Francis J. Dudley

The Effect of Selective Intestinal Decontamination on the Hyperdynamic Circulatory State in Cirrhosis: A Randomized Trial

Background Patients with advanced hepatic cirrhosis often experience a hyperdynamic circulatory state, possibly because nitric oxide is released by intestinal bacteria. Selective intestinal decontamination with norfloxacin may decrease nitric oxide release by altering the intestinal flora. Contribution This randomized, double-blind, placebo-controlled cross-over study shows that norfloxacin administration is associated with decreased serum endotoxin levels, attenuation of increased cardiac output, and maintenance of systemic vascular resistance in cirrhotic patients. Implications Norfloxacin may be useful in treating the hyperdynamic circulatory state in cirrhotic patients. Although the mechanism may not be explicitly proven by this study, it may involve a decrease in nitric oxide production. The Editors Advanced hepatic cirrhosis, regardless of its cause, is accompanied by a hyperdynamic circulatory state characterized by high cardiac output, low systemic vascular resistance, and low mean arterial pressure (1). This vasodilatory state contributes to the pathogenesis of many complications of chronic liver disease, such as portal hypertension, ascites formation, and the hepatopulmonary and hepatorenal syndromes. The mechanism responsible for the vasodilatory state is thought to be largely due to the increased vascular production of nitric oxide (2, 3). We hypothesize that in cirrhosis, the overgrowth or translocation of intestinal bacteria results in increased levels of circulating endotoxin or cytokines, such as tumor necrosis factor-, and enhances nitric oxide production by inducing either nitric oxide synthase II (4) or III (5) expression. Consistent with this hypothesis, we have previously demonstrated (6) that selective intestinal decontamination with the fluoroquinolone antibiotic norfloxacin successfully inhibits some of the nitric oxidemediated effects that are evident in patients with alcohol-related cirrhosis. We further tested this hypothesis by examining the effects of selective intestinal decontamination using the antibiotic norfloxacin on systemic and regional hemodynamics in a group of cirrhotic patients in a randomized, double-blind, placebo-controlled, crossover study. Methods Patients The study group consisted of 14 well-characterized men with alcohol-related hepatic cirrhosis (mean age [SD], 53 2 years). In 12 patients, cirrhosis was proven by biopsy, and in 2 patients, results of clinical, laboratory, and imaging studies were consistent with the presence of cirrhosis (7). Table 1 presents patient characteristics and ChildPugh grades. The cirrhotic patients studied were recruited through the Liver Clinic, Department of Gastroenterology, Alfred Hospital, Melbourne, Australia, which has more than 800 patients in its database. Patients attending the clinic on a regular basis and who had prolonged periods of abstinence from alcohol, as determined by corroborative history and random plasma alcohol determinations, were eligible for the study. To maintain homogeneity, we selected only stable patients with alcohol-related cirrhosis. All recruited patients had negative results for viral hepatitis (hepatitis B surface antigen and hepatitis C virus antibody), autoantibodies (smooth-muscle antibody, antinuclear antibody, and antimitochondrial antibody), and metabolic disorders (iron studies, serum ceruloplasmin levels, and 1-antitrypsin levels). The severity of the liver disease varied among the patients and was representative of the eligible pool. Table 1. Demographic and Clinical Data for Study Participants Fourteen healthy male volunteers (mean age [SD], 47 4 years) served as an age- and sex-matched normal control group for the systemic and forearm hemodynamic studies. This group was recruited by advertisement. Before the start of the study, all participants underwent a thorough screening process, involving a detailed medical history, physical examination, 12-lead electrocardiography, and routine laboratory tests (including a complete blood count, renal function tests, and liver function tests). All participants tested negative for hepatitis B surface antigen and hepatitis C virus antibody by enzyme-linked immunosorbent assay, had normal liver function test results, and had negative plasma ethanol levels. Because the study was invasive, it was considered unethical to subject healthy controls to additional invasive studies; therefore, only baseline hemodynamic data were obtained and no antibiotic capsules or placebo was administered. We obtained written informed consent from all participants, who were aware of their right to withdraw from the study at any time. The study protocol conformed with the principles outlined in the Declaration of Helsinki and was approved by the Alfred Hospitals ethics committee. All participants had abstained from alcohol from 5 days to more than 6 months before the study. They also refrained from consuming caffeine-containing beverages the night before and the day of the study; plasma caffeine levels were not obtained, however. None of the study participants were smokers. All vasoactive medication was stopped 5 days before the study. Study Design Upon entry into the double-blind, placebo-controlled, crossover study, patients with cirrhosis were randomly assigned to receive either norfloxacin (400 mg twice daily) or a placebo capsule of identical appearance for 28 days (Figure 1). The patients were then switched to the alternate capsules, which were continued for an additional 28 days. Study medication was dispensed and adherence monitored by pill count and corroborative history by the Clinical Trials Section of the Pharmacy Department at Alfred Hospital. In the cirrhotic patients, all study measures and procedures were performed twice, once after the first 28 days of treatment and again 28 days after use of the alternate capsules began. Because the study was invasive, it was considered unethical to perform initial baseline studies before randomization. Since the half-life of norfloxacin ranges from 2.5 to 8 hours (normal half-life in end-stage renal disease, 6 hours), little carryover effect was anticipated or observed in this study and in our previous study of similar design (6). Figure 1. Flow chart of the study design. The 14 healthy controls were recruited to allow a cross-sectional comparison of physiologic differences between cirrhotic patients and healthy controls. Healthy controls received no treatment intervention. Study Measures and Procedures On the day of each study, patients rested supine in a quiet environment that was maintained at 22 C. The left brachial artery was cannulated with a 3-French, 5-cm catheter (Cook, Sydney, Australia). A blood sample was obtained for measurement of endotoxin, renin, and aldosterone levels. After a rest period of 30 minutes, baseline intraarterial blood pressure (Model 90603, Spacelabs, Inc., Redmond, Washington) was recorded. Then, baseline forearm blood flow was determined by venous occlusion plethysmography (Medasonic, Mountain View, California). This entailed the use of two blood pressure cuffs, one placed distally around the wrist and the other proximally on the upper arm; in addition, an alloy-filled, double-stranded strain gauge was placed on the widest part of the forearm. The occlusion pressure was 40 to 50 mm Hg at the proximal end and 200 mm Hg at the distal end. Recordings were obtained for 10 out of every 20 seconds over 3 minutes. Immediately after this study period, participants were prepared for regional catheterization studies. Regional Catheter Studies Cardiac Output Central venous and right heart catheterization was performed under direct fluoroscopic (Model DC 12MB-1, Toshiba Corp., Tochigi-Ken, Japan) and pressure waveform (Model 90603, Spacelabs, Inc.) guidance by using an 8.5-French percutaneous introducing sheath (product no. AK-09802-A, Arrow International, Reading, Pennsylvania) inserted into a medial antecubital vein. To evaluate cardiac output, a 7-French, 110-cm thermodilution balloon catheter (product no. Al-07067, Arrow International) interphased with a cardiac output monitor (model 15055A, Hewlett Packard, Andover, Massachusetts) was guided into the pulmonary artery, where rapid boluses of dextrose solution were injected into the right atrium via a proximal port. The values reported for cardiac output are the means of a minimum of three determinations differing by less than 10%. Hepatic Venous Pressure Gradient A 7-French Cournand catheter was passed via the antecubital sheath into the hepatic vein under fluoroscopic control. The hepatic venous pressure gradient (the difference between wedged hepatic vein pressure and free hepatic vein pressure) was recorded. The average of three measurements is reported. Plasma Samples Endotoxin Blood samples were collected in ice-cold, pyrogen-free tubes containing heparin (Greiner Bio-One, Inc., Longwood, Florida) and subjected to centrifugation for 15 minutes at 4 C. Plasma was stored at 20 C until analyzed by the chromogenic Limulus lysate assay, based on the methods first introduced by Iwanaga and colleagues (8). In brief, this involved kinetic chromogenic analysis assessment by using the Endochrome-K test (Charles River Laboratories, Charleston, South Carolina) and a kinetic microplate reader (Elx 808iu Microplate Reader, Bio-Tek Instruments, Winooski, Vermont) that was read at 405 nm. This assay has been previously used successfully in several studies to detect endotoxin in cirrhotic patients (9-11). To eliminate interassay variability, the assays were performed as a single batch in triplicate on appropriately stored plasma obtained at the time of the original studies. The intraassay variability was less than 10%. Renin and Aldosterone Levels Renin and aldosterone levels were determined by using standard commercially available radioimmunoassay techniques. Glomerular Filtration Rate Glomerular filtration rate was determined by using the radioisotope [3H]i

Hepatic Hydrothorax in the Absence of Clinical Ascites: Diagnosis and Management

Two cases of right hepatic hydrothorax occurring in the absence of clinical ascites are reported. Diagnosis was confirmed by the intraperitoneal and intrapleural injection of radioisotope 99mTc-sulfur colloid that demonstrated the one-way transdiaphragmatic flow of fluid from the peritoneal to pleural cavities. In contrast, radioisotope injected into the peritoneal cavity of 5 patients with pleural effusions secondary to pulmonary or cardiac disease failed to traverse the diaphragm and localize in the pleural space. Medical therapy with salt and water restriction and diuretics resulted in both of the patients with hepatic hydrothorax developing signs of intravascular volume depletion without significant mobilization of the pleural fluid. Thoracotomy allowed identification of the diaphragmatic defects that were repaired by chemical and traumatic pleurodesis followed by postoperative peritoneal and pleural drainage. This therapy resulted in complete resolution of the pleural effusions, which have not recurred despite the subsequent development of clinical ascites in both patients.

Chronic hepatitis C: Effect of alcohol on hepatic activity and viral titre

BACKGROUND/AIMS Alcohol and the hepatitis C virus have been postulated to interact to adversely affect the natural history of patients with chronic liver disease. The aim of this study was to examine the effect of alcohol on hepatitic activity and serum HCV RNA levels in patients with chronic hepatitis C. METHODS Forty-five consecutive patients with chronic hepatitis C were classified according to alcohol intake over the 3-month period preceding study entry: group 1 (n = 23), > 10 g alcohol/day; group 2 (n = 22), < or = 10 g alcohol/day. Hepatitic activity and alcohol intake were assessed at study entry and, following moderation of alcohol intake, after a mean follow-up period of 4.4 +/- 0.2 months. RESULTS Hepatitic activity was significantly greater in the patients who consumed > 10 g of alcohol/day. Moderation of alcohol consumption in patients consuming > 10 g/day resulted in a significant decrease in both disease activity (p = 0.0002) and viral RNA titre (p = 0.018); there was no change over the study period in patients with a consistently low alcohol intake. CONCLUSION The results support the hypotheses that, in patients with chronic hepatitis C, alcohol aggravates hepatic injury, increases viral load and adversely affects the natural history of the associated liver disease.

Hepatic extraction of morphine is impaired in cirrhosis

SummaryUsing hepatic vein catheterization this study has provided the first direct measurement of morphine hepatic extraction in 8 controls and 8 cirrhotics.The extraction ratio was 0.52 in the control group and was reduced by 25% in the cirrhotics. This reduction is due to impaired enzyme capacity rather than reduced blood flow. The effect of cirrhosis is less than that reported in similar studies of high clearance oxidized drugs and this lends support to the concept that glucuronidation may be relatively spared in cirrhosis.A discrepancy between the systemic clearance and the hepatic clearance provides indirect support for extra-hepatic metabolism of morphine.

Impaired Reactivity of the Peripheral Vasculature to Pressor Agents in Alcoholic Cirrhosis

BACKGROUND Studies of the in vivo vascular reactivity of the peripheral circulation to pressor agents in cirrhosis have produced conflicting results, possibly because of changes in mean arterial pressure that make it difficult to clearly separate peripheral and central effects. The aim of the present study was to assess the reactivity of the forearm circulation to pressor agents in vivo without activating central control systems. METHODS Forearm blood flow was measured by venous occlusion strain gauge plethysmography in the basal state and during the infusion of subpressor doses of norepinephrine and angiotensin II into the brachial artery in 10 male patients with well-compensated alcoholic cirrhosis and 10 male age-matched controls. Plasma renin activity and aldosterone and angiotensin II concentrations were assayed. Forearm and systemic sympathetic nervous system activity was estimated using a norepinephrine spillover technique. RESULTS Basal forearm blood flow, renin angiotensin aldosterone system activity, and forearm sympathetic nervous system activity were similar in both the control and cirrhotic groups. The cirrhotic patients showed an impaired response to both norepinephrine and angiotensin II. CONCLUSIONS There is impaired reactivity of the peripheral vasculature to pressor agents in cirrhosis, indicating that the control of vascular tone is disturbed even in well-compensated cirrhosis.

Interaction between oral propranolol and hydralazine

Seven healthy subjects were given oral propranolol (1 mg/kg) alone or in combination with hydralazine 25, 50, or 100 mg on separate occasions. Hydralazine induced variable increases in the peak concentrations (p < 0.05) and in the area under the propranolol concentration: time curves (p < 0.02) without change in the recovery of 14C‐propranolol/metabolites in urine or in the systemic clearance of propranolol; i.e., oral hydralazine enhanced the systemic availability of propranolol by alteration of “first‐pass” (hepatic) clearance. The results indicate the possibility of change in presystemic clearance without reciprocal change in systemic clearance and also suggest that propranolol or any other high‐clearance drug should be administered in fixed relationship to hydralazine and other drugs capable of altering “first‐pass” hepatic extraction (either due to alteration in splanchnic blood flow or competitive inhibition of metabolism) if large variations in plasma concentrations are to be avoided.

Reduction of first‐pass hepatic clearance of propranolol by food

Exposure to drugs, chemicals, and hormones from the gut depends on initial (first‐pass) liver extraction and subsequent removal from the general circulation. We show that food decreases first‐pass liver extraction of a marker drug, propranolol, during first‐pass transit. This effect may apply to other drugs, environmental chemicals, and nutritional and gut‐hormone balance.

Effect of Fluoroquinolone on the Enhanced Nitric Oxide-Induced Peripheral Vasodilation Seen in Cirrhosis

Advanced liver disease is associated with low vascular resistance, which may contribute to complications that adversely affect survival [1]. The mechanisms responsible for this phenomenon are poorly defined and limit options for treatment. Injury of the vascular system can induce isoform II of nitric oxide synthase, resulting in excess nitric oxide [2]. In cirrhosis, intestinal bacterial overgrowth occurs [3] and portosystemic shunts may allow intestinal bacteria and endotoxin to enter the systemic circulation [4]. Endotoxemia may indirectly promote excessive nitric oxide and vasodilation through induction of nitric oxide synthase isoform II [5]. We tested the hypothesis that norfloxacin, a fluoroquinolone antibiotic, indirectly blocks the augmented effects of nitric oxide. Methods We recruited 9 patients with alcohol-related liver disease (mean age, 54 3 years) and 10 controls (mean age, 56 4 years). All patients with cirrhosis were clinically classified as having well-compensated Pugh grade A disease; cirrhosis was proven by biopsy in 8 patients. All patients had evidence of portal hypertension and esophageal varices. Controls had no evidence of liver disease on history or on physical and laboratory examination (including liver function tests), and they tested negative for hepatitis B surface antigen and hepatitis C virus by enzyme-linked immunosorbent assay. The study protocol was approved by the Alfred Group of Hospitals Ethics Committee, which conforms to the Declaration of Helsinki, and participants gave written, informed consent. In a double-blind, crossover protocol, patients with cirrhosis were randomly allocated to receive either norfloxacin (400 mg twice daily) or placebo for 4 weeks each. Responses to acetylcholine and the nitric oxide synthase inhibitor NG -monomethyl-L-arginine were studied twice, once after the first 4 weeks of treatment and once after patients crossed over to the other treatment. No washout period separated the treatments. Because the half-life of norfloxacin ranges from 3.5 to 8 hours (normal half-life in end-stage renal disease, 6 hours), little carry-over was expected. We assumed that the extent of bacterial recolonization that had occurred during the 4-week placebo period was similar to that during the original infection. Control forearm vascular responses were obtained from untreated, healthy participants. All participants refrained from drinking caffeinated beverages the night before and on the day of study. Controls refrained from drinking alcoholic beverages for 5 to 7 days before the study, and patients with cirrhosis abstained from alcohol from 5 days to more than 6 months before the study. Patients with cirrhosis stopped using vasoactive medication exactly 5 days before the study day. Experiments were performed in a quiet room kept at 22C. The left brachial artery was cannulated by inserting a 3-French, 5-cm catheter (Cook, Sydney, Australia) under local anesthesia with 1% lidocaine (Astra, Sydney, Australia) and full aseptic conditions. We obtained a 10-mL sample of arterial blood and measured the intra-arterial blood pressure (Biosensors International Proprietary, Ltd., Singapore, linked with Spacelabs, Inc., Redmond, Washington). A sealed, alloy-filled, double-stranded strain gauge was used with a plethysmograph (Medasonic, Mountain View, California). Recordings were done for 10 seconds every 20 seconds. The occlusion pressure was 40 to 50 mm Hg at the proximal end and 200 mm Hg at the distal end. Forearm vascular resistance was calculated as the mean arterial pressure (mm Hg) divided by the forearm blood flow (mL/100 mL per minute). We measured responses to acetylcholine (9.25, 18.5, and 37 g/min) and NG -monomethyl-L-arginine (1, 2, and 4 mol/min). After an initial equilibration period of 60 seconds, the average of three flow measurements was used as a measure of basal blood flow. Each drug was infused at 2 mL/min for at least 2 minutes (for acetylcholine) or 5 minutes (for NG -monomethyl-L-arginine) or until the response of three flow measurements reached a plateau. The average of these measurements was used as the measure of drug-induced flow. Rest periods of 5 to 10 minutes between concentrations and 15 minutes between drug administrations were allowed. No effect on either systemic blood pressure or heart rate was observed (Spacelabs, Inc.). Plasma was stored at 20C until the time of analysis. When thawed, plasma was deproteinized by spinning in Microcon 10 microconcentrators (Amicon, Bedford, Massachusetts) for 150 minutes at 9000 g. Conversion of nitrate to nitrite by using nitrate reductase was tracked by adding Griess reagents (Cayman Chemical Co., Ann Arbor, Michigan). Total nitrite concentration was determined by photometric measurement of the absorbance at 540 nm. Results are expressed as the mean SE or the median (25th, 75th percentiles) when the data were not normally distributed. The Student t-test (paired or unpaired as appropriate) was used to compare the two groups. Dose-response curves were compared by using two-way analysis of variance (with repeated measures where necessary, such as to compare cirrhotic norfloxacin recipients with cirrhotic placebo recipients). These tests were followed by post hoc t-tests with the appropriate corrections for multiple comparisons (Bonferroni t-test, Student-Newman-Keuls test, or Dunn test). We used only data obtained from the nine patients with cirrhosis who participated in the full study. Results Basal forearm blood flow was higher in patients with cirrhosis who received placebo than in controls (3.69 0.27 mL/100 mL per minute and 2.26 0.39 mL/100mL per minute; P = 0.014). Mean arterial pressure did not differ between cirrhotic patients and controls. Derived basal forearm vascular resistance was lower in patients with cirrhosis who received placebo than in controls, but this difference was not statistically significant (27.2 2.56 units and 52.8 11.18 units; P = 0.065). Forearm blood flow returned to normal with norfloxacin (norfloxacin compared with control, 2.64 0.31 mL/100 mL per minute and 2.26 0.39 mL/100mL per minute [P > 0.2]; placebo compared with norfloxacin, P = 0.097). Administration of NG -monomethyl-L-arginine decreased forearm blood flow in all participants, but this effect was greater in patients with cirrhosis (Figure 1, top). Because values for basal forearm blood flow and vascular resistance were altered in patients with cirrhosis, responses to NG -monomethyl-L-arginine were also expressed as the percentage of basal values (Figure 1, bottom). Responses to NG -monomethyl-L-arginine returned to normal with norfloxacin (Figure 1). Figure 1. Effect of NG -monomethyl-L-arginine. Top. Bottom. P P Acetylcholine increased blood flow responses in all participants, but this effect was greater in patients with cirrhosis (Figure 2, top). These responses remained substantially greater in patients with cirrhosis when expressed as a percentage of basal values. Norfloxacin had no effect on response to acetylcholine (Figure 2). Figure 2. Effect of acetylcholine. Top. Bottom. P Plasma total nitrate values were elevated in patients with cirrhosis who received placebo (35.7 mol/L [range, 29.8 to 37.1 mol/L]) compared with controls (22.1 mol/L [range, 16.5 to 23.1 mol/L]; P = 0.015). Norfloxacin had no effect on these values (33.54 5.93 mol/L in norfloxacin recipients compared with 35.7 mol/L in placebo recipients; P > 0.2). A two-sample t-test on the effect of placebo in the norfloxacin-placebo sequence compared with the effect of placebo in the placebo-norfloxacin sequence found no difference between the two placebo phases. In addition, within-patient differences of the two sequences were examined by using paired t-tests; no sequence or period effect was found. Discussion We found that peripheral vasodilation in patients with cirrhosis was associated with increased response to NG -monomethyl-L-arginine. This result indicates that nitric oxide has a vasodilatory role. Our major new finding was that basal forearm blood flow and the increased NG -monomethyl-L-arginine response in patients with cirrhosis returned to normal with the fluoroquinolone antibiotic norfloxacin. Peripheral vasodilation in patients with cirrhosis was characterized by significantly higher basal forearm blood flow. Because N (G) -monomethyl-L-arginine is a specific inhibitor of nitric oxide synthase [7], the increased response in these patients suggests increased nitric oxide synthesis. Given the potent vasodilatory action of nitric oxide, this effect is consistent with the increased basal flow that we saw. Patients with cirrhosis have intestinal bacterial overgrowth [3]. This condition, along with portal hypertension and portosystemic shunts, can lead to systemic bacteremia or endotoxemia. This, in turn, may induce isoform II of nitric oxide synthase and lead to increased nitric oxide production and vasodilation [5]. To test this hypothesis, we administered norfloxacin, an agent commonly used to prevent bacterial infections in cirrhosis [8]. Our finding that cirrhotic patients had an augmented response to the endothelium-dependent muscarinic agonist acetylcholine has been reported with methacholine, a similar agonist [9]. These agonists promote release of nitric oxide through activation of endothelial-constitutive isoform III of nitric oxide synthase [2]. Because NG -monomethyl-L-arginine blocks both isoforms of nitric oxide, our finding of enhanced vasoconstriction with NG -monomethyl-L-arginine is consistent with the hypothesis that patients with cirrhosis have enhanced release of endothelial-constitutive isoform III of nitric oxide synthase. On the other hand, because norfloxacin had no effect on enhanced responses to acetylcholine, the nitric oxide responsible for peripheral vasodilation and enhanced response to NG -monomethyl-L-arginine responses seems associated with isoform II of nitric oxide synthase. From this finding, we inferred

Glomerular hyperfiltration in patients with well-compensated alcoholic cirrhosis

BACKGROUND Peripheral and splanchnic arteriolar tone is often decreased in patients with cirrhosis. The responsible circulating vasodilator(s) would be expected to also lower renal vascular resistance. To examine this possibility we have undertaken a hemodynamic study of the renal circulation in patients with stable, well characterized, compensated cirrhosis and in healthy controls of similar age and sex. METHODS Clearance techniques were used to assess splanchnic and renal hemodynamics and hepatocellular function. RESULTS Renal vascular resistance was significantly reduced in the cirrhotic patients (P = 0.048) and was accompanied by a significant (P = 0.014) and proportional (r = -0.45; P = 0.016) increase in glomerular filtration rate. The hepatic extraction of indocyanine green, a measure of functional intrahepatic portasystemic shunts, was the only independent predictor of glomerular filtration rate (r = -0.65; P = 0.002). CONCLUSIONS The results support the hypothesis that, in patients with cirrhosis, the presence of portasystemic shunts results in an increased delivery of endogenous vasodilator(s) into the systemic circulation where their principal action on the renal circulation is to preferentially decrease afferent arteriolar tone. The resultant glomerular hyperfiltration may contribute to the pathogenesis of cirrhotic glomerulosclerosis.

Indomethacin-induced renal dysfunction in patients with well-compensated cirrhosis

BACKGROUND Patients with cirrhosis and ascites are especially sensitive to the adverse renal effects of indomethacin-induced inhibition of prostaglandin synthesis. The aim of this study was to determine whether indomethacin affects renal function in patients with well-compensated cirrhosis. METHODS Clearance techniques were used to assess renal hemodynamics and sodium and water homeostasis. RESULTS The oral administration of 50 mg of indomethacin to well-compensated patients with alcoholic cirrhosis was followed by a significant decrease in glomerular filtration rate (GFR) and effective renal plasma flow because of a preferential increase in afferent arteriolar tone. Indomethacin was both antidiuretic and antinatriuretic due principally to decreased free water clearance and increased proximal tubular reabsorption of sodium. The acute changes in renal function were not sustained. Patients with a high basal GFR were particularly sensitive to the adverse renal effects of indomethacin. CONCLUSIONS This study indicates that in patients with well-compensated cirrhosis renal prostaglandins are functionally active and may contribute to the pathogenesis of glomerular hyperfiltration. Nonsteroidal anti-inflammatory drugs should be used with caution in all patients with cirrhosis.