis. A. Mild (1 +) phlebosclerosis with surrounding alcoholic hepatitis, peri venular fibrosis, and mild compression of terminal hepatic venule lumen. (Masson's trichrome, x 250; AFIP negative ). B. Moderate (2 +) phlebosclerosis. The terminal hepatic venule lumen (arrows) is severely compromised but still recognizable. (Masson's trichrome, x250; AFIP negative ). C. Marked (3+) phlebosclerosis. The centrilobular region is entirely scarred, and the terminal hepatic venule can no longer be identified. (Masson's trichrome, x160; AFIP negative ). lumen of the vessel, and the va changes could be seen within (Figure 3A). The Masson trichrome stain was almost as good for demonstrating the lesions (Figure 3B), but they were often not apparent with other stains. The va lesions were morphologically similar whether or not they occurred in terminal hepatic venules, intercalated veins, or portal veins (Figures 3-5). In some cases the lesions appeared fresh and quite proliferative with a loose stroma and plump fibroblasts (Figure 6). Often the proliferative lesions contained spindle cells with eosinophilic cytoplasm suggestive of myofibroblasts. Older appearing lesions were frequent in cases with established, inactive cirrhosis. These were more fibrotic, and often there was total occlusion of the lumen (Figures 3C and 4B). Recanalization of the lumen in such cases was common. Recent thrombi were not found. Affected portal vessels seldom showed > 1/3 occlusion of the lumen and were never totally occluded (Figure 5). Terminal hepatic venules and intercalated veins with va lesions often showed thickening of the wall with reduplication of elastic fibers (Figures 3 and 4). nearly 3/4 of cirrhotic livers (Table 3); only 36 cases (25.9%) had no va lesions. Subtotal occlusions were present in 38 cases (27.3%), and at least one totally occluded terminal hepatic venule or intercalated vein was found in 46.8% of patients with cirrhosis. Veno-occlusive lesions were found in slightly more than one-half of the cases with precirrhotic alcoholic hepatitis (Table 3). In 7 cases (14.6%) total occlusions were found, while in another 18 cases (37.5%) there were subtotal occlusions of terminal hepatic venules or intercalated veins. Mild intimal proliferation was found in occasional portal vein branches, in the absence of va lesions in terminal hepatic venules or intercalated veins, in 6 cases with cirrhosis, and in 2 cases with precirrhotic alcoholic hepatitis. All 8 cases had portal hypertension. Since it seemed possible that mild portal vein thickening could have resulted from the portal hypertension unrelated to whatever might have caused va lesions in hepatic venules and intercalated veins, these 8 cases were combined with the cases with no va lesions in subsequent analysis of the data. No Va-type lesions were seen in hepatic arteries in any case. Incidence of Vena-Occlusive Lesions in Alcoholic Liver Disease Relationship of Vascular Lesions to Other Pathologic Changes In precirrhotic alcoholic hepatitis, the va lesions were associated with several other changes Either subtotal or total occlusions of terminal hepatic venules or intercalated veins were found in

6 October 1982 VENO-OCCLUSIVE LESIONS IN THE ALCOHOLIC 791 Figure 3. Alcoholic veno-occlusive lesions in terminal hepatic venules. A. Subendothelial proliferation of loose connective tissue producing subtotal (2+) occlusion of lumen. Note reduplication of elastic fibers. (Movat pentachrome stain x 160; AFIP negative ). B. Subtotal occlusion; arrows indicate original lumen. (Masson trichrome stain x 160; AFIP negative ). C. Total occlusion by dense, fibrotic connective tissue. (Movat's pentachrome, x250; AFIP negative ). D. Entrance point of a subtotally occluded terminal hepatic venule into an intercalated vein. (Movat's pentachrome, x60; AFIP negative ). related to alcoholic liver disease (Table 4). The severity of the VO lesion was strongly associated with the degree of both fibrosis (p < 0.005) and phlebosclerosis (p < 0.025). There was weaker association of VO lesions with the severity of alcoholic hepatitis as measured by the numbers of Mallory bodies and neutrophils. This did not reach statistical significance for Mallory bodies (p = 0.16), but was significant for numbers of neutrophils (p < 0.05). Fat was inversely associated with VO lesions in that

7 792 GOODMAN AND ISHAK GASTROENTEROLOGY Vol. 83, No. 4 Figure 4. Veno-occlusive lesions in intercalated veins. A. Subtotally occluded intercalated vein. Arrows indicate entrance points of three occluded terminal hepatic venules. (Movat's pentachrome, X60; AFIP negative ). B. Totally ocduded, fibroti c intercalated vein. (Movat's pentachrome, x60; AFIP negative ). patients with marked steatosis were less likely to have total or subtotal occlusions than patients with le~ser degrees of fat. The 139 cases with cirrhosis showed no signifi- Figure 5. Veno-occlusive lesion in portal vein branch. The vein reveals eccentric intimal proliferation with subtotal ocdusion. (Masson 's trichrome, x 100; AFip n egative ). cant association of VO lesions with the severity of superimposed alcoholic hepatitis, fat accumulation, or hemosiderosis. However it was noted that VO lesions were more common in small cirrhotic livers

8 October 1982 VENO-OCCLUSIVE LESIONS IN THE ALCOHOLIC 793 Figure 6. Vena-occlusive lesion in a terminal hepatic venule. A proliferative lesion with loose stroma and darkly stained and elongated myofibroblastlike cells. (Masson's trichrome, x250; AFIP negative ). than in large livers (Table 5). Liver weights were recorded for 136 of the cases with cirrhosis. There were 48 livers that weighed 1500 g or less, and only 10.4% of these had no VO lesions demonstrated, while 25% showed subtotal occlusions and 64.6% had total occlusions. This was significantly different (p < 0.01) from the 88 cases with livers >1500 g in Table 4. Association of Veno-Occlusive Lesions With Other Pathologic Findings in Precirrhotic Alcoholic Hepatitis No Veno- Total Subtotal Occlusive Pathologic occlusions occlusions lesions finding (n = 7) (n = 18) (n = 23) Fibrosis % 56.5% % 44.4% 39.1% p < % 22.2% 4.4% Phlebosclerosis % 56.5% % 27.8% 34.8% p < % 27.8% 8.7% Mallory bodies % 30.4% % 50.0% 52.1% p = % 38.9% 17.4% Neutrophils % 38.9% 52.2% % 50.0% 39.1% p < % 11.1% 8.7% Fat % 22.2% 8.7% % 27.8% 8.7% p = % 50.0% 82.6% which there were no VO lesions in 34.1%, subtotal occlusions in 29.5%, and total occlusions in 36.4%. Relationship of Vascular Lesions and Other Pathologic Changes to Portal Hypertension Stigmata of portal hypertension (esophageal varices, ascites, or splenomegaly) were found in 132 of the 139 cases with cirrhosis. The remaining 7 were not obviously different, except for the fact that they all died of causes unrelated to liver disease. Four of the 7 had VO lesions, and in 3 of these there were totally occluded vessels. Six of the 7 showed some degree of active alcoholic hepatitis superimposed on cirrhosis. Signs of portal hypertension were present in 23 of the 48 patients with precirrhotic alcoholic hepatitis. The degrees of fibrosis, phlebosclerosis, and VO change were all significantly associated with portal hypertension (Table 6). The various pathologic changes tended to be associated with one another, so Table 5. Veno-Occlusive Lesions and Weight of Livers With Cirrhosiso Liver weight Mean Vascular weight No. No. lesion (g) Range S1500 g >1500 g Total occlusions Subtotal occlusions No veno-occlusive lesions an = 136.

9 794 GOODMAN AND ISHAK GASTROENTEROLOGY Vol. 83, No.4 Table 6. Association of Veno-Occlusive Lesions, Phlebosclerosis, and Fibrosis With Signs of Portal Hypertension in Forty-Eight Cases of Precirrhotic Alcoholic Hepatitis Veno-occlusive lesions Total occlusion Subtotal occlusion None Phlebosclerosis Fibrosis Portal hypertension 85.7% (6/7) 55.6% (10/18) 30.4% (7/23) 91.7% (11/12) 53.3% (8/15) 19.5% (4/21) 77.8% (7/9) 60.0% (12/20) 21.1% (4/19) p < 0.05 p < p < that it was not possible to determine the exact contribution of each pathologic finding to portal hypertension. Among precirrhotic patients with evidence of portal hypertension, 81.8% with 3+ phlebosclerosis had some degree of va change (Table 7). Incidence of Veno-Occlusive Lesions in Nonalcoholic Liver Disease Since Va-type lesions can be found in occasional cases of cirrhosis from causes other than alcoholic liver disease, 15 cases of end-stage primary biliary cirrhosis (PBC) and 9 cases of cirrhosis due to chronic hepatitis B were examined for the presence of vascular lesions. Total occlusions were found in only 26.7% of PBC and 11.1% of viral-induced cirrhosis (Table 8), which was less frequent than the 46.8% found in alcoholics. Subtotal occlusions were somewhat more common, but in most cases these only showed a mild (1 +) degree of intimal proliferation. Although the number of nonalcoholic cases was small, the differences in frequency between the alcoholic and nonalcoholic groups reached statistical significance (p < 0.05). Table 7. Association of Veno-Occlusive Lesions With Phlebosclerosis in Precirrhotic Alcoholic Hepatitis With Portal Hypertension Q Phlebosclerosis an = 23. No. of cases with veno-occlusive lesions (subtotal or total) No. of cases without venoocclusive lesions Table 8. Incidence of Veno-Occlusive Lesions in Cirrhosis of Different Causes Primary Alcoholic biliary Viral cirrhosis cirrhosis cirrhosis Number Total occlusions 46.8% 26.7% 11.1% Subtotal occlusions 27.3% 40.0% 44.4% No veno-occlusive 25.9% 33.3% 44.4% lesions Discussion The propensity of alcohol-related liver damage to involve zone 3 of the Rappaport acinus (central region of the classical lobule ) is well recognized. Edmondson et al. (1) in 1963 pointed out the important pathophysiologic significance of the obliteration of terminal hepatic venules by the process of phlebosclerosis. This has been reiterated by many authors (2-5) and is now an established concept. In contrast to phlebosclerosis, phlebitis and va lesions have only rarely been noted in alcoholic liver disease. Edmondson et al. (1) in the original description of sclerosing hyaline necrosis illustrated 1 case with phlebitis. The occurrence of va lesions in alcoholics is mentioned by Scheuer (3) and illustrated by Patrick and McGee (4) in their respective textbooks, but no comment is made regarding the possible significance of this finding. Popper et al. (5) noted that va lesions may occasionally be seen, but suggested that they are rare and seldom of significant functional importance. In the present study, the frequent occurrence and functional importance of phlebosclerosis in alcoholic liver disease was again confirmed. The fact that va lesions were also a very frequent finding, being present to some degree in 74.1% of patients with cirrhosis and 52.1% of those with alcoholic hepatitis, was unexpected. We had not anticipated finding va lesions in so many cases. It is possible that with more extensive sampling, lesions might have been found in even more cases, since in a number of patients only one or two lesions were present on the available sections. Thus, the figures represent the minimum incidence in this population. The cause and pathogenesis of va lesions in alcoholic liver disease remains speculative at this time. In cases with cirrhosis, at least some of the lesions, especially those in intercalated veins, may result from decreased blood flow. Since obliteration of terminal hepatic venules by the process of phlebosclerosis would cause less blood to be delivered to vessels distal to the point of obstruction, the reduced flow might stimulate an intimal proliferation that

10 October 1982 VENO-OCCLUSIVE LESIONS IN THE ALCOHOLIC 795 would reduce the caliber of these vessels to a size appropriate for that amount of blood. This could explain the finding of va lesions in occasional cases of cirrhosis due to other causes such as chronic viral hepatitis, primary biliary cirrhosis, and Wilson's disease. It is also noteworthy that in none of these nonalcoholic liver diseases are va lesions found in the precirrhotic stages. On the other hand, the propensity of precirrhotic alcoholic liver disease to involve the perivenular regions with sinusoidal fibrosis might explain the occurrence of lesions in the terminal hepatic venules. This could also explain why va lesions are found so much more frequently in alcoholic liver disease than in other diseases in which the brunt of the injury is periportal, such as primary biliary cirrhosis or chronic active hepatitis. The finding that va lesions are highly correlated with fibrosis (Table 4) supports this mechanism. Veno-occlusive lesions could also result from a toxic injury to the endothelium of affected vessels. Leakage of plasma into the intima with subsequent ingrowth of mesenchymal cells, organization, and fibrosis might then produce these lesions. Similar mechanisms have been postulated for VOD of other causes such as Jamaican bush tea disease (9) and radiation injury (10), and could also be the cause of drug-induced VOD (11-16). The toxic injury could be due to ethanol, but alcoholic beverages also contain many other compounds including nitrosamines, mycotoxins, and other chemically active substances (17-18), any of which might be the injurious agent. Dfferences in the concentration of the injurious agent in different alcoholic beverages could explain why not all cases of alcoholic cirrhosis show va lesions. Veno-occlusive lesions could also be a reaction to the severe inflammation that may be associated with alcoholic liver disease. Intimal proliferation and luminal narrowing are frequently found in many parts of the body in vessels near areas of inflammation, such as gastrointestinal ulcers, ulcerated and necrotic tumors, and gangrenous limbs. Alcoholic va lesions could be a similar phenomenon in veins that drain the inflamed liver. The finding that neutrophilic infiltration was significantly associated with severity of va change (Table 4) supports this mechanism. Whether the phlebitis that was observed in some cases was related to this is not entirely clear, but it is possible that this represents a similar phenomenon. Veno-occlusive lesions in alcoholics, although morphologically similar to other forms of VOD, differ in several respects. Acute VOD, as seen with pyrrolizidine alkaloid or other toxic injury, characteristically shows severe congestion with necrosis of the parenchyma. Frequently, there is extravasation of erythrocytes into the intima of affected veins. These are not seen in alcoholic liver disease, which may partly explain why the vascular lesions have been so often overlooked. The lack of congestion may be due in part to the fact that alcoholic va lesions tend to occur in livers with fibrosis and also to the fact that alcohol-related liver damage occurs over a relatively long period of time. In contrast, fibrosis is only seen in toxic VOD during the chronic stage of the disease after extensive damage due to vascular occlusion has occurred. Thus, the differences between alcoholic and other forms of VOD may reflect the chronicity of the disease as well as the concurrent changes due to alcohol. The clinical importance of the various vascular lesions in alcoholics lies in their contribution to the development of portal hypertension. Since the lesions tend to occur together, it is not possible to deduce the exact role that each type of vascular alteration may have played. Nevertheless, it seems probable that both phlebosclerosis and va lesions contribute to portal hypertension. The relative contribution of the two lesions may vary from case to case. In some cases, va lesions were quite extensive and appeared to have played an important role, whereas in others only occasional va lesions were seen, and their role in portal hypertension was less clear. Factors other than vascular lesions probably also playa role, since there were a few cases of portal hypertension in precirrhotic alcoholic hepatitis with only mild phlebosclerosis and no va lesions (Table 7). On the other hand, it is possible that va lesions may play a greater role than suspected from their number. Total occlusion of one intercalated vein could block the venous outflow from many terminal hepatic venules and thus could have a relatively large effect on the development of portal hypertension. Likewise, since va lesions were sometimes widely scattered, there were undoubtedly cases in which more extensive sampling would have turned up lesions where the available sections showed none. Thus, the effect of alcoholic va lesions may be much greater than their extent would suggest. Hepatic venography, which might better define the extent and distribution of va lesions in such cases, is thus a promising area for future study. Another possible effect of va lesions which could have clinical significance is the apparent association of the lesions with atrophy in livers with cirrhosis (Table 5). Small cirrhotic livers are more likely to have va lesions than large cirrhotic livers. We have even seen occasional cases in which one part of the liver shows an early cirrhosis while sections from another part show an advanced cirrhosis with small, atrophic nodules. In such cases, the va lesions are present only in the atrophic, highly cirrhotic areas.

11 796 GOODMAN AND ISHAK GASTROENTEROLOGY Vol. 83, No. 4 Thus it is possible that VO lesions in cirrhotic livers contribute to atrophy with loss of functioning parenchyma and could consequently make the patient prone to hepatocellular failure and hepatic encephalopathy. Evidence to support this was sought in our cases, but death in hepatic coma was not more frequent in patients with small cirrhotic livers. Nevertheless the possibility remains that this could be a contributing factor in some patients. In summary, we have described in detail two vascular lesions, phlebitis and VO change, which are more common in alcoholic liver disease than has been previously recognized. The importance of phlebosclerosis and possible contribution of VO lesions to the development of portal hypertension have been demonstrated. The association of VO lesions with other findings related to alcoholic liver disease suggest possible factors in the formation of the vascular lesions, but the exact cause and pathogenesis of VO lesions in the alcoholic remain areas for future study. References 1. Edmondson HA, Peters RL, Reynolds TB, et al. Sclerosing hyaline necrosis of the liver in the chronic alcoholic: a recognizable syndrome. Ann Intern Med 1963;59: Christoffersen P, Poulsen H. Alcoholic liver disease. In: MacSween RNM, Anthony PP, Scheuer PJ, eds. Pathology of the liver. Edinburgh: Churchill Livingstone; 1979: Scheuer PJ. Liver disease of the alcoholic. In: Liver biopsy interpretation. London: Bailliere Tindall, 1980: Patrick RS, McGee JOD. Alcoholic liver disease. In: Biopsy pathology of the liver. Philadelphia: J.B. Lippincott Co., 1980: Popper H, Thung SN, Gerber MA. Pathology of alcoholic liver diseases. Sem Liver Dis 1981;1: Reynolds TB. Portal hypertension. In: Schiff L, ed. Diseases of the liver. Philadelphia: J.B. Lippincott, Co., 1975: Karasawa T, Chedid A. SclerOSing hyaline necrosis in noncirrhotic chronic alcoholic hepatitis. Am J Clin Pathol 1976; Galambos JT. Pathogenesis. In: Cirrhosis. Philadelphia: W.B. Saunders, Co., 1979: Bras G, Jelliffe DB, Stuart KL. Veno-occlusive disease of the liver with non-portal type of cirrhosis occurring in Jamaica. Arch PathoI1954;57: Fajardo LF, Colby TV. Pathogenesis of veno-occlusive liver disease after radiation. Arch Pathol Lab Med 1980;104: Marubbio AT, Danielson B. Hepatic veno-occlusive disease in a renal transplant patient receiving azathioprine. Gastroenterology 1975;69: Griner PF, Elbadawi A, Packman C.H. Veno-occlusive disease of the liver after chemotherapy. Ann Intern Med 1976; 85: Asbury RF, Rosenthal SN, Descalzi ME, et al. Hepatic venoocclusive disease due to OTIC. Cancer 1980;45: Shulman HM, McDonald GB, Matthews 0, et al. An analysis of hepatic veno-occlusive disease and centrilobular hepatic degeneration following bone marrow transplantation. Gastroenterology 1980;79: Mcintyre RE, Magidson JG, Austin GE, et al. Fatal venoocclusive disease of the liver following high dose l,3-bis (2- chloroethyl)-l-nitrosourea (BCNU) and autologous bone marrow transplantation. Am J Clin PathoI1981;75: Gill RA, Onstad GR, Cardamone JM, et al. Hepatic venoocclusive disease caused by 6-thioguanine. Ann Intern Med 1982;96: Tuyns AJ, Castegnaro M, Toussaint G, et al. Recherches concernant les facteurs etiologiques du cancer de l'oesophage dans I'ouest de la France. Bull Cancer (Paris) 1980;67: Schoental R. Relationships of Fusarium mycotoxins to disorders and tumors associated with alcoholic drinks. Nutr Cancer 1980;2:88-92.