- Email: [email protected]
Soluble TNF-R1, but not tumor necrosis factor alpha, predicts the 3-month mortality in patients with alcoholic hepatitis
Journal of Hepatology 41 (2004) 229–234 www.elsevier.com/locate/jhep
Soluble TNF-R1, but not tumor necrosis factor alpha, predicts the 3-month mortality in patients with alcoholic hepatitisq Laurent Spahr1,*, Emile Giostra1, Jean-Louis Frossard1, Solange Bresson-Hadni1, Laura Rubbia-Brandt2, Antoine Hadengue1 1
Gastroenterology and Hepatology, University Hospital, Geneva, Switzerland 2 Clinical Pathology, University Hospital, Geneva, Switzerland
Background/Aims: In alcoholic hepatitis (AH), soluble TNFa receptor-1 (sTNF-R1) is increased. Elevated TNFa predicts mortality, but infection influences TNFa values. In patients with AH, we determined the prognostic value of TNFa, sTNF-R1, and lipopolysaccharide binding protein (LBP) and CD14, both involved in endotoxemia-associated inflammation. Methods: One hundred and eight cirrhotic patients (Pugh score 10 [6 –13]) and biopsy-proven AH (Maddrey’s DF <32: n 5 46; $ 32: n 5 62) without associated infection were included within 8 days of admission and followed-up for 3 months. Cytokines were measured using specific immunoassays. Patients with severe AH received steroids. Results: Twenty four patients died at a median time of 35 days (range: 3– 89). The overall survival was 78%. Multivariate Cox regression analysis showed that sTNF-R1 was an independent predictor of mortality, (OR 4.33: 95% CI [1.12 – 16.75]). Pugh’s score (P 5 0.618), Maddrey’s DF (P 5 0.182), creatinine (P 5 0.197), TNFa (P 5 0.319), LBP (P 5 0.362), and CD14 (P 5 0.347) were not related to survival. Conclusions: In patients with AH, sTNF-R1 measured at admission is an independent predictor of survival at 3 months. Provided that TNF-R1 mediates the cytotoxic actions of TNFa, these results support the concept of dysregulated TNFa metabolism in AH. q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Keywords: Alcoholic hepatitis; TNFa; Soluble TNF receptor 1; Mortality; Clinical study; Multivariate analysis; Lipopolysaccharide binding protein; Maddrey’s discriminant function; CD14 1. Introduction Alcoholic hepatitis (AH) is a frequent inflammatory liver disease in which a number of pathogenetic mechanisms are involved [1]. In particular, dysregulated inflammatory and immune mechanisms are believed to result from ethanol-associated increased intestinal permeability [2 – 4] and a depressed clearance capacity of the liver reticuloendothelial system [1]. Accordingly, endotoxaemia is transiently elevated in patients with excessive alcohol intake [3]. Endotoxin, or lipopolysaccharide, stimulates Received 22 October 2003; received in revised form 5 April 2004; accepted 13 April 2004; available online 18 May 2004 q Presented in part at the AASLD session of Digestive Disease Week in San Francisco, California, May 2002. * Corresponding author. Tel.: þ 41-223729340; fax: þ 41-223729366. E-mail address: [email protected] (L. Spahr). Abbreviations: sTNF-R1, soluble TNFa receptor-1; TNFa, tumor necrosis factor a; LBP, lipopolysaccharide-binding protein; AH, alcoholic hepatitis.
Kupffer cells to produce proinflammatory cytokines [5,6], thus contributing to the severe inflammatory response that characterizes AH [1,7]. The action of endotoxin on target cells is mediated by two glycoproteins, lipopolysaccharide binding protein (LBP) and CD14. Lipopolysaccharide binding protein, the synthesis of which is increased in the presence of endotoxemia [8,9], acts as a carrier for endotoxin molecules, and has been proposed as a good marker of long-term exposure to endotoxin [10]. The endotoxin– LBP complex then binds to CD14, a membrane-bound receptor on Kupffer cells, to activate the cellular production of inflammatory mediators [11]. In CD14-deficient mice [12], alcoholic liver injury is reduced. In patients with alcoholic liver disease, circulating concentrations of the lipoproteins LBP and CD14 are elevated [8,11], with LBP correlating to serum CD14 in patients with cirrhosis [10]. Elevated TNFa is a typical feature of AH [13 – 17]. TNFa mediates its varied biological responses by binding to two cell surface receptors that signal different cellular
0168-8278/$30.00 q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhep.2004.04.028
230
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
pathways: TNF-R1 (also named p55, according to its molecular mass of 55 kD) and TNF-R2 (also named p75) [18,19]. TNF-R1 is the main inductor of cytotoxicity, promoting a number of biological responses ranging from NF-kB activation to cell death. Accordingly, mice deficient in TNF-R1 [20] are protected against alcoholic liver injury. Soluble form of both receptors are present constitutively in serum, derived after receptor shedding after cellular activation by stimuli such as TNFa production [21]. Thus, sTNFR serum level is considered a reliable indicator of TNFa system activation [22,23]. In patients with AH, sTNF-R1 plasma values increase with disease severity [24]. High plasma TNFa in patients hospitalised for AH has been shown to predict a poor outcome in several studies [13 –15]. However, interpretation of elevated TNFa values is difficult in the presence of bacterial infections [25] or cirrhosis [26], both conditions known to affect circulating TNFa levels. Thus, the prognostic significance of TNFa in studies in which such confounding factors have not been considered, should be questioned. Finally, recent trials that aimed at antagonizing TNFa using monoclonal antibody did not demonstrate evident survival benefit [16,17], but a possible increased mortality (S. Naveau, personal communication). Therefore, the prognostic role of TNFa, soluble receptor 1 and endotoxin-associated proteins should be clarify in patients with AH. We decided to determine the serum concentrations and prognostic value of TNFa, sTNF-R1, CD14 and LBP obtained early after hospital admission in a cohort of 108 patients with cirrhosis and AH defined in pathological terms, in the absence of associated severe bacterial infection.
2. Patients and methods
Fig. 1. Algorithm of patients’ selection.
AH. All patients with histologically-proven severe AH received a 28-day course of prednisolone, as recommended [29], in association with oral norfloxacin in the presence of low-protein ascites. A severe infection (spontaneous bacterial peritonitis in two patients, pneumonia in two patients, urinary tract-associated sepsis in one patient) associated to AH was detected at hospital admission. These patients were excluded. The study protocol was conform to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethical committee of the Hoˆpitaux Universitaires de Gene`ve. Written informed consent was obtained from all patients and controls.
2.2. Liver biopsy Liver biopsy was performed in all patients within 7 days of hospitalisation by the transjugular route, due to the presence of coagulopathy and/or ascites. Histopathological examination was performed by a single histopathologist (L. R-B) expert in liver diseases. The diagnosis of AH was based on definite criteria established by international experts [30]. Thus, ballooning degeneration of hepatocytes and necrosis, Mallory bodies, fatty changes, cholestasis and parenchymal infiltration with neutrophils [31] were present in all patients’ liver biopsy.
2.3. Outcome 2.1. Patients One hundred and eight patients with AH included between January 2000 and December 2002 in a single institution formed the basis of this study. All patients had underlying cirrhosis, an observation that prevails in European countries [27]. The algorithm of patients’ selection is described in Fig. 1, and demographic data are given in Table 1. The Maddrey’s discriminant function (DF), a strong prognostic indicator of disease severity, is calculated by integrating the prothrombin time and serum bilirubin in the formula [28]: 4,6 £ [prothrombin time—control time (in seconds)] þ serum bilirubin (in mmol/l)/17. A score $32 defines a severe
We determined the outcome of all patients (survival or death) over a 3-month period using hospital records and telephone inquiry to general practitioners. We chose arbitrarily this follow-up period of 3 months because increased mortality following severe AH has been reported to persist beyond the treatment period [32].
2.4. Biological values All blood samples were collected at the time of liver biopsy in a fasting state, and transported immediately on ice to the laboratory. After
Table 1 Patients characteristics
n Gender (M/F) Age (years, range) Pugh’s score Maddrey’s DF Serum creatinine (mmol/l)
Controls
AH and DF , 32
AH and DF $ 32
12 8/4 52 (46–59) 8 (7–9) – 64 (58–75)
46 31/15 54 (41–69) 8 (6–12) 20 (15–29) 67 (54–101)
62 39/23 57 (34– 73) 11 (8–13)* 50 (32– 112)** 77 (48– 194)*
Abbreviations: DF: Maddrey’s discriminant function. Note: controls are patients with abstinent alcoholic cirrhosis. *P , 0:05 vs. controls and vs. patients with AH and DF , 32; **P , 0:01 vs. controls and vs. patients with AH and DF , 32 (Bonferroni’s correction for multiple testing of data).
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234 centrifugation for 15 min at 5 8C, aliquots were made under pyrogen-free conditions and the samples kept frozen at 270 8C until assayed. We used commercially available specific immunoassays for the measurement in duplicate of TNFa, sTNF-R1, CD14 (R&D Systems Europe, Abingdon, UK) and LBP (Bioreba AG, Reinach, Switzerland). The immunoassay used for sTNF-R1 is not influenced by the presence of TNFa and thus determines the total amount of the soluble receptors in the sample. Except for sTNF-R1 values which were assessed both in peripheral and hepatic venous blood, cytokines were measured in peripheral blood only. We decided to measure sTNF-R1 only, as it reflects hepatocellular injury better than sTNF-R2 [21–23]. For TNFa, the minimum detectable concentration was less than 4 pg/ml, and the intra- and interassay variability were 4.2 and 4.6%, respectively. For sTNF-R1, the minimum detectable dose was less than 3 pg/ml. Intra- and interassay variability were 2.9 and 3.7%, respectively. The minimum detection level for LBP was 1 ng/ml, with a measurable concentration range of 0.4 to 100 ng/ml. For CD14, the sensitivity threshold was 125 pg/ml, and the intra- and interassay variability were 5.2 and 4.8%, respectively. We measured serum creatinine in all patients, as renal function has been shown to influence the prognosis in AH [33,34].
3. Statistical analysis Results are summarized as medians and ranges. Due to the non-normal distribution of values, group comparisons were performed by the non-parametric Kruskal Wallis method, with Bonferroni’s post hoc analysis to correct significance levels for multiple testing of data. Survival curves were estimated using the Kaplan – Meier method. The prognostic significance of the variables in the univariate analysis was determined using the log rank test. Rating for each variable was determined according to the median value in the group. We used the Cox regression analysis (Cox’s proportional hazard model) with the forward-stepwise method [35] (likelihood ratio) and enter and remove limits of P , 0:05 and P . 0:10; respectively. To improve model fit in the multivariate analysis, we performed a logarithmic transformation of all variables prior to introduce them in the model as continuous variables. The assumptions of the Cox model including the assumption of proportional hazards over time were assessed. In view of the limited number of events in the study period, the assumption of proportionality does not seem to be grossly violated. Correlation analyses were performed by using the Spearman rank correlation test. To assess the use of sTNF-R1 serum value in a clinical setting, we constructed a receiver operating characteristic (ROC) curve by plotting
231
Fig. 2. Box-whisker plots (median, upper and lower quartiles and range) showing elevated serum values of TNFa in patients with severe AH (DF $ 32, upper left), high sTNF-R1 in patients with severe (DF $ 32) and non-severe AH (<32, upper right) compared to patients with abstinent alcoholic cirrhosis. Note: a difference between LBP in controls and in patients with AH was found to be significant ðP < 0:03Þ; but Bonferroni’s post hoc test removes this significance. *P < 0:0001 vs. controls and vs. patients with non-severe AH.
the sensitivity of sTNF-R1 against the reverse specificity (1 2 specificity). All statistical tests were performed by using the Statistical Package for the Social Sciences, version 9.0 for Windows (SPSS, Chicago, IL, USA).
4. Results 4.1. Cytokines Cytokines values are given in Table 2 and illustrated in Fig. 2. Serum TNFa levels were elevated in patients with severe AH compared to patients with non-severe AH and patients with abstinent cirrhosis. Patients with severe and non-severe AH showed higher serum sTNF-R1 values compared to controls. Circulating concentrations of LBP and CD14, however, were similar between groups. 4.2. Correlations In patients with AH, a weak but statistically significant
Table 2 Cytokines values
TNF (pg/ml) sTNF-R1 (PV) (pg/ml) sTNF-R1 (HV) (pg/ml) LBP (ng/ml) CD14 (pg/ml)
Controls
AH and DF , 32
AH and DF $ 32
1.7 (0.5–4.3) 1943 (1111–2201) – 26.6 (11.4–46.4) 1588 (1321–2166)
1.9 (0.4– 9.5) 2480 (360–5771)* 2574 (900–6132) 47.9 (18.8 –73.3) 2065 (1105–2806)
7.4 (1.3–34.6)** 3541 (1707–11747)** 3518 (1330–11176)*** 44.6 (15–167.2) 1964 (907–3465)
Abbreviations: DF: Maddrey’s discriminant function. *P , 0:01 vs. controls; **P , 0:01 vs. controls and vs. patients with AH and DF , 32 (Bonferroni’s corrections for multiple testing of data); ***P , 0:01 vs. patients with AH and DF , 32.
232
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
correlation was found between TNFa and sTNF-R1 (r ¼ 0:33; P , 0:001), as well as between TNFa and the Maddrey’s DF at hospital admission (r ¼ 0:35; P , 0:001). In the subgroup of patients with severe AH, however, sTNF-R1 and TNFa were not correlated ðr ¼ 0:05Þ: Concentrations of sTNF-R1 in peripheral and hepatic venous blood were significantly correlated (r ¼ 0:795; P , 0:0001). 4.3. Outcome
Table 3 Univariate analysis Variable Pugh’s score Maddrey’s DF sCreatinine TNFa sTNF-1
The overall survival rate was 78% at 3 months (Fig. 3). Twenty four patients (2 in the non-severe group, 22 in the severe group) died at a median time of 35 days (range: 3 –89) following hospital admission. Except for 4 patients who died after hospital discharge, all deaths occurred during the hospitalisation. The causes of death were progressive liver failure in 14, variceal haemorrhage in 2, severe infection in 4, hepatorenal syndrome in 2, and cardiac failure in one patient. The survival rate at 28 days was 90%, consistent with data from a recent meta-analysis [29]. 4.4. Univariate analysis Univariate analysis (Table 3) identified the Pugh’s score, Maddrey’s DF, serum creatinine, TNFa and sTNF-R1, but not LBP neither CD14, as factors associated with mortality at 3 months.
LBP CD14
Number of patients ,10 $10 ,32 $32 ,72 $7 ,4.5 $4.5 ,3000 $3000 ,45 $45 ,2000 $2000
34 74 46 62 47 61 59 49 48 60 47 61 53 55
P value
0.0006 0.0007 0.01 0.025 0.0018 0.20 0.199
ðP ¼ 0:362Þ; and CD14 ðP ¼ 0:347Þ were not related to survival (Table 4). 4.6. Sensitivity and specificity-ROC curve The ROC curve for serum sTNF-R1 in the group of patients with AH is illustrated in Fig. 4. The AUC was 0.765 (SE 0.065), P ¼ 0:003: At a serum sTNF-R1 level of 4835 pg/ml, the sensitivity in predicting a poor outcome was 89%, while the specificity was 75%. 4.7. Alcoholism
4.5. Multiple regression analysis Multivariate Cox regression analysis showed that sTNFR1 was an independent predictor of mortality at 3 months (OR 4.33: 95% CI [1.12 – 16.75]). Other variables such as the Pugh’s score ðP ¼ 0:618Þ; Maddrey’s DF ðP ¼ 0:182Þ; serum creatinine ðP ¼ 0:197Þ; TNFa ðP ¼ 0:319Þ; LBP
Fig. 3. Cumulative survival at 3 months of 108 patients with cirrhosis and AH (severe and non-severe form).
During the study period, 25 patients (12 in the severe AH group, 13 in the non-severe AH group; total ¼ 23%)
Fig. 4. ROC curve showing the ability of serum sTNF-R1 to predict mortality at 3 months in patients with cirrhosis and AH. Some points of sTNF-R1 values are depicted along the curve.
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
233
Table 4 Cox regression analysis Variable
Regression coefficient b
SE ðbÞ
P coefficient
OR (95% CI)
LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ
1.253 1.055 0.767 0.317 1.465 20.363 20.892
2.515 0.791 0.595 0.318 0.690 0.399 0.949
0.618 0.182 0.197 0.319 0.034 0.362 0.347
3.5 (0.025–484.66) 2.87 (0.61–13.537) 2.15 (0.671– 6.916) 1.37 (0.737– 2.56) 4.33 (1.119– 16.748) 0.69 (0.318– 1.52) 0.41 (0.064– 2.633)
Pugh’s score Maddrey’s DF sCreatinine (mmol/l) TNFa (pg/ml) sTNF-R1 (pg/ml) LBP (ng/ml) CD14 (pg/ml)
resumed a regular alcohol consumption which was documented by review of clinics records and/or plasma ethanol determination. 5. Discussion In this well characterized cohort of patients with biopsyproven AH and no severe associated bacterial infection, we show that serum level of sTNF-R1 measured shortly after hospital admission is an independent predictor of 3-month mortality. These results highlight the elevated residual mortality in AH in spite of a 28-day steroid course in patients with the severe form [32], and also reminds that death may occur in patients with non-severe AH [27,29], in relation with decompensated cirrhosis. In our patients with AH and no bacterial infection at hospital admission, TNFa serum concentration was not associated to a poor outcome. This observation in apparent opposition with results of previous studies [13 – 15] may be confusing with regard to the accepted key role of TNFa in AH. However, elevated level and independent prognostic value of sTNF-R1, the key mediator of TNFa bioactivity [21 – 23], is in line with the central role of TNFa in this inflammatory liver disease. Serum creatinine was not an independent prognostic indicator. This may be surprising, in the view of the known deleterious role of renal failure in patients with severe alcoholic liver injury [33,34]. The majority of our patients had normal or mildly elevated values of serum creatinine, a finding that may explain the absence of statistical significance in the model. We demonstrate high circulating concentrations of sTNF-R1 in patients with cirrhosis and AH with the severe and non-severe form. Our results are in line with data from Naveau et al. [24] who reported highest values in patients with severe AH. In addition, we show for the first time that high sTNF-R1 at hospital admission is an independent predictor of mortality during a 3-month follow-up. In our patients with AH, elevated serum concentrations of sTNFR1 may be interpreted either as a marker of TNFa-mediated inflammation, or as the presence of a tentative regulatory mechanism to down-regulate the inflammatory process [36]. Soluble TNF-R1 originates from the shedding of the extracellular domain both from target cell’s surface and
activated neutrophils [37] consecutive to inflammatory stimuli. In line with this, sTNF receptor is increased following endotoxemia [38], a common situation in alcoholic liver disease [39], and concentration correlates to disease activity [24]. Provided that cytotoxic activity of TNFa is produced subsequent to binding of TNF-R1 to target cells, increased sTNF-R1 should correlate with TNFa levels. In the present study, we observe a significant but weak correlation ðr ¼ 0:33Þ between serum TNFa and sTNF-R1 concentrations, suggesting that factors such as the underlying cirrhosis may influence circulating concentrations of cytokines [26]. Increased sTNF-R1 is considered as a naturally occurring substance generated in the course of inflammation to inhibit the biological effects of TNFa, which may be last several weeks [14] in the setting of AH. In this aspect, high serum sTNF-R1 as an independent predictor of short-term mortality suggests the overproduction of this cytokine antagonist is not sufficient to counteract the deleterious effects of TNFa. Thus, anti cytokine therapy derived from soluble TNF receptor subunits that has been shown to benefit patients with rheumatic disease [40] should be tested in patients with AH. Surprisingly, we observe similar serum levels of LBP and CD14 between groups of patients with cirrhosis and AH and control patients with abstinent alcoholic cirrhosis. Consistent with the abnormal gut permeability in alcoholic cirrhosis [4], chronic exposure of liver macrophages to portal blood endotoxemia [8] is associated to elevated LBP and CD14 serum concentrations. In our patients, we cannot conclude on the absence of participation of these two cytokines in the pathogenesis of AH, as their concentrations have not been assessed in liver and kupffer cells. In conclusion, we show in this well characterized cohort of patients with cirrhosis and biopsy-proven AH that the serum level of sTNF-R1, but not TNFa, obtained at hospital admission are independent predictors of 3-month mortality. Nevertheless, it has to be emphasized that the number of patients included may have introduce a type II error that can explain the absence of significance for other variables such as the Maddrey’s DF, Pugh’s score and serum creatinine. Provided that sTNF-R1 is the soluble form of the cell surface receptor mediating the cytotoxic activity of TNFa, these results support the concept of dysregulated TNFa metabolism in AH.
234
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
References [1] McClain CJ, Barve S, Deaciuc I, Kugelmas M, Hill D. Cytokines in alcoholic liver disease. Semin Liver Dis 1999;19:205 –219. [2] Parlesak A, Scha¨fer C, Schu¨tz T, Bode JC, Bode C. Increased intestinal permeability to macromolecules and endotoxemia in patients with chronic alcohol abuse in different stages of alcoholinduced liver disease. J Hepatol 2000;32:742 –747. [3] Fukui H, Brauner B, Bode JC, Bode C. Plasma endotoxin concentrations in patients with alcoholic and non-alcoholic liver disease: reevaluation with an improved chromogenic assay. J Hepatol 1991;12:162– 169. [4] Keshavarzian A, Holmes EW, Patel M, Iber F, Fields JZ, Pethkar S. Leaky gut in alcoholic cirrhosis: a possible mechanism for alcoholinduced liver damage. Am J Gastroenterol 1999;94:200–207. [5] Tilg H, Wilmer A, Vogel W, Herold M, No¨lchen B, Judmaier G, et al. Serum levels of cytokines in chronic liver diseases. Gastroenterology 1992;103:264 –274. [6] Khoruts A, Stahnke L, McClain CJ, Logan G, Allen JI. Circulating tumor necrosis factor, interleukin-1 and interleukin-6 concentrations in chronic alcoholic patients. Hepatology 1991;13:267 –276. [7] McClain CJ, Barve Sh, Barve S, Deaciuc I, Hill DB. Tumor necrosis factor and alcoholic liver disease. Alcohol Clin Exp Res 1998;22: 248S–252S. [8] Scha¨fer C, Parlesak A, Schu¨tt C, Bode JC, Bode C. Concentrations of lipopolysaccharide-binding protein, bactericidal/permeabilityincreasing protein, soluble CD14 and plasma lipids in relation to endotoxemia in patients with alcoholic liver disease. Alcohol Alcohol 2002;37:81 –86. [9] Uesugi T, Froh M, Arteel GE, Bradford BU, Wheeler MD, Ga¨bele E, et al. Role of lipopolysaccharide-binding protein in early alcoholinduced liver injury in mice. J Immunol 2002;168:2963–2969. [10] Albillos A, de la Hera A, Gonzalez M, Moya JL, Calleja JL, Monserrat J, et al. Increased lipolysaccharide binding protein in cirrhotic patients with marked immune and hemodynamic derangement. Hepatology 2003;37:208 –217. [11] Urbaschek R, McCuskey RS, Rudi V, Becker KP, Sticker F, Urbaschek B, et al. Endotoxin, endotoxin-neutralizing-capacity, sCD14, sICAM-1, and cytokines in patients with various degrees of alcoholic liver disease. Alcohol Clin Exp Res 2001;25:261 –268. [12] Yin M, Bradford BU, Wheeler MD, Uesugi T, Froh M, Goyert SM, et al. Reduced early alcohol-induced liver injury in CD14-deficient mice. J Immunol 2001;166:4737– 4742. [13] Felver ME, Mezey E, McGuire M, Mitchell MC, Herlong HF, Veech GA, et al. Plasma tumor necrosis factor alpha predicts decreased longterm survival in severe alcoholic hepatitis. Alcohol Clin Exp Res 1990;14:255 –259. [14] Rodriguez-Rodriguez E, Gonzalez-Reimers E, Santolaria-Fernandez F, Milena-Abril A, Rodriguez-Moreno F, Oramas-Rodriguez J, et al. Cytokine levels in acute alcoholic hepatitis: a sequential study. Drugs Alcohol Depend 1995;39:23–27. [15] Bird GLA, Sheron N, Goka J, Alexander GJ, Williams RS. Increased plasma tumor necrosis factor in severe alcoholic hepatitis. Ann Intern Med 1990;112:917–920. [16] Spahr L, Rubbia-Brandt L, Frossard JL, Giostra E, Rougemont AL, Pugin J, et al. Combination of steroids with infliximab or placebo in severe alcoholic hepatitis: a randomized controlled pilot study. J Hepatol 2002;37:448– 455. [17] Tilg H, Jalan R, Kaser A, Davies NA, Offner FA, Hodges SJ, et al. Anti-tumor necrosis factor-alpha monoclonal antibody therapy in severe alcoholic hepatitis. J Hepatol 2003;38:419 –425. [18] Papadakis KA, Targan SR. Tumor necrosis factor: biology and therapeutic inhibitors. Gastroenterology 2000;119:1148–1157. [19] Bradham CA, Plumpe J, Manns MP, Brenner DA, Trautwein C. Mechanisms of hepatic toxicity: TNF-induced liver injury. Am J Physiol 1998;275:G387–G392.
[20] Yin M, Wheeler MD, Kono H, Bradford BU, Gallucci RM, Luster MI, et al. Essential role of tumor necrosis factor a in alcohol-induced liver injury in mice. Gastroenterology 1999;117:942–952. [21] Beutler B, van Huffel C. Unraveling function in the TNF ligand and receptor families. Science 1994;264:667–668. [22] Leist M, Gantner F, Jilg S, Wendel A. Activation of the 55 kDa TNF receptor is necessary and sufficient for TNF-induced liver failure, hepatocyte apoptosis, and nitrite release. J Immunol 1995;154: 1307–1316. [23] Tartaglia LA, Weber RF, Figari IS, Reynolds C, Palladino MA, Goeddel DV. The two different receptors for tumor necrosis factor mediate distinct cellular responses. Proc Natl Acad Sci USA 1991;88: 9292–9296. [24] Naveau S, Emilie D, Balian A, Grangeot-Keros L, Borotto E, Portier A, et al. Plasma levels of soluble tumor necrosis factor receptors p55 and p75 in patients with alcoholic liver disease of increasing severity. J Hepatol 1998;28:778 –784. [25] Byl B, Roucloux I, Crusiaux A, Dupont E, Deviere J. Tumor necrosis factor a and interleukin 6 plasma levels in infected cirrhotic patients. Gastroenterology 1993;104:1492–1497. [26] Simpson KJ, Lukacs NW, Colletti L, Strieter RM, Kunkel SL. Cytokines and the liver. J Hepatol 1997;27:1120 –1132. [27] Pessione F, Ramond MJ, Peters L, Pham BN, Batel P, Rueff B, et al. Five-year survival predictive factors in patients with excessive alcohol intake and cirrhosis. Effect of alcoholic hepatitis, smoking and abstinence. Liver Int 2003;23:45–53. [28] Carithers RL, Franklin Herlong H, Diehl AM, Shaw EW, Combes B, Fallon HJ, et al. Methylprednisolone therapy in patients with severe alcoholic hepatitis. Ann Intern Med 1989;110:685–690. [29] Mathurin P, Mendenhall CL, Carithers RL, Ramond MJ, Maddrey WC, Garstide P, et al. Corticosteroids improve short-term survival in patients with severe alcoholic hepatitis (AH): individual data analysis of the last three randomized placebo controlled double blind trials of corticosteroids in severe AH. J Hepatol 2002;36:480 –487. [30] International group, Alcoholic liver disease: morphological manifestations. Review by an International Group. Lancet 1981;1:707– 711. [31] Bautista AP. Neutrophilic infiltration in alcoholic hepatitis. Alcohol 2002;27:17–21. [32] Mathurin P, Abdelnour M, Ramond M-J, Carbonell N, Fartoux L, Serfaty L, et al. Early change in bilirubin levels is an important prognostic factor in severe alcoholic hepatitis treated with prednisolone. Hepatology 2003;38:1363– 1369. [33] Mutimer DJ, Burra P, Neuberger JM, Hubscher S, Buckels JA, Mayer AD, et al. Managing severe alcoholic hepatitis complicated by renal failure. Q J Med 1993;86:649– 656. [34] Fujimoto M, Uemura M, Kojima H, Ishii Y, Ann T, Sakurai S, et al. Prognostic factors in severe alcoholic liver injury. Nara Liver Study Group. Alcohol Clin Exp Res 1999;23:33S–38S. [35] Christensen E. Multivariate survival analysis using Cox’s regression model. Hepatology 1987;7:1346–1358. p. 649–56. [36] Van Zee KJ, Kohno T, Fischer E, Rock CS, Moldawer LL, Lowry SF. Tumor necrosis factor soluble receptors circulate during experimental and clinical inflammation and can protect against excessive tumor necrosis factor a in vitro and in vivo. Proc Natl Acad Sci USA 1992; 89:4845–4849. [37] Porteu F, Nathan C. Shedding of tumor necrosis factor receptors by activated human neutrophils. J Exp Med 1990;172:599–607. [38] Spinas GA, Keller U, Brockhaus M. Release of soluble receptors for tumor necrosis factor (TNF) in relation to circulating TNF during experimental endotoxinemia. J Clin Invest 1992;90:533–536. [39] Fujimoto M, Uemura M, Nakatani Y, Tsujita S, Hoppo K, Tamagawa T, et al. Plasma endotoxin and serum cytokine levels in patients with alcoholic hepatitis: relation to severity of liver disturbance. Alcohol Clin Exp Res 2000;24:48S –54S. [40] Reimold AM. New indications for treatment of chronic inflammation by TNF-alpha blockade. Am J Med Sci 2003;325:75 –92.
Soluble TNF-R1, but not tumor necrosis factor alpha, predicts the 3-month mortality in patients with alcoholic hepatitisq Laurent Spahr1,*, Emile Giostra1, Jean-Louis Frossard1, Solange Bresson-Hadni1, Laura Rubbia-Brandt2, Antoine Hadengue1 1
Gastroenterology and Hepatology, University Hospital, Geneva, Switzerland 2 Clinical Pathology, University Hospital, Geneva, Switzerland
Background/Aims: In alcoholic hepatitis (AH), soluble TNFa receptor-1 (sTNF-R1) is increased. Elevated TNFa predicts mortality, but infection influences TNFa values. In patients with AH, we determined the prognostic value of TNFa, sTNF-R1, and lipopolysaccharide binding protein (LBP) and CD14, both involved in endotoxemia-associated inflammation. Methods: One hundred and eight cirrhotic patients (Pugh score 10 [6 –13]) and biopsy-proven AH (Maddrey’s DF <32: n 5 46; $ 32: n 5 62) without associated infection were included within 8 days of admission and followed-up for 3 months. Cytokines were measured using specific immunoassays. Patients with severe AH received steroids. Results: Twenty four patients died at a median time of 35 days (range: 3– 89). The overall survival was 78%. Multivariate Cox regression analysis showed that sTNF-R1 was an independent predictor of mortality, (OR 4.33: 95% CI [1.12 – 16.75]). Pugh’s score (P 5 0.618), Maddrey’s DF (P 5 0.182), creatinine (P 5 0.197), TNFa (P 5 0.319), LBP (P 5 0.362), and CD14 (P 5 0.347) were not related to survival. Conclusions: In patients with AH, sTNF-R1 measured at admission is an independent predictor of survival at 3 months. Provided that TNF-R1 mediates the cytotoxic actions of TNFa, these results support the concept of dysregulated TNFa metabolism in AH. q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. Keywords: Alcoholic hepatitis; TNFa; Soluble TNF receptor 1; Mortality; Clinical study; Multivariate analysis; Lipopolysaccharide binding protein; Maddrey’s discriminant function; CD14 1. Introduction Alcoholic hepatitis (AH) is a frequent inflammatory liver disease in which a number of pathogenetic mechanisms are involved [1]. In particular, dysregulated inflammatory and immune mechanisms are believed to result from ethanol-associated increased intestinal permeability [2 – 4] and a depressed clearance capacity of the liver reticuloendothelial system [1]. Accordingly, endotoxaemia is transiently elevated in patients with excessive alcohol intake [3]. Endotoxin, or lipopolysaccharide, stimulates Received 22 October 2003; received in revised form 5 April 2004; accepted 13 April 2004; available online 18 May 2004 q Presented in part at the AASLD session of Digestive Disease Week in San Francisco, California, May 2002. * Corresponding author. Tel.: þ 41-223729340; fax: þ 41-223729366. E-mail address: [email protected] (L. Spahr). Abbreviations: sTNF-R1, soluble TNFa receptor-1; TNFa, tumor necrosis factor a; LBP, lipopolysaccharide-binding protein; AH, alcoholic hepatitis.
Kupffer cells to produce proinflammatory cytokines [5,6], thus contributing to the severe inflammatory response that characterizes AH [1,7]. The action of endotoxin on target cells is mediated by two glycoproteins, lipopolysaccharide binding protein (LBP) and CD14. Lipopolysaccharide binding protein, the synthesis of which is increased in the presence of endotoxemia [8,9], acts as a carrier for endotoxin molecules, and has been proposed as a good marker of long-term exposure to endotoxin [10]. The endotoxin– LBP complex then binds to CD14, a membrane-bound receptor on Kupffer cells, to activate the cellular production of inflammatory mediators [11]. In CD14-deficient mice [12], alcoholic liver injury is reduced. In patients with alcoholic liver disease, circulating concentrations of the lipoproteins LBP and CD14 are elevated [8,11], with LBP correlating to serum CD14 in patients with cirrhosis [10]. Elevated TNFa is a typical feature of AH [13 – 17]. TNFa mediates its varied biological responses by binding to two cell surface receptors that signal different cellular
0168-8278/$30.00 q 2004 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhep.2004.04.028
230
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
pathways: TNF-R1 (also named p55, according to its molecular mass of 55 kD) and TNF-R2 (also named p75) [18,19]. TNF-R1 is the main inductor of cytotoxicity, promoting a number of biological responses ranging from NF-kB activation to cell death. Accordingly, mice deficient in TNF-R1 [20] are protected against alcoholic liver injury. Soluble form of both receptors are present constitutively in serum, derived after receptor shedding after cellular activation by stimuli such as TNFa production [21]. Thus, sTNFR serum level is considered a reliable indicator of TNFa system activation [22,23]. In patients with AH, sTNF-R1 plasma values increase with disease severity [24]. High plasma TNFa in patients hospitalised for AH has been shown to predict a poor outcome in several studies [13 –15]. However, interpretation of elevated TNFa values is difficult in the presence of bacterial infections [25] or cirrhosis [26], both conditions known to affect circulating TNFa levels. Thus, the prognostic significance of TNFa in studies in which such confounding factors have not been considered, should be questioned. Finally, recent trials that aimed at antagonizing TNFa using monoclonal antibody did not demonstrate evident survival benefit [16,17], but a possible increased mortality (S. Naveau, personal communication). Therefore, the prognostic role of TNFa, soluble receptor 1 and endotoxin-associated proteins should be clarify in patients with AH. We decided to determine the serum concentrations and prognostic value of TNFa, sTNF-R1, CD14 and LBP obtained early after hospital admission in a cohort of 108 patients with cirrhosis and AH defined in pathological terms, in the absence of associated severe bacterial infection.
2. Patients and methods
Fig. 1. Algorithm of patients’ selection.
AH. All patients with histologically-proven severe AH received a 28-day course of prednisolone, as recommended [29], in association with oral norfloxacin in the presence of low-protein ascites. A severe infection (spontaneous bacterial peritonitis in two patients, pneumonia in two patients, urinary tract-associated sepsis in one patient) associated to AH was detected at hospital admission. These patients were excluded. The study protocol was conform to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethical committee of the Hoˆpitaux Universitaires de Gene`ve. Written informed consent was obtained from all patients and controls.
2.2. Liver biopsy Liver biopsy was performed in all patients within 7 days of hospitalisation by the transjugular route, due to the presence of coagulopathy and/or ascites. Histopathological examination was performed by a single histopathologist (L. R-B) expert in liver diseases. The diagnosis of AH was based on definite criteria established by international experts [30]. Thus, ballooning degeneration of hepatocytes and necrosis, Mallory bodies, fatty changes, cholestasis and parenchymal infiltration with neutrophils [31] were present in all patients’ liver biopsy.
2.3. Outcome 2.1. Patients One hundred and eight patients with AH included between January 2000 and December 2002 in a single institution formed the basis of this study. All patients had underlying cirrhosis, an observation that prevails in European countries [27]. The algorithm of patients’ selection is described in Fig. 1, and demographic data are given in Table 1. The Maddrey’s discriminant function (DF), a strong prognostic indicator of disease severity, is calculated by integrating the prothrombin time and serum bilirubin in the formula [28]: 4,6 £ [prothrombin time—control time (in seconds)] þ serum bilirubin (in mmol/l)/17. A score $32 defines a severe
We determined the outcome of all patients (survival or death) over a 3-month period using hospital records and telephone inquiry to general practitioners. We chose arbitrarily this follow-up period of 3 months because increased mortality following severe AH has been reported to persist beyond the treatment period [32].
2.4. Biological values All blood samples were collected at the time of liver biopsy in a fasting state, and transported immediately on ice to the laboratory. After
Table 1 Patients characteristics
n Gender (M/F) Age (years, range) Pugh’s score Maddrey’s DF Serum creatinine (mmol/l)
Controls
AH and DF , 32
AH and DF $ 32
12 8/4 52 (46–59) 8 (7–9) – 64 (58–75)
46 31/15 54 (41–69) 8 (6–12) 20 (15–29) 67 (54–101)
62 39/23 57 (34– 73) 11 (8–13)* 50 (32– 112)** 77 (48– 194)*
Abbreviations: DF: Maddrey’s discriminant function. Note: controls are patients with abstinent alcoholic cirrhosis. *P , 0:05 vs. controls and vs. patients with AH and DF , 32; **P , 0:01 vs. controls and vs. patients with AH and DF , 32 (Bonferroni’s correction for multiple testing of data).
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234 centrifugation for 15 min at 5 8C, aliquots were made under pyrogen-free conditions and the samples kept frozen at 270 8C until assayed. We used commercially available specific immunoassays for the measurement in duplicate of TNFa, sTNF-R1, CD14 (R&D Systems Europe, Abingdon, UK) and LBP (Bioreba AG, Reinach, Switzerland). The immunoassay used for sTNF-R1 is not influenced by the presence of TNFa and thus determines the total amount of the soluble receptors in the sample. Except for sTNF-R1 values which were assessed both in peripheral and hepatic venous blood, cytokines were measured in peripheral blood only. We decided to measure sTNF-R1 only, as it reflects hepatocellular injury better than sTNF-R2 [21–23]. For TNFa, the minimum detectable concentration was less than 4 pg/ml, and the intra- and interassay variability were 4.2 and 4.6%, respectively. For sTNF-R1, the minimum detectable dose was less than 3 pg/ml. Intra- and interassay variability were 2.9 and 3.7%, respectively. The minimum detection level for LBP was 1 ng/ml, with a measurable concentration range of 0.4 to 100 ng/ml. For CD14, the sensitivity threshold was 125 pg/ml, and the intra- and interassay variability were 5.2 and 4.8%, respectively. We measured serum creatinine in all patients, as renal function has been shown to influence the prognosis in AH [33,34].
3. Statistical analysis Results are summarized as medians and ranges. Due to the non-normal distribution of values, group comparisons were performed by the non-parametric Kruskal Wallis method, with Bonferroni’s post hoc analysis to correct significance levels for multiple testing of data. Survival curves were estimated using the Kaplan – Meier method. The prognostic significance of the variables in the univariate analysis was determined using the log rank test. Rating for each variable was determined according to the median value in the group. We used the Cox regression analysis (Cox’s proportional hazard model) with the forward-stepwise method [35] (likelihood ratio) and enter and remove limits of P , 0:05 and P . 0:10; respectively. To improve model fit in the multivariate analysis, we performed a logarithmic transformation of all variables prior to introduce them in the model as continuous variables. The assumptions of the Cox model including the assumption of proportional hazards over time were assessed. In view of the limited number of events in the study period, the assumption of proportionality does not seem to be grossly violated. Correlation analyses were performed by using the Spearman rank correlation test. To assess the use of sTNF-R1 serum value in a clinical setting, we constructed a receiver operating characteristic (ROC) curve by plotting
231
Fig. 2. Box-whisker plots (median, upper and lower quartiles and range) showing elevated serum values of TNFa in patients with severe AH (DF $ 32, upper left), high sTNF-R1 in patients with severe (DF $ 32) and non-severe AH (<32, upper right) compared to patients with abstinent alcoholic cirrhosis. Note: a difference between LBP in controls and in patients with AH was found to be significant ðP < 0:03Þ; but Bonferroni’s post hoc test removes this significance. *P < 0:0001 vs. controls and vs. patients with non-severe AH.
the sensitivity of sTNF-R1 against the reverse specificity (1 2 specificity). All statistical tests were performed by using the Statistical Package for the Social Sciences, version 9.0 for Windows (SPSS, Chicago, IL, USA).
4. Results 4.1. Cytokines Cytokines values are given in Table 2 and illustrated in Fig. 2. Serum TNFa levels were elevated in patients with severe AH compared to patients with non-severe AH and patients with abstinent cirrhosis. Patients with severe and non-severe AH showed higher serum sTNF-R1 values compared to controls. Circulating concentrations of LBP and CD14, however, were similar between groups. 4.2. Correlations In patients with AH, a weak but statistically significant
Table 2 Cytokines values
TNF (pg/ml) sTNF-R1 (PV) (pg/ml) sTNF-R1 (HV) (pg/ml) LBP (ng/ml) CD14 (pg/ml)
Controls
AH and DF , 32
AH and DF $ 32
1.7 (0.5–4.3) 1943 (1111–2201) – 26.6 (11.4–46.4) 1588 (1321–2166)
1.9 (0.4– 9.5) 2480 (360–5771)* 2574 (900–6132) 47.9 (18.8 –73.3) 2065 (1105–2806)
7.4 (1.3–34.6)** 3541 (1707–11747)** 3518 (1330–11176)*** 44.6 (15–167.2) 1964 (907–3465)
Abbreviations: DF: Maddrey’s discriminant function. *P , 0:01 vs. controls; **P , 0:01 vs. controls and vs. patients with AH and DF , 32 (Bonferroni’s corrections for multiple testing of data); ***P , 0:01 vs. patients with AH and DF , 32.
232
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
correlation was found between TNFa and sTNF-R1 (r ¼ 0:33; P , 0:001), as well as between TNFa and the Maddrey’s DF at hospital admission (r ¼ 0:35; P , 0:001). In the subgroup of patients with severe AH, however, sTNF-R1 and TNFa were not correlated ðr ¼ 0:05Þ: Concentrations of sTNF-R1 in peripheral and hepatic venous blood were significantly correlated (r ¼ 0:795; P , 0:0001). 4.3. Outcome
Table 3 Univariate analysis Variable Pugh’s score Maddrey’s DF sCreatinine TNFa sTNF-1
The overall survival rate was 78% at 3 months (Fig. 3). Twenty four patients (2 in the non-severe group, 22 in the severe group) died at a median time of 35 days (range: 3 –89) following hospital admission. Except for 4 patients who died after hospital discharge, all deaths occurred during the hospitalisation. The causes of death were progressive liver failure in 14, variceal haemorrhage in 2, severe infection in 4, hepatorenal syndrome in 2, and cardiac failure in one patient. The survival rate at 28 days was 90%, consistent with data from a recent meta-analysis [29]. 4.4. Univariate analysis Univariate analysis (Table 3) identified the Pugh’s score, Maddrey’s DF, serum creatinine, TNFa and sTNF-R1, but not LBP neither CD14, as factors associated with mortality at 3 months.
LBP CD14
Number of patients ,10 $10 ,32 $32 ,72 $7 ,4.5 $4.5 ,3000 $3000 ,45 $45 ,2000 $2000
34 74 46 62 47 61 59 49 48 60 47 61 53 55
P value
0.0006 0.0007 0.01 0.025 0.0018 0.20 0.199
ðP ¼ 0:362Þ; and CD14 ðP ¼ 0:347Þ were not related to survival (Table 4). 4.6. Sensitivity and specificity-ROC curve The ROC curve for serum sTNF-R1 in the group of patients with AH is illustrated in Fig. 4. The AUC was 0.765 (SE 0.065), P ¼ 0:003: At a serum sTNF-R1 level of 4835 pg/ml, the sensitivity in predicting a poor outcome was 89%, while the specificity was 75%. 4.7. Alcoholism
4.5. Multiple regression analysis Multivariate Cox regression analysis showed that sTNFR1 was an independent predictor of mortality at 3 months (OR 4.33: 95% CI [1.12 – 16.75]). Other variables such as the Pugh’s score ðP ¼ 0:618Þ; Maddrey’s DF ðP ¼ 0:182Þ; serum creatinine ðP ¼ 0:197Þ; TNFa ðP ¼ 0:319Þ; LBP
Fig. 3. Cumulative survival at 3 months of 108 patients with cirrhosis and AH (severe and non-severe form).
During the study period, 25 patients (12 in the severe AH group, 13 in the non-severe AH group; total ¼ 23%)
Fig. 4. ROC curve showing the ability of serum sTNF-R1 to predict mortality at 3 months in patients with cirrhosis and AH. Some points of sTNF-R1 values are depicted along the curve.
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
233
Table 4 Cox regression analysis Variable
Regression coefficient b
SE ðbÞ
P coefficient
OR (95% CI)
LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ LðnÞ
1.253 1.055 0.767 0.317 1.465 20.363 20.892
2.515 0.791 0.595 0.318 0.690 0.399 0.949
0.618 0.182 0.197 0.319 0.034 0.362 0.347
3.5 (0.025–484.66) 2.87 (0.61–13.537) 2.15 (0.671– 6.916) 1.37 (0.737– 2.56) 4.33 (1.119– 16.748) 0.69 (0.318– 1.52) 0.41 (0.064– 2.633)
Pugh’s score Maddrey’s DF sCreatinine (mmol/l) TNFa (pg/ml) sTNF-R1 (pg/ml) LBP (ng/ml) CD14 (pg/ml)
resumed a regular alcohol consumption which was documented by review of clinics records and/or plasma ethanol determination. 5. Discussion In this well characterized cohort of patients with biopsyproven AH and no severe associated bacterial infection, we show that serum level of sTNF-R1 measured shortly after hospital admission is an independent predictor of 3-month mortality. These results highlight the elevated residual mortality in AH in spite of a 28-day steroid course in patients with the severe form [32], and also reminds that death may occur in patients with non-severe AH [27,29], in relation with decompensated cirrhosis. In our patients with AH and no bacterial infection at hospital admission, TNFa serum concentration was not associated to a poor outcome. This observation in apparent opposition with results of previous studies [13 – 15] may be confusing with regard to the accepted key role of TNFa in AH. However, elevated level and independent prognostic value of sTNF-R1, the key mediator of TNFa bioactivity [21 – 23], is in line with the central role of TNFa in this inflammatory liver disease. Serum creatinine was not an independent prognostic indicator. This may be surprising, in the view of the known deleterious role of renal failure in patients with severe alcoholic liver injury [33,34]. The majority of our patients had normal or mildly elevated values of serum creatinine, a finding that may explain the absence of statistical significance in the model. We demonstrate high circulating concentrations of sTNF-R1 in patients with cirrhosis and AH with the severe and non-severe form. Our results are in line with data from Naveau et al. [24] who reported highest values in patients with severe AH. In addition, we show for the first time that high sTNF-R1 at hospital admission is an independent predictor of mortality during a 3-month follow-up. In our patients with AH, elevated serum concentrations of sTNFR1 may be interpreted either as a marker of TNFa-mediated inflammation, or as the presence of a tentative regulatory mechanism to down-regulate the inflammatory process [36]. Soluble TNF-R1 originates from the shedding of the extracellular domain both from target cell’s surface and
activated neutrophils [37] consecutive to inflammatory stimuli. In line with this, sTNF receptor is increased following endotoxemia [38], a common situation in alcoholic liver disease [39], and concentration correlates to disease activity [24]. Provided that cytotoxic activity of TNFa is produced subsequent to binding of TNF-R1 to target cells, increased sTNF-R1 should correlate with TNFa levels. In the present study, we observe a significant but weak correlation ðr ¼ 0:33Þ between serum TNFa and sTNF-R1 concentrations, suggesting that factors such as the underlying cirrhosis may influence circulating concentrations of cytokines [26]. Increased sTNF-R1 is considered as a naturally occurring substance generated in the course of inflammation to inhibit the biological effects of TNFa, which may be last several weeks [14] in the setting of AH. In this aspect, high serum sTNF-R1 as an independent predictor of short-term mortality suggests the overproduction of this cytokine antagonist is not sufficient to counteract the deleterious effects of TNFa. Thus, anti cytokine therapy derived from soluble TNF receptor subunits that has been shown to benefit patients with rheumatic disease [40] should be tested in patients with AH. Surprisingly, we observe similar serum levels of LBP and CD14 between groups of patients with cirrhosis and AH and control patients with abstinent alcoholic cirrhosis. Consistent with the abnormal gut permeability in alcoholic cirrhosis [4], chronic exposure of liver macrophages to portal blood endotoxemia [8] is associated to elevated LBP and CD14 serum concentrations. In our patients, we cannot conclude on the absence of participation of these two cytokines in the pathogenesis of AH, as their concentrations have not been assessed in liver and kupffer cells. In conclusion, we show in this well characterized cohort of patients with cirrhosis and biopsy-proven AH that the serum level of sTNF-R1, but not TNFa, obtained at hospital admission are independent predictors of 3-month mortality. Nevertheless, it has to be emphasized that the number of patients included may have introduce a type II error that can explain the absence of significance for other variables such as the Maddrey’s DF, Pugh’s score and serum creatinine. Provided that sTNF-R1 is the soluble form of the cell surface receptor mediating the cytotoxic activity of TNFa, these results support the concept of dysregulated TNFa metabolism in AH.
234
L. Spahr et al. / Journal of Hepatology 41 (2004) 229–234
References [1] McClain CJ, Barve S, Deaciuc I, Kugelmas M, Hill D. Cytokines in alcoholic liver disease. Semin Liver Dis 1999;19:205 –219. [2] Parlesak A, Scha¨fer C, Schu¨tz T, Bode JC, Bode C. Increased intestinal permeability to macromolecules and endotoxemia in patients with chronic alcohol abuse in different stages of alcoholinduced liver disease. J Hepatol 2000;32:742 –747. [3] Fukui H, Brauner B, Bode JC, Bode C. Plasma endotoxin concentrations in patients with alcoholic and non-alcoholic liver disease: reevaluation with an improved chromogenic assay. J Hepatol 1991;12:162– 169. [4] Keshavarzian A, Holmes EW, Patel M, Iber F, Fields JZ, Pethkar S. Leaky gut in alcoholic cirrhosis: a possible mechanism for alcoholinduced liver damage. Am J Gastroenterol 1999;94:200–207. [5] Tilg H, Wilmer A, Vogel W, Herold M, No¨lchen B, Judmaier G, et al. Serum levels of cytokines in chronic liver diseases. Gastroenterology 1992;103:264 –274. [6] Khoruts A, Stahnke L, McClain CJ, Logan G, Allen JI. Circulating tumor necrosis factor, interleukin-1 and interleukin-6 concentrations in chronic alcoholic patients. Hepatology 1991;13:267 –276. [7] McClain CJ, Barve Sh, Barve S, Deaciuc I, Hill DB. Tumor necrosis factor and alcoholic liver disease. Alcohol Clin Exp Res 1998;22: 248S–252S. [8] Scha¨fer C, Parlesak A, Schu¨tt C, Bode JC, Bode C. Concentrations of lipopolysaccharide-binding protein, bactericidal/permeabilityincreasing protein, soluble CD14 and plasma lipids in relation to endotoxemia in patients with alcoholic liver disease. Alcohol Alcohol 2002;37:81 –86. [9] Uesugi T, Froh M, Arteel GE, Bradford BU, Wheeler MD, Ga¨bele E, et al. Role of lipopolysaccharide-binding protein in early alcoholinduced liver injury in mice. J Immunol 2002;168:2963–2969. [10] Albillos A, de la Hera A, Gonzalez M, Moya JL, Calleja JL, Monserrat J, et al. Increased lipolysaccharide binding protein in cirrhotic patients with marked immune and hemodynamic derangement. Hepatology 2003;37:208 –217. [11] Urbaschek R, McCuskey RS, Rudi V, Becker KP, Sticker F, Urbaschek B, et al. Endotoxin, endotoxin-neutralizing-capacity, sCD14, sICAM-1, and cytokines in patients with various degrees of alcoholic liver disease. Alcohol Clin Exp Res 2001;25:261 –268. [12] Yin M, Bradford BU, Wheeler MD, Uesugi T, Froh M, Goyert SM, et al. Reduced early alcohol-induced liver injury in CD14-deficient mice. J Immunol 2001;166:4737– 4742. [13] Felver ME, Mezey E, McGuire M, Mitchell MC, Herlong HF, Veech GA, et al. Plasma tumor necrosis factor alpha predicts decreased longterm survival in severe alcoholic hepatitis. Alcohol Clin Exp Res 1990;14:255 –259. [14] Rodriguez-Rodriguez E, Gonzalez-Reimers E, Santolaria-Fernandez F, Milena-Abril A, Rodriguez-Moreno F, Oramas-Rodriguez J, et al. Cytokine levels in acute alcoholic hepatitis: a sequential study. Drugs Alcohol Depend 1995;39:23–27. [15] Bird GLA, Sheron N, Goka J, Alexander GJ, Williams RS. Increased plasma tumor necrosis factor in severe alcoholic hepatitis. Ann Intern Med 1990;112:917–920. [16] Spahr L, Rubbia-Brandt L, Frossard JL, Giostra E, Rougemont AL, Pugin J, et al. Combination of steroids with infliximab or placebo in severe alcoholic hepatitis: a randomized controlled pilot study. J Hepatol 2002;37:448– 455. [17] Tilg H, Jalan R, Kaser A, Davies NA, Offner FA, Hodges SJ, et al. Anti-tumor necrosis factor-alpha monoclonal antibody therapy in severe alcoholic hepatitis. J Hepatol 2003;38:419 –425. [18] Papadakis KA, Targan SR. Tumor necrosis factor: biology and therapeutic inhibitors. Gastroenterology 2000;119:1148–1157. [19] Bradham CA, Plumpe J, Manns MP, Brenner DA, Trautwein C. Mechanisms of hepatic toxicity: TNF-induced liver injury. Am J Physiol 1998;275:G387–G392.
[20] Yin M, Wheeler MD, Kono H, Bradford BU, Gallucci RM, Luster MI, et al. Essential role of tumor necrosis factor a in alcohol-induced liver injury in mice. Gastroenterology 1999;117:942–952. [21] Beutler B, van Huffel C. Unraveling function in the TNF ligand and receptor families. Science 1994;264:667–668. [22] Leist M, Gantner F, Jilg S, Wendel A. Activation of the 55 kDa TNF receptor is necessary and sufficient for TNF-induced liver failure, hepatocyte apoptosis, and nitrite release. J Immunol 1995;154: 1307–1316. [23] Tartaglia LA, Weber RF, Figari IS, Reynolds C, Palladino MA, Goeddel DV. The two different receptors for tumor necrosis factor mediate distinct cellular responses. Proc Natl Acad Sci USA 1991;88: 9292–9296. [24] Naveau S, Emilie D, Balian A, Grangeot-Keros L, Borotto E, Portier A, et al. Plasma levels of soluble tumor necrosis factor receptors p55 and p75 in patients with alcoholic liver disease of increasing severity. J Hepatol 1998;28:778 –784. [25] Byl B, Roucloux I, Crusiaux A, Dupont E, Deviere J. Tumor necrosis factor a and interleukin 6 plasma levels in infected cirrhotic patients. Gastroenterology 1993;104:1492–1497. [26] Simpson KJ, Lukacs NW, Colletti L, Strieter RM, Kunkel SL. Cytokines and the liver. J Hepatol 1997;27:1120 –1132. [27] Pessione F, Ramond MJ, Peters L, Pham BN, Batel P, Rueff B, et al. Five-year survival predictive factors in patients with excessive alcohol intake and cirrhosis. Effect of alcoholic hepatitis, smoking and abstinence. Liver Int 2003;23:45–53. [28] Carithers RL, Franklin Herlong H, Diehl AM, Shaw EW, Combes B, Fallon HJ, et al. Methylprednisolone therapy in patients with severe alcoholic hepatitis. Ann Intern Med 1989;110:685–690. [29] Mathurin P, Mendenhall CL, Carithers RL, Ramond MJ, Maddrey WC, Garstide P, et al. Corticosteroids improve short-term survival in patients with severe alcoholic hepatitis (AH): individual data analysis of the last three randomized placebo controlled double blind trials of corticosteroids in severe AH. J Hepatol 2002;36:480 –487. [30] International group, Alcoholic liver disease: morphological manifestations. Review by an International Group. Lancet 1981;1:707– 711. [31] Bautista AP. Neutrophilic infiltration in alcoholic hepatitis. Alcohol 2002;27:17–21. [32] Mathurin P, Abdelnour M, Ramond M-J, Carbonell N, Fartoux L, Serfaty L, et al. Early change in bilirubin levels is an important prognostic factor in severe alcoholic hepatitis treated with prednisolone. Hepatology 2003;38:1363– 1369. [33] Mutimer DJ, Burra P, Neuberger JM, Hubscher S, Buckels JA, Mayer AD, et al. Managing severe alcoholic hepatitis complicated by renal failure. Q J Med 1993;86:649– 656. [34] Fujimoto M, Uemura M, Kojima H, Ishii Y, Ann T, Sakurai S, et al. Prognostic factors in severe alcoholic liver injury. Nara Liver Study Group. Alcohol Clin Exp Res 1999;23:33S–38S. [35] Christensen E. Multivariate survival analysis using Cox’s regression model. Hepatology 1987;7:1346–1358. p. 649–56. [36] Van Zee KJ, Kohno T, Fischer E, Rock CS, Moldawer LL, Lowry SF. Tumor necrosis factor soluble receptors circulate during experimental and clinical inflammation and can protect against excessive tumor necrosis factor a in vitro and in vivo. Proc Natl Acad Sci USA 1992; 89:4845–4849. [37] Porteu F, Nathan C. Shedding of tumor necrosis factor receptors by activated human neutrophils. J Exp Med 1990;172:599–607. [38] Spinas GA, Keller U, Brockhaus M. Release of soluble receptors for tumor necrosis factor (TNF) in relation to circulating TNF during experimental endotoxinemia. J Clin Invest 1992;90:533–536. [39] Fujimoto M, Uemura M, Nakatani Y, Tsujita S, Hoppo K, Tamagawa T, et al. Plasma endotoxin and serum cytokine levels in patients with alcoholic hepatitis: relation to severity of liver disturbance. Alcohol Clin Exp Res 2000;24:48S –54S. [40] Reimold AM. New indications for treatment of chronic inflammation by TNF-alpha blockade. Am J Med Sci 2003;325:75 –92.