Increased Plasma Tumor Necrosis Factor in Severe Alcoholic Hepatitis George L.A. Bird, MB; Nick Sheron, MB; A.K. John Goka, MB; Graeme J. Alexander, MB; and Roger S. Williams, MD
Study Objective: To determine whether elevated tumor necrosis factor levels contribute to the clinical manifestations and complications of severe acute alcoholic hepatitis and to evaluate the relation between tumor necrosis factor and plasma levels of endotoxin and interleukin-1/3. Design: Prospective, controlled study. Setting: The liver unit of a university teaching hospital. Patients: We studied 21 patients with acute severe alcoholic hepatitis. There were four control groups: patients with inactive alcoholic cirrhosis, alcoholic persons without liver disease, patients with impaired renal function, and normal subjects. Measurements and Main Results: With one exception, patients with alcoholic hepatitis had higher tumor necrosis factor levels (mean, 26.3 ng/L; 95% CI, 21.7 to 30.9) than normal subjects (6.4 ng/L; CI, 5.4 to 7.4). Patients who subsequently died had a higher tumor necrosis factor level (34.7 ng/L; CI, 27.8 to 41.6) than survivors (16.6 ng/L; CI, 14.0 to 19.2). In patients with alcoholic hepatitis, tumor necrosis factor levels correlated positively with serum bilirubin (r = 0.74; P = 0.0009) and serum creatinine (r = 0.81; P = 0.0003). Patients with alcoholic hepatitis had higher tumor necrosis factor levels than patients with inactive alcoholic cirrhosis (11.1 ng/L; CI, 8.9 to 13.3) and severely alcoholic persons without liver disease (6.4 ng/L; CI, 5.0 to 7.8). Patients with abnormal renal function had lower tumor necrosis factor levels (14.1 ng/L; CI, 5.4 to 22.8) than patients with alcoholic hepatitis. Serial samples obtained during a 1-week period from patients with alcoholic hepatitis showed no significant change in tumor necrosis factor when patients who died were compared with survivors. No correlation was found between tumor necrosis factor and plasma endotoxin. Levels of interleukin-1/3 did not exceed 20 ng/L. Conclusions: Elevations in tumor necrosis factor in alcoholic hepatitis are most marked in severe cases, suggesting that tumor necrosis factor plays a role in the pathogenesis.
Approximately 57% of patients hospitalized because of alcoholic liver disease have evidence of diffuse hepatitis-like injury (alcoholic hepatitis) (1). In severe cases, the associated clinical syndrome typically includes fever, hepatic encephalopathy, neutrophilia, coagulopathy, and renal failure (2). Mortality within a month of hospitalization for severely ill patients can be as high as 60% (3). The exact mechanism of hepatocyte damage in alcoholic hepatitis is unclear, but several lines of evidence suggest that immune-mediated factors may be important in either initiating or perpetuating liver damage (4). Most recently, interest has centered on tumor necrosis factor, a major immune modulator secreted by monocytes, macrophages, and lymphocytes. Biologic actions of tumor necrosis factor include the induction of fever, neutrophilia, and hypotension associated with gram-negative sepsis (5), which are also many of the clinical and laboratory features of acute alcoholic hepatitis. Furthermore, in animal models of another toxininduced liver injury (galactosamine hepatitis), tumor necrosis factor has been implicated as a mediator of liver cell necrosis, although the mechanism of action is not known (6). To determine whether tumor necrosis factor mediates some of the manifestations of alcoholic hepatitis, we studied the relation between plasma tumor necrosis factor levels and extrahepatic manifestations of disease severity, serum bilirubin levels, and short-term survival. We also measured plasma endotoxin, a potent stimulator of tumor necrosis factor release from macrophages (7) and possibly a trigger for the increase in plasma tumor necrosis factor levels, and plasma interleukin-1/3, which also mediates fever and neutrophilia and is released in response to endotoxin (8).
Materials and Methods Participants
Annals of Internal Medicine. 1990;112:917-920. From King's College Hospital and King's College School of Medicine and Dentistry, Denmark Hill, London, United Kingdom. For current author addresses, see end of text.
We studied three groups of alcoholic patients. Group 1 comprised 21 consecutive patients with alcoholic hepatitis (9 women, 12 men) admitted for emergency treatment during a 5-month period. In 15 patients, the diagnosis of alcoholic hepatitis was confirmed by histologic analysis of specimens obtained by percutaneous biopsy or at autopsy: Samples showed hepatocyte swelling, Mallory hyaline bodies, and an inflammatory infiltrate. Six patients did not have a biopsy because they had either a prolonged prothrombin time or ascites. Their diagnosis was based on the following clinical and laboratory findings: hepatomegaly; a serum aspartate aminotransferase level of 1.34 to 4.17 /xkat/L (80 to 250 U/L) (normal range, 0.10 to 0.70 /Ltkat/L [5 to 40 U/L]); no serologic evidence of hepatitis A, hepatitis B, or autoimmune chronic active hepatitis; and a history of a daily alcohol intake of more than 80 g/d, which was a diagnostic criterion for all patients in the series. © 1990 American College of Physicians
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
917
Table 1. Clinical and Laboratory Characteristics of Patients with Alcoholic Hepatitis (n = 21) Variable
Value n(%)
Mortality at 6 weeks, n (%) Temperature higher than 37.0 °C, n(%) Documented infection, n (%) Endotoxic shock, n (%) Neutrophil count, x 109IL Serum creatinine, [imollL Serum bilirubin, yjnollL
11(52) 11(52) 14 (67) 3(14) 16.9 ± 1.45* 279 ± 33.3* 320 ±31.0*
* Mean ± SE.
Isotope scans in these 6 patients showed the "wipe-out" pattern typical of severe acute alcoholic hepatitis. Endotoxic shock was defined as a systolic blood pressure of below 80 mm Hg for at least 1 hour despite adequate filling pressures as determined by hemodynamic monitoring. Documented infection was defined as a positive bacterial or fungal culture of blood, urine, ascitic fluid, sputum, or vaginal swabs during the 6-day study period. Ascitic fluid was obtained using a sterile technique, and samples for culture and microscopy were collected in sterile glass bottles, with additional samples cultured in blood culture bottles using the Bactec system (Becton-Dickinson and Company, Towson, Maryland) (9). In the absence of a positive microbiologic culture, spontaneous bacterial peritonitis was defined by a neutrophil count in ascitic fluid of more than 250 x 106/L. Of the 21 patients with alcoholic hepatitis, 10 died within 6 weeks of admission, 8 from hepatocellular failure (seven of these cases were complicated by sepsis) and 2 from uncontrollable upper gastrointestinal hemorrhage. The remaining 11 patients recovered and were discharged. In 10 patients, one or more pathogens were isolated during the 6-day period. These 10 patients included 4 of 8 patients who were diagnosed as having spontaneous bacterial peritonitis on the basis of a raised ascitic neutrophil count. Three patients developed endotoxic shock (Table 1). The second group comprised ten clinically stable patients (two women, eight men) who were attending the outpatient follow-up clinic; these patients had histologically confirmed inactive alcoholic cirrhosis and had abstained from alcohol use for at least 1 year. The third group comprised ten patients (3 women, 7 men) who were admitted to the alcohol treatment unit for detoxification; these patients had no clinical evidence
of liver disease. All patients had been consuming more than 80 grams of alcohol daily for at least 1 year before admission. Some of the patients had mild increases in serum transaminase levels, which returned to the normal range within a few weeks of admission. Ten healthy laboratory staff (5 men, 5 women) served as normal control subjects. Because many of the patients with alcoholic hepatitis had hepatorenal failure, another control group was formed to evaluate the effect of impaired renal function on tumor necrosis factor levels; this group comprised ten patients (three woman, seven men) with an elevated serum creatinine level (284 ±64.6 /Ltmol/L [mean ± SE]; normal range, 45 to 105 /imol/L) or decreased creatinine clearance caused by chronic renal disease. Five patients had chronic glomerulonephritis; two patients, lupus nephritis; two patients, diabetic nephropathy; and one patient, hypertensive nephropathy. Assays Peripheral blood samples from patients with alcoholic hepatitis were taken using a sterile technique on alternate days starting before the end of the first week after admission. Three samples were obtained from each patient, except when death or discharge (two and three patients, respectively) rendered this impossible. In alcoholic persons without hepatitis and in the other control subjects, only one blood sample was taken. Blood samples for tumor necrosis factor, endotoxin, and interleukin-ljS assays were collected in heparinized glass tubes on ice and immediately spun down at 4 °C before being stored at -20 °C in 1-mL aliquots. Plasma tumor necrosis factor levels were determined using a commercial radioimmunoassay (IRE-Medgenix, Fleurus, Belgium). The plasma endotoxin level was determined using the chromogenic LAL assay (Kabi Vitrum, Stockholm, Sweden), and the plasma interleukin-1 level using a commercially available enzyme-linked immunosorbent assay (ELISA) (Cistron Biotechnology, Pine Brook, New Jersey). In our laboratory, this ELISA allows the detection of plasma interleukin-lj3 at levels above 20 ng/L and quantitative measurement at levels above 100 ng/L. Statistical Analysis Group data are expressed as mean with 95% CI or as mean ± SE. Comparison between patient groups was also made using the Mann-Whitney and Spearman rank tests. A difference was regarded as significant if P < 0.05. Analysis of covariates was done by both univariate and multivariate analysis. A twoway analysis of variance was used to study the change in the tumor necrosis factor level over time. Analyses were carried
Figure 1. Plasma tumor necrosis factor (TNF) level (mean values for study period) in patients with alcoholic hepatitis, persons with inactive alcoholic cirrhosis, alcoholic persons without liver disease, patients with decreased creatinine clearance, and normal subjects. Open circles represent patients who died.
918
15 June 1990 • Annals of Internal Medicine • Volume 112 • Number 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
out on a Dell microcomputer (Dell Computer Corporation, Austin, Texas) using a BMDP program (Statistical Software Inc., Los Angeles, California). Results Plasma Tumor Necrosis Factor Patients with alcoholic hepatitis (26.3 ng/L; CI, 21.7 to 30.9) had higher plasma tumor necrosis factor levels than all but one of the normal subjects (6.4 ng/L; CI, 5.4 to 7.4) for both the first and subsequent samples (P = 0.00008). Patients with alcoholic hepatitis also had significantly higher levels of plasma tumor necrosis factor when compared with patients who had inactive cirrhosis (11.1 ng/L; CI, 8.9 to 13.3; P = 0.008); alcoholic persons without liver disease (6.4 ng/L; CI, 5.0 to 7.8; P = 0.0006); and patients with chronic renal disease (14.1 ng/L; CI, 5.4 to 22.8; P = 0.03). Plasma tumor necrosis factor levels were higher in the patients with alcoholic hepatitis who subsequently died (34.7 ng/L; CI, 27.8 to 41.6) than in survivors (16.6 ng/L; CI, 14.0 to 19.2; P = 0.001). In patients with alcoholic hepatitis, plasma tumor necrosis factor levels correlated positively with temperature (r = 0.61; P = 0.02), peripheral blood neutrophil count (r = 0.79; P = 0.0008), serum creatinine (r = 0.81, P = 0.0003), and serum bilirubin (r = 0.74, P = 0.004). Multivariate analysis, using logistic regression, did not show a statistically significant relation between death and any of the above factors except tumor necrosis factor level. Among patients with alcoholic hepatitis, those with infection had higher levels of plasma tumor necrosis factor (30.3 ng/L; CI, 24.8 to 35.8) when compared with uninfected patients (16.2 ng/L; CI, 12.6 to 19.8; P = 0.011); however, tumor necrosis factor levels were not higher in three patients with endotoxic shock. Similarly, no significant difference in tumor necrosis factor levels was found when patients in whom an organism was isolated were compared with those who were diagnosed with sepsis on the basis of a high ascitic neutrophil count alone. Those who died showed a trend toward increasing plasma tumor necrosis factor levels during the study period (mean increase, 5.7 ng/L), whereas survivors showed a small decrease (mean decrease, 1.3 ng/L); however, this relative increase in patients who died was not statistically significant at either 3 or 6 days. Plasma Endotoxin and Interleukin-1/3 No correlation was found between tumor necrosis factor and plasma endotoxin levels in patients with alcoholic hepatitis (74.3 ng/L; CI, 32.6 to 116.0) and normal control subjects (24.7 ng/L; CI, 12.2 to 37.2). The plasma interleukin-lj3 level did not exceed 20 ng/L in either patients with alcoholic hepatitis or normal subjects. Discussion The direct measurement of plasma tumor necrosis factor has enabled us to extend previous in-vitro work, which suggested that there is increased production of
tumor necrosis factor by peripheral blood mononuclear cells from patients with alcoholic hepatitis (10). Not only were plasma tumor necrosis factor levels considerably elevated in patients with alcoholic hepatitis, but levels were significantly higher in those who died. Patients with compensated alcoholic cirrhosis who had minimal hepatocyte damage, as judged histologically and biochemically, showed a much smaller increase in plasma tumor necrosis factor levels compared with normal subjects; in alcoholic persons without liver disease, the range of plasma tumor necrosis factor levels was similar to the normal range. Administration of exogenous tumor necrosis factor causes many clinical and biochemical features that are also seen in cases of severe alcoholic hepatitis. These include fever, neutrophilia, shock, and biochemical abnormalities such as a depressed serum albumin level and alterations in zinc and iron metabolism (11). That the severity of alcoholic hepatitis, as measured by several clinical and biochemical indices, correlates well with the plasma tumor necrosis factor level raises the possibility that tumor necrosis factor acts as the predominant mediator and has a role in initiating or perpetuating liver damage in alcoholic hepatitis. The biologic activities of interleukin-1/3 overlap with those of tumor necrosis factor, in that both can cause pyrogenic activity, neutrophil and B-cell activation, and induction of hepatic-phase proteins; however, elevated levels of interleukin-1/3 were not detected in our patients, and it is unlikely that it was the predominant mediator of the indices of severity studied. Tumor necrosis factor can cause liver cell necrosis when injected into mice, but its mode of action is not clear (12); in acute liver failure secondary to hepatitis B and non-A, non-B viruses, where immune mechanisms are also thought to cause hepatocyte necrosis, increased production of tumor necrosis factor has been reported (13). Infection is common in patients with acute liver failure (14), and abnormalities in lymphokine activity including tumor necrosis factor may be related more to the consequences of infection than to hepatocyte necrosis. Work from our unit confirms that in liver failure, bacterial and fungal infections are associated with greatly increased production of tumor necrosis factor by mononuclear cells from peripheral blood (de la Mata M. Personal communication). However, patients in this series in whom a pathogenic organism was isolated did not have significantly higher levels of plasma tumor necrosis factor than those with either clinical indices of infection or a high ascitic neutrophil count only. Furthermore, a much weaker correlation was found between documented infection and plasma tumor necrosis factor than between tumor necrosis factor and either serum bilirubin or mortality, thus suggesting that the elevated tumor necrosis factor levels seen in our study have a more direct relation to the underlying hepatocellular damage and necrosis. In this respect, our study agrees with those that showed that tumor necrosis factor levels correlate with features indicative of disease severity and mortality in patients with meningococcal meningitis (15) and malaria (16). Our findings are also consistent with those of the recent study by McClain and Cohen (10). These investigators showed that com-
15 June 1990 • Annals of Internal Medicine • Volume 112 • Number 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
919
pared with normal subjects, patients with alcoholic hepatitis in whom infection was absent had increased production of tumor necrosis factor by peripheral blood mononuclear cells when stimulated with endotoxin. Endotoxin is recognized as one of the most potent mediators of tumor necrosis factor release (7), but no relation was found between plasma endotoxin and plasma tumor necrosis factor in our patients. Patients with alcoholic liver disease frequently have portal vein and systemic endotoxemia in the absence of sepsis detectable by routine microbiologic techniques (17, 18), possibly as a result of defective Kupffer cell function (19). Nevertheless, it seems unlikely that the high levels of tumor necrosis factor in our patients were triggered by endotoxin release. Our observations do not exclude the possibility that the effect of endotoxin on tumor necrosis factor release may be delayed, or that high endotoxin levels may cause a short period of tumor necrosis factor release, followed by a prolonged period during which tumor necrosis factor release is suppressed. Other mediators of tumor necrosis factor release include the Sendai and influenza viruses (20, 21), but that these viruses mediated factor release in our patients seems unlikely. Alcohol does not promote tumor necrosis factor release when incubated with peripheral blood mononuclear cells (10), and hepatocyte necrosis secondary to hepatotoxins, such as paracetamol (acetaminophen), has not been shown to stimulate tumor necrosis factor release (de la Mata M. Personal communication). Tumor necrosis factor has been detected in urine, and it is therefore possible that increased plasma tumor necrosis factor arises secondary to impaired renal function, although preliminary studies have not shown a relation between plasma and urine tumor necrosis factor levels (Sheron N. Personal communication). Patients with impaired creatinine clearance had significantly higher plasma tumor necrosis factor levels than normal control subjects, possibly because of underlying disease activity or decreased creatinine clearance. In patients with severe alcoholic hepatitis, impaired renal function could have contributed to the increased tumor necrosis factor levels, but the plasma tumor necrosis factor level was significantly higher in these patients than in patients with decreased creatinine clearance alone. According to the most recent trial, steroid treatment for severe alcoholic hepatitis appears to be beneficial in a select group of patients (22), and it has been suggested that such therapy exerts its effect by reducing T-cell activity and cytokine production (23). Further studies on the effect of steroid treatment on plasma tumor necrosis factor and its mechanisms of action in patients with alcoholic hepatitis would therefore be of considerable interest. In addition, the protective effect of neutralizing antibodies to tumor necrosis factor when given to animals dying from septicemic shock (24) suggests that antitumor necrosis factor could offer another possible approach to therapy in alcoholic hepatitis.
920
15 J u n e 1990 • Annals
of Internal
Medicine
Acknowledgment: tistical analysis.
The authors thank Karen Hallyer for advice on sta-
Grant Support: Dr. Sheron is supported by the Wellcome Trust. Requests for Reprints: Roger Williams, MD, Liver Unit, King's College Hospital, Denmark Hill, London SE5 9RS, United Kingdom. Current Author Addresses: Drs. Bird, Sheron, Goka, Alexander, and Williams: Liver Unit, King's College Hospital, Bessemer Road, London SE5 9RS, United Kingdom. References 1. Orrego H, Blake JE, Blendis LM, Medline A. Prognosis of alcoholic cirrhosis in the presence and absence of alcoholic hepatitis. Gastroenterology. 1987;92:208-14. 2. Maddrey WC. Alcoholic hepatitis: clinicopathologic features and therapy. Semin Liv Dis. 1988;8:91-102. 3. Theodossi A, Eddleston AL, Williams R. Controlled trial of methylprednisolone therapy in severe acute alcoholic hepatitis. Gut. 1982; 23:75-9. 4. Immunological abnormalities in alcoholic liver disease [Editorial]. Lancet. 1983;2:605-6. 5. Thiele DL. Tumor necrosis factor, the acute phase response and the pathogenesis of alcoholic liver disease. Hepatology. 1989;9:497-9. 6. Lehmann V, Freudenberg MA, Galanos C. Lethal toxicity of lipopolysaccharide and tumor necrosis factor in normal and D-galactosamine-treated mice. J Exp Med. 1987;165:657-63. 7. Beutler B, Cerami A. Cachectin: more than a tumor necrosis factor. N Engl J Med. 1987;316:379-85. 8. Le J, Vilcek J. Tumor necrosis factor and interleukin 1: cytokines with multiple overlapping biological activities. Lab Invest. 1987; 56:234-48. 9. Runyon BA, Canawati HN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology. 1988;95:1351-5. 10. McClain CJ, Cohen DA. Increased tumor necrosis factor production by monocytes in alcoholic hepatitis. Hepatology. 1989;9:349-51. 11. Remick DG, Kunkel RG, Larrick JW, Kunkel SL. Acute in vivo effects of human recombinant tumor necrosis factor. Lab Invest. 1987;56:583-90. 12. Gresser I, Woodrow D, Moss J, Maury C, Tavernier J, Fiers W. Toxic effects of recombinant tumor necrosis factor in suckling mice. Comparison with interferons alphatoeta. Am J Pathol. 1987;128: 13-8. 13. Muto Y, Nouri-Aria KT, Meager A, Alexander GJ, Eddleston AL, Williams R. Enhanced tumour necrosis factor and interleukin-1 in fulminant hepatitic failure. Lancet. 1988;2:72-4. 14. Canalese JC, Gove CD, Gimson AE, Wilkinson SP, Wardle EN, Williams R. Reticuloendothelial system and hepatocytic function in fulminant hepatic failure. Gut. 1982;23:265-9. 15. Waage A, Halstensen A, Espevik T. Association between tumour necrosis factor in serum and fatal outcome in patients with meningococcal disease. Lancet. 1987;2:355-7. 16. Grau GE, Taylor TE, Molyneux ME, et al. Tumor necrosis factor and disease severity in children with falciparum malaria. N Engl J Med. 1989;320:1586-91. 17. Nolan JP. Endotoxin, reticuloendothelial function, and liver injury. Hepatology. 1981;1:458-65. 18. Bigatello LM, Broitman SA, Fattori L, et al. Endotoxemia, encephalopathy, and mortality in cirrhotic patients. Am J Gastroenterol. 1987;82:11-5. 19. Yazima Y. Endotoxaemia in liver disease: detection by a quantitative assay using chromogenic substrate with perchromic acid pretreatment. Acta Hepatol Japan. 1986;27:234-44. 20. Aderka D, Holtmann H, Toker L, Hahn T, Wallach D. Tumor necrosis factor induction by Sendai virus. J Immunol. 1986; 136: 2938-42. 21. Beutler B, Krochin N, Milsark IW, et al. Induction of cachectin (tumor necrosis factor) synthesis by influenza virus: deficient production by endotoxin-resistant (C3H/HeJ) macrophages [Abstract]. Clin Res. 1986;34:491. 22. Carithers RL Jr, Herlong HF, Diehl AM, et al. Methyl prednisolone therapy in patients with severe alcoholic hepatitis. A randomized multicenter trial. Ann Intern Med. 1989;110:685-90. 23. Black M, Tavill AS. Corticosteroids in severe alcoholic hepatitis. Ann Intern Med. 1989;110:677-80. 24. Tracey KJ, Fong Y, Hesse DG, et al. Anti-cachectin/TNF monoclonal antibodies prevent septic shock during lethal bacteraemia. Nature. 1987;330:662-4.
• V o l u m e 112 • N u m b e r 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
Study Objective: To determine whether elevated tumor necrosis factor levels contribute to the clinical manifestations and complications of severe acute alcoholic hepatitis and to evaluate the relation between tumor necrosis factor and plasma levels of endotoxin and interleukin-1/3. Design: Prospective, controlled study. Setting: The liver unit of a university teaching hospital. Patients: We studied 21 patients with acute severe alcoholic hepatitis. There were four control groups: patients with inactive alcoholic cirrhosis, alcoholic persons without liver disease, patients with impaired renal function, and normal subjects. Measurements and Main Results: With one exception, patients with alcoholic hepatitis had higher tumor necrosis factor levels (mean, 26.3 ng/L; 95% CI, 21.7 to 30.9) than normal subjects (6.4 ng/L; CI, 5.4 to 7.4). Patients who subsequently died had a higher tumor necrosis factor level (34.7 ng/L; CI, 27.8 to 41.6) than survivors (16.6 ng/L; CI, 14.0 to 19.2). In patients with alcoholic hepatitis, tumor necrosis factor levels correlated positively with serum bilirubin (r = 0.74; P = 0.0009) and serum creatinine (r = 0.81; P = 0.0003). Patients with alcoholic hepatitis had higher tumor necrosis factor levels than patients with inactive alcoholic cirrhosis (11.1 ng/L; CI, 8.9 to 13.3) and severely alcoholic persons without liver disease (6.4 ng/L; CI, 5.0 to 7.8). Patients with abnormal renal function had lower tumor necrosis factor levels (14.1 ng/L; CI, 5.4 to 22.8) than patients with alcoholic hepatitis. Serial samples obtained during a 1-week period from patients with alcoholic hepatitis showed no significant change in tumor necrosis factor when patients who died were compared with survivors. No correlation was found between tumor necrosis factor and plasma endotoxin. Levels of interleukin-1/3 did not exceed 20 ng/L. Conclusions: Elevations in tumor necrosis factor in alcoholic hepatitis are most marked in severe cases, suggesting that tumor necrosis factor plays a role in the pathogenesis.
Approximately 57% of patients hospitalized because of alcoholic liver disease have evidence of diffuse hepatitis-like injury (alcoholic hepatitis) (1). In severe cases, the associated clinical syndrome typically includes fever, hepatic encephalopathy, neutrophilia, coagulopathy, and renal failure (2). Mortality within a month of hospitalization for severely ill patients can be as high as 60% (3). The exact mechanism of hepatocyte damage in alcoholic hepatitis is unclear, but several lines of evidence suggest that immune-mediated factors may be important in either initiating or perpetuating liver damage (4). Most recently, interest has centered on tumor necrosis factor, a major immune modulator secreted by monocytes, macrophages, and lymphocytes. Biologic actions of tumor necrosis factor include the induction of fever, neutrophilia, and hypotension associated with gram-negative sepsis (5), which are also many of the clinical and laboratory features of acute alcoholic hepatitis. Furthermore, in animal models of another toxininduced liver injury (galactosamine hepatitis), tumor necrosis factor has been implicated as a mediator of liver cell necrosis, although the mechanism of action is not known (6). To determine whether tumor necrosis factor mediates some of the manifestations of alcoholic hepatitis, we studied the relation between plasma tumor necrosis factor levels and extrahepatic manifestations of disease severity, serum bilirubin levels, and short-term survival. We also measured plasma endotoxin, a potent stimulator of tumor necrosis factor release from macrophages (7) and possibly a trigger for the increase in plasma tumor necrosis factor levels, and plasma interleukin-1/3, which also mediates fever and neutrophilia and is released in response to endotoxin (8).
Materials and Methods Participants
Annals of Internal Medicine. 1990;112:917-920. From King's College Hospital and King's College School of Medicine and Dentistry, Denmark Hill, London, United Kingdom. For current author addresses, see end of text.
We studied three groups of alcoholic patients. Group 1 comprised 21 consecutive patients with alcoholic hepatitis (9 women, 12 men) admitted for emergency treatment during a 5-month period. In 15 patients, the diagnosis of alcoholic hepatitis was confirmed by histologic analysis of specimens obtained by percutaneous biopsy or at autopsy: Samples showed hepatocyte swelling, Mallory hyaline bodies, and an inflammatory infiltrate. Six patients did not have a biopsy because they had either a prolonged prothrombin time or ascites. Their diagnosis was based on the following clinical and laboratory findings: hepatomegaly; a serum aspartate aminotransferase level of 1.34 to 4.17 /xkat/L (80 to 250 U/L) (normal range, 0.10 to 0.70 /Ltkat/L [5 to 40 U/L]); no serologic evidence of hepatitis A, hepatitis B, or autoimmune chronic active hepatitis; and a history of a daily alcohol intake of more than 80 g/d, which was a diagnostic criterion for all patients in the series. © 1990 American College of Physicians
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
917
Table 1. Clinical and Laboratory Characteristics of Patients with Alcoholic Hepatitis (n = 21) Variable
Value n(%)
Mortality at 6 weeks, n (%) Temperature higher than 37.0 °C, n(%) Documented infection, n (%) Endotoxic shock, n (%) Neutrophil count, x 109IL Serum creatinine, [imollL Serum bilirubin, yjnollL
11(52) 11(52) 14 (67) 3(14) 16.9 ± 1.45* 279 ± 33.3* 320 ±31.0*
* Mean ± SE.
Isotope scans in these 6 patients showed the "wipe-out" pattern typical of severe acute alcoholic hepatitis. Endotoxic shock was defined as a systolic blood pressure of below 80 mm Hg for at least 1 hour despite adequate filling pressures as determined by hemodynamic monitoring. Documented infection was defined as a positive bacterial or fungal culture of blood, urine, ascitic fluid, sputum, or vaginal swabs during the 6-day study period. Ascitic fluid was obtained using a sterile technique, and samples for culture and microscopy were collected in sterile glass bottles, with additional samples cultured in blood culture bottles using the Bactec system (Becton-Dickinson and Company, Towson, Maryland) (9). In the absence of a positive microbiologic culture, spontaneous bacterial peritonitis was defined by a neutrophil count in ascitic fluid of more than 250 x 106/L. Of the 21 patients with alcoholic hepatitis, 10 died within 6 weeks of admission, 8 from hepatocellular failure (seven of these cases were complicated by sepsis) and 2 from uncontrollable upper gastrointestinal hemorrhage. The remaining 11 patients recovered and were discharged. In 10 patients, one or more pathogens were isolated during the 6-day period. These 10 patients included 4 of 8 patients who were diagnosed as having spontaneous bacterial peritonitis on the basis of a raised ascitic neutrophil count. Three patients developed endotoxic shock (Table 1). The second group comprised ten clinically stable patients (two women, eight men) who were attending the outpatient follow-up clinic; these patients had histologically confirmed inactive alcoholic cirrhosis and had abstained from alcohol use for at least 1 year. The third group comprised ten patients (3 women, 7 men) who were admitted to the alcohol treatment unit for detoxification; these patients had no clinical evidence
of liver disease. All patients had been consuming more than 80 grams of alcohol daily for at least 1 year before admission. Some of the patients had mild increases in serum transaminase levels, which returned to the normal range within a few weeks of admission. Ten healthy laboratory staff (5 men, 5 women) served as normal control subjects. Because many of the patients with alcoholic hepatitis had hepatorenal failure, another control group was formed to evaluate the effect of impaired renal function on tumor necrosis factor levels; this group comprised ten patients (three woman, seven men) with an elevated serum creatinine level (284 ±64.6 /Ltmol/L [mean ± SE]; normal range, 45 to 105 /imol/L) or decreased creatinine clearance caused by chronic renal disease. Five patients had chronic glomerulonephritis; two patients, lupus nephritis; two patients, diabetic nephropathy; and one patient, hypertensive nephropathy. Assays Peripheral blood samples from patients with alcoholic hepatitis were taken using a sterile technique on alternate days starting before the end of the first week after admission. Three samples were obtained from each patient, except when death or discharge (two and three patients, respectively) rendered this impossible. In alcoholic persons without hepatitis and in the other control subjects, only one blood sample was taken. Blood samples for tumor necrosis factor, endotoxin, and interleukin-ljS assays were collected in heparinized glass tubes on ice and immediately spun down at 4 °C before being stored at -20 °C in 1-mL aliquots. Plasma tumor necrosis factor levels were determined using a commercial radioimmunoassay (IRE-Medgenix, Fleurus, Belgium). The plasma endotoxin level was determined using the chromogenic LAL assay (Kabi Vitrum, Stockholm, Sweden), and the plasma interleukin-1 level using a commercially available enzyme-linked immunosorbent assay (ELISA) (Cistron Biotechnology, Pine Brook, New Jersey). In our laboratory, this ELISA allows the detection of plasma interleukin-lj3 at levels above 20 ng/L and quantitative measurement at levels above 100 ng/L. Statistical Analysis Group data are expressed as mean with 95% CI or as mean ± SE. Comparison between patient groups was also made using the Mann-Whitney and Spearman rank tests. A difference was regarded as significant if P < 0.05. Analysis of covariates was done by both univariate and multivariate analysis. A twoway analysis of variance was used to study the change in the tumor necrosis factor level over time. Analyses were carried
Figure 1. Plasma tumor necrosis factor (TNF) level (mean values for study period) in patients with alcoholic hepatitis, persons with inactive alcoholic cirrhosis, alcoholic persons without liver disease, patients with decreased creatinine clearance, and normal subjects. Open circles represent patients who died.
918
15 June 1990 • Annals of Internal Medicine • Volume 112 • Number 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
out on a Dell microcomputer (Dell Computer Corporation, Austin, Texas) using a BMDP program (Statistical Software Inc., Los Angeles, California). Results Plasma Tumor Necrosis Factor Patients with alcoholic hepatitis (26.3 ng/L; CI, 21.7 to 30.9) had higher plasma tumor necrosis factor levels than all but one of the normal subjects (6.4 ng/L; CI, 5.4 to 7.4) for both the first and subsequent samples (P = 0.00008). Patients with alcoholic hepatitis also had significantly higher levels of plasma tumor necrosis factor when compared with patients who had inactive cirrhosis (11.1 ng/L; CI, 8.9 to 13.3; P = 0.008); alcoholic persons without liver disease (6.4 ng/L; CI, 5.0 to 7.8; P = 0.0006); and patients with chronic renal disease (14.1 ng/L; CI, 5.4 to 22.8; P = 0.03). Plasma tumor necrosis factor levels were higher in the patients with alcoholic hepatitis who subsequently died (34.7 ng/L; CI, 27.8 to 41.6) than in survivors (16.6 ng/L; CI, 14.0 to 19.2; P = 0.001). In patients with alcoholic hepatitis, plasma tumor necrosis factor levels correlated positively with temperature (r = 0.61; P = 0.02), peripheral blood neutrophil count (r = 0.79; P = 0.0008), serum creatinine (r = 0.81, P = 0.0003), and serum bilirubin (r = 0.74, P = 0.004). Multivariate analysis, using logistic regression, did not show a statistically significant relation between death and any of the above factors except tumor necrosis factor level. Among patients with alcoholic hepatitis, those with infection had higher levels of plasma tumor necrosis factor (30.3 ng/L; CI, 24.8 to 35.8) when compared with uninfected patients (16.2 ng/L; CI, 12.6 to 19.8; P = 0.011); however, tumor necrosis factor levels were not higher in three patients with endotoxic shock. Similarly, no significant difference in tumor necrosis factor levels was found when patients in whom an organism was isolated were compared with those who were diagnosed with sepsis on the basis of a high ascitic neutrophil count alone. Those who died showed a trend toward increasing plasma tumor necrosis factor levels during the study period (mean increase, 5.7 ng/L), whereas survivors showed a small decrease (mean decrease, 1.3 ng/L); however, this relative increase in patients who died was not statistically significant at either 3 or 6 days. Plasma Endotoxin and Interleukin-1/3 No correlation was found between tumor necrosis factor and plasma endotoxin levels in patients with alcoholic hepatitis (74.3 ng/L; CI, 32.6 to 116.0) and normal control subjects (24.7 ng/L; CI, 12.2 to 37.2). The plasma interleukin-lj3 level did not exceed 20 ng/L in either patients with alcoholic hepatitis or normal subjects. Discussion The direct measurement of plasma tumor necrosis factor has enabled us to extend previous in-vitro work, which suggested that there is increased production of
tumor necrosis factor by peripheral blood mononuclear cells from patients with alcoholic hepatitis (10). Not only were plasma tumor necrosis factor levels considerably elevated in patients with alcoholic hepatitis, but levels were significantly higher in those who died. Patients with compensated alcoholic cirrhosis who had minimal hepatocyte damage, as judged histologically and biochemically, showed a much smaller increase in plasma tumor necrosis factor levels compared with normal subjects; in alcoholic persons without liver disease, the range of plasma tumor necrosis factor levels was similar to the normal range. Administration of exogenous tumor necrosis factor causes many clinical and biochemical features that are also seen in cases of severe alcoholic hepatitis. These include fever, neutrophilia, shock, and biochemical abnormalities such as a depressed serum albumin level and alterations in zinc and iron metabolism (11). That the severity of alcoholic hepatitis, as measured by several clinical and biochemical indices, correlates well with the plasma tumor necrosis factor level raises the possibility that tumor necrosis factor acts as the predominant mediator and has a role in initiating or perpetuating liver damage in alcoholic hepatitis. The biologic activities of interleukin-1/3 overlap with those of tumor necrosis factor, in that both can cause pyrogenic activity, neutrophil and B-cell activation, and induction of hepatic-phase proteins; however, elevated levels of interleukin-1/3 were not detected in our patients, and it is unlikely that it was the predominant mediator of the indices of severity studied. Tumor necrosis factor can cause liver cell necrosis when injected into mice, but its mode of action is not clear (12); in acute liver failure secondary to hepatitis B and non-A, non-B viruses, where immune mechanisms are also thought to cause hepatocyte necrosis, increased production of tumor necrosis factor has been reported (13). Infection is common in patients with acute liver failure (14), and abnormalities in lymphokine activity including tumor necrosis factor may be related more to the consequences of infection than to hepatocyte necrosis. Work from our unit confirms that in liver failure, bacterial and fungal infections are associated with greatly increased production of tumor necrosis factor by mononuclear cells from peripheral blood (de la Mata M. Personal communication). However, patients in this series in whom a pathogenic organism was isolated did not have significantly higher levels of plasma tumor necrosis factor than those with either clinical indices of infection or a high ascitic neutrophil count only. Furthermore, a much weaker correlation was found between documented infection and plasma tumor necrosis factor than between tumor necrosis factor and either serum bilirubin or mortality, thus suggesting that the elevated tumor necrosis factor levels seen in our study have a more direct relation to the underlying hepatocellular damage and necrosis. In this respect, our study agrees with those that showed that tumor necrosis factor levels correlate with features indicative of disease severity and mortality in patients with meningococcal meningitis (15) and malaria (16). Our findings are also consistent with those of the recent study by McClain and Cohen (10). These investigators showed that com-
15 June 1990 • Annals of Internal Medicine • Volume 112 • Number 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
919
pared with normal subjects, patients with alcoholic hepatitis in whom infection was absent had increased production of tumor necrosis factor by peripheral blood mononuclear cells when stimulated with endotoxin. Endotoxin is recognized as one of the most potent mediators of tumor necrosis factor release (7), but no relation was found between plasma endotoxin and plasma tumor necrosis factor in our patients. Patients with alcoholic liver disease frequently have portal vein and systemic endotoxemia in the absence of sepsis detectable by routine microbiologic techniques (17, 18), possibly as a result of defective Kupffer cell function (19). Nevertheless, it seems unlikely that the high levels of tumor necrosis factor in our patients were triggered by endotoxin release. Our observations do not exclude the possibility that the effect of endotoxin on tumor necrosis factor release may be delayed, or that high endotoxin levels may cause a short period of tumor necrosis factor release, followed by a prolonged period during which tumor necrosis factor release is suppressed. Other mediators of tumor necrosis factor release include the Sendai and influenza viruses (20, 21), but that these viruses mediated factor release in our patients seems unlikely. Alcohol does not promote tumor necrosis factor release when incubated with peripheral blood mononuclear cells (10), and hepatocyte necrosis secondary to hepatotoxins, such as paracetamol (acetaminophen), has not been shown to stimulate tumor necrosis factor release (de la Mata M. Personal communication). Tumor necrosis factor has been detected in urine, and it is therefore possible that increased plasma tumor necrosis factor arises secondary to impaired renal function, although preliminary studies have not shown a relation between plasma and urine tumor necrosis factor levels (Sheron N. Personal communication). Patients with impaired creatinine clearance had significantly higher plasma tumor necrosis factor levels than normal control subjects, possibly because of underlying disease activity or decreased creatinine clearance. In patients with severe alcoholic hepatitis, impaired renal function could have contributed to the increased tumor necrosis factor levels, but the plasma tumor necrosis factor level was significantly higher in these patients than in patients with decreased creatinine clearance alone. According to the most recent trial, steroid treatment for severe alcoholic hepatitis appears to be beneficial in a select group of patients (22), and it has been suggested that such therapy exerts its effect by reducing T-cell activity and cytokine production (23). Further studies on the effect of steroid treatment on plasma tumor necrosis factor and its mechanisms of action in patients with alcoholic hepatitis would therefore be of considerable interest. In addition, the protective effect of neutralizing antibodies to tumor necrosis factor when given to animals dying from septicemic shock (24) suggests that antitumor necrosis factor could offer another possible approach to therapy in alcoholic hepatitis.
920
15 J u n e 1990 • Annals
of Internal
Medicine
Acknowledgment: tistical analysis.
The authors thank Karen Hallyer for advice on sta-
Grant Support: Dr. Sheron is supported by the Wellcome Trust. Requests for Reprints: Roger Williams, MD, Liver Unit, King's College Hospital, Denmark Hill, London SE5 9RS, United Kingdom. Current Author Addresses: Drs. Bird, Sheron, Goka, Alexander, and Williams: Liver Unit, King's College Hospital, Bessemer Road, London SE5 9RS, United Kingdom. References 1. Orrego H, Blake JE, Blendis LM, Medline A. Prognosis of alcoholic cirrhosis in the presence and absence of alcoholic hepatitis. Gastroenterology. 1987;92:208-14. 2. Maddrey WC. Alcoholic hepatitis: clinicopathologic features and therapy. Semin Liv Dis. 1988;8:91-102. 3. Theodossi A, Eddleston AL, Williams R. Controlled trial of methylprednisolone therapy in severe acute alcoholic hepatitis. Gut. 1982; 23:75-9. 4. Immunological abnormalities in alcoholic liver disease [Editorial]. Lancet. 1983;2:605-6. 5. Thiele DL. Tumor necrosis factor, the acute phase response and the pathogenesis of alcoholic liver disease. Hepatology. 1989;9:497-9. 6. Lehmann V, Freudenberg MA, Galanos C. Lethal toxicity of lipopolysaccharide and tumor necrosis factor in normal and D-galactosamine-treated mice. J Exp Med. 1987;165:657-63. 7. Beutler B, Cerami A. Cachectin: more than a tumor necrosis factor. N Engl J Med. 1987;316:379-85. 8. Le J, Vilcek J. Tumor necrosis factor and interleukin 1: cytokines with multiple overlapping biological activities. Lab Invest. 1987; 56:234-48. 9. Runyon BA, Canawati HN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology. 1988;95:1351-5. 10. McClain CJ, Cohen DA. Increased tumor necrosis factor production by monocytes in alcoholic hepatitis. Hepatology. 1989;9:349-51. 11. Remick DG, Kunkel RG, Larrick JW, Kunkel SL. Acute in vivo effects of human recombinant tumor necrosis factor. Lab Invest. 1987;56:583-90. 12. Gresser I, Woodrow D, Moss J, Maury C, Tavernier J, Fiers W. Toxic effects of recombinant tumor necrosis factor in suckling mice. Comparison with interferons alphatoeta. Am J Pathol. 1987;128: 13-8. 13. Muto Y, Nouri-Aria KT, Meager A, Alexander GJ, Eddleston AL, Williams R. Enhanced tumour necrosis factor and interleukin-1 in fulminant hepatitic failure. Lancet. 1988;2:72-4. 14. Canalese JC, Gove CD, Gimson AE, Wilkinson SP, Wardle EN, Williams R. Reticuloendothelial system and hepatocytic function in fulminant hepatic failure. Gut. 1982;23:265-9. 15. Waage A, Halstensen A, Espevik T. Association between tumour necrosis factor in serum and fatal outcome in patients with meningococcal disease. Lancet. 1987;2:355-7. 16. Grau GE, Taylor TE, Molyneux ME, et al. Tumor necrosis factor and disease severity in children with falciparum malaria. N Engl J Med. 1989;320:1586-91. 17. Nolan JP. Endotoxin, reticuloendothelial function, and liver injury. Hepatology. 1981;1:458-65. 18. Bigatello LM, Broitman SA, Fattori L, et al. Endotoxemia, encephalopathy, and mortality in cirrhotic patients. Am J Gastroenterol. 1987;82:11-5. 19. Yazima Y. Endotoxaemia in liver disease: detection by a quantitative assay using chromogenic substrate with perchromic acid pretreatment. Acta Hepatol Japan. 1986;27:234-44. 20. Aderka D, Holtmann H, Toker L, Hahn T, Wallach D. Tumor necrosis factor induction by Sendai virus. J Immunol. 1986; 136: 2938-42. 21. Beutler B, Krochin N, Milsark IW, et al. Induction of cachectin (tumor necrosis factor) synthesis by influenza virus: deficient production by endotoxin-resistant (C3H/HeJ) macrophages [Abstract]. Clin Res. 1986;34:491. 22. Carithers RL Jr, Herlong HF, Diehl AM, et al. Methyl prednisolone therapy in patients with severe alcoholic hepatitis. A randomized multicenter trial. Ann Intern Med. 1989;110:685-90. 23. Black M, Tavill AS. Corticosteroids in severe alcoholic hepatitis. Ann Intern Med. 1989;110:677-80. 24. Tracey KJ, Fong Y, Hesse DG, et al. Anti-cachectin/TNF monoclonal antibodies prevent septic shock during lethal bacteraemia. Nature. 1987;330:662-4.
• V o l u m e 112 • N u m b e r 12
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19706/ by a University of California San Diego User on 06/20/2017
