High Concentrations of Soluble Tumor Necrosis Factor Receptors in Ascites TILOANDUS,l * VOLKER GROSS,^ * AXEL HOLSTEGE,~ * MARGIT OTT,l * MARLIES WEBER,l HARALD GAL LA TI,^ WOLFGANG GEROK~ AND J ~ G E SCHOLMERICH' N *
MARTINA
DAVID,'
lDepartment of Internal Medicine, University of Freiburg, W-7800 Freiburg, Germany; and 'F. Hoffmann La Roche AG, Pharmaceutical Research-New Technologies, Basel, Switzerland
Ascites and plasma concentrations of soluble tumor receptor p75 levels were usually higher in plasma than necrosis factor receptors p55 and p75 were measured in ascites in patients with uncomplicated hepatic in a prospective study in 34 patients (35occasions of ascites. The concentrations of both tumor necrosis ascites)with hepatic (5 infected and 21 uninfected) and factor receptors correlated well in ascites (r = 0.83, malignancy-related (9) ascites. All patients had high p < 0.001) and plasma (r = 0.85, p < 0.001) but only concentrations of both soluble tumor necrosis factor weakly with tumor necrosis factor-a levels in ascites receptors in ascites and plasma; these were about 500 (p55: r = 0.32, p = 0.03; p75: r = 0.29, p = 0.047) and times higher than the corresponding tumor necrosis not with tumor necrosis factor-a in plasma. The high soluble tumor necrosis factor receptor factor-aconcentrations. Ascites levels of soluble tumor necrosis factor receptors p55 and soluble tumor ne- concentrationsfound in ascitesmay influence the local crosis factor receptors p75 were significantly elevated bioavailability of tumor necrosis factor and might have in patients with malignancy-related (p55: 26.0 k 8.6 an impact on the treatment of peritoneal carcinoman g / e p76 20.5 k 17.4 ng/ml; mean S.D.) and in- tosis with recombinant tumor necrosis factor. Furfected ascites (p55: 25.1 k 10.9 ng/ml, p75 22.6 k 11.0 thermore, the determinationof soluble tumor necrosis ng/ml) compared with patients with uncomplicated factor receptors could be of value for the differential hepatic ascites (p55: 10.1 4.4 ng/ml; p75: 6.0 2.6 diagnosis of ascites. (HEPATOLOGY 1992;16749-755.) ng/ml). Patients with infected or malignancy-related ascites also showed higher soluble tumor necrosis Tumor necrosis factor (TNF1-a (cachedin) and factor receptor concentrations in plasma than did TNF-P (lymphotoxin) are structurally and functionally patients with plain hepatic ascites. Succeesful antibiotic treatment of peritonitis reduced soluble tumor related proteins produced mainly by activated macroneed factor receptor p55 and p75 ascites levels in phages and lymphocytes, respectively (1-3).Both cythree patients from 24.2 15.2 ng/d to 10.7 1.9 tokines bind to the same surface receptors and mediate ng/ml and from 20.2 k 14.4 ng/ml to 7.5 k 1.8 ndml, a wide variety of biological effects, including cytotoxicity respectively. Soluble tumor necrosis factor receptors against tumors and virus-infected cells, inflammation, p55 and p75 at cutoff levels of 16.5 ng/ml and 9.5 ng/ml, wasting, shock and stimulation of various immune respectively, differentiated between infected or ma- effector cells (1-4). Two distinct T N F receptors that lignant and plain hepatic ascites with diagnostic accu- belong to the nerve growth factorlepidermal growth racies of 94% and 89%, respectively. They did not factorflow-density lipoprotein family, with molecular differentiate between infected and malignant ascites. ) 75,000 Da (~751,have The concentrations of soluble tumor necrosis factor masses of 55,000 Da ( ~ 5 5and (5-9). Soluble inhibitory forms been purified and cloned receptor pS5 were usually higher in ascites than in plasma in all subgroups of patients. Levels of soluble of TNF receptors have been found in human serum and tumor necrosis factor receptor p75 were also higher in urine (10-15). In patients with viral and alcoholic hepatitis or ascites than in plasma in patients with malignant and infected ascites. but soluble tumor necrosis factor cirrhosis, elevated plasma levels of TNF-a (16-18) and
*
*
*
Received December 9, 1991; accepted April 22, 1992. This work was supported by the Deutsche Forschungagerneinschaft (SFB 154 and a "Heisenberg-Stipendium" to VG). *Current address: Medizinische Klinik und Powdinik I, Universitat Regemburg, W-8400Regensburg, Germany. $Current address: F. Hoffiann La Roche AG, Pharmaceutical Research-New Technologies, Basel, Switzerland. Address reprint requests to: Dr. Tilo Andus, Medizinische Klinik und Poliklinik I, Universitiit Regemburg, F r a u Josef Strauss Nee 11,Postfach 10 06 62, W-8400€&gemburg,Germany.
31/1/99252
altered production of TNF-a in lipopolysaccharidestimulated peripheral blood monocytes have been found (19-23). TNF-a production correlated with histological activity index (22), and TNF-a plasma levels correlated with the long-term survival rates of patients with severe alcoholic hepatitis (24). Recently, increased numbers of TNF receptors have been found on peripheral blood monocytes of patients with chronic HBV infection (25). In ascitic fluid of mice injected with different hybridomas (26) and of four patients with spontaneous bacterial peritonitis (SBP) (27), elevated TNF-a levels have been detected.
749
750
ANDUS ET AL.
HEPATOLOGY
TABLE1. Patient characteristics Ascites Characteristics
No. of patients Age
0.I)"
Child classification (A/B/C) Positive bacterial blood cultures Antibiotic treatment Temperature (Celsius);" Ascitic leukocytes (per ml)" Ascitic total protein (gm/dl)" Ascitic LDH [U/L]" Ascitic cholesterol (mg/dl)" Ascitic fibronectin (pg/ml)" Ascitic lactate (mmoUL)" Ascitic pH" Serum bilirubin (mg/dl)" Prothrombin time (quick; %)" Serum creatinine (mg/dl)" Deaths during hospital stay
Hepatic
Infected
21 56.0 f 12.0 41918 2 5 37.0 f 0.8 195 t 128 1.84 -t 1.33 51 f 36 31.5 -t 28.9 97.4 55.4 2.1 f 0.9 7.46 -t 0.06 8.9 f 9.6 50 f 18 1.62 1.14 4
5 56.6 f 6.3 0/1/4 1 3 37.0 f 0.4 4,980 2 1,24@ 2.55 f 0.26 188 f 112 20.5 t 2.9 105 -t 32.1 6.6 2 1.8' 7.21 f 0.07b 12.2 f 6.9 35 f 4 2.04 f 0.65 3
*
Malignant 9 64
f
9.4
0 0 36.7 f 0.5 2,410 f 3,64W 4.11 f 1.42d 225 f 135d 122 2 75.5d 200 f 68.1d 2.8 f 0.8 7.40 f 0.06f 0.7 f 0.7d 77 f 2od 0.80 2 1.40 1
"Data expressed as mean 2 S.D. bHepaticascites/infeded ascites: p < 0.01. 'Hepatic ascites/malignant ascites: p < 0.01. dHepatic ascites/malignant ascites: p < 0.005. "Hepaticascites/infeded-ascites: p < 0.05. fHepatic ascites/malignant ascites: p < 0.05.
Nothing is known about soluble TNF receptors in ascites. In this prospective study, we determined the concentrations of soluble TNF receptors p55 and p75, both in ascitic fluid and plasma of 34 patients with ascites of hepatic or malignancy-related origin. PATZENTS AND METHODS Patients. We prospectively studied 34 patients (18men and 16 women) with ascites from June to December 1990. Paracentesis was performed for diagnostic or therapeutic reasons. The patients were divided into the following three groups. Group 1consisted of 21 patients (11men and 10 women) with hepatic ascites caused by chronic parenchymal liver disease. Eighteen had alcoholic cirrhosis and three had cirrhosis due to chronic viral hepatitis B. One of these patients also had an HCC. Peritoneal carcinomatosiswas excluded in all cases by ascitic fluid cytological study and sonography and, in some cases, additionally by computed tomography or laparoscopy. Group 2 comprised five patients (three men and two women) with hepatic ascites and SBP or secondary bacterial peritonitis (two had Enterococcus fizecalis, one had Escherichia coli, one had Staphylococcus aureus and E. fizecalis and one had Candida albicans). Group 3 comprised nine patients (four men and five women) with malignancy-related ascites. Diagnosis of malignancyrelated ascites was confirmed by positive cytological appearance of ascites, computed tomography, laparoscopy and/or biopsy. Ovarian carcinoma was found in five patients, nonHodgkin's lymphoma was found in two patients, gastric carcinoma was found in one patient and malignant melanoma with liver metastasis was found in one patient. One woman was hospitalized twice, first with uncomplicated hepatic ascites and several weeks later with perfor-
ated cholecystitis. Therefore she was included in groups 1 and 2. Laboratory data are given in Table 1. Patients with peritonitis had higher levels of ascites leukocytes than did those with plain hepatic ascites. They also had significantly higher concentrations of lactate and lactate dehydrogenase (LDH) and lower ascitic fluid pH values. Three of the five patients (60%) with peritonitis died, whereas only 4 of 21 patients (19%) with plain hepatic ascites died during the observation period. The subgroup of patients with malignant ascites had significantly higher levels of leukocytes, total protein, LDH, cholesterol and fibronectin in the ascitic fluid. Kidney function did not differ significantly between the three groups (Table 1). Methods. Ascites and blood were obtained aseptically,cooled on ice immediatelyand centrifuged at 4"C. Aliquots of cell-free . samples were stored at - 20" C for up to 3 wk before analysis. No significant degradation of TNF receptors occurred during this time. In two patients with malignancy-related ascites, cytokine concentrations were not determined in plasma because there was not enough plasma. Concentrations of TNF receptors p55 and p75 in ascites and plasma were measured in duplicate with enzyme-linked immunological-biological binding assays. Microtiter plates were sensibilized with 1 pg of noninhibitory monoclonal (mouse) antibodies against soluble TNF receptor p55 (clone: htr 20) and soluble TNF receptor p75 (clone: utr 4), respectively, in a volume of 100 p1 sodium phosphate, pH 6.5, per well for 24 hr at 15" to 25" C. After blocking with BSA (10 mglml), samples and standards were added with TNF-a-peroxidase conjugate and incubated for 24 hr at 15" to 25" C. The TNF receptor was bound immunologically to the immobilized monoclonal antibodies and is labeled biologically with the TNF-a peroxidase. After this immunological and biological reaction, the unbound material was removed by a washing step and the quantity of peroxidase bound to the microtiter plate
Vol. 16, No. 3, 1992
75 1
HIGH CONCENTRATIONS OF SOLUBLE T N F RECEPTORS IN ASCITES
was measured enzymatically. The resulting color intensity was determined photometrically; it was directly proportional to the soluble TNF receptor concentrations in the sample. The TNF receptor (p55 or p75) assay did not cross-react with the other TNF receptor (p55 or p75) or with interleukin (ILI-la; IL-lp; IL-2; TNF-a; interferon (IFN)-a; IFN--y; or platelet-derived growth factors AA, AB or BB. The detection limit was 100 p g / d for soluble TNF receptor p55 and 100 pg/ml for soluble TNF receptor p75. The variation between duplicate samples was 3%to 6%; the interassay variation was 5%to 10%.Normal plasma ranges for both soluble TNF receptors are 1.0 5 0.5 ng/ml (mean ? 2 S.D.). Normal plasma ranges were established from 50 healthy coworkers of Hoffmann-La Roche (25 men and 25 women; age range = 18 to 65 yr). TNF-a was measured with an ELISA from R & D Systems (Minneapolis, MN); it did not cross-react with IL-la, IL-lp, IL-2, IL-3, IL-4, IL-6, TNF-@,transforming growth factor-@, platelet-derived growth factor, granulocyte-macrophage colony-stimulating factor or fibroblast growth factor. The detection limit was 4.8 pg/ml. The interassay variation was less than 15%. Purified human recombinant TNF-cr, TNF receptor p55 and TNF receptor p75 were obtained from F. Hoffmann La Roche (Basel, Switzerland). Ascitic fluid bacterial cultures were analyzed after bedside inoculation into aerobic and anaerobic blood culture bottles as suggested by Runyon et d.(28). LDH, cholesterol, lactate, pH, bilirubin, prothrombin time, creatinine, total protein and fibronectin were measured with established methods. Sfubistical Analysis. Data are expressed as mean t S.D. The nonparametric Mann-Whitney U test and the Wilcoxon rank-sum test were used to compare data between groups. Correlations were calculated with the nonparametric Spearman rank-regression analysis. p Values (two-tailed, corrected for ties) of less than 0.05 were considered significant. Sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy were calculated with standard procedures.
RESULTS Concentrations of Soluble TNF Receptors in Various Forms of Ascites. High concentrations of soluble TNF
receptors p55 and p75 were detected in all ascites samples (from 1.5 ng/ml to 60 ng/ml) and plasma samples (from 2.5 ng/ml t o 35 ng/ml). In all patients molar soluble TNF receptor concentrations were about 500 times higher than molar TNF-a concentrations (TNF-a in hepatic ascites: 5.2 ? 8.1 pg/ml; TNF-a in infected ascites: 23.8 -+ 12.8 pg/ml; TNF-(rin malignant ascites: 27.0 42.3 pg/ml). Furthermore, the mean concentrations of soluble TNF receptors p55 and p75 were elevated about 2.5 and 3.5 times, respectively, in infected hepatic ascites (p55: 25.1 -+ 10.9 ng/ml, p = 0.0041; p75: 22.6 k 11.0 ng/ml, p = 0.001) and in malignant ascites (p55: 26.0 8.6 ng/ml, p < 0.0001; p75: 20.5 t 17.4 ng/ml, p = 0.0003) compared with uncomplicated hepatic ascites (p55: 10.1 +- 4.4 ng/ml; p75: 6.0 t 2.6 ng/ml) (Fig. 1 A and B). Essentially, no difference was found for the ascites soluble TNF receptor concentrations between patients with infected and malignant ascites. The plasma concentrations of soluble TNF receptors p55 and p75 also tended to be
*
*
higher in patients with infected hepatic (p55: 12.7 -+ 10.2 ng/ml, p > 0.05; p75: 16.4 5 10.7 ng/ml, p > 0.05) or malignant ascites (p55: 13.8 ? 11.4 ng/ml, p > 0.05; p75: 18.2 k 11.8 ng/ml, p > 0.051, but showed a large overlap with the plasma concentration of patients with uncomplicated hepatic ascites (Fig. 1C and D). Little overlap was seen between the ascites soluble TNF receptor concentrations in different subgroups of patients. Therefore ascitic fluid concentrations of soluble TNF receptors p55 and p75 differentiated well between malignant or infected ascites and uncomplicated hepatic ascites. Cutoff levels of 16.5 ng/ml and 9.5 ng/ml for soluble TNF receptor p55 and soluble TNF receptor p75 resulted in diagnostic accuracies of 94% and 89% for the separation between malignant or infected and uncomplicated hepatic ascites, respectively (Table 2). Interestingly, in uncomplicated hepatic ascites the concentrations of soluble TNF receptor p75 were significantly higher in plasma than in ascites (ratio of ascites/plasma = 0.75 & 0.64), whereas soluble TNF receptor p55 showed significantly higher concentrations in ascites (ratio of ascites/plasma = 2.10 t 1.13). In patients with malignant or infected hepatic ascites, ratios between ascites and plasma were higher for both soluble TNF receptors p55 (infected: 3.45 % 3.42, malignant: 5.12 t 7.52) and p75 (infected: 1.79 1.35, malignant: 1.45 k 1.47) because in these groups levels of soluble TNF receptors were more elevated in ascites than in plasma. In plasma the ratios between soluble TNF receptors p75 and p55 were 1.84 k 0.68, 1.76 ? 0.59 and 1.61 -+ 0.59 in patients with plain hepatic ascites, SBP and malignancy-related ascites, respectively. In ascites these ratios were 0.89 0.08, 0.69 t 0.60 and 0.78 t 0.57.
*
*
Correlations Between Soluble TNF Receptors pbb and p76, TNF-aand Ascites Protein. Values of soluble TNF receptors p55 and p75 correlated well in ascites (r = 0.83, p < 0.001) and in plasma (r = 0.85, p < 0.001). No significant correlation was found between the concentrations in ascites and plasma for soluble TNF receptors p55 (r = 0.24, p = 0.09) and p75 ( r = 0.25, p = 0.08). TNF-a in ascites correlated only weakly with soluble TNF receptors p55 (r = 0.32, p = 0.03) and p75 (r = 0.29, p = 0.047) in ascites. No correlation was found between soluble TNF receptor p55 (r = 0.21, p = 0.12) or p75 (r = 0.26, p = 0.074) and TNF-a in plasma. Both soluble TNF receptors correlated weakly with protein levels in ascites (soluble TNF receptor p55: r = 0.55, p = 0.001; soluble TNF receptor p75: r = 0.51, p = 0.002) and with IL-6 levels in ascites soluble TNF receptor p55: r = 0.53, p = 0.001; soluble TNF receptor p75: r = 0.63, p < 0.001). Neither soluble TNF receptor concentration correlated with the concentrations of IL-1 and IL-8 in ascites. The measurement of TNF-a was not impaired by binding to soluble TNF receptors; the addition of different amounts of human recombinant TNF receptor p55 or human recombinant TNF receptor p55 or human recombinant TNF receptor p75 (0, 1, 10 and
752
ANDUS ET AL.
A Ascites sTNF-R 50
B
p66 [ng/ml]
0
0
401-
I
0
30 -
lol 1 "I.
20 -
0
0
00
0
0
0
HEPATOLOGY
Ascites sTNF-R p75 Ing/mll
60
000
,~oooo
Hepatic
Infected
Malignant
Hepatic
Infected
Malignant
C Plasma sTNF-R p65 Ing/mll
40
Plasma sTNF-R p75 [ng/ml]
X
30 -
X
xx
30 X X
20
-
20
X
X X X X
X X X
10 X X
0'
X X
8 0 Hepatic
Infected
Malignant
Hepatic
Infected
Malignant
F'IG. 1. Scattergram distribution of (A, B) ascites and (C, D) plasma concentrations of soluble TNF receptors p55 (A, C ) and p75 (B, D) in 21 patients with uncomplicated hepatic ascites, 5 patients with infected hepatic ascites and 9 patients with malignant ascites. Means S.D. are indicated by the vertical lines.
*
TABLE 2. Diagnostic value of soluble TNF receptors p55 and p75 and TNF-a in ascitic fluid in differentiating infected and malignant ascites from uncomplicatedhepatic ascites
Discrimination value
Soluble TNF receptor p66 (16.6 nglml)
Soluble TNF receptor p76 (9.6 ng/ml)
Sensitivity Specificity Positive predictive value Negative predictive value Diagnostic accuracy
86% 95% 93% 93% 94%
79% 86% 92% 87% 89%
50 ng/ml) to human recombinant TNF-a (0 to 2,000 pg/ml) allowed exact measurement of TNF-a. ELISA of TNF-a was not affected by addition of TNF receptor p55 or TNF receptor p75 in the concentrations mentioned above. Also, neither TNF receptor enzymelinked immunological-biological binding assay was significantly affected by the addition of human recombinant TNF-a in concentrations ranging from 10 pg/ml to 1 n g / d (data not shown). Decrease of Soluble TNF Receptor Concentrations in Ascites After Antibiotic !l'reatment. In three patients with hepatic ascites we measured soluble TNF receptors
in ascites and in plasma during ascites infection and after antibiotic treatment. As shown in Figure 2, the elevated concentrations of soluble TNF receptors p55 (24.2 k 15.2 ng/ml) and p75 (20.2 k 14.4 n g / d ) found in infected ascitic fluid decreased (p55: 10.7 k 1.89 ng/ml, p75: 7.50 k 1.80 ng/ml) after successful antibiotic treatment. Plasma soluble TNF receptor levels did not show similar changes. DISCUSSION
We found high levels of soluble TNF receptors in all patients with ascites. Soluble TNF receptors derived from the membrane-bound proteins have molecular masses of about 30,000 Da (10-15). TNF-a and TNF-P form tightly packed trimers with molecular masses of 45,000 and 65,000 Da, respectively (3). Recently it was shown that three molecules of soluble TNF receptor p55 bind to one trimer of TNF-a or TNF-P (29). From these data it can be calculated that soluble TNF receptors are present in a 500-fold molar excess over TNF-a in ascites and plasma. These large amounts of soluble TNF receptors may affect the bioavailability of TNF in ascites and plasma. The effect of soluble TNF receptors on the bioavailability of TNF is complex. Both soluble TNF receptors inhibit binding of TNF to the cellular re-
753
HIGH CONCENTRATIONS OF SOLUBLE TNF RECEPTORS IN ASCITES
Vol. 16, No. 3, 1992
B
A
Ascites
Plasma
Plasma
Ascites
40 -
301
I
't
'
'\\,
+
-
+
-
Infection FIG.2. Comparison of the concentrations of soluble TNF receptors (A) p55 and (B)p75 in ascites and plasma in three patients before and after succeesful antibiotic treatment of bacterial peritonitis.
ceptors and reduce the biological effects of TNF in a dose-dependent manner (10-12, 29). However, it was recently shown that low concentrations of soluble TNF receptors (resulting in soluble TNF receptor/TNF ratios of up to 20) stabilize the trimeric structure and activity of TNF in uitro (30). Therefore it was proposed that soluble TNF receptors may act as carriers for TNF or may even augment the effect of TNF by prolonging its function in closed compartments where clearance of the TNFholuble TNF receptor complexes is slow (30). The finding that intraarticular injection of 1 kg of soluble TNF receptor acts in an antiinflammatory manner in a rat antigen-induced arthritis model (31) indicates that high concentrations of soluble TNF receptor inhibit the function of TNF in uiuo even in compartments such as the synovial space. But without data on the clearance of TNFholuble TNF receptor complexes from the peritoneal cavity, the exact effect of soluble TNF receptors on the bioavailability of TNF in ascites cannot be assessed. Soluble TNF receptor p75 concentrations were higher than those of soluble TNF receptor p55 concentrations in most of the ascites samples, whereas in most of the plasma samples soluble TNF receptor p55 was predominant. This may be caused by local differences in soluble TNF receptor expression and secretion or by different clearance rates and mechanisms in ascites and plasma. Statistically significant correlation is present between soluble TNF receptor p55 (r = 0.55, p = 0.001) and soluble TNF receptor p75 (r = 0.51, p = 0.002) with total protein in ascites. However, because the mean ascites-to-plasma ratios of soluble TNF receptors in infected (3.45 and 1.79) or malignant ascites (5.12 and 1.45) were much higher than the respective mean ratios
for total protein (0.32 and 0.611, it can be concluded that the increased soluble TNF receptor concentrations do not reflect only an increased peritoneal permeability but also an increased local synthesis. Because nearly all mammalian cells express TNF receptors (3), the origin of soluble TNF receptors in ascites remains to be identified. Also, the mechanisms leading to the increased concentrations of soluble TNF receptors in malignant or infected ascites are still unclear. Some knowledge exists about the regulation of membrane bound TNF receptors. Receptor expression can be stimulated by IFN-y (321, IL-1p (331, IL-2 (341, IL-4 (33) and IL-6 (33)and in cancer (15). Cellular TNF receptors are down-regulated by TNF (351, IL-1 (361, IFN-y (37), LPS and phorbolester (PMA). The most likely mechanism for the release of soluble TNF receptors is proteolytic processing, which leads to shedding of soluble TNF receptors from the cell membranes (8, 38). Soluble TNF receptors are released by activated neutrophils (391, mononuclear blood cells (381, fibroblasts (38)and cell lines such as HL-60 (38).The release of soluble TNF receptors from human neutrophils is increased by formyl methionyl leucyl phenylalanine (FMLP), C5a, GM-CSF, PMA and A23187 (39); the release of soluble TNF receptors from HL-60 was stimulated by TNF-a, TNF-p and PMA (38). Bacterial products such as FMLP may be the primary inducers of soluble TNF receptor release from neutrophils in infected ascites. TNF may be involved in the increased soluble TNF receptor release in malignant ascites. The fact that we did not find a close correlation between TNF and soluble TNF receptors in ascites does not rule out this possibility because soluble TNF receptors are released from mononuclear cells several hours after
754
ANDUS ET AL.
stimulation (38). We found a statistically significant correlation of soluble TNF receptors with ascites levels of IL-6 in our patients. This might be explained by regulation of IL-6 similar to that of soluble TNF receptors or by the fact that IL-6 may be involved in the formation of soluble TNF receptors in ascites. The modulatory function of soluble TNF receptors may be an important, protective physiological regulatory mechanism that minimizes tissue destruction caused by overproduction of TNF during acute infection. Although TNF usually contributes to the protection of the organism against infectious agents and to healing from injury, high concentrations become pernicious in certain pathological situations such as septic shock (1-31, resulting in more severe destruction than that induced by the pathogen itself. Regulators of the biological activity of TNF-such as the detected soluble TNF receptors, which counteract potentially harmful effects of TNF -are therefore essential in maintaining physiological homeostasis during infection and malignancy. On the other hand, release of soluble TNF receptors may be a defensive mechanism of malignant or infected cells to protect themselves from cytotoxic effects of TNF. Similarly, high concentrations of soluble TNF receptors in malignant ascites might diminish bioavailability of locally administered TNF in the treatment of cancer (14). Because of the low number of bacteria in ascites, diagnosis of infection by culture of ascitic fluid is often difficult or impossible (40). The fact that concentrations of both soluble TNF receptors are significantly higher in infected and malignant ascites and return to lower values after successful treatment of peritonitis could be helpful in such cases, especially if patients have already been treated with antibiotic drugs. Larger prospective studies are required to compare the diagnostic value of soluble TNF receptors with that of neutrophil counts in infected ascites and with that of ascitic proteins such as fibronectin in malignant ascites. In conclusion, our data show that soluble TNF receptors are present in high amounts in ascites and plasma of patients with ascites of different origins. Soluble TNF receptors are present in ascites in a 500-fold excess over TNF-a. The high soluble TNF receptor concentrations in ascites may attenuate the biological activity of TNF. Significantlyhigher amounts of soluble TNF receptors p55 and p75 are found in ascitic fluid of patients with infected or malignant ascites than in ascites of patients with plain hepatic ascites. REFERENCES 1. Beutler B, Cerami A. The biology of cachectidTNF: a primary mediator of the host response. Annu Rev Immunol 1989;7: 625-655. 2. Tracey KJ, Vlassara H, Cerami A. Cachectinhumor necrosis factor. Lancet 1989;1:1122-1126. 3. Vilcek J, Lee TH. Tumor necrosis factor: new insights into the molecular mechanisms of its multiple actions. J Biol Chem 1991;266:7313-7316. 4. Carswell EA, Old U,Kassel RL, Green S, Fiore N, Williamson B. An endotoxin-induced serum factor that causes necrosis of tumors. Proc Natl Acad Sci USA 1975;72:3666-3670.
HEPATOLOGY
5. Loetscher H, Pan YC, Lahm HW, Gentz R, Brockhaus M, Tabuchi H, Lesslauer W. Molecular cloningand expression of the human 55 kd tumor necrosis factor receptor. Cell 1990;61:351-359. 6. Schall TJ, Lewis M, Koller KJ, Lee A, Rice GC, Wong GHW, Gatanaga T, et al. Molecular cloning and expression of a receptor for human tumor necrosis factor. Cell 1990;61:361-370. 7. Smith CA, Davei T, Anderson D, Solam L, Beckmann MP, Jerzy R, Dower SK, et al. A receptor for tumor necrosis factor defined an unusual family of cellular and viral proteins. Science 1990;248: 1019-1023. 8. Nophar Y, Kemper 0, Brakebusch C, Engelmann H, Zwang R, Aderka D, Holtmann H, et al. Soluble forms of tumor necrosis factor receptors (TNF-Rs): The cDNA for the type I TNF-R, cloned using amino acid sequence data of its soluble form, encodes both the cell surface and a soluble form of the receptor. EMBO J 1990;9:3269-3278. 9. Loetscher H, Schlaeger EJ, Lahm HW, Pan Y-CE, Lesslauer W, Brockhaus M. Purification and partial amino acid sequence analysis of two distinct tumor necrosis factor receptors from HL60 cells. J Biol Chem 1990;265:20131-20138. 10. Olsson I, Lantz M, Nilsson E, Peetre C, Thysell H, Grubb A, Adolf G. Isolation and characterization of a tumor necrosis factor binding protein from urine. Eur J Haematol 1989;42:270-275. 11. Seckinger P, Isaaz S, Dayer J-M. Purification and biologic characterization of a specific tumor necrosis factor a inhibitor. J Biol Chem 1989;265:11966-11973. 12. Engelmann H, Novick D, Wallach D. Two tumor necrosis factor-binding proteins purified from human urine. J Biol Chem 1990;265:1531-1536. 13. Schiitze S, Scheurich P, Pfizenmaier K, Kronke M. Tumor necrosis factor signal transduction: tissue-specific serine phosphorylation of a 26-kDa cytosolic protein. J Biol Chem 1989;264: 3562-3567. 14. Gatanaga T, Lentz R, Mansunuka I, Tomich J , Jeffes EWB 111, Baird M, Granger G. Identification of TNF-LT blocking factor(s) in serum and ultrafiltrates of human cancer patients. Lymphkine Res 1990;9:225-229. 15. Gatanaga T, Hwang C, Kohr W, Cappucini F, Lucci JA 111, Jeffes EWB, h n t z R, et al. Purification and characterization of an inhibitor (soluble tumor necrosis factor receptor) for tumor necrosis factor and lymphotoxin obtained from the serum ultrafiltrates of human cancer patients. Proc Natl Acad Sci USA 1990;87:8781-8784. 16. Bird GLA, Sheron N, Goka AKJ, Alexander GJ, Williams RS. Increased plasma tumor necrosis factor in severe alcoholic hepatitis. Ann Intern Med 1990;112:917-920. 17. Khoruts A, StahnkeL, 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. 18. Sheron N, Lau J, Daniels H, Goka J, Eddleston A, Alexander GJM, Williams R. Increased production of tumor necrosis factor alpha in chronic hepatitis B virus infection. J Hepatol 1991;12:241-245. 19. McClain CJ, Cohen DA. Tumor necrosis factor in alcoholic hepatitis. Gastroenterology 1989;97:1056-1057. 20. McClain CJ, Cohen DA. Increased tumor necrosis factor production by monocytes in alcoholic hepatitis. HEPATOLOGY 1989;9: 349-351. 21. Deviere J , Content J , Denys C, Vandenbussche P, Schandene L, Wybran J , Dupont E. Excessive in uitro bacterial lipopolysaccharide-inducedproduction of monokines in cirrhosis. 1990;11:628-634. HEPATOLOGY 22. Yoshioka K, Kakumu S, Arao M, Tsutsumi Y, Inoue M. Tumor necrosis factor-a production by peripheral blood mononuclear cells of patients with chronic liver disease. HEPATOLOGY1989;lO: 769-773. 23. Miizes G, Deak G, Lang I, Gonzalez-Cabello R, Gergely P, Feher J. Depressed monocyte production of interleukin-1 and tumor necrosis factor-alpha in patients with alcoholic liver cirrhosis. Liver 1989;9:302-306. 24 Felver ME, Mezey E, McGuire M, Mitchell MC, Herlong HF, Veech GA, Veech RL. Plasma tumor necrosis factor a predicts decreased long-term survival in severe alcoholic hepatitis. Alcohol Clin Exp Res 1990;77:221-225.
Vol. 16, No. 3, 1992
HIGH CONCENTRATIONS OF SOLUBLE TNF RECEPTORS IN ASCITES
25. Lau JYN, Sheron N, Nouri-Aria KR, Alexander GJM, Williams R. Increased tumor necrosis factor-a receptor number in chronic hepatitis B virus infection. HEPATOLDGY 1991;14:44-50. 26. Gearing AJ, Leung H, Bird CR, Thorpe R. Presence of the inflammatory cytokines IL-1, TNF, and IL-6 in preparations of monoclonal antibodies. Hybridoma 1989;8:361-367. 27. Deviere J , Content J, Crusiaux A, Dupont E. IL-6 and TNFa in ascitic fluid during spontaneous bacterial peritonitis. Dig Dis Sci 1991;36:123-124. 28. Runyon BA, Umland ET, Merlin T. Inoculation of blood culture bottles with ascitic fluid. Improved detection of spontaneous bacterial peritonitis. Arch Intern Med 1987;147:73-75. 29. Loetscher H, Gentz R, Zulauf M, Lustig A, Tabuchi H, Schlaeger EJ, Brockhaus M, et al. Recombinant 55-kDa tumor necrosis factor (TNF) receptor: stoichiometry of binding to TNF alpha and TNF beta and inhibition of TNF activity. J Biol Chem 1991;266: 18324-18329. 30. Aderka D, Engelmann H, Maor Y, Brakebysch C, Wallach D. Stabilization of the bioactivity of tumor necrosis factor by its soluble receptors. J Exp Med 1992;175:323-329. 31. Mohler KM, Torrance DT, Young DM, Callis G, Roux ER, Jacobs CA. Immunotherapeutic potential of soluble cytokine receptors in inflammatory disease [Abstract]. FASEB J 1992;6:A1086. 32. Aggarwal BB, Eessalu TE, Hass PE. Characterization of receptors for human tumor necrosis factor and their regulation by gammainterferon. Nature 1985;318:665-667.
755
33. Dett CA, Gatanaga M, Ininns EK, Cappuccini F, Yamamoto RS, Granger GA, Gatanaga T. Enhancement of lymphokine-activated T killer cell tumor necrosis factor receptor mRNA transcription, tumor necrosis factor receptor membrane expression, and tumor necrosis factorflymphotoxin release by IL-lB, IL-4 and IL-6 in vitro. J Immunol 1991;146:1522-1526. 34. Owen-Schaub LB, Crump WL 111, Morin GI, Grimm EA. Regulation of lymphocyte tumor necrosis factor receptors by IL-2. J Immunol 1989;135:2236-2241. 35. Tsujimoto M, Vilcek J. Tumor necrosis factor-induced downregulation of its receptors in HeLa cells. J Biochem 1987;102:15711577. 36. Holtmann H, Wallach D. Downregulation of the receptors of tumor necrosis factor by interleukin 1 and 4p-phorbol-12myriState-13-acetate.J Immunol 1987;139:1161-1167. 37. Drapier J-C, Wietzerbin J. IFN-gamma reduces specific binding of tumor necrosis factor on murine macrophages. J Immunol 1991;146:1198-1203. 38. Lantz M, Gullberg U, Nilsson E, Olsson I. Characterization in vitro of a human tumor necrosis factor-binding protein: a soluble form of a tumor necrosis factor receptor. J Clin Invest 1990;86: 1396-1402. 39. Porteu F, Nathan C. Shedding of tumor necrosis factor receptors by activated human neutrophils. J Exp Med 1990;172:599-607. 40. Hoefs JK. Spontaneous bacterial peritonitis: prevention and therapy. HEPATOLOGY 1990;12:776-781.
MARTINA
DAVID,'
lDepartment of Internal Medicine, University of Freiburg, W-7800 Freiburg, Germany; and 'F. Hoffmann La Roche AG, Pharmaceutical Research-New Technologies, Basel, Switzerland
Ascites and plasma concentrations of soluble tumor receptor p75 levels were usually higher in plasma than necrosis factor receptors p55 and p75 were measured in ascites in patients with uncomplicated hepatic in a prospective study in 34 patients (35occasions of ascites. The concentrations of both tumor necrosis ascites)with hepatic (5 infected and 21 uninfected) and factor receptors correlated well in ascites (r = 0.83, malignancy-related (9) ascites. All patients had high p < 0.001) and plasma (r = 0.85, p < 0.001) but only concentrations of both soluble tumor necrosis factor weakly with tumor necrosis factor-a levels in ascites receptors in ascites and plasma; these were about 500 (p55: r = 0.32, p = 0.03; p75: r = 0.29, p = 0.047) and times higher than the corresponding tumor necrosis not with tumor necrosis factor-a in plasma. The high soluble tumor necrosis factor receptor factor-aconcentrations. Ascites levels of soluble tumor necrosis factor receptors p55 and soluble tumor ne- concentrationsfound in ascitesmay influence the local crosis factor receptors p75 were significantly elevated bioavailability of tumor necrosis factor and might have in patients with malignancy-related (p55: 26.0 k 8.6 an impact on the treatment of peritoneal carcinoman g / e p76 20.5 k 17.4 ng/ml; mean S.D.) and in- tosis with recombinant tumor necrosis factor. Furfected ascites (p55: 25.1 k 10.9 ng/ml, p75 22.6 k 11.0 thermore, the determinationof soluble tumor necrosis ng/ml) compared with patients with uncomplicated factor receptors could be of value for the differential hepatic ascites (p55: 10.1 4.4 ng/ml; p75: 6.0 2.6 diagnosis of ascites. (HEPATOLOGY 1992;16749-755.) ng/ml). Patients with infected or malignancy-related ascites also showed higher soluble tumor necrosis Tumor necrosis factor (TNF1-a (cachedin) and factor receptor concentrations in plasma than did TNF-P (lymphotoxin) are structurally and functionally patients with plain hepatic ascites. Succeesful antibiotic treatment of peritonitis reduced soluble tumor related proteins produced mainly by activated macroneed factor receptor p55 and p75 ascites levels in phages and lymphocytes, respectively (1-3).Both cythree patients from 24.2 15.2 ng/d to 10.7 1.9 tokines bind to the same surface receptors and mediate ng/ml and from 20.2 k 14.4 ng/ml to 7.5 k 1.8 ndml, a wide variety of biological effects, including cytotoxicity respectively. Soluble tumor necrosis factor receptors against tumors and virus-infected cells, inflammation, p55 and p75 at cutoff levels of 16.5 ng/ml and 9.5 ng/ml, wasting, shock and stimulation of various immune respectively, differentiated between infected or ma- effector cells (1-4). Two distinct T N F receptors that lignant and plain hepatic ascites with diagnostic accu- belong to the nerve growth factorlepidermal growth racies of 94% and 89%, respectively. They did not factorflow-density lipoprotein family, with molecular differentiate between infected and malignant ascites. ) 75,000 Da (~751,have The concentrations of soluble tumor necrosis factor masses of 55,000 Da ( ~ 5 5and (5-9). Soluble inhibitory forms been purified and cloned receptor pS5 were usually higher in ascites than in plasma in all subgroups of patients. Levels of soluble of TNF receptors have been found in human serum and tumor necrosis factor receptor p75 were also higher in urine (10-15). In patients with viral and alcoholic hepatitis or ascites than in plasma in patients with malignant and infected ascites. but soluble tumor necrosis factor cirrhosis, elevated plasma levels of TNF-a (16-18) and
*
*
*
Received December 9, 1991; accepted April 22, 1992. This work was supported by the Deutsche Forschungagerneinschaft (SFB 154 and a "Heisenberg-Stipendium" to VG). *Current address: Medizinische Klinik und Powdinik I, Universitat Regemburg, W-8400Regensburg, Germany. $Current address: F. Hoffiann La Roche AG, Pharmaceutical Research-New Technologies, Basel, Switzerland. Address reprint requests to: Dr. Tilo Andus, Medizinische Klinik und Poliklinik I, Universitiit Regemburg, F r a u Josef Strauss Nee 11,Postfach 10 06 62, W-8400€&gemburg,Germany.
31/1/99252
altered production of TNF-a in lipopolysaccharidestimulated peripheral blood monocytes have been found (19-23). TNF-a production correlated with histological activity index (22), and TNF-a plasma levels correlated with the long-term survival rates of patients with severe alcoholic hepatitis (24). Recently, increased numbers of TNF receptors have been found on peripheral blood monocytes of patients with chronic HBV infection (25). In ascitic fluid of mice injected with different hybridomas (26) and of four patients with spontaneous bacterial peritonitis (SBP) (27), elevated TNF-a levels have been detected.
749
750
ANDUS ET AL.
HEPATOLOGY
TABLE1. Patient characteristics Ascites Characteristics
No. of patients Age
0.I)"
Child classification (A/B/C) Positive bacterial blood cultures Antibiotic treatment Temperature (Celsius);" Ascitic leukocytes (per ml)" Ascitic total protein (gm/dl)" Ascitic LDH [U/L]" Ascitic cholesterol (mg/dl)" Ascitic fibronectin (pg/ml)" Ascitic lactate (mmoUL)" Ascitic pH" Serum bilirubin (mg/dl)" Prothrombin time (quick; %)" Serum creatinine (mg/dl)" Deaths during hospital stay
Hepatic
Infected
21 56.0 f 12.0 41918 2 5 37.0 f 0.8 195 t 128 1.84 -t 1.33 51 f 36 31.5 -t 28.9 97.4 55.4 2.1 f 0.9 7.46 -t 0.06 8.9 f 9.6 50 f 18 1.62 1.14 4
5 56.6 f 6.3 0/1/4 1 3 37.0 f 0.4 4,980 2 1,24@ 2.55 f 0.26 188 f 112 20.5 t 2.9 105 -t 32.1 6.6 2 1.8' 7.21 f 0.07b 12.2 f 6.9 35 f 4 2.04 f 0.65 3
*
Malignant 9 64
f
9.4
0 0 36.7 f 0.5 2,410 f 3,64W 4.11 f 1.42d 225 f 135d 122 2 75.5d 200 f 68.1d 2.8 f 0.8 7.40 f 0.06f 0.7 f 0.7d 77 f 2od 0.80 2 1.40 1
"Data expressed as mean 2 S.D. bHepaticascites/infeded ascites: p < 0.01. 'Hepatic ascites/malignant ascites: p < 0.01. dHepatic ascites/malignant ascites: p < 0.005. "Hepaticascites/infeded-ascites: p < 0.05. fHepatic ascites/malignant ascites: p < 0.05.
Nothing is known about soluble TNF receptors in ascites. In this prospective study, we determined the concentrations of soluble TNF receptors p55 and p75, both in ascitic fluid and plasma of 34 patients with ascites of hepatic or malignancy-related origin. PATZENTS AND METHODS Patients. We prospectively studied 34 patients (18men and 16 women) with ascites from June to December 1990. Paracentesis was performed for diagnostic or therapeutic reasons. The patients were divided into the following three groups. Group 1consisted of 21 patients (11men and 10 women) with hepatic ascites caused by chronic parenchymal liver disease. Eighteen had alcoholic cirrhosis and three had cirrhosis due to chronic viral hepatitis B. One of these patients also had an HCC. Peritoneal carcinomatosiswas excluded in all cases by ascitic fluid cytological study and sonography and, in some cases, additionally by computed tomography or laparoscopy. Group 2 comprised five patients (three men and two women) with hepatic ascites and SBP or secondary bacterial peritonitis (two had Enterococcus fizecalis, one had Escherichia coli, one had Staphylococcus aureus and E. fizecalis and one had Candida albicans). Group 3 comprised nine patients (four men and five women) with malignancy-related ascites. Diagnosis of malignancyrelated ascites was confirmed by positive cytological appearance of ascites, computed tomography, laparoscopy and/or biopsy. Ovarian carcinoma was found in five patients, nonHodgkin's lymphoma was found in two patients, gastric carcinoma was found in one patient and malignant melanoma with liver metastasis was found in one patient. One woman was hospitalized twice, first with uncomplicated hepatic ascites and several weeks later with perfor-
ated cholecystitis. Therefore she was included in groups 1 and 2. Laboratory data are given in Table 1. Patients with peritonitis had higher levels of ascites leukocytes than did those with plain hepatic ascites. They also had significantly higher concentrations of lactate and lactate dehydrogenase (LDH) and lower ascitic fluid pH values. Three of the five patients (60%) with peritonitis died, whereas only 4 of 21 patients (19%) with plain hepatic ascites died during the observation period. The subgroup of patients with malignant ascites had significantly higher levels of leukocytes, total protein, LDH, cholesterol and fibronectin in the ascitic fluid. Kidney function did not differ significantly between the three groups (Table 1). Methods. Ascites and blood were obtained aseptically,cooled on ice immediatelyand centrifuged at 4"C. Aliquots of cell-free . samples were stored at - 20" C for up to 3 wk before analysis. No significant degradation of TNF receptors occurred during this time. In two patients with malignancy-related ascites, cytokine concentrations were not determined in plasma because there was not enough plasma. Concentrations of TNF receptors p55 and p75 in ascites and plasma were measured in duplicate with enzyme-linked immunological-biological binding assays. Microtiter plates were sensibilized with 1 pg of noninhibitory monoclonal (mouse) antibodies against soluble TNF receptor p55 (clone: htr 20) and soluble TNF receptor p75 (clone: utr 4), respectively, in a volume of 100 p1 sodium phosphate, pH 6.5, per well for 24 hr at 15" to 25" C. After blocking with BSA (10 mglml), samples and standards were added with TNF-a-peroxidase conjugate and incubated for 24 hr at 15" to 25" C. The TNF receptor was bound immunologically to the immobilized monoclonal antibodies and is labeled biologically with the TNF-a peroxidase. After this immunological and biological reaction, the unbound material was removed by a washing step and the quantity of peroxidase bound to the microtiter plate
Vol. 16, No. 3, 1992
75 1
HIGH CONCENTRATIONS OF SOLUBLE T N F RECEPTORS IN ASCITES
was measured enzymatically. The resulting color intensity was determined photometrically; it was directly proportional to the soluble TNF receptor concentrations in the sample. The TNF receptor (p55 or p75) assay did not cross-react with the other TNF receptor (p55 or p75) or with interleukin (ILI-la; IL-lp; IL-2; TNF-a; interferon (IFN)-a; IFN--y; or platelet-derived growth factors AA, AB or BB. The detection limit was 100 p g / d for soluble TNF receptor p55 and 100 pg/ml for soluble TNF receptor p75. The variation between duplicate samples was 3%to 6%; the interassay variation was 5%to 10%.Normal plasma ranges for both soluble TNF receptors are 1.0 5 0.5 ng/ml (mean ? 2 S.D.). Normal plasma ranges were established from 50 healthy coworkers of Hoffmann-La Roche (25 men and 25 women; age range = 18 to 65 yr). TNF-a was measured with an ELISA from R & D Systems (Minneapolis, MN); it did not cross-react with IL-la, IL-lp, IL-2, IL-3, IL-4, IL-6, TNF-@,transforming growth factor-@, platelet-derived growth factor, granulocyte-macrophage colony-stimulating factor or fibroblast growth factor. The detection limit was 4.8 pg/ml. The interassay variation was less than 15%. Purified human recombinant TNF-cr, TNF receptor p55 and TNF receptor p75 were obtained from F. Hoffmann La Roche (Basel, Switzerland). Ascitic fluid bacterial cultures were analyzed after bedside inoculation into aerobic and anaerobic blood culture bottles as suggested by Runyon et d.(28). LDH, cholesterol, lactate, pH, bilirubin, prothrombin time, creatinine, total protein and fibronectin were measured with established methods. Sfubistical Analysis. Data are expressed as mean t S.D. The nonparametric Mann-Whitney U test and the Wilcoxon rank-sum test were used to compare data between groups. Correlations were calculated with the nonparametric Spearman rank-regression analysis. p Values (two-tailed, corrected for ties) of less than 0.05 were considered significant. Sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy were calculated with standard procedures.
RESULTS Concentrations of Soluble TNF Receptors in Various Forms of Ascites. High concentrations of soluble TNF
receptors p55 and p75 were detected in all ascites samples (from 1.5 ng/ml to 60 ng/ml) and plasma samples (from 2.5 ng/ml t o 35 ng/ml). In all patients molar soluble TNF receptor concentrations were about 500 times higher than molar TNF-a concentrations (TNF-a in hepatic ascites: 5.2 ? 8.1 pg/ml; TNF-a in infected ascites: 23.8 -+ 12.8 pg/ml; TNF-(rin malignant ascites: 27.0 42.3 pg/ml). Furthermore, the mean concentrations of soluble TNF receptors p55 and p75 were elevated about 2.5 and 3.5 times, respectively, in infected hepatic ascites (p55: 25.1 -+ 10.9 ng/ml, p = 0.0041; p75: 22.6 k 11.0 ng/ml, p = 0.001) and in malignant ascites (p55: 26.0 8.6 ng/ml, p < 0.0001; p75: 20.5 t 17.4 ng/ml, p = 0.0003) compared with uncomplicated hepatic ascites (p55: 10.1 +- 4.4 ng/ml; p75: 6.0 t 2.6 ng/ml) (Fig. 1 A and B). Essentially, no difference was found for the ascites soluble TNF receptor concentrations between patients with infected and malignant ascites. The plasma concentrations of soluble TNF receptors p55 and p75 also tended to be
*
*
higher in patients with infected hepatic (p55: 12.7 -+ 10.2 ng/ml, p > 0.05; p75: 16.4 5 10.7 ng/ml, p > 0.05) or malignant ascites (p55: 13.8 ? 11.4 ng/ml, p > 0.05; p75: 18.2 k 11.8 ng/ml, p > 0.051, but showed a large overlap with the plasma concentration of patients with uncomplicated hepatic ascites (Fig. 1C and D). Little overlap was seen between the ascites soluble TNF receptor concentrations in different subgroups of patients. Therefore ascitic fluid concentrations of soluble TNF receptors p55 and p75 differentiated well between malignant or infected ascites and uncomplicated hepatic ascites. Cutoff levels of 16.5 ng/ml and 9.5 ng/ml for soluble TNF receptor p55 and soluble TNF receptor p75 resulted in diagnostic accuracies of 94% and 89% for the separation between malignant or infected and uncomplicated hepatic ascites, respectively (Table 2). Interestingly, in uncomplicated hepatic ascites the concentrations of soluble TNF receptor p75 were significantly higher in plasma than in ascites (ratio of ascites/plasma = 0.75 & 0.64), whereas soluble TNF receptor p55 showed significantly higher concentrations in ascites (ratio of ascites/plasma = 2.10 t 1.13). In patients with malignant or infected hepatic ascites, ratios between ascites and plasma were higher for both soluble TNF receptors p55 (infected: 3.45 % 3.42, malignant: 5.12 t 7.52) and p75 (infected: 1.79 1.35, malignant: 1.45 k 1.47) because in these groups levels of soluble TNF receptors were more elevated in ascites than in plasma. In plasma the ratios between soluble TNF receptors p75 and p55 were 1.84 k 0.68, 1.76 ? 0.59 and 1.61 -+ 0.59 in patients with plain hepatic ascites, SBP and malignancy-related ascites, respectively. In ascites these ratios were 0.89 0.08, 0.69 t 0.60 and 0.78 t 0.57.
*
*
Correlations Between Soluble TNF Receptors pbb and p76, TNF-aand Ascites Protein. Values of soluble TNF receptors p55 and p75 correlated well in ascites (r = 0.83, p < 0.001) and in plasma (r = 0.85, p < 0.001). No significant correlation was found between the concentrations in ascites and plasma for soluble TNF receptors p55 (r = 0.24, p = 0.09) and p75 ( r = 0.25, p = 0.08). TNF-a in ascites correlated only weakly with soluble TNF receptors p55 (r = 0.32, p = 0.03) and p75 (r = 0.29, p = 0.047) in ascites. No correlation was found between soluble TNF receptor p55 (r = 0.21, p = 0.12) or p75 (r = 0.26, p = 0.074) and TNF-a in plasma. Both soluble TNF receptors correlated weakly with protein levels in ascites (soluble TNF receptor p55: r = 0.55, p = 0.001; soluble TNF receptor p75: r = 0.51, p = 0.002) and with IL-6 levels in ascites soluble TNF receptor p55: r = 0.53, p = 0.001; soluble TNF receptor p75: r = 0.63, p < 0.001). Neither soluble TNF receptor concentration correlated with the concentrations of IL-1 and IL-8 in ascites. The measurement of TNF-a was not impaired by binding to soluble TNF receptors; the addition of different amounts of human recombinant TNF receptor p55 or human recombinant TNF receptor p55 or human recombinant TNF receptor p75 (0, 1, 10 and
752
ANDUS ET AL.
A Ascites sTNF-R 50
B
p66 [ng/ml]
0
0
401-
I
0
30 -
lol 1 "I.
20 -
0
0
00
0
0
0
HEPATOLOGY
Ascites sTNF-R p75 Ing/mll
60
000
,~oooo
Hepatic
Infected
Malignant
Hepatic
Infected
Malignant
C Plasma sTNF-R p65 Ing/mll
40
Plasma sTNF-R p75 [ng/ml]
X
30 -
X
xx
30 X X
20
-
20
X
X X X X
X X X
10 X X
0'
X X
8 0 Hepatic
Infected
Malignant
Hepatic
Infected
Malignant
F'IG. 1. Scattergram distribution of (A, B) ascites and (C, D) plasma concentrations of soluble TNF receptors p55 (A, C ) and p75 (B, D) in 21 patients with uncomplicated hepatic ascites, 5 patients with infected hepatic ascites and 9 patients with malignant ascites. Means S.D. are indicated by the vertical lines.
*
TABLE 2. Diagnostic value of soluble TNF receptors p55 and p75 and TNF-a in ascitic fluid in differentiating infected and malignant ascites from uncomplicatedhepatic ascites
Discrimination value
Soluble TNF receptor p66 (16.6 nglml)
Soluble TNF receptor p76 (9.6 ng/ml)
Sensitivity Specificity Positive predictive value Negative predictive value Diagnostic accuracy
86% 95% 93% 93% 94%
79% 86% 92% 87% 89%
50 ng/ml) to human recombinant TNF-a (0 to 2,000 pg/ml) allowed exact measurement of TNF-a. ELISA of TNF-a was not affected by addition of TNF receptor p55 or TNF receptor p75 in the concentrations mentioned above. Also, neither TNF receptor enzymelinked immunological-biological binding assay was significantly affected by the addition of human recombinant TNF-a in concentrations ranging from 10 pg/ml to 1 n g / d (data not shown). Decrease of Soluble TNF Receptor Concentrations in Ascites After Antibiotic !l'reatment. In three patients with hepatic ascites we measured soluble TNF receptors
in ascites and in plasma during ascites infection and after antibiotic treatment. As shown in Figure 2, the elevated concentrations of soluble TNF receptors p55 (24.2 k 15.2 ng/ml) and p75 (20.2 k 14.4 n g / d ) found in infected ascitic fluid decreased (p55: 10.7 k 1.89 ng/ml, p75: 7.50 k 1.80 ng/ml) after successful antibiotic treatment. Plasma soluble TNF receptor levels did not show similar changes. DISCUSSION
We found high levels of soluble TNF receptors in all patients with ascites. Soluble TNF receptors derived from the membrane-bound proteins have molecular masses of about 30,000 Da (10-15). TNF-a and TNF-P form tightly packed trimers with molecular masses of 45,000 and 65,000 Da, respectively (3). Recently it was shown that three molecules of soluble TNF receptor p55 bind to one trimer of TNF-a or TNF-P (29). From these data it can be calculated that soluble TNF receptors are present in a 500-fold molar excess over TNF-a in ascites and plasma. These large amounts of soluble TNF receptors may affect the bioavailability of TNF in ascites and plasma. The effect of soluble TNF receptors on the bioavailability of TNF is complex. Both soluble TNF receptors inhibit binding of TNF to the cellular re-
753
HIGH CONCENTRATIONS OF SOLUBLE TNF RECEPTORS IN ASCITES
Vol. 16, No. 3, 1992
B
A
Ascites
Plasma
Plasma
Ascites
40 -
301
I
't
'
'\\,
+
-
+
-
Infection FIG.2. Comparison of the concentrations of soluble TNF receptors (A) p55 and (B)p75 in ascites and plasma in three patients before and after succeesful antibiotic treatment of bacterial peritonitis.
ceptors and reduce the biological effects of TNF in a dose-dependent manner (10-12, 29). However, it was recently shown that low concentrations of soluble TNF receptors (resulting in soluble TNF receptor/TNF ratios of up to 20) stabilize the trimeric structure and activity of TNF in uitro (30). Therefore it was proposed that soluble TNF receptors may act as carriers for TNF or may even augment the effect of TNF by prolonging its function in closed compartments where clearance of the TNFholuble TNF receptor complexes is slow (30). The finding that intraarticular injection of 1 kg of soluble TNF receptor acts in an antiinflammatory manner in a rat antigen-induced arthritis model (31) indicates that high concentrations of soluble TNF receptor inhibit the function of TNF in uiuo even in compartments such as the synovial space. But without data on the clearance of TNFholuble TNF receptor complexes from the peritoneal cavity, the exact effect of soluble TNF receptors on the bioavailability of TNF in ascites cannot be assessed. Soluble TNF receptor p75 concentrations were higher than those of soluble TNF receptor p55 concentrations in most of the ascites samples, whereas in most of the plasma samples soluble TNF receptor p55 was predominant. This may be caused by local differences in soluble TNF receptor expression and secretion or by different clearance rates and mechanisms in ascites and plasma. Statistically significant correlation is present between soluble TNF receptor p55 (r = 0.55, p = 0.001) and soluble TNF receptor p75 (r = 0.51, p = 0.002) with total protein in ascites. However, because the mean ascites-to-plasma ratios of soluble TNF receptors in infected (3.45 and 1.79) or malignant ascites (5.12 and 1.45) were much higher than the respective mean ratios
for total protein (0.32 and 0.611, it can be concluded that the increased soluble TNF receptor concentrations do not reflect only an increased peritoneal permeability but also an increased local synthesis. Because nearly all mammalian cells express TNF receptors (3), the origin of soluble TNF receptors in ascites remains to be identified. Also, the mechanisms leading to the increased concentrations of soluble TNF receptors in malignant or infected ascites are still unclear. Some knowledge exists about the regulation of membrane bound TNF receptors. Receptor expression can be stimulated by IFN-y (321, IL-1p (331, IL-2 (341, IL-4 (33) and IL-6 (33)and in cancer (15). Cellular TNF receptors are down-regulated by TNF (351, IL-1 (361, IFN-y (37), LPS and phorbolester (PMA). The most likely mechanism for the release of soluble TNF receptors is proteolytic processing, which leads to shedding of soluble TNF receptors from the cell membranes (8, 38). Soluble TNF receptors are released by activated neutrophils (391, mononuclear blood cells (381, fibroblasts (38)and cell lines such as HL-60 (38).The release of soluble TNF receptors from human neutrophils is increased by formyl methionyl leucyl phenylalanine (FMLP), C5a, GM-CSF, PMA and A23187 (39); the release of soluble TNF receptors from HL-60 was stimulated by TNF-a, TNF-p and PMA (38). Bacterial products such as FMLP may be the primary inducers of soluble TNF receptor release from neutrophils in infected ascites. TNF may be involved in the increased soluble TNF receptor release in malignant ascites. The fact that we did not find a close correlation between TNF and soluble TNF receptors in ascites does not rule out this possibility because soluble TNF receptors are released from mononuclear cells several hours after
754
ANDUS ET AL.
stimulation (38). We found a statistically significant correlation of soluble TNF receptors with ascites levels of IL-6 in our patients. This might be explained by regulation of IL-6 similar to that of soluble TNF receptors or by the fact that IL-6 may be involved in the formation of soluble TNF receptors in ascites. The modulatory function of soluble TNF receptors may be an important, protective physiological regulatory mechanism that minimizes tissue destruction caused by overproduction of TNF during acute infection. Although TNF usually contributes to the protection of the organism against infectious agents and to healing from injury, high concentrations become pernicious in certain pathological situations such as septic shock (1-31, resulting in more severe destruction than that induced by the pathogen itself. Regulators of the biological activity of TNF-such as the detected soluble TNF receptors, which counteract potentially harmful effects of TNF -are therefore essential in maintaining physiological homeostasis during infection and malignancy. On the other hand, release of soluble TNF receptors may be a defensive mechanism of malignant or infected cells to protect themselves from cytotoxic effects of TNF. Similarly, high concentrations of soluble TNF receptors in malignant ascites might diminish bioavailability of locally administered TNF in the treatment of cancer (14). Because of the low number of bacteria in ascites, diagnosis of infection by culture of ascitic fluid is often difficult or impossible (40). The fact that concentrations of both soluble TNF receptors are significantly higher in infected and malignant ascites and return to lower values after successful treatment of peritonitis could be helpful in such cases, especially if patients have already been treated with antibiotic drugs. Larger prospective studies are required to compare the diagnostic value of soluble TNF receptors with that of neutrophil counts in infected ascites and with that of ascitic proteins such as fibronectin in malignant ascites. In conclusion, our data show that soluble TNF receptors are present in high amounts in ascites and plasma of patients with ascites of different origins. Soluble TNF receptors are present in ascites in a 500-fold excess over TNF-a. The high soluble TNF receptor concentrations in ascites may attenuate the biological activity of TNF. Significantlyhigher amounts of soluble TNF receptors p55 and p75 are found in ascitic fluid of patients with infected or malignant ascites than in ascites of patients with plain hepatic ascites. REFERENCES 1. Beutler B, Cerami A. The biology of cachectidTNF: a primary mediator of the host response. Annu Rev Immunol 1989;7: 625-655. 2. Tracey KJ, Vlassara H, Cerami A. Cachectinhumor necrosis factor. Lancet 1989;1:1122-1126. 3. Vilcek J, Lee TH. Tumor necrosis factor: new insights into the molecular mechanisms of its multiple actions. J Biol Chem 1991;266:7313-7316. 4. Carswell EA, Old U,Kassel RL, Green S, Fiore N, Williamson B. An endotoxin-induced serum factor that causes necrosis of tumors. Proc Natl Acad Sci USA 1975;72:3666-3670.
HEPATOLOGY
5. Loetscher H, Pan YC, Lahm HW, Gentz R, Brockhaus M, Tabuchi H, Lesslauer W. Molecular cloningand expression of the human 55 kd tumor necrosis factor receptor. Cell 1990;61:351-359. 6. Schall TJ, Lewis M, Koller KJ, Lee A, Rice GC, Wong GHW, Gatanaga T, et al. Molecular cloning and expression of a receptor for human tumor necrosis factor. Cell 1990;61:361-370. 7. Smith CA, Davei T, Anderson D, Solam L, Beckmann MP, Jerzy R, Dower SK, et al. A receptor for tumor necrosis factor defined an unusual family of cellular and viral proteins. Science 1990;248: 1019-1023. 8. Nophar Y, Kemper 0, Brakebusch C, Engelmann H, Zwang R, Aderka D, Holtmann H, et al. Soluble forms of tumor necrosis factor receptors (TNF-Rs): The cDNA for the type I TNF-R, cloned using amino acid sequence data of its soluble form, encodes both the cell surface and a soluble form of the receptor. EMBO J 1990;9:3269-3278. 9. Loetscher H, Schlaeger EJ, Lahm HW, Pan Y-CE, Lesslauer W, Brockhaus M. Purification and partial amino acid sequence analysis of two distinct tumor necrosis factor receptors from HL60 cells. J Biol Chem 1990;265:20131-20138. 10. Olsson I, Lantz M, Nilsson E, Peetre C, Thysell H, Grubb A, Adolf G. Isolation and characterization of a tumor necrosis factor binding protein from urine. Eur J Haematol 1989;42:270-275. 11. Seckinger P, Isaaz S, Dayer J-M. Purification and biologic characterization of a specific tumor necrosis factor a inhibitor. J Biol Chem 1989;265:11966-11973. 12. Engelmann H, Novick D, Wallach D. Two tumor necrosis factor-binding proteins purified from human urine. J Biol Chem 1990;265:1531-1536. 13. Schiitze S, Scheurich P, Pfizenmaier K, Kronke M. Tumor necrosis factor signal transduction: tissue-specific serine phosphorylation of a 26-kDa cytosolic protein. J Biol Chem 1989;264: 3562-3567. 14. Gatanaga T, Lentz R, Mansunuka I, Tomich J , Jeffes EWB 111, Baird M, Granger G. Identification of TNF-LT blocking factor(s) in serum and ultrafiltrates of human cancer patients. Lymphkine Res 1990;9:225-229. 15. Gatanaga T, Hwang C, Kohr W, Cappucini F, Lucci JA 111, Jeffes EWB, h n t z R, et al. Purification and characterization of an inhibitor (soluble tumor necrosis factor receptor) for tumor necrosis factor and lymphotoxin obtained from the serum ultrafiltrates of human cancer patients. Proc Natl Acad Sci USA 1990;87:8781-8784. 16. Bird GLA, Sheron N, Goka AKJ, Alexander GJ, Williams RS. Increased plasma tumor necrosis factor in severe alcoholic hepatitis. Ann Intern Med 1990;112:917-920. 17. Khoruts A, StahnkeL, 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. 18. Sheron N, Lau J, Daniels H, Goka J, Eddleston A, Alexander GJM, Williams R. Increased production of tumor necrosis factor alpha in chronic hepatitis B virus infection. J Hepatol 1991;12:241-245. 19. McClain CJ, Cohen DA. Tumor necrosis factor in alcoholic hepatitis. Gastroenterology 1989;97:1056-1057. 20. McClain CJ, Cohen DA. Increased tumor necrosis factor production by monocytes in alcoholic hepatitis. HEPATOLOGY 1989;9: 349-351. 21. Deviere J , Content J , Denys C, Vandenbussche P, Schandene L, Wybran J , Dupont E. Excessive in uitro bacterial lipopolysaccharide-inducedproduction of monokines in cirrhosis. 1990;11:628-634. HEPATOLOGY 22. Yoshioka K, Kakumu S, Arao M, Tsutsumi Y, Inoue M. Tumor necrosis factor-a production by peripheral blood mononuclear cells of patients with chronic liver disease. HEPATOLOGY1989;lO: 769-773. 23. Miizes G, Deak G, Lang I, Gonzalez-Cabello R, Gergely P, Feher J. Depressed monocyte production of interleukin-1 and tumor necrosis factor-alpha in patients with alcoholic liver cirrhosis. Liver 1989;9:302-306. 24 Felver ME, Mezey E, McGuire M, Mitchell MC, Herlong HF, Veech GA, Veech RL. Plasma tumor necrosis factor a predicts decreased long-term survival in severe alcoholic hepatitis. Alcohol Clin Exp Res 1990;77:221-225.
Vol. 16, No. 3, 1992
HIGH CONCENTRATIONS OF SOLUBLE TNF RECEPTORS IN ASCITES
25. Lau JYN, Sheron N, Nouri-Aria KR, Alexander GJM, Williams R. Increased tumor necrosis factor-a receptor number in chronic hepatitis B virus infection. HEPATOLDGY 1991;14:44-50. 26. Gearing AJ, Leung H, Bird CR, Thorpe R. Presence of the inflammatory cytokines IL-1, TNF, and IL-6 in preparations of monoclonal antibodies. Hybridoma 1989;8:361-367. 27. Deviere J , Content J, Crusiaux A, Dupont E. IL-6 and TNFa in ascitic fluid during spontaneous bacterial peritonitis. Dig Dis Sci 1991;36:123-124. 28. Runyon BA, Umland ET, Merlin T. Inoculation of blood culture bottles with ascitic fluid. Improved detection of spontaneous bacterial peritonitis. Arch Intern Med 1987;147:73-75. 29. Loetscher H, Gentz R, Zulauf M, Lustig A, Tabuchi H, Schlaeger EJ, Brockhaus M, et al. Recombinant 55-kDa tumor necrosis factor (TNF) receptor: stoichiometry of binding to TNF alpha and TNF beta and inhibition of TNF activity. J Biol Chem 1991;266: 18324-18329. 30. Aderka D, Engelmann H, Maor Y, Brakebysch C, Wallach D. Stabilization of the bioactivity of tumor necrosis factor by its soluble receptors. J Exp Med 1992;175:323-329. 31. Mohler KM, Torrance DT, Young DM, Callis G, Roux ER, Jacobs CA. Immunotherapeutic potential of soluble cytokine receptors in inflammatory disease [Abstract]. FASEB J 1992;6:A1086. 32. Aggarwal BB, Eessalu TE, Hass PE. Characterization of receptors for human tumor necrosis factor and their regulation by gammainterferon. Nature 1985;318:665-667.
755
33. Dett CA, Gatanaga M, Ininns EK, Cappuccini F, Yamamoto RS, Granger GA, Gatanaga T. Enhancement of lymphokine-activated T killer cell tumor necrosis factor receptor mRNA transcription, tumor necrosis factor receptor membrane expression, and tumor necrosis factorflymphotoxin release by IL-lB, IL-4 and IL-6 in vitro. J Immunol 1991;146:1522-1526. 34. Owen-Schaub LB, Crump WL 111, Morin GI, Grimm EA. Regulation of lymphocyte tumor necrosis factor receptors by IL-2. J Immunol 1989;135:2236-2241. 35. Tsujimoto M, Vilcek J. Tumor necrosis factor-induced downregulation of its receptors in HeLa cells. J Biochem 1987;102:15711577. 36. Holtmann H, Wallach D. Downregulation of the receptors of tumor necrosis factor by interleukin 1 and 4p-phorbol-12myriState-13-acetate.J Immunol 1987;139:1161-1167. 37. Drapier J-C, Wietzerbin J. IFN-gamma reduces specific binding of tumor necrosis factor on murine macrophages. J Immunol 1991;146:1198-1203. 38. Lantz M, Gullberg U, Nilsson E, Olsson I. Characterization in vitro of a human tumor necrosis factor-binding protein: a soluble form of a tumor necrosis factor receptor. J Clin Invest 1990;86: 1396-1402. 39. Porteu F, Nathan C. Shedding of tumor necrosis factor receptors by activated human neutrophils. J Exp Med 1990;172:599-607. 40. Hoefs JK. Spontaneous bacterial peritonitis: prevention and therapy. HEPATOLOGY 1990;12:776-781.
