Hepatic Kupffer cell blockade reduces mortality of acute hemorrhagic pancreatitis in mice

Inflammatory cytokines derived from the liver may cause distant organ failure and death m severe pancreatms To minmize hver cytokme release, we studied the effects of Kupffer cell blockade on the mortality rate and severity of mflammatlon m a model of that disease Thirty nuce were dlvlded mto three groups Group 1 received gadohmum chloride (1 mg/l00 g mtravenously), which blocks Kupffer cell acuvlty, and regular food Groups 2 and 3 were fed a chohne-deficient, etionine-supplemented diet and developed severe pancreatitls Group 2 (control) received intravenous sahne solution, and group 3 recewed gadohmum chlonde Animals were lulled at 72 hours Serum levels of tumor necrosis factor-a and mterlenkm-1p, mterleukmd, and mterlenkm-10 were determined by enzyme-hnked lmmunosorbent assay Lung neutrophll mfiltration was assessed by myeloperoxldase assay Pancreatic mflammauon was scored m a blinded manner In a separate experiment, mortahty rates were determmed m saline- and gadohnmm-treated animals (n = 100) Gadohmium reduced the levels of all the cytokmes and lung myeloperoxldase (P <0 0.5) Gadolmium also reduced the mortality rate (52% vs 86%,P<0 001) However, the degree of pancreatic mflammaaon was unchanged by gadohnmm treatment These data support the hypothesis that mortality m severe pancreauus may m part be related to the secondary release of hepauc cytokmes. Dr Gloor IS supported by a grant from the Swiss Nauonal Science Foundation This work was supported by a Veterans Admmlstranon Merit Review Grant (H A R) Presented at the Thuty-Eighth Annual Meetmg of The Society for Surgery of the Ahmentary Tract, Washington, D C, May ll–14,1997     

Failure of Kupffer cell blockade to prevent disseminated intravascular coagulation in endotoxemic rats despite improved survival

Objective: Studies were conducted to evaluate the impact of gadolinium chloride (GdCl₃), an agent which blocks the phagocytosis of liver macrophages (Kupffer cells, KC), on the coagulation system and on mortality in a model of rats subjected to a lethal dose of Escherichia coli lipopolysaccharide (LPS) (10 mg/kg body weight, intravenously). Methods: Rats were either pretreated with GdCl₃ (10 mg/kg, i.v., 48 h and 24 h prior to LPS exposure) or saline vehicle. A variety of coagulation parameters such as activated partial prothrombin time (aPTT), fibrinogen, systemic platelet count, antithrombin III (AT III), and activities of factors V, VII, and XII were monitored in the early (1 h) and late time course (16 h) following administration of E.coli LPS. Results: The administration of LPS resulted in the development of disseminated intravascular coagulation (DIC) and was associated with a mortality rate of 47% within 16 h. Blockade of KC by GdCl₃ completely abolished LPS-related mortality (0%). However, despite improved survival, GdCl₃ failed to prevent laboratory and clinical signs of DIC. GdCl₃ per se even contributed to coagulatory and fibrinolytic disorders. Conclusion: These results confirm reports on the protective potential of GdCl₃ pretreatment in experimental endotoxemia. However, the present study does not support the concept of DIC as a strong prognostic criterion for the outcome of sepsis and septic shock. Furthermore, the results presented suggest a minor role for KC in LPS-mediated activation of coagulation and indicate an involvement of KC in LPS-associated lethality independent of the coagulation system. Received: 29 October 1997     

Thymic stromal lymphopoietin mediates the host response and increases mortality during sepsis

Background

Sepsis and subsequent multiorgan system failure is associated with high rates of mortality and morbidity. Thymic stromal lymphopoietin (TSLP) is a cytokine that can be produced by keratinocytes and epithelial cells. Primarily, TSLP has been shown to promote counter-inflammatory processes. However, its potential expression or role in the pathogenesis of sepsis is largely unexplored. We hypothesized that TSLP is expressed during sepsis and TSLP blockade would alter the immune response and mortality.

Materials and methods

Mice underwent cecal ligation and puncture (CLP) to produce a physiologically relevant murine model for sepsis. Cohorts were either treated with neutralizing TSLP antibodies or isotype controls before the CLP to determine changes in survival, bacterial loads, cytokine levels, and neutrophil function.

Results

It was observed that TSLP levels peaked at 6 h and remained detectable up to 48 h after CLP. Mice pretreated with neutralizing TSLP showed decreased mortality and bacterial load after CLP. Additionally, we determined that septic mice pretreated with the anti-TSLP antibody had increased tumor necrosis factor alpha and oxidative burst as well as increased interleukin 17 and neutrophil numbers compared with mice pretreated with isotype controls.

Conclusions

TSLP levels peak early but are sustained during the first 48 h of sepsis. We speculate that TSLP blunts the neutrophil response resulting in increased bacterial load and mortality.     

Beta-adrenergic blockade increases the hepatic extraction of glucose in sepsis

To determine the relationship between hepatic glucose clearance and elevated epinephrine levels in sepsis, dogs with gangrenous cholecystitis were anesthetized and received either propranolol hydrochloride (mean dose, 0.29 mg/kg) or saline solution before intraduodenal glucose injection (2.5 g/kg). The amounts of glucose, insulin, and glucagon in the portal vein, the hepatic artery, and the hepatic vein were determined from the concentrations and the blood flows in these vessels over a two-hour period. Normal dogs served as controls. The amounts of glucose, insulin, and glucagon reaching the livers of both septic groups were the same. However, propranolol treatment increased the percent of glucose extracted by the liver without affecting the extractions of insulin or glucagon. Propranolol reverses the limitation of hepatic glucose extraction in sepsis by a direct effect. Whether the extracted glucose is utilizable as an energy substrate needs to be established.     

Kupffer cell modulation of the systemic immune response

The effects of global hepatic injury and of Kupffer cell activation on systemic immunity were studied in an in vivo rat model, using the diameters of the delayed-type hypersensitivity (DTH) response to keyhole limpet hemocyanin and of a subcutaneous Staphylococcus aureus abscess as measures of systemic immunoresponsiveness. Hepatic injury with carbon tetrachloride resulted in significant suppression of the DTH score (5.5 +/- 0.7 vs 8.8 +/- 0.8 mm). Kupffer cell activation with intraportal Escherichia coli was likewise suppressive (DTH score, 4.4 +/- 0.5 vs 6.1 +/- 0.4 mm for animals receiving systemic E coli); the magnitude of this suppression correlated with the numbers of organisms extracted by the liver. Conversely, Kupffer cell ablation with carrageenan lessened the immunosuppressive effects of anesthesia and surgery (DTH score, 8.5 +/- 0.9 vs 6.8 +/- 0.6 mm for controls; S aureus abscess, 4.1 +/- 0.4 vs 5.7 +/- 0.4 mm for controls). These results indicate that Kupffer cells can modulate the systemic immune response and suggest that gram-negative portal bacteremia with resultant Kupffer cell activation may contribute to the immunologic derangements characteristic of trauma and critical surgical illness.     

Kupffer cell ablation improves hepatic microcirculation after trauma and sepsis

BACKGROUND: Macrophages undergo maladaptive alterations after trauma. In this study, we assessed the role of Kupffer cells in hepatic microcirculatory response to endothelin-1 (ET-1) after femur fracture (FFx) and cecal ligation and puncture (CLP). METHODS: Sprague-Dawley rats (200-300 g) underwent sham, FFx, CLP, or FFx + CLP. To ablate Kupffer cells, group 1 animals were treated with gadolinium chloride, and group 2 animals received saline. Hepatic microcirculation was assessed by intravital microscopy. Liver mitochondrial redox state and tissue oxygen (tPo2) were determined by NADH and ruthenium fluorescence, respectively. Liver damage was estimated by alanine aminotransferase levels. Differences were assessed using analysis of variance followed by Student-Newman-Keuls post hoc test. RESULTS: After 10 minutes of ET-1, CLP and FFx + CLP caused significant reduction in hepatic perfusion index (2.5-fold and 5-fold vs. sham, p < 0.05, respectively), redox state (36% and 45% vs. sham, p < 0.01, respectively), tPo2 (10% and 12% vs. sham, p < 0.05, respectively), and more liver damage compared with sham and FFx-treated animals. Kupffer cell depletion restored microcirculation, redox state, and tPo2 and abrogated hepatocellular damage. CONCLUSION: Kupffer cells contribute directly to hepatic microcirculatory dysfunction and liver injury after inflammatory stress. Furthermore, Kupffer cell depletion ameliorates the microcirculatory perturbations of trauma and sepsis. Thus, modulation of Kupffer cell response may prove beneficial.     

Peritoneal fluid: a potential mechanism of systemic neutrophil priming in experimental intra-abdominal sepsis

BackgroundRecent studies suggest that peritoneal fluid (PF) may be an important mediator of inflammation. The aim of this study was to test the hypothesis that PF may drive systemic inflammation in intra-abdominal sepsis by representing a priming agent for neutrophils.MethodsPF was collected 12 hours after the initiation of intra-abdominal sepsis in swine. Naive human neutrophils were primed with PF before treatment with N-formyl-Met-Leu-Phe or phorbol 12-myristate 13-acetate to elucidate receptor-dependent and receptor-independent mechanisms of neutrophil activation. Flow cytometry was used to quantify neutrophil surface adhesion marker expression of integrins and selectins and superoxide anion production. Additionally, proinflammatory cytokines were quantified in PF.ResultsPF primed neutrophils via receptor-dependent and receptor-independent mechanisms. There were significant increases in the proinflammatory cytokines interleukin-6 and tumor necrosis factor–α in PF correlating with the development of intra-abdominal sepsis.ConclusionsPF represents a priming agent for naive polymorphonuclear cells in intra-abdominal sepsis. This may be secondary to increased levels of proinflammatory cytokines. Strategies to reduce the amount of PF may decrease the systemic inflammatory response by reducing a priming agent for neutrophils.     

Enteral feeding preserves mucosal immunity despite in vivo MAdCAM-1 blockade of lymphocyte homing

OBJECTIVE: To determine the influence of route of nutrition on gut mucosal addressin cellular adhesion molecule-1 (MAdCAM-1) expression and the effect of MAdCAM-1 blockade on gut-associated lymphoid tissue (GALT) lymphocyte populations and established respiratory antibacterial immunity. SUMMARY BACKGROUND DATA: Lymphocytes, sensitized to antigens in Peyer's patches, migrate via mesenteric lymph nodes and home to intestinal lamina propria. MAdCAM-1 located on endothelial cells regulates this trafficking. Experimentally, parenteral nutrition (PN) decreases GALT cell mass and mucosal immunity when compared with enteral feeding. METHODS: In experiment 1, MAdCAM-1 expression was quantified in 32 mice after 4 days of feeding chow, a complex diet, intragastric (IG)-PN, or PN. In experiment 2, MAdCAM-1 was measured in 102 mice 0, 4, 8, 12, 24, 48, or 72 hours after starting PN and at 0, 4, 8, 12, 24, or 48 hours after reinstituting chow following 5 days of PN. In experiment 3, 56 mice received chow, PN, chow + MECA-367 (anti-MAdCAM-1 mAb), or chow + Isotype control Ab (IsoAb) for 5 days, followed by Peyer's patches, lamina propria, and intraepithelial lymphocyte yield with respiratory and intestinal IgA levels. In experiment 4, 10 days after Pseudomonas immunization, mice received chow + MECA-367 or chow + IsoAb for 4 days followed by 1.2 x 108 Pseudomonas intratracheally. RESULTS: Diet and route affect MAdCAM-1 expression (chow > complex diet > IG-PN > PN). Decreased MAdCAM-1 expression occurred within hours of starting PN in Peyer's patches, but not mesenteric lymph nodes or the intestine, and recovered quickly with enteral refeeding. MAdCAM-1 blockade reduced all GALT populations. Blockade had little effect on IgA levels and partially impaired the late response of established respiratory immunity. CONCLUSIONS: Enteral feeding affects MAdCAM-1 expression. Complete MAdCAM-1 blockade reduces GALT lymphocytes to PN levels, but the chow feeding stimulus preserves IgA and early antibacterial resistance, implying the existence of non-MAdCAM-1 mechanisms to preserve mucosal immunity.     

Propofol attenuates Kupffer cell activation during hypoxia-reoxygenation

PURPOSE: We undertook a study to determine whether propofol may attenuate Kupffer cell (KC) activation, thus protecting the cells against hypoxia-reoxygenation injury through the modulation of intracellular calcium ([Ca2+]i). METHODS: [Ca2+]i, the expression of tumour necrosis factor (TNF)-alpha mRNA, and KC viability were measured in response to hypoxia-reoxygenation following pretreatment with propofol 0.5 and 5 microg.mL(-1) (Groups P1 and P2, respectively) or without propofol (Group HRC). KCs were isolated and cultured from male Sprague-Dawley rats. KCs were incubated under an atmosphere of hypoxia (95% N2 + 5% CO2) for 60 min with subsequent 120 min reoxygenation (95% air + 5% CO2). [Ca2+]i for the first 12 min after reoxygenation, TNF-alpha mRNA, and KC viability at the end of reoxygenation in groups P1 and P2 were compared with those of HRC. RESULTS: The increase of [Ca2+]i from the baseline was attenuated in P1 (96.6 +/- 6.9%) and P2 (96.1 +/- 5.4%) compared with HRC (143.8 +/- 11.5%), (P < 0.001), with no difference between P1 and P2. The expression of TNF-alpha mRNA increased only in HRC during hypoxia-reoxygenation. KC viability increased in P1 (97.5 +/- 2.6%) and P2 (94.6 +/- 2.9%), compared with HRC (89.9 +/- 1.4%), (P < 0.005), with no difference between P1 and P2. CONCLUSION: The results indicate that propofol attenuates KC activation and protects KC from hypoxia-reoxygenation injury at clinically relevant concentrations. This attenuation seems to result from inhibition of [Ca2+]i increase in KC.     

Dietary omega-3 fatty acids decrease mortality and Kupffer cell prostaglandin E2 production in a rat model of chronic sepsis

We tested the hypothesis that substitution of omega-3 fat for dietary omega-6 fat would reduce mortality and decrease Kupffer cell prostaglandin E2 (PGE2) production in a rat model of chronic sepsis. Rats were fed via gastrostomy for 12 days with isonitrogenous, isocaloric diets containing 15% of calories as either safflower oil (omega-6) or a 10:1 mixture of menhaden oil (omega-3) and safflower oil. After five days of feeding, animals received an intra-abdominal abscess of defined bacterial content. Survivors were killed on post-laparotomy day 6 in conjunction with liver perfusion and protease liver digestion for Kupffer cell isolation. Kupffer cell PGE2 production was measured by radioimmunoassay after 18 hours of cell culture and again after stimulation with 0 LPS, 10 ng/ml LPS, and 10 micrograms/LPS. Mortality was decreased in menhaden oil-fed animals compared with safflower oil-fed animals (16% vs. 35%). Kupffer cell PGE2 production was decreased in menhaden oil-fed animals at 18 hours (354 +/- 54 vs. 570 +/- 95 pg/0.1 ml; p = 0.09) and after stimulation with 10 micrograms/ml LPS (140 +/- 41 vs. 288 +/- 45 pg/0.1 ml; p = 0.03) compared with safflower oil-fed animals.     

Synergistic effects of nafamostat mesilate rinse and Kupffer cell blockade for rat liver preservation

Abstract  We investigated the efficacy of a new rinse solution containing nafamostat mesilate (NM) (a serine protease inhibitor) for liver preservation with modulation of Kupffer cell function. Orthotopic liver transplantation (OLT) was performed in male Lewis rats after 24 h of cold storage in University of Wisconsin organ preservation solution. After OLT, survival was determined, together with assays of blood chemistry, tissue NM metabolites, and histology 3 h after OLT. NM rinse was found to have a cytoprotective effect on liver parenchymal cells, based on enzyme data showing that NM rinse reduced the release of serum alanine aminotransferase significantly in comparison with saline rinse ( P < 0.05). However, the effect was not sufficient to improve the survival rate. In contrast, when the donor was treated with gadolinium chloride 24–30 h before graft harvest, NM rinse improved the survival rate to around 80 % compared with 25 % for saline. The assay of NM metabolites in grafted liver tissue showed that pretreatment of the donor rats with GdCl₃ delayed the degeneration of NM in the liver tissue. These data demonstrate that NM rinse and Kupffer cell blockade exert synergistic effects, leading to increased survival after cold-preserved liver transplantation.     

Kupffer cell blockade prevents induction of portal venous tolerance in rat cardiac allograft transplantation

Pretransplant portal venous (pv) administration of donor antigen induces allospecific partial tolerance. Although the involved mechanism has not been defined, antigen presentation by Kupffer cells (KC) in the liver is considered to be critical. We evaluated the effect of KC blockade on this pv tolerance induction in Buffalo (RT1b) rats receiving Lewis (RT1(1] cardiac heterotopic allografts. Control rats received no treatment, while experimental animals received 25 X 10(6) ultraviolet B-irradiated (12,000 J/m2) donor spleen cells via either the iv (systemic intravenous) or the pv routes 7 days before transplantation. Gadolinium chloride (GdCl3), a rare earth metal known to inhibit KC phagocytosis, was given (7 mg/kg) 1 and 2 days before pv preimmunization. Cardiac graft prolongation was obtained by pv (MST = 13.3 +/- 1.9 days, n = 6, vs control = 7.3 +/- 0.5 days, n = 6; P less than 0.001) but not by iv preimmunization (7.7 +/- 0.7 days, n = 6, NS vs control). KC blockade abolished the pv tolerance, as indicated by abrogation of graft prolongation (PV + GdCl3 = 8.0 +/- 0.8 days, n = 6, NS vs control). These findings suggest that effective alloantigen uptake by KC in the liver is essential for the induction of pv tolerance in rat cardiac transplantation.     

Amitriptyline therapy increases electrocardiographic changes during reversal of neuromuscular blockade

Induction of anesthesia is associated with an increased incidence of cardiac arrhythmias in patients maintained on amitriptyline medication. This study presents additional evidence showing that, in experimental animals (cats), amitriptyline treatment also produces significant ST-T wave and conduction abnormalities during neostigmine reversal of neuromuscular blockade.     

Contribution of portal systemic shunt to Kupffer cell dysfunction in obstructive jaundice

Previous animal models of biliary tract obstruction have shown that hepatic phagocytic activity is impaired secondary to Kupffer cell dysfunction. Biliary tract obstruction leads to portal hypertension and an associated portal systemic shunt. Forty-eight Sprague-Dawley rats were studied to determine the contribution of portal systemic shunt to Kupffer cell dysfunction after 21 days of obstructive jaundice or sham operation. Liver uptake of radiolabeled Escherichia coli decreased from 76.1 +/- 1.4% (sham) to 63.1 +/- 6.1% in the common duct ligation (CDL) rats (P less than 0.05); lung uptake increased from 4.0 +/- 0.6% (sham) to 20.2 +/- 4.5 (CDL) (P less than 0.05). Portal systemic shunt, determined using radioactive microspheres, increased from 2.0 +/- 1.0% (sham) to 46.6 +/- 13.1% (CDL), P less than 0.05. Although a significant portal systemic shunt does exist in this 21-day model of obstructive jaundice, it does not appear to be the only mechanism underlying Kupffer cell dysfunction.     

Hepatocellular dysfunction persists during early sepsis despite increased volume of crystalloid resuscitation.

Although hepatocellular dysfunction occurs early in sepsis despite fluid resuscitation, it is unknown if an increased volume of resuscitation protects hepatocellular function. To study this, rats were subjected to sepsis by cecal ligation and puncture (CLP). These and sham-treated rats then received either 3 or 6 mL/100 g BW normal saline subcutaneously. Studies were performed at 5 hours (i.e., early sepsis) or 20 hours (late sepsis) after CLP. Hepatic blood flow was determined by radioactive microspheres, 3H-galactose clearance technique, and laser Doppler flowmetry in both groups. Active hepatocellular function (i.e., Vmax and Km) was assessed by an in vivo indocyanine green clearance technique. The results indicate that: (1) hepatic blood flow increased markedly in early sepsis; (2) Vmax and Km decreased significantly at 5 hours and 20 hours after CLP; and (3) the increased volume of fluid resuscitation did not improve the depressed active hepatocellular function 5 hours following CLP. Cardiac output and hepatic microcirculation, however, were significantly increased in early sepsis. These results confirm the notion that the depression in hepatocellular function in early sepsis is not the result of any reduction of hepatic perfusion. The dissociation of increased hepatic blood flow from depressed hepatocellular function remains despite the larger volume of resuscitation. The hepatocellular dysfunction that occurs even in early sepsis cannot be corrected simply by increasing the volume of crystalloid resuscitation.     

Kupffer cells are responsible for producing inflammatory cytokines and hepatocellular dysfunction during early sepsis.

BACKGROUND: Although hepatocellular dysfunction occurs early after the onset of sepsis, the mechanism responsible for this remains unknown. We tested the hypothesis that the reduction in Kupffer cell (KC) numbers prior to the onset of sepsis prevents the occurrence of hepatocellular dysfunction and reduces levels of the proinflammatory cytokines IL-1beta and IL-6 during the early stage of polymicrobial sepsis. MATERIALS AND METHODS: The number of KC in male adult rats was reduced in vivo by intravenous injection of gadolinium chloride 48 h before the induction of sepsis. KC-reduced and KC-normal rats were then subjected to cecal ligation and puncture (CLP, i.e., a model of polymicrobial sepsis) or sham operation followed by administration of normal saline solution. At 5 h after CLP (the early stage of sepsis), hepatocellular function [i.e., the maximal velocity of clearance (Vmax) and efficiency of active transport (Km) of indocyanine green] was measured using a fiber-optic catheter and in vivo hemoreflectometer. Whole blood was drawn to measure plasma levels of IL-1beta and IL-6 using enzyme-linked immunosorbent assays. RESULTS: Hepatocellular function was depressed and the circulating levels of IL-1beta and IL-6 were increased significantly at 5 h after CLP. KC reduction by prior administration of gadolinium chloride, however, prevented the occurrence of hepatocellular dysfunction and the upregulation of IL-1beta and IL-6. CONCLUSIONS: The KC-derived proinflammatory cytokines IL-1beta and IL-6 appear to be directly or indirectly responsible for producing hepatocellular dysfunction during early sepsis. Thus, pharmacologic agents that downregulate the production of one or both of these proinflammatory cytokines in the liver may be useful for maintaining hepatocellular function during the early stage of sepsis.     

Kupffer cell tumor necrosis factor-alpha production is suppressed during liver regeneration

Mammalian liver regeneration following resection invokes intrinsic hepatic responses which result in rapid tissue repair. The role of soluble immune cytokines in this phenomenon is not known. The capacity of Kupffer cells (KC) from regenerating liver to produce the potent cytokine TNF-alpha was evaluated. Twenty-four hours after 70% partial hepatectomy (PHx) or sham operation, Kupffer cells were harvested from collagenase-digested Wistar-Furth rat livers and purified (greater than 95% by phagocytosis) by adherence. Following overnight culture with or without the cyclooxygenase inhibitor indomethacin (10 microM), 5 x 10(5) KC were repleted with fresh media with or without 2.5 micrograms/ml lipopolysaccharide (LPS). Supernatant TNF-alpha activities (units/ml) were measured using the L929 fibroblast lysis assay. With LPS, sham KC TNF-alpha levels were significantly higher (P less than 0.001) than those for PHx KC. Indomethacin significantly increased PHx KC TNF-alpha levels, but did not affect those for sham KC, suggesting autoregulation by arachidonic acid cyclooxygenase metabolites following PHx. We conclude that KC TNF-alpha production is suppressed following PHx by a mechanism apparently regulated by eicosanoid metabolism. During the stress of hepatic regeneration, a coordinated limitation of excessive TNF-alpha responses by PHx liver KC may naturally protect the host.     

Kupffer cell blockade, tumour necrosis factor secretion and survival following endotoxin challenge in experimental biliary obstruction

BACKGROUND: Gram-negative sepsis and its sequelae frequently complicate invasive procedures in patients with obstructive jaundice. In response to endotoxin, Kupffer cells secrete tumour necrosis factor (TNF), a pivotal early mediator of sepsis. An investigation was carried out into the specific role of Kupffer cell TNF secretion following endotoxin challenge in obstructive jaundice. METHODS: Survival following intraperitoneal administration of endotoxin (2.0, 0.02 and 0.0002 mg per 100 g) was determined in rats following bile duct ligation (BDL) or sham operation. Plasma TNF concentration was quantified following endotoxin administration (0.0002 mg per 100 g) at 1, 2 and 6 h. Subsequently, the effect of Kupffer cell blockade by gadolinium chloride on survival and plasma TNF concentration was assessed. RESULTS: Jaundiced animals showed a significantly increased mortality rate following intraperitoneal injection of endotoxin 2.0 mg per 100 g (BDL 100 per cent versus sham 0 per cent) and 0.02 mg per 100 g (BDL 70 per cent versus sham 0 per cent; P = 0. 002, Fisher's exact test). Median plasma TNF concentration was significantly greater in jaundiced animals 1 h after endotoxin administration (BDL 943 (interquartile range (i.q.r.) 211-3900) pg/ml versus sham 64 (i.q.r. 47-127) pg/ml; P = 0.002, Mann-Whitney U test). Kupffer cell blockade with gadolinium chloride increased the survival rate following endotoxin administration in BDL animals (BDL-GdCl3 100 per cent versus BDL-saline 40 per cent; P = 0.0003, Fisher's exact test) and decreased median plasma levels of TNF (BDL-GdCl3 88 (i.q.r. 0-1065) pg/ml versus BDL-saline 16 550 (1255-29 360) pg/ml; P = 0.002, Mann-Whitney U test). CONCLUSION: Kupffer cell blockade improved survival and suppressed systemic TNF activity after endotoxin challenge. In obstructive jaundice, hypersecretion of TNF by Kupffer cells may supplement systemic cytokine production and be responsible for significant complications.