HEPATIC LESIONS IN CHRONIC INFLAMMATORY BOWEL DISEASE

Transcription

1 GASTROENTEROLOGY Copyright 1967 by The Williams & Wilkins Co. Vol. 52, No. 2, Part 1 Printed in U.S.A. HEPATIC LESIONS IN CHRONIC INFLAMMATORY BOWEL DISEASE I. Clinical correlations with liver biopsy diagnoses in 103 patients ERL DORDAL, M.D., SEYMOUR GLAGOV, M.D., AND JOSEPH B. KIRSNER, M.D., PH.D. DepaTtments of Medicine and Pathology, University of Chicago, Chicago, Illinois Increasing attention has been directed to the state of the liver in patients with chronic inflammatory bowel disease, for it has become apparent that complicating hepatic disease may playa significant clinical role. 1-5 It has been suggested that the liver disorder may be part of a general systemic process also reflected in the intestinal disease, or that it may result from the chronic intestinal derangement and the accompanying malnutrition, portal bacteremia, or absorption of toxic materials, or both of the latter. Blood transfusions and therapeutic drugs also have been implicated. Some reports characterizing the hepatic lesions have been based entirely on the examination of postmortem material. 6, 7 However, the terminal changes noted in autopsy material may be quite different from those observed in liver biopsy speci- Received August 1, Accepted S eptember 27, Presented in part a t the 67th Annual Meeting of the American Gastroenterological Association, Montreal, Canada, May Address requests for reprints to: Dr. Erl Dordal, Department of Medicine, University of Chicago, 950 E. 59th Street, Chicago, Illinois This investigation was supported in part by United States Public Health Service Medical Research Grant 1136 from the National Institute of Arthritis and Metabolic Diseases, National Institutes of Health. Dr. Glagov is an E stablished Investigator of the American Heart Association. The authors wish to express their gratitude to Dr. Henry Rappaport, Professor of Pathology and Director of the Laboratory of Surgical Pathology at the University of Chicago, for his cooperation in making available the excellent histological preparations without which this study could not have been undertaken. 239 mens from living patients. A more complete understanding of the pathogenesis and natural history of the liver disease associated with chronic intestinal inflamma.tion could result from the examination of liver tissue obtained during various phases of the clinical course. Hopefully, information furnished by such a study would permit a more accurate clinical estimate of the severity of the liver disease and thus lead to more effective treatment and eventually to prevention of the complication. The present study was concerned with the relationship of liver biopsy diagnoses to the clinical course of inflammatory bowel disease in patients treated at the University of Chicago. The severity of the intestinal disorder, the therapeutic program, and the laboratory determinations at the time of liver biopsy w ere evaluated. The results of this study support the opinion that liver disease is a common and potentially serious complication of chronic inflammatory bowel disease and that moderately severe morphological abnormalities may be present despite biochemically normal liver function. It is shown that the severity of the liver lesions is not easily correlated with estimates of the severity of the bowel disease by current methods and that t he choice and duration of therapy were not related to the hepatic biopsy findings in any consistent manner. Although serum hepatitis occurred, the most frequent lesions were not those usually ascribed to the transfusion of infectious agents, nor was there any evidence from biopsy cultures of chronic infection of the liver.

2 240 DORDAL ET AL. Vol. 52, No.2, Part 1 colitis, regional a gn o e ~, i n c l u d u ~ nc. at g e ri v e ententls, and IleocolItIS, were established by the usual clinical, radiological, and proctoscopic criteria. Of these 750 patients 103 were subjected to liver biopsy (table 1).' Some had biopsies on more than one occasion, so a total of 122 specimens was available for examination. Initially, patients with inflammatory bowel disease were biopsied only when clinical findings consistent with hepatic disease were present; these included abnormal values for liver function tests or abnormal appearance of the liver at laparotomy. As awareness of the frequency of associated hepatic disease increased, patients with more subtle abnormalities of liver function were biopsied. 1\1ore recently, the presence of chronic intestinal inflammatory disease has been considered sufficient indication for liver biopsy. Individual differences among attending physicians in the degr.ee of concern with hepatic changes m bowel disease, the failure of some patients to consent to biopsy, and the presence of contrnindications to the biopsy procedure necessarily resulted in an incomplete sample. Case study. Analysis included estimates of the probable time of onset and the date of diagnosis of the bowel disease. Evaluation of the severity of the bowel disease was based on clinical evidences of intestinal bleeding, diarrhea and fever, general state of health and roentgenological and proctoscopic findings. Roentgenological descriptions were accepted as reported by the radiologist. All coexistent diseases. and abnormalities, including specified allergies, were noted. A complete record was made of all known TABLE 1. Liver biopsies in patients with inflammatory bowel disease at the University of Chicago f rom 1942 to 1964 Diagnosis Chronic ulcerative colitis.... Ileocolitis Regional enteritis. Total No. of patients hiopsied Materials and Methods therapeutic agents, including the quantity, frequency, and duration of administration of Case selection. During the 22-year period corticosteroids and adrenocorticotropic hormone, blood transfusions, and other intrave from January 1, 1942, to December patients were hospitalized at the U n i v e r ~ nous infusions, antibiotics, sulfonamides and sity of Chicago Hospitals and Clinics for supportive medications such as n a r c o tan- i c ~ chronic inflammatory bowel disease. The di ticholinergics, or tranquilizers. ' All available white blood cell counts and hemoglobin and serum alkaline phosphatase values were tabulated. The results of other laboratory examinations, in most instances performed within a few days of liver biopsy, were noted for each patient. These included: serum bilirubin, transaminase, cholesterol, and c h o l e s ester ~ e r ovalues; l serum total protein, ~ l b u and m m globulin, determinations; cepha- 1m and thymol flocculation, thymol turbidity, and Bromsulphalein retention tests serum calcium, phosphorus, blood urea nitrogen, and carotene levels; differential white blood cell c o ~ n hematocrit t s, levels, erythrocyte sediment ~ t lrates, O n reticulocyte counts, and prothrombill time determinations, lupus erythematosus preparations, sheep and latex agglutination tests and serological tests for syphilis. ' Liver biopsies. The percutaneous technique, mostly by means of the Franklin modification. of the Vim-Silverman needle, was used to obtam 104 specimens. A few w ere obtained with the Menghini needle. In 69 instances, a small fragment of the liver biopsy specimen and any blood or debris adherent to the biopsy needle were removed for microbiological culture in media for aerobic and anaerobic organisms, as well as for acid-fast bacilli and fungi. Eighteen specimens were obtained during laparotomy as wedge resections ; no cultures were made from these. Tissue for histological study was fixed in 10% neutral buffered formalin. Sections were stained with hematoxylin and eosin, the Heidenhain aniline blue connective tissue stain, and Wilder's ret!culum stain. In most instances, Perl's PrussIan blue test for iron was performed. Tissue sections were studied as unknowns with regard to modification of hepatic archit e c e ~, abnormalities u r in the portal triads, and various hepatocellular changes. Each finding was g r ~ d!or e d ea?h specimen. Finally, the anatomical diagnostic term which best summarized. the predominant pathological changes was assigned to each biopsy. Some slides were submitted again to the examiner, without his k n o ~ l eas. d a g e check, on the consistency of gradmg and diagnosis. In this paper anatomical liver disease is characterized by' the main pathological diagnosis only.

3 February 1967 HEPATIC LESIONS IN BOWEL DISEASE. I 241 Results Liver Biopsy Diagnoses The following analysis is based on the first biopsy diagnosis made on each patient. The distribution of liver biopsy diagnoses in the 103 patients is presented in table 2. Diagnoses included the following: cirrhosis or bridging portal hepatofibrosis, 20 patients; pericholangitis, 33 patients; fatty change, 22 patients; nonspecific reactive changes, 22 patients; viral hepatitis, 5 patients; amyloidosis, 1 patient. There were 19 repeat biopsies: seven were from 6 patients with cirrhosis; 9 followed an original diagnosis of pericholangitis or fatty change; in 3 patients, the diagnoses included viral hepatitis. Cirrhosis or bridging portal hepatofibrosis. Extensive fibrosis with modification of the normal lobular architecture was observed in biopsy specimens of 20 patients. In 13, the fibrosis appeared predominantly as bands bridging recognizable remnants of portal fields and often sparing central veins (fig. 1). Regenerative nodules and evidence of focal collapse were common, but the predominant pattern was a relatively uniform, diffuse, bridging portal fibrosis' usually with prominent bile ductule proliferation. Little or no cholestasis was present, but the clearly systematized bridging fibrosis was most reminiscent of the structural changes characterizing obstructive biliary cirrhosis. The descriptive term "portal-biliary type" of cirrhosis or bridging portal hepatofibrosis was used for these cases. In two instances, diffuse intralobular fibrous septae were associated with complete obliteration of hepatic architecture; these changes were characteristic of nutritional or alcoholic cirrhosis. Evidence of lobular and multilobular zones of collapse with triads identifiable within fibrous bands and patchy conservation of lobules was seen in five biopsies; these were considered to represent postnecrotic or posthepatitic cirrhosis. The distribution of the types of cirrhosis noted is given in table 2. Thus cirrhosis or bridging portal hepatofibrosis was not an uncommon finding in TABLE 2. Liver biopsy diagnoses in patients with inflammatory bowel disease Diagnosis Cirrhosis or bridging portal hepatofibrosis.... Portal-biliary cirrhosis or bridging portal hepatofibrosis... Postnecrotic or posthepatitic cirrhosis.... Nutritional cirrhosis Pericholangitis (triaditis).... Little inflammatory exudate.... Moderate inflammatory exudate... Marked inflammatory exudate.... No exudate.... Fatty change Focal..... Moderate Marked Fatty liver Nonspecific reactive change Minimal.... Moderately severe Severe.... Viral hepatitis Amyloidosis Total (initial biopsies).... a Total patients for this diagnosis. No. of patients 20a a a inflammatory bowel disease; the most frequent form was a relatively uniform diffuse bridging portal fibrosis a"ssociated with some degree of hepatic reconstruction. Pericholangitis. A moderate to marked degree of inflammatory cell infiltration in more or less intact portal fields was the principal feature in 30 biopsies (fig. 2). Three contained moderate portal fibrosis without cellular exudates which was presumed to be related to previous pericholangitis. Table 2 shows the distribution of the pericholangitis by intensity of the exudate. Portal sclerosis ranged from slight increases in collagen to marked fibrosis, with or without periportal extension of fine fibrous strands, but without bridging.

4 242 DORDAL ET AL. Vol. 52, No.2, Part 1 FIG. 1. Microscopic appearance of the liver of a patient with cirrhosis. Reconstruction of hepatic architecture was advanced but relatively uniform; bridging portal hepatofibrosis was a prominent feature (H & E, X 83). Bile duct proliferation often was a prominent feature. Focal fatty change and other nonspecific cellular changes were frequent in patients with pericholangitis. Portal inflammatory cells were mainly lymphocytes and a few plasma cells. In some cases, polymorphonuclear granulocytes exceeded mononuclear forms and the process then was considered to be acute. Pericholangitis or triaditis was the most frequent hepatic abnormality in the present series. There was no obvious relationship between the degree of portal fibrosis and the degree or type of inflammatory exudate. Fatty change. Biopsies with this main diagnosis showed distinct vacuolization of hepatocytes (fig. 3). If there was no associated pericholangitis, biopsies with minimal or "focal" fatty change were included in this group; specimens showing moderate or severe fatty change with minimal triaditis also were included. All cases of marked fatty change and "fatty liver" were grouped as fatty change, regardless of the degree of triaditis. Altogether, 22 specimens had the main diagnosis of fatty change: two showed complete fatty metamorphosis or fatty liver, five showed fatty vacuolization or more than half of the parenchymal cells (marked fatty change), 11 showed fatty change of one-fourth to one-half of the parenchymal cells (moderate fatty change); four contained only scattered small groups of fat-filled cells (focal fatty change). Hepatic fatty change was not infrequent in liver biopsies of patients with inflammatory bowel disease but was less frequent than has been noted in some autopsy series. 6 7 N onspecijic reactive change. Degenerative changes other than fatty change and not associated with any specific primary or secondary hepatic disease were noted in 22 biopsies (table 2). Findings s uch as

5 FIG. 2. Portal field in a liver biopsy which showed moderately severe pericholangitis. Inflammatory cells are abundant (H & E, X 455). FIG. 3. Moderate fatty change was the predominant change in this liver biopsy of a patient with ulcerative colitis. Focal triadal inflammation was minimal (H & E, X 175). 243

6 244 DORDAL ET AL. Vol. 52, No.2, Part 1 parenchymal cell hydropic swelling, hyperpigmentation, or altered cytoplasmic affinity for hematotoxylin or eosin dyes, nuclear atypia and vacuolization, scattered necroses involving a few cells, and unusual prominence of von Kupffer cells occurred in varying degree. In eight biopsies, these changes were judged as severe, in five as moderate, and in nine as minimal. Other biopsy diagnoses. Viral hepatitis: seven biopsies, all from patients with ulcerative colitis (5 men and 2 women), showed the characteristic morphological findings of viral hepatitis. In each instance, clinical features typical of this disease were present at the time of biopsy. Five probably had serum hepatitis since symptoms followed blood transfusions after a suitable incubation period. Two of the patients had had a previous liver biopsy, 1 showing marked fatty change and the other mild chronic pericholangitis. Amyloidosis: deposition of hyaline metachromatic material in the walls of blood vessels and sinusoids, consistent with the diagnosis of amyloidosis, was found in the biopsy of 1 patient. Subsequent studies revealed deposition of amyloid in other organs. Other significant biopsy findings. Granulomas: lesions interpreted as inflammatory granulomas were found in the biopsies of eight patients. In some instances, groups of histiocytes and giant cells of the Langhans type were rimmed by lymphocytes; caseation was not noted. Other less cellular, focal, nodular lesions contained collagen; focal calcification was noted in several TABLE 3. Interval liver biopsy diagnoses in patients with pericholangitis Degree of pericholangitis at first biopsy Interval between biopsies wk Degree of pericholangitis at second biopsy Severe chronic 2 Moderate chronic Severe acute and 3 Moderate chronic chronic Severe chronic 52 Moderate chronic Acute and chronic 95 None (moderately severe reactive change) Mild chronic 300 None (severe reactive change) large sclerotic lesions. Cirrhosis, hepatofibrosis, fatty change, pericholangitis, or nonspecific reactive change also was noted in biopsies containing granulomas. Carcinoma: two biopsy specimens contained metastatic carcinoma and mild chronic pericholangitis. Successive biopsies. Two or more biopsies were performed on 18 patients at intervals ranging from 1 month to 19 years. In three instances, one of the two interval biopsies showed characteristic viral hepatitis without evidence of underlying chronic liver disease. Acute or subacute changes possibly related to inflammatory bowel disease were obscured by the hepatitic lesions so that comparisons with previous or subsequent biopsy findings could not be made. In 6 patients, the initial biopsy revealed cirrhosis; repeat biopsies at intervals of 1 month to 19 years revealed no significant change in either the type or severity of the process. Initial and subsequent biopsy findings in patients with pericholangitis are compared in table 3. All of 5 patients with an initial biopsy of pericholangitis had less severe lesions in the second biopsy. Three with severe acute and chronic pericholangitis initially showed less disease subsequently (2, 3, and 52 weeks later). Two with acute and chronic pericholangitis and mild pericholangitis at first showed only nonspecific reactive changes later (at 2 years and 6 years, respectively). Fatty change was diagnosed in 4 patients who had repeat biopsies. Two patients with marked and moderate fatty change had the same diagnosis on specimens obtained 6 and 13 weeks later, respectively. A patient with fatty liver initially still showed marked fatty change 48 weeks later; another patient with moderate fatty change at first had a nearly normal liver without fat after 1 year. The findings thus indicate that neither pericholangitis nor fatty change was necessarily persistent or progressive lesions. Microbiological Examination of Liver Biopsy Material Of the 69 cultures of needle biopsy fragments, 52 produced no growth. Each of 15 cultures produced only one organism.

7 February 1967 HEPATIC LESIONS IN BOWEL DISEASE. I 245 Because the strains found usually are not pathogens and because of the minute growth or appearance in the broth only, these were classified as contaminants. Species included diphtheroids, Staphylococcus albus, and Penicillium strains. Micrococcus was found in three instances, but in broth only. One culture grew Paracolobactrum aerogenoides; the corresponding hepatic lesion was severe reactive change. A fragment from a biopsy which showed fatty liver grew out anaerobic nonhemolytic Streptococcus. Repeat biopsy of the same patient showed the same hepatic morphology but the culture was negative. The technical limitations of the culture method were probably not the basis for the consistently negative results, for pathogens were demonstrated by the same technique in patients with well documented ascending cholangitis without inflammatory bowel disease. Relation of Clinical Features of Bowel Disease to Liver Biopsy Diagnoses Type of inflammatory bowel disease. Of the 20 patients with cirrhosis or bridging portal hepatofibrosis, 15 had ulcerative colitis, 4 had ileocolitis, and 1 had regional enteritis. All of the patients with portalbiliary type of cirrhosis or bridging portal hepatofibrosis had ulcerative colitis or ileocolitis. Patients with pericholangitis included 22 with ulcerative colitis, 8 with ileocolitis, and 3 with regional enteritis. Fifteen patients with fatty change had ulcerative colitis, 1 had ileocolitis, and 6 had regional enteritis. The diagnosis of nonspecific reactive change was made in 18 patients with ulcerative colitis, 2 with ileocolitis, and 2 with regional enteritis. Although the total number of patients is too small to permit a statistical anaylsis, it should be noted that bridging portal hepatofibrosis was not noted in regional enteritis and that 6 with regional enteritis had fatty change. Five of the 8 patients with hepatic granulomas had regional enteritis or ileocolitis. Three patients had a clinical diagnosis of ulcerative colitis, but granulomas were demonstrated in histological sections of the colons of 2 of these; the colon of the 3rd has not been biopsied. Sex and age distribution of patients for each liver biopsy diagnosis. The 103 biopsied patients included 60 men and 43 women. The average age at the time of biopsy was, for men, 36.6 years, for women, 36.4 years, and, for the entire group, 36.5 years. Cirrhosis of the portal-biliary type was more common in men. The average age of both men and women with a diagnosis of fatty change was notably higher than that of any other biopsy diagnostic group. Duration of inflammatory bowel disease. The time interval between the onset of symptoms attributable to inflammatory bowel disease and the liver biopsy ranged from less than 1 year to 32 years, w.ith an average of 8 years. The time between a definite diagnosis of inflammatory bowel disease and liver biopsy ranged from less than 1 year to 26 years, with an average of approximately 6 years. Liver biopsy was performed within 1 year of the original diagnosis of inflammatory bowel disease in 17 instances. In 1 case, the diagnosis of inflammatory bowel disease was made 5 years after the onset of symptoms and 3 years after the initial liver biopsy. However, proctoscopy had been performed at about the time of the open liver biopsy and the description recorded by the proctoscopist is that of ulcerative colitis. The normal appearance of the colon on X-ray and at laparotomy may have caused the attending physician to underestimate the proctoscopic findings. Without the previous proctoscopic demonstration, this case would have been considered to 1:)e liver disease, preceding the development of ulcerative colitis. There were no significant differences in average time elapsed between onset of symptoms and biopsy for any of the biopsy diagnoses, although the longest time interval was in patients with fatty change and the shortest was in those individuals with nonspecific reactive change. Clinical course of inflammatory bowel disease. The clinical severity of the bowel disease was evaluated on the basis of such factors as bleeding, diarrhea, fever, therapeutic regimen, and general state of health. Cirrhosis and bridging portal hepatofibrosis. At the time of liver biopsy, colitis could be considered severe in only 1 pa-

8 246 DORD.4L ET AL. Vol. 52, No.2, Part 1 tient. The majority had mild symptoms. Proctoscopic examination at the time of biopsy revealed essentially normal mucosa in 2 of 13 patients so studied; the remainder showed healing or mild to moderately severe disease. Roentgenological examination indicated involvement of the entire colon in 5 and of the descending colon only in 1; 3 had radiologically normal colons. None of these patients had surgery for bowel disease before biopsy, but 1 subsequently underwent ileostomy and 5 had venous shunting procedures to relieve portal hypertension. Two had undergone laparotomy, 1 for appendicitis and the other for intussusception. Two were operated on for suspected biliary tract disease. Of these, 1 had a cholecystectomy; no stones were present and there was only minimal evidence of cholecystitis. The other patient had a preoperative diagnosis of common duct stone; although no stones were found, biopsy of the common duct revealed severe inflammation and fibrosis. The patient remained jaundiced despite repeated exploration of the biliary tract, and he eventually died. At autopsy, a carcinoma of the bile duct was found deep in the hilus of the liver. Four of the 20 patients with cirrhosis or marked hepatofibrosis have died. Pericholangitis. In most instances, bowel disease activity at the time of liver biopsy was considerably greater than in patients with cirrhosis or bridging portal hepatofibrosis. Twelve of the 33 patients with pericholangitis had undergone surgery for the inflammatory bowel disease; procedures varied from resection of the ileum to combined abdominal perineal resection of the colon and rectum. Eight patients had other operations including appendectomy (3), hemorrhoidectomy (2), cholecystectomy (1), lysis of adhesions (1), and an adrenalectomy for metastatic breast carcinoma (1). Of the 22 patients in this category who had ulcerative colitis, proctoscopic examination revealed mild to moderate activity in 20 and inactivity in 2. Roentgenological examination indicated involvement of the entire colon in 7, transverse and descending colon in 2, descending colon only in 2, and rectosigmoid only in 3; 5 had normal colon X-ray studies. Although some individuals with acute pericholangitis had relatively active bowel disease, others appeared to have no active disease at the time of biopsy. Of the 8 patients with ileocolitis in the pericholangitis groups, there was roentgen evidence of activity in all except the 1 patient with an ileostomy. Of the 3 patients with regional enteritis, 1 had apparently normal X-ray findings and 1 had involvement of the terminal ileum with fistula formation; no recent X-rays of the 3rd were available. Four of the patients who had pericholangitis have died. One died of hepatic failure following viral hepatitis, 1 of carcinoma of the colon, and another of carcinoma of the breast. The 4th patient died of heart failure associated with chronic lung disease. Fatty change. The activity of the bowel disease was comparable to that found in patients with pericholangitis. There were 14 patients with chronic ulcerative colitis. Proctoscopy generally revealed more activity than was noted in patients with pericholangitis. Roentgenological examination revealed total involvement of the colon in 6 patients, transverse colon in 2, descending colon only in 2, and rectosigmoid only in 2; radiologically normal colons were demonstrated in 2 individuals. Of the two patients with ileocolitis and fatty change, only 1 had recent radiological demonstration of involvement of the terminal ileum and ascending colon. Three of the 6 patients with regional enteritis and fatty change had recent abnormal X-ray findings. Of the 22 patients with the diagnosis of fatty change, 6 had bowel surgery not related to their bowel disease, all prior to liver biopsy. Only 1 patient in this group had died, the cause being ventricular fibrillation and cardiac arrest. Nonspecific reactive changes. The range of bowel disease activity was comparable in all respects with that noted in patients with pericholangitis. Eighteen of the patients had chronic ulcerative colitis; proctoscopic findings ranged from "essentially normal" to "very severe." Roentgenological examination revealed involvement of the entire colon in 10, transverse colon in 3, descending and sigmoid colon in 3, and

9 Febmary 1967 HEPATIC LESIONS IN BOWEL DISEASE. I 247 rectosigmoid in 1; 4 patients had no radiologically demonstrable disease. Three patients underwent surgery related to their inflammatory bowel disease prior to biopsy; 7 had other types of abdominal surgery. Recent X-ray studies of the 2 patients with regional enteritis and of the 2 with ileocolitis revealed persistent characteristic abnormalities. Relation of Other Diseases to Liver Biopsy Diagnoses Only 12 patients who did not have viral hepatitis at the time of biopsy gave histories of viral hepatitis. No relationship between previous hepatitis and liver biopsy diagnosis category was noted; however, 3 of the 5 patients with postnecrotic cirrhosis gave histories consistent with previous hepatitis. Roentgenographic studies of the biliary tree were performed on 49 patients. Four patients had radiological evidence of cholelithiasis; 11 had faintly visualizing gall bladders, and 1 had a nonvisualizing gall bladder. Nine patients with a previous cholecystectomy included 4 with definite cholecystitis without stones. Pericholangitis and biliary portal hepatofibrosis were found in the presence of a normal extrahepatic biliary system. However, the biliary system frequently was abnormal. An individual with ileocolitis had amyloid deposits in the liver and was found to have widespread amyloidosis at autopsy 1 year later. Neither the presence of diabetes mellitus (8 patients) nor the admitted consumption of alcohol (31 patients) appeared to be related to any particular liver biopsy diagnosis group. A clinical history of allergy was more common in patients with biliary type cirrhosis than in the other groups; 50% of these patients had allergies as compared to less than 35% for the other groups. Relation of Hematological and Serological Findings to Liver Biopsy Diagnoses The range of hemoglobin concentrations, the average hemoglobin concentration, and the proportion of anemic patients were approximately the same for each liver biopsy category. Although most patients had high white blood cell counts during the course of their bowel disease, leukocytosis was not a prominent finding at the time of biopsy. Forty-one patients had lupus erythematosus cell preparations, including some subj ects in each of the liver diagnostic categories. All of the preparations were negative. Serological tests for syphilis were performed on 94 patients. The single positive test occurred in a patient who had a previous clinical diagnosis of syphilis. R elation of Therapeutic Agents to Liver Biopsy Diagnoses Patients in each of the liver biopsy diagnosis categories received one or more of the therapeutic agents most commonly administered to patients with inflammatory bowel disease, i.e., corticosteroid compounds, sulfonamides, and blood transfusions (table 4). Except for the relatively small number of patients with cirrhosis who received blood, there was a comparable proportion of the patients in each diagnostic category receiv.ing each therapeutic agent. Furthermore, there were patients in each liver biopsy group that had not received one or another of the therapeutic agents, and there were some who TABLE 4. Therapy related to liver biopsy diagnosis Liver biopsy diagnosis Total no. of patients No. of patients given Blood Corticosteroids Sulfonamides CirrhOSis (all types) or bridging hepatofibrosis (35%) 12 (60%) 14 (70% ) Peri cholangitis , (60%) 21 (63%) 27 (82%) Fatty change (50%) 16 (73%) 20 (91%) Heactive change (45%) 13 (59%) 19 (86%)

10 248 DORDAL ET AL. Vol. 52, No.2, Part 1 had not received any. Medications such as deodorized tincture of opium, antibiotics, tranquilizers, and anticholinergics were used so widely and in such varying degree that no meaningful correlation of biopsy diagnoses with individual drugs was feasible. Correlation of Laboratory Tests of Hepatic Function with Liver Biopsy Diagnoses Serum alkaline phosphatase. Repeated determinations over months or years revealed wide variation in serum alkaline phosphatase values in individual patients with inflammatory bowel disease. The results were analyzed in two ways: (1) serum alkaline phosphatase values near the time of the biopsy were correlated with the biopsy findings, and (2) all values during the entire clinical course were charted for each patient, in an attempt to discover possible long term trends and to correlate these with the biopsy findings. Serum alkaline phosphatase near the time of biopsy. The findings are presented in figure 4. The average of serum alkaline phosphatase values determined at the time of biopsy for the 13 patients with portal- " E '" 50 - ~ ~ ~ 20 '" II) co o LIVER FUNCTION STUDIES IN PATIENTS WITH INflAMMATORY BOWEL DISEASE BSP RETENTION ALKALIN E PHOSPHATASE I -.- I , I " " 4 J 1... A i'5 '* C PC FC RC C PC FC RC "' 0'" ~ 0( 1: 30" '" o 1: z I o ::; 0( o.;{ '" lo => "' '" II) FIG. 4. Serum Bromsulphalein retention and alkaline phosphatase levels in patients with chronic inflammatory bowel disease. C, cirrhosis; PC, pericholangitis; FC, fatty change; RC, nonspecific reactive change. In general, severe liver lesions corresponded to elevated values for both of these tests of hepatic function. biliary type cirrhosis or bridging portal hepatofibrosis was 15.5 Bodansky units; 1 subject had a normal alkaline phosphatase level (less than 5.0 Bodansky units). Average of the values of the 33 patients in the pericholangitis group was 13.3 Bodansky units; levels ranged from normal (15 patients) to 54 units. Ten of the 22 patients with fatty change had normal alkaline phosphatase values at the time of biopsy; the average level was 5.1 units. In the group with nonspecific reactive changes, the average alkaline phosphatase value at the time of biopsy was 6 Bodansky units; 13 of the 22 patients had normal levels. Serum alkaline phosphatase throughout the course of the disease. Individual patients showed great variation in serum alkaline phosphatase levels during the course of their inflammatory bowel disease. Serial determination of the serum alkaline phosphatase revealed a decrease in the serum level with improvement in the bowel disease during 31 episodes in 24 patients. The improvement in the bowel disease and the decrease in the serum alkaline phosphatase were associated with the administration of corticosteroids in 24 of the 31 episodes. Fourteen patients had 21 instances of a rise in the serum alkaline phosphatase with an exacerbation of the bowel disease. Twenty-nine patients had a relatively normal alkaline phosphatase, regardless of exacerbations and remissions of the bowel disease. Seven patients continued to have relatively stable but abnormal serum alkaline phosphatase values as the bowel disease varied in activity. There were nine instances (not included above) in which serial determinations of alkaline phosphatase indicated a progressive rise following the administration of blood. The rise sometimes occurred, although the patient improved clinically following the transfusion. The rise occurred within a very few days of transfusion, and the patient did not subsequently develop viral hepatitis. There were more instances in which no change was noted after a blood transfusion, but, when the increases did occur, they were so striking as to be noteworthy. Thus, although the serum alkaline phosphatase

11 February 1967 HEPATIC LESIONS IN BOWEL DISEASE. I 249 levels frequently corresponded with the clinical activity of the bowel disease, the levels remained stable in many patients despite clinical variation. Bromsulphalein (ESP) retention test. The distribution of 45-min BSP retention values for the biopsied patients is given in figure 4. Seven patients with cirrhosis of the portal-biliary type or bridging portal hepatofibrosis had BSP tests; only 1 had normal results. Serum bilirubin. Serum bilirubin levels were known in almost all cases at or near the time of biopsy. Only 23 patients had elevated values; 5 of these had viral hepatitis, 10 had cirrhosis or bridging portal hepatofibrosis, and 4 had pericholangitis. Of the 4 with pericholangitis, 1 also had hepatic involvement by metastatic carcinoma, another probably had a common duct stone, and another had an abdominal abscess with hepatic granulomata. The 4th was a patient who also had elevated serum cholesterol and xanthelasma and was considered to have the characteristic clinical features of primary biliary cirrhosis. Three patients with fatty change on liver biopsy and elevated serum bilirubin were very ill and had other toxic hepatic changes. Eight patients with histological cholestasis in the liver biopsy tissue had no corresponding elevation of serum bilirubin; 15 had both histological cholestasis and elevated serum bilirubin; 6 had histological cholestasis on the initial biopsy without elevated serum bilirubin. Serum transaminase. Thirty-one patients had serum transaminase determinations. The highest serum transaminase recorded in a patient without viral hepatitis was 157 units (normal = 40 units); the liver biopsy diagnosis was nonspecific reactive change. Fourteen patients had elevated values, but the range of values was comparable in each of the liver biopsy diagnosis categories. There was no obvious relationship between the clinical activity of the bowel disease and transaminase levels. Serum cholesterol. Ninety-five subjects had serum cholesterol determinations. Fifty-nine had values considered to be in the normal range (under 200 mg per 100 ml). Twenty-two (14 males and 8 females) had serum cholesterol concentrations between 200 and 300 mg per 100 ml; 11 had values greater than 300 but less than 400 mg per 100 ml; 3 patients had levels greater than 400 mg per 100 m!. The highest value was 1000 mg per 100 ml in the patient with other clinical and laboratory findings consistent with the clinical syndrome of primary biliary cirrhosis. Eleven of 20 patients with cirrhosis, 11 of 29 patients with pericholangitis, 7 of 21 patients with fatty change, and 7 of 20 patients with reactive change had values over 200 mg per 100 m!. Serum proteins. All of the patients had serum protein determinations at or near the time of biopsy. The average serum albumin concentration was 3.6 g per 100 ml in each of the liver biopsy diagnosis categories. There were 56 patients with globulin levels between 1.1 and 3.0 g per 100 ml and 41 with higher values. Twentyseven of these had between 3.0 and 4.0 g per 100 ml; 14 had more than 4.0 g per 100 m!. The average serum globulin in cases of cirrhosis or bridging portal hepatofibrosis was 3.3 g per 100 ml; in pericholangitis it was 3.2 g per 100 ml; in fatty change it was 2.7 g per 100 ml; and in nonspecific reactive change it was 2.8 g per 100 m!. Although patients with cirrhosis and pericholangitis had slightly higher average globulin levels than the others, the range was approximately the same in all liver biopsy diagnosis groups. This finding suggests that other factors such as protein loss in the stool may have a greater influence than the liver disease upon the serum protein levels of patients with chronic inflammatory bowel disease: Serum carotene. Carotene concentrations were determined on sera of 63 patients. Average serum carotene in individuals with cirrhosis or bridging portal hepatofibrosis was 149 mg per 100 m!. Subjects with pericholangitis averaged 176 mg per 100 ml; those with fatty change averaged 99 mg per 100 ml; and those with nonspecific reactive change averaged 110 mg per 100 m!. Nine of the 17 pa-

12 250 DORDAL ET AL. Vol. 52, No.2, Part 1 tients with fatty change, who had a serum carotene determination, had values less than 100 mg per 100 ml. The relatively low serum carotene values in patients with fatty change or nonspecific reactive change may reflect nutritional abnormalities which contribute also to the liver changes. Flocculation tests. The results of cephalin flocculation, thymol flocculation, and thymol turbidity tests were available for 88 patients. Ten of the 16 results on patients with cirrhosis or bridging portal hepatofibrosis were abnormal; 10 of the 30 tested patients with pericholangitis had elevated flocculations. However, only 4 of 22 patients with fatty change and 4 of 20 tested patients with nonspecific reactive change had abnormal results. Although flocculation tests lack specificity, elevated values in the patients tested seemed to reflect relatively more severe liver disease. Blood urea nitrogen. Ten of the 98 patients with recorded data concerning blood urea nitrogen had values higher than 15 mg per 100 ml (normal value = 9 mg per 100 ml); the highest level was 21 mg per 100 ml. Six of the patients with elevated values had clinically diagnosed chronic renal disease; 1 of these had chronic membranous glomerulonephritis on renal biopsy and another a perinephric abscess. Discussion Liver biopsy diagnoses made on specimens obtained from 103 patients with inflammatory bowel disease could be grouped into four main categories: cirrhosis or bridging portal hepatofibrosis, pericholangitis, fatty change, and nonspecific reactive change. The most common, pericholangitis, was characterized by abundant inflammatory exudate in portal fields often accompanied by bile duct proliferation, and portal fibrosis ranged from a slight increase in collagen to stellate portal enlargement with porto-portal bridging. The relatively uniform cirrhosis of the portalbiliary type seen in biopsies of 13 patients appeared to be an extension of the portal fibrogenic process. Interval biopsies in individual patients which might have depicted a progression of changes from pericholangitis to cirrhosis were not available. However, all stages in the transition from mild portal fibrosis to cirrhosis could be found in the biopsy material taken as a whole. The findings on repeat biopsy of some of the patients with pericholangitis indicate that the hepatic fibrosis may not necessarily be progressive. These observations are in general agreement with those of other investigators who have studied the hepatic disorders as- 80ciated with chronic inflammatory bowel disease. s However, our material included a relatively large number of cases of cirrhosis and portal hepatofibrosis. The descriptive terms cirrhosis of the portalbiliary type and bridging portal hepatofibrosis were selected to emphasize that the sclerosing process is diffuse and relatively uniform, and that fibrosis can be quite advanced despite the persistence of distinguishable lobules. This type of systematized alteration of hepatic architecture is not characteristic of either postnecrotic or nutritional cirrhosis but is the type usually noted in biliary cirrhosis. In a recent report, Warren et al.9 have discussed the association of sclerosing cholangitis with ulcerative colitis; the necessity for strict criteria in the diagnosis of sclerosing cholangitis was emphasized. Some of our histological findings correspond with those noted by these authors, but the state of the common bile duct had not been adequately investigated in our subjects. Nevertheless, there were patients in our series who, at operation or autopsy, showed evidence of sclerosing cholangitis. However, it should be emphasized that portal fibrosis, biliary type cirrhosis, or pericholangitis may be present in patients with apparently normal extrahepatic biliary passages. Stauffer et ap have stated that hepatic cholestasis is frequently associated with ulcerative colitis. Their case selection was largely determined by the presence of clinical jaundice. Jaundice was not a major criterion for biopsy in our series. Mistilis et a1. 5 divided their biopsied patients into symptomatic and asymptomatic categories; the larger, asymptomatic group did not have clinically prominent jaundice. Granulomas in the liver seemed to re-

13 February 1967 HEPATIC LESIONS IN BOWEL DISEASE. I 251 flect granulomas in the bowel. The etiology of the hepatic granulomas is no more apparent than those in the bowel. They may be nonspecific inflammatory reactions but are more likely caused by the same agent responsible for the similar bowel le SIOns. No particular type of predominant hepatic histological change could be related to either the duration or clinical severity of the bowel disease as revealed by proctoscopic or radiological examination. Although this may indicate that current methods for the estimation of the severity of inflammatory bowel disease are inadequate, it is also conceivable that some poorly understood aspects of the bowel disease are detrimental to the liver even when bowel symptoms are minimal. Study of biopsy material of liver and intestine obtained simultaneously from the same patient might be instructive in this regard. Hepatic changes in patients with other bowel diseases such as chronic diverticulitis, parasitic infections, dysentery, and carcinoma should also be investigated. Portal bacteremia has been demonstrated in ulcerative colitis. 10 However, liver biopsy cultures were almost entirely negative, indicating that the liver probably does not become chronically infected. It is, of course, possible that very low grade infections could not be detected by our method or that products of dead bacteria induce the hepatic inflammatory reaction. If the liver is not chronically infected, it may be preferable to direct therapy exclusively against the bowel disease, especially since some of the broad spectrum antibiotics which have been recommended for treatment of portal triaditis may intensify the inflammatory bowel diseasep, 12 Whether bacteria lodged in the liver are alive or dead, portal bacteremia may be a relatively insignificant factor in hepatic injury compared to the quantity and variety of toxic substances which may be absorbed from the ulcerated, infected intestine. Further study of the portal blood stream in inflammatory bowel disease is needed. Tests which have been used to detect possible autoimmune processes, including the serological tests for syphilis, lupus erythematosus cell preparation, and serum globulin level, did not correlate with any of the liver biopsy diagnoses. No single therapeutic agent or combination of agents corresponded to the development of any particular type of liver damage; there was no relationship between the severity of the liver disease and either the duration of therapy or dose levels of therapeutic agents. If drug hypersensitivities or idiosyncrasies occurred, these did not produce any characteristic systemic reactions or hepatic morphological changes. Serum alkaline phosphatase and Bromsulphalein retention were valuable in the diagnosis of liver disease. However. the observation that these tests gave values in some patients with histologically manifest hepatic damage indicates that liver biopsy may be necessary for the evaluation of the liver in patients with chronic intestinal disease. The grouping of the liver biopsy diagnoses, according to the predominant hepatic change, permitted evaluation of the possible relationships between individual factors in the accompanying chronic inflammatory bowel disease and the particular hepatic lesions. Thus each group could be compared with the other, despite the lack of normal controls; such normal controls as are available in human biopsy series are characteristically inadequate and cannot meet any set of exact criteria. Since various types of hepatic damage occurred with any given bowel lesion, factors other than the mere presence of inflammation must be operative. Thus an evaluation of the specificity of the hepatic changes corresponding to particular factors in inflammatory bowel disease could be explored, using other clinical states. Some or all of the changes could be associated with other chronic illnesses, specifically systemic diseases or precise metabolic alterations as yet undefined. It is difficult to establish any pathogenetic relationship between inflammatory bowel disease and liver lesions while the n ~ r m a l

14 252 DORDAL ET AL. Vol. 52, No.2, Part 1 etiology of the bowel disease remains unknown. Colectomy has been proposed for the alleviation of severe liver disease in patients with ulcerative colitis. However, it has not yet been shown that. the liver disease progresses or becomes irreversible because of persistent bowel disease activity. The severity of the intestinal disease, per se, probably remains the best criterion for a decision regarding surgery. Should the relationship of the liver disease to the bowel disease be more clearly defin ed at some future time, liver biopsy findings could be utilized in establishing the time for operative intervention, regardless of the apparent clinical severity of the bowel disease. However, should the lesions in the liver or common duct, or both, and the intestine be independent manifestations of an underlying systemic disease, the removal of the colon or other inflamed bowel could not be expected to alter the course of the liver disease. Summary Liver biopsies of 103 patients with inflammatory bowel disease were classified according to the diagnostic term which best described the hepatic changes. Pericholangitis was the main biopsy diagnosis in 33 patients, fatty change in 22, nonspecific reactive change in 22, and cirrhosis or bridging portal hepatofibrosis in 20. The most frequent form of cirrhosis was a relatively uniform, systematized alteration of hepatic architecture resembling the pattern usually noted in biliary cirrhosis. All stages between mild portal fibrosis and cirrhosis were seen in the biopsy material as a whole. Therapeutic agents used in treating inflammatory bowel disease could not be related to either the type or severity of liver lesions. Intensity or type of bowel disease, as defined by clinical criteria of severity or duration, did not correlate well with either the type or severity of the liver disease. Bacteriological cultures of material from 69 of the liver biopsies were essentially negative, suggesting that infection in the liver may not be a factor in the hepatic changes. Serum Bromsulphalein retention and alkaline phosphatase levels were the best laboratory measures of the presence of hepatic lesions. Since it was not possible to define any consistent relationship between chronic bowel disease and liver injury, and since liver function tests may be normal in the presence of hepatic injury, liver biopsy must be utilized to determine the type and severity of the liver disease. No specific recommendations concerning therapeutic measures can be made at this time; adequate treatment of the bowel disease appears to be the best available measure for the prevention and control of the hepatic complications. REFERENCES 1. Palmer, W. L., J. B. Kirsner, M. B. Goldgraber, and S. S. Fuentes Diseases of the liver in chronic ulcerative colitis. Amer. J. Med. 36: Palmer, W. L., J. B. Kirsner, M. B. Goldgraber, and S. S. Fuentes Disease of the liver in regional enteritis. Amer. J. Med. Sci. 246: Stauffer, M. H., W. C. Sauer, W. H. Dearing, and A. H. Baggenstoss The spectrum of cholestatic hepatic disease. J. A. M. A. 191: Mistilis, S. P Pericholangitis and ulcerative colitis. 1. Pathology, etiology and pathogenesis. Ann. Intern. Med. 63: Mistilis, S. P., A. P. Skyring, and S. J. M. Goulston Pericholangitis and ulcerative colitis. II. Clinical aspects. Ann. Intern. Med. 63: Monto, A. S The liver in ulcerative disease of the intestinal tract: functional and anatomical changes. Ann. Intern. Med. 50: Jones, G. W., A. H. Baggenstoss, and J. A. Bargen Hepatic lesions and dysfunction associated with chronic ulcerative colitis. Amer. J. Med. Sci. 221: Vinnik, 1. E., and F. Kern, Jr Liver diseases in ulcerative colitis. Arch. Intern. Med. (Chicago) 112: Warren, K. W., S. Athanassiades, and J. 1. Monge Primary sclerosing cholangitis. Amer. J. Surg. 111: Brooke, B. N., and G. Slaney Portal