Effects of Systemic Prostaglandin E1 on Hepatic AminoAcid-Nitrogen Metabolism in Patients With Cirrhosis

ANDREA FABBRI, GIAMPAOLO BIANCHI, MARA BRIZI, ELISABETTA BUGIANESI, DONATELLA MAGALOTTI, MARCO ZOLI,AND GIULIO MARCHESINI

Prostaglandins of the E (PGE) series have long beenconsidered ‘‘catabolic’’ hormones, but recent data suggestthat they may be secreted in critically ill patients tocounteract stress hormones, stimulating protein synthesis.Their use is under scrutiny to improve hepatic microcircula-tion and as cytoprotective agents. We tested the effects ofPGE1 on hepatic and whole-body nitrogen metabolism ineight patients with cirrhosis. Urea-nitrogen synthesis rate,a-amino-nitrogen levels, and nitrogen exchange were mea-sured in the basal, postabsorptive state and in response tocontinuous alanine infusion, in paired experiments, duringsuperinfusion of PGE1 or saline. Splanchnic and systemichemodynamics were assessed by echo-Doppler at the begin-ning and at the end of each experiment. PGE1 produced arapid fall in plasma amino acids and in urea-nitrogensynthesis rate, as well as a positive nitrogen exchange. Theslope of the regression of a-amino-nitrogen levels onurea-nitrogen synthesis rate, a measure of liver cell meta-bolic activity, was not affected, but the regression line wasshifted rightward, suggesting a nitrogen-sparing effect ofPGE1. Mesenteric artery and portal flow were unchanged,whereas femoral artery flow increased by 30%. Insulin andglucagon levels were not systematically different. We con-clude that PGE1 reduces hepatic urea synthesis rate, indepen-dent of hormones and/or hepatic flow, possibly acting at theperipheral level on amino acid transport, thus reducingamino acid supply to the liver. The resulting net nitrogensparing might be the basis for the beneficial effect of PGE1

in clinical hepatology. (HEPATOLOGY 1998;27:815-821.)

Prostaglandins (PGs) are produced locally and act as inter-or intracellular modulators of biochemical processes in the tis-sues where they are formed. They have significant effects onamino acid/protein metabolism, but results are conflicting.1-6

Because of their increase in inflammatory processes andcatabolic stress, it has long been supposed that PGs might be

mediators of accelerated muscle proteolysis during traumaand sepsis.1,2 However, suppression of PG synthesis byindomethacin failed to decrease the elevated protein degrada-tion, usually observed in these patients.3 Whereas cytokines,such as interleukin-1 and tumor necrosis factor, have a roleon net protein breakdown,7 there is a growing line ofevidence arguing that PGs might be secreted to counteractcatabolic mediators, acting synergically with insulin to stimu-late tissue repair, stimulating muscle protein synthesis,4

and/or inhibiting muscle proteolysis.5-6

In the last few years, there has been a rising interest in thetherapeutic use of PGs in clinical hepatology. The hepatocyteshave specific binding sites for PGs of the E series (PGE1 andPGE2).8 Several investigations suggested that PGE may havebeneficial effects in patients with severe acute liver injury,9 aswell as in acute rejection, following liver transplantation.Now significant results were obtained in primary nonfunc-tion.11,12 The mechanism(s) for cytoprotection and improvedliver function are not well defined.13 Changes in splanchnichemodynamics might be involved. In patients with cirrhosis,PGE infusion into the superior mesenteric artery increasedportal and hepatic blood flow,14 whereas inhibitors of PGEsynthesis decreased hepatic blood flow.15 No data are avail-able on possible direct effects of PGs on hepatocellularfunction and metabolic activities.

The control of urea synthesis is a key point in whole-bodynitrogen economy in humans. The amount of nitrogendisposed of in the liver under hormonal control may directlyaffect peripheral nitrogen metabolism, favoring or blockingthe outpouring of amino acid-N from peripheral tissues.16

The process of urea-N formation may be quantified, afterstandardization for substrate availability during amino aciddrive, by the slope of the regression of urea-N synthesis rateon the corresponding a-amino-nitrogen concentration, theso-called functional hepatic nitrogen clearance (FHNC).16

Under such experimental conditions, the process is indepen-dent of hepatic blood flow. The technique proved useful to studythe effects of disease, hormones, drugs, and dietary changes onthe dynamics of amino acid–derived urea synthesis. Also, PGshave been tested. Experimental studies suggested that they mayhave a permissive influence on surgical stress-induced ureasynthesis,17,18 but no data were obtained in humans.

In the present study, we assessed the effects of short-terminfusion of a PGE1 analogue (alprostadil-a-ciclodestrin) onthe hepatic conversion of amino-N and on urea synthesis in agroup of patients with cirrhosis, under controlled conditionsof substrate availability induced by continuous amino acidinfusion. Splanchnic and peripheral hemodynamics were alsomeasured to test PG activity on circulatory parameters and

Abbreviations: PG, prostaglandin; FHNC, functional hepatic nitrogen clearance;UNSR, urea-nitrogen synthesis rate; TBW, total body water.

From the Dipartimento di Medicina Interna, Cardioangiologia, Epatologia andCattedra di Malattie del Metabolismo, Universita di Bologna, Policlinico S. Orsola,Bologna, Italy.

Received April 22, 1997; accepted November 18, 1997.Supported by Funds from Ministero dell’Universita e della Ricerca Scientifica

(MURST), Rome, Italy, Fondi ex 60%, 1997.Address reprint requests to: Giulio Marchesini, M.D., Dipartimento di Medicina

Interna, Cardioangiologia, Epatologia, Universita di Bologna, Policlinico S. Orsola, 9,Via Massarenti, I-40138 Bologna, Italy. Fax: 39-51-340877.

Copyright r 1998 by the American Association for the Study of Liver Diseases.0270-9139/98/2703-0025$3.00/0

815

the possible relationship between hemodynamic changes andmetabolic effects.

PATIENTS AND METHODS

Subjects

Eight patients with histologically documented cirrhosis andstable clinical conditions were studied. All patients were men, aged46 to 66 years (median, 58 years), with cirrhosis of alcoholic (fourcases) or hepatitis C virus origin (four cases). Their clinical andlaboratory data, including the galactose elimination capacity19 andChild-Pugh class,20 are reported in Table 1.

Seven patients had esophageal varices; the last patient hadsuffered from episodes of variceal bleeding 6 months before thestudy, had been repeatedly treated with endoscopic sclerosis, andvarices were completely eradicated. Three patients had mild ascitesat ultrasonography, which was not clinically evident at the time ofexperiments, but all were treated with diuretics (spironolactone[100-200 mg/d] and/or furosemide [25 mg/d]).

Clinical evidence of encephalopathy was present in one case, butall were actively treated with lactulose (15-30 g/d). Patients withalcoholic cirrhosis had abstained from alcohol for at least 1 year beforethe study, and two cases had stopped drinking alcohol 2 years before.

All subjects were in fairly good nutritional conditions, the bodymass index being on average 25.3 [2.9] kg/m2 (range, 22.5-30.3kg/m2). Renal function was in the normal range (plasma creatinine,1.3 mg/dL), and there was no evidence of previous or actualendocrine diseases and/or complicating disorders.

All studies were performed in the course of hospital admission.During the study period and in the 3 days before the study, allpatients consumed a standard hospital diet to provide 30 to 35 kcaland 0.8 to 1.0 g protein per kilogram of body weight.

Subjects gave informed consent to take part in the study, whichwas submitted to and approved by the Ethical Committee forHuman Studies operating in our hospital.

Methods

Metabolic Study. Urea-N synthesis rate (UNSR) was measuredusing a simplified procedure21 in the course of five consecutive90-minute periods, including the fasting postabsorptive state andduring variable a-amino-N concentrations attained by intravenousalanine infusion, as described by Vilstrup16 (Fig. 1). Alanine(Ajinomoto Co. Inc., Tokyo, Japan; 10% wt/vol water solution) wasinfused at a constant rate of 2 mmol/kg/h for 4.5 hours (from time 08to 2708). Blood samples were obtained from a vein of the contralat-eral arm every 45 minutes, starting 90 minutes before alanineinfusion (time 2908). A final blood sample was obtained 90 minutesafter alanine infusion stop (time 3608). Urine was collected quantita-tively by voiding in the five consecutive periods (every second bloodsampling). Subjects were not fed during the course of the test.

The whole study was performed under superinfusion of saline(control experiment) or a PGE1 analogue (Alprostadil-a-ciclode-strin, Schwarz Pharma AG, Monheim, Germany), in random pairedexperiments performed at 3- to 5-day intervals. PGE1 was infused ata constant rate of 30 µg/h in isotonic solution throughout theexperimental period. The infusion rate was calibrated to yieldplasma concentrations of 5 to 6 pg/mL, i.e., approximately threetimes basal values (normal plasma levels: 1.2-1.8 pg/mL).22

During the experimental period, urine flow was stimulated byperoral water or saline infusion, to keep diuresis above 2 mL/min.This was attained in all subjects (mean diuresis: 3.0 [1.1] mL/min).The total amount of water or saline given in paired experiments wasapproximately the same, i.e., <2,000 mL, and the differencebetween paired experiments was ,200 mL. Mild fluid retention wasobserved in a few experiments, but it never exceeded 1 L (,2.5% ofbody water).

There were no side-effects or complications in relation to theinfusion of alanine. In relation to PGE1 infusion, one patient showedprogressive flushing and two experienced erythema over the infusedvein, but the effects were well tolerated and the experiments werecompleted.

The UNSR during each 90-minute period was measured as thesum of urea-N excretion rate in urine and accumulation of urea-N inthe urea space, assumed to equal total body water (TBW), as16:

UNSR 5 (E 1 A)/(1 2 L),

where E 5 (urine flow [L/h]) 3 (urinary urea-N [mmol/L]); A 5(change in blood urea-N [mmol/L/h]) 3 (TBW [L]); L 5 (fractionalloss of newly formed urea in the gut). TBW was calculated by meansof a nomogram, based on age, height, and weight.23 Intestinal loss ofurea-N caused by bacterial hydrolysis was taken to be 0.26.24

In each experiment, FHNC was calculated as the slope of thelinear regression of UNSR on the corresponding average a-amino-Nconcentration during each time period (mean of a-amino-N valuesmeasured at the beginning and at the end of each urine collection).

The stoichiometric balance between infused amino acid-N andurea-N appearance rate (the nitrogen exchange in millimoles perhour) was calculated as the difference between the alanine-Ninfusion rate, corrected for urinary a-amino-N excretion anda-amino-N accumulation, and urea-N appearance (urea-N excretion 1urea-N accumulation in TBW, not corrected for gut hydrolysis). Thevolume of distribution of a-amino-N was assumed to equal TBW.

In our laboratory, normal values of galactose elimination capacityare .6.0 mg/kg/min,25 and FHNC is .25 L/h.21 Repeated measure-ments of the two tests in the same subject vary within 610% and615%, respectively.21

Hemodynamic Study. The echo-Doppler measurements of splanch-nic and peripheral hemodynamics were performed before thebeginning of the experiment and at the end of alanine infusion,

TABLE 1. Clinical and Laboratory Data of Patients With Cirrhosis

CaseNo.

Age(yr)

Etiology ofCirrhosis

Albumin(g/L)

ProthrombinActivity (%)

Bilirubin(mg/dL)

Galactose Elimination(mg/kg/min)

Child-Pugh(score)

Ammonia(mmol/L)

Varices(grade)*

1 59 Alcohol 30.0 28 8.2 3.2 11 48 I2 58 HCV 25.0 30 1.7 3.9 10 21 III3 51 Alcohol 31.8 59 1.6 4.0 7 21 I4 66 Alcohol 36.6 61 1.5 3.5 7 34 I5 46 HCV 36.7 62 1.6 3.7 6 21 I6 61 HCV 26.2 54 0.9 4.3 8 19 07 46 Alcohol 39.5 65 2.2 3.3 7 19 I8 59 HCV 30.7 50 3.3 3.4 8 31 IIMean [SD] 32.0 [5.3] 57 [13] 2.6 [2.4] 3.7 [0.4] 8 [2] 27 [10]Normal values .35.0 .80 ,1.0 .6.0 — ,35 —

Abbreviation: HCV, hepatitis C virus.*0, absent; I, small; II, medium; and III, large.

816 FABBRI ET AL. HEPATOLOGY March 1998

using an equipment that combines a real-time electronic sectorscanner and an echo-Doppler unit (AU4-Idea, ESAOTE Ansaldo,Genova, Italy), as previously reported.26 We measured blood veloc-ity and flow in the superior mesenteric artery and in the portal vein.Heart rate, mean arterial pressure, and femoral artery velocity andflow were also recorded. In our department, the intra-assay andinterassay variability in the measurement of splanchnic flows iswithin 8.0% and 7.4%, respectively.

Laboratory Procedures

Urea-N concentration in plasma and urine was measured by theurease Berthelot method.27 Total a-amino-N was assayed by thedinitrofluorobenzene method.28 All analyses were performed inbatches, in duplicate or triplicate, to minimize the analytical error.The intra-assay coefficients of variation are: urea, 61.5%; a-amino-N, 62%. Plasma amino acid profile was measured byninhydrin reaction after ion-exchange chromatography29 at baseline(2908) and at the beginning (08) and the end of alanine infusion(2708), with a coefficient of variation ,5%.

Plasma insulin was measured by an immunoenzymometric assay(AIA-PACK IRI, AIA-1200 system, Tosoh Co., Tokyo, Japan). Plasmaglucagon was measured by radioimmunoassay (Glucagon kit, Bio-data-Serono, Guidonia, Italy). Glucose was measured enzymatically.

Galactose was determined enzymatically (Test Combination Ga-lactose, Boehringer GmbH, Mannheim, Germany).

Statistical Analysis

All analyses were performed on a personal computer by means ofStatView II program (Abacus Concepts, Inc., Berkeley, CA). Differ-ences between paired data were analyzed by the paired t test.Differences in serial determination of the same parameters in pairedexperiments were also tested for significance using repeated-measures ANOVA. Linear correlation analysis between variables wasperformed by the least-squares method. Data in the text and in thetables are given as mean [SD], whereas, in the figures, they are givenas means 6 2 SE.

RESULTS

Metabolic Study

Control Experiment. Fasting values of plasma a-amino-N,glucose, insulin, and ammonia at the beginning of alanineinfusion were in the normal range. Glucagon was increasedby approximately 50% in comparison with values measuredin our laboratory in healthy subjects (Table 2). Alanineinfusion increased a-amino-N levels fourfold; glucose andinsulin did not change significantly, whereas ammonia andglucagon doubled.

Plasma urea-N decreased slightly in the fasting preinfusionperiod, and increased thereafter during nitrogen infusion,reaching a maximum value of 16.7 [SD 2.8] mmol/L soonafter the end of alanine infusion (Fig. 2). Also, urinary urea-Nexcretion progressively increased to a maximum value of 67[20] mmol/h in the last observation period. As a result, UNSRincreased to a maximum value of 146 [22] mmol/h, and, inindividual experiments, it strictly correlated with the averagea-amino-N level (r values ranging from 0.77 to 0.99). FHNCwas on average 22.1 [5.2] L/h (Fig. 3). In individual subjects,the theoretical a-amino-N concentration corresponding toUNSR 5 0 varied from 0.75 to 3.03 mmol/L (on average 1.45[0.74] mmol/L) (Fig. 4).

In the fasting, preinfusion period, in the absence of anyextra nitrogen delivery, the apparent N exchange was nega-tive, due to urea-N excretion exceeding the negative plasmaurea-N and a-amino-N accumulation (Fig. 5, Table 3). Atpeak a-amino-N concentrations in the last period of alanineinfusion, the apparent N exchange was slightly positive,because N infusion exceeded the sum of urea-N excretion,plasma urea-N accumulation, and plasma a-amino-N accumu-lation (Table 3).

Individual plasma amino acid concentrations (not reportedin detail) decreased slightly in the fasting preinfusion period(2908 to 08), and did not change systematically in the courseof alanine infusion, from time 08 to 2708. Their sum,excluding alanine, increased slightly from 2.76 [0.36] to 2.97[0.24] mmol/L (not significant).

PG Infusion. After 90 minutes of PGE1 infusion, glucoseincreased by nearly 1 mmol/L, and remained 0.5 to 1.0mmol/L higher throughout the experimental period. Basalglucagon at the beginning of alanine infusion was notdifferent from values measured in the control experiment, butthe glucagon response to alanine was 30% higher. Insulin wasin the normal range and did not change. a-Amino-N concen-

FIG. 1. Experimental protocol of paired studies of urea-N synthesis ratein response to alanine infusion with/without PGE1 infusion.

TABLE 2. Glucose, a-Amino-N, Insulin, and Glucagon Concentrations at Baseline (time 2908), at the Beginning (time 08), and at the End of AlanineInfusion (time 2708) in the Course of the Paired Experiments Performed in Cirrhotic Patients With/Without PGE1 Infusion

Control Experiment Prostaglandin Infusion

Time 2908 Time 08 Time 2708 Time 2908 Time 08 Time 2708

a-Amino-N 3.2 [0.4] 3.1 [0.3] 8.6 [1.3]* 3.2 [0.2] 2.8 [0.3]† 8.6 [1.5]*Ammonia ND 27 [10] 48 [30]* ND 25 [0.6] 51 [25]*Glucose 5.5 [1.5] 5.3 [0.8] 5.0 [0.4] 5.5 [1.0] 6.1 [1.1]† 6.0 [1.0]†Insulin ND 114 [70] 177 [96]* ND 112 [65] 160 [112]Glucagon ND 79 [27] 136 [41]* ND 87 [14] 181 [42]*†

NOTE. Data are mean [SD]. Glucose and a-amino-nitrogen concentrations are in mmol/L; insulin and glucagon are in pmol/L; and ammonia is in µmol/L.Abbreviation: ND, not determined.*Significantly different from time 0 value.†Significantly different from the corresponding value in the control experiment.

HEPATOLOGY Vol. 27, No. 3, 1998 FABBRI ET AL. 817

trations were 0.4 to 0.5 mmol/L lower than those observed inthe control experiment.

PGE1 infusion produced a prompt decrease in plasma urea(by 1.4 mmol/L) (Fig. 2). During alanine infusion, plasmaurea-N increased progressively, but values remained nearly1.0 mmol/L lower than those observed in the control experi-ment. Unexpectedly, urinary urea-N excretion also was lowerduring most periods (Fig. 2), as was UNSR (by 20 to 40mmol/h).

FHNC did not change significantly in comparison withvalues measured in the control experiment (Fig. 3), but therelationship was rightward shifted (Fig. 4). In individual

subjects, the theoretical a-amino-N concentration correspond-ing to UNSR 5 0 varied from 1.72 to 4.63 mmol/L (onaverage 3.07 [1.22] mmol/L; P vs. control experiment: ,.01).

The apparent N exchange, measured in the fasting preinfu-sion period in the absence of any N supply, was positive(Table 3, Fig. 5), due to the significantly larger plasma urea-Ndeaccumulation and a 30% decrease in urinary urea-Nexcretion. At peak a-amino-N concentration, in the lastperiod of alanine infusion, the apparent N exchange wassignificantly more positive than in the absence of PGE1, thedifference being mostly due to a 35% decrease in plasmaurea-N accumulation.

Plasma amino acid profile at baseline (2908) was notdifferent from that observed in the control experiment. PGE1

infusion produced a prompt reduction of most acidic and

FIG. 2. Time-course of plasma urea-N concentration and urinary urea-Nexcretion in cirrhosis in the control experiment (saline infusion, [s] and[h]) or PGE1 infusion (j). The plasma urea-N time courses are significantlydifferent (time 3 treatment, repeated-measurement ANOVA: P 5 .0007).*Statistically different from the control experiment (P , .05).

FIG. 3. FHNC in the control experiment (saline infusion) and inresponse to PGE1 infusion. Data of individual patients are connected by aline. The horizontal bars represent mean values.

FIG. 4. The dynamics of a-amino-N to urea-N conversion in patientswith cirrhosis, in relation to (s) saline or (j) PGE1 infusion. UNSRincreases with increasing a-amino-N concentrations, and the slope of theregression is the FHNC. The continuous lines represent the regression,calculated on the basis of average a and b coefficients, in a range ofa-amino-N concentrations attained in the course of the experiments. Theequations of the regressions are respectively: saline infusion: UNSR 5 230.8[13.0] 1 22.1 [5.2] 3 a-amino-N; PGE1 infusion: UNSR 5 269.9 [37.9] 123.2 [7.1] 3 a-amino-N. Paired analysis reveals no differences in the slope(b), whereas the a coefficient is significantly lower during PGE1 infusion(P , .05).

FIG. 5. Apparent N exchange in the (s, left) control experiment and inresponse to (j, right) PGE1 infusion during the basal period, before alanineinfusion, and at peak a-amino-N concentrations, at the end of alanineinfusion. Horizontal bars represent mean values. PGE1 infusion significantlyincreased N exchange in all subjects, and mean values in paired experiments,both in the basal period and at peak a-amino-N concentration, weresignificantly different (P , .05).

818 FABBRI ET AL. HEPATOLOGY March 1998

neutral amino acids. At time 08, they were lower in compari-son with the control experiment. In particular, taurine wassignificantly lower by 66%, alanine by 44%, methionine by32%, glutamine by 28%, arginine by 26%, and the sum ofbranched-chain amino acids by 13% (P vs. control experi-ment: ,.05). In the course of alanine infusion, individualplasma amino acid concentrations (excluding alanine) in-creased, their sum changing from 2.09 [0.26] mmol/L to 2.64[0.26] (P , .01).

Hemodynamic Study

Portal venous and mesenteric artery flow did not change inthe course of the control experiment (Table 4), whereasfemoral artery flow increased by only 50 mL/min (,10%),possibly as a result of infusion-increased extracellular vol-ume.

PGE1 infusion produced a slight increase in heart rate, anonsignificant reduction in mean arterial pressure, a signifi-cant increase in femoral artery blood flow (by nearly 30%),and no effect on mesenteric artery and portal blood flow.

DISCUSSION

The present study shows that PGE1 infusion significantlyinfluences amino acid-N economy in patients with cirrhosis,leading to a significant nitrogen sparing. These effects do not

depend on changes in hepatic hemodynamics, or on hormone-mediated regulation of urea synthesis.

These conclusions are based on the changes in plasmalevels and urinary excretion of urea-N and on total bodyN-exchange, suggesting that, during PGE1 infusion, theproduction of urea in the liver is reduced. If the block were athepatic level, then plasma amino acids would accumulate inplasma, both in the fasting state and in response to alanine.Because amino acids were reduced in comparison with thecontrol experiment, PGE1 is likely to act through a block inamino acid outpouring from both the liver and peripheraltissues, and/or via an increased amino acid uptake for proteinsynthesis.

The methodology of the present study, i.e., the measure-ment of the dynamics of hepatic urea synthesis duringstandardized conditions of substrate availability,16 is the samepreviously used to measure the effects of hormones or drugsin patients with cirrhosis30,31 and in other physiological32 andpathological conditions.17,33-35 The assumptions underlyingthe technique have been dealt with extensively in previousarticles.16,25 In the calculation of UNSR, intestinal hydrolysiswas considered a fixed fraction of total urea-N excretion onthe basis of the average values derived from the literature.24 Inour study, all patients at the time of study were takinglactulose, which reduces gut urea hydrolysis.36 This may leadto overestimation of urea synthesis rate, but it is not likely tobe relevant in paired experiments, because no changes inlactulose treatment occurred, and short-term PGE1 infusionis not expected to affect intestinal hydrolysis of urea per se.

PGs were reported to increase renal blood flow in experi-mental animals,37 and there was evidence of improved renalfunction in cirrhosis following misoprostol administration.38

Renal flow was not measured in the present study, butincreased renal flow would be expected to increase urinaryurea excretion. This did not happen in the present study;urine flow was not different in paired experiments, andurinary urea-N excretion was on average reduced by 25% byPGE1 infusion (33.5 [SD 10.4] mmol/h vs. 43.9 [9.8] in thecontrol experiments; P , .05).

The process of urea synthesis is linearly dependent onsubstrate availability. The present methodology was devel-oped to study nitrogen metabolism as the regression betweena-amino-N and urea-N synthesis rate at different steps ofsubstrate availability; the slope measures the efficiency of theliver to transform amino-N into urea-N. PGE1 infusion didnot significantly change the slope, but produced a significantrightward shift of the dose-response curve (Fig. 3), mathemati-cally expressed by an increase of the a-amino-N concentra-tion corresponding to UNSR 5 0, i.e., the theoretical plasmalevels of nitrogen substrates for which the process of ureasynthesis is switched off. Whereas, in the absence of PGE1

urea-N synthesis is maintained unless a-amino-N levels donot fall below 1.45 mmol/L, during PGE1 infusion, theprocess comes to a stop at a-amino-N levels over 3 mmol/L.Such levels are in the physiological range in fasting subjects,when a-amino-N concentrations are maintained solely byproteolysis. Changes in the dynamics of the process by simpleshift of the relationship between a-amino-N and UNSR havebeen previously shown in humans in response to catabolicevents, such as increased thyroid hormones34 or growthhormone deprivation,39 where the curve is left-shifted. Therightward shift produced by PGE1 infusion is clearly anexample of a metabolic state prone to preserve nitrogen both

TABLE 3. Calculation of the Apparent Nitrogen Exchange in the FastingState, Before Alanine Infusion, and at Peak Alanine Concentrations in

Paired Experiments With/Without PGE1 Infusion

Control Experiment Prostaglandin Infusion

Basal Period Peak Alanine Basal Period Peak Alanine

N infusion rate — 155 [23] — 155 [23]Plasma a-amino-N accu-

mulation 210 [4] 40 [16] 210 [6] 32 [24]Plasma urea-N accumu-

lation 219 [9] 61 [9] 244 [13]* 40 [32]Urinary urea-N excretion 37 [13] 47 [15] 27 [12] 43 [24]Apparent N exchange 28 [6] 7 [14] 27 [12]* 40 [10]*

NOTE. Data are mean [SD] in mmol/h.*Significantly different from the corresponding value in the control

experiment.

TABLE 4. Heart Rate, Mean Arterial Pressure, and Peripheral andSplanchnic Hemodynamic Values in Paired Experiments With/Without

PGE1 Infusion

Control Experiment Prostaglandin Infusion

Before End Before End

Heart rate (beats/min) 73 [18] 73 [19] 72 [19] 77 [17]*

Mean arterial pres-sure (mm Hg) 91 [2] 94 [2] 90 [7] 86 [6]

Femoral artery flow(mL/min) 821 [269] 869 [242]* 812 [262] 1,039 [230]*†

Mesenteric arteryflow (mL/min) 662 [128] 647 [114] 643 [111] 698 [137]

Portal flow (mL/min) 1,423 [580] 1,367 [624] 1,392 [648] 1,270 [600]

NOTE. Data are mean [SD].*Significantly different from the corresponding preinfusion value.†Significantly different from the corresponding value in the control

experiment.

HEPATOLOGY Vol. 27, No. 3, 1998 FABBRI ET AL. 819

at low a-amino-N concentrations, when urea formation iscompletely suppressed, as well as during nitrogen supply,when less urea is formed at any a-amino-N concentration.

The metabolic effects of PGE1 were confirmed by comput-ing the apparent N exchange, a measure of the total amountof nitrogen irreversibly transferred from the a-amino-N poolinto the urea-N pool, where nitrogen can no longer be reused.This calculation is limited by non–steady-stateness, and maygive only a qualitative idea of total nitrogen losses. Asexpected, basal N exchange, measured in the period beforealanine infusion, was negative, because no extra nitrogen wasgiven to meet urinary urea-N losses. During alanine infusion,the apparent N exchange also remained around 0 at peaka-amino-N concentrations (Fig. 5), as previously demon-strated in cirrhosis.40 This is indicative of any metaboliccondition characterized by increased catabolism, amino acidconsumption, and nitrogen wasting. During PGE1 infusion,the apparent N exchange was positive, also in the basal periodwhen no amino acid-N was supplied. This was obtainedthrough the combined effects of plasma urea-N deaccumula-tion (expressed by a significant reduction of plasma urea-Nconcentration) and lower-than-normal urinary urea-N excre-tion. It is concluded that PGE1 infusion promptly reducesurea formation, and the effect is maintained during alanineinfusion, because the apparent N exchange also was signifi-cantly more positive at peak a-amino-N concentrations.

Plasma amino acids decreased significantly in the first 90minutes of PGE1 infusion, reaching a level significantly lowerthan in the control experiment. Stiegler et al.6 measured thearterial-deep venous difference of plasma amino acids duringarterial infusion of PGE1 in healthy subjects. In their experi-ence, fasting amino acid balance became significantly positiveduring PGE1 infusion, in keeping with an inhibition ofproteolysis or a stimulation of protein synthesis. Experimen-tal data showed that PGE2 increase the intracellular transportof glutamine by activation of the hepatic system N, as well asthe transport of small neutral amino acids, e.g., alanine, byactivation of the ubiquitous system A.41 Similar data were alsoobtained on system y(1)-mediated L-arginine transport.42

These results strongly support our finding of positive appar-ent N exchange. Whether this is caused by decreased aminoacid efflux or, more likely, increased amino acid uptake andnonoxidative disposal, and whether the primary site is theliver or peripheral tissues, requires further investigation.Because the efficiency of the liver in the conversion of aminoacids into urea—expressed by the slope of the regressionbetween a-amino-N and urea-N synthesis rate—is un-changed, a peripheral mechanism seems more likely.34,39 Thishypothesis is strongly supported by the significant reductionof branched-chain amino acids, which are expected to escapeliver uptake and are mainly used in the periphery.43

Hormones are known to regulate the kinetics of the processand might theoretically be responsible for the effects of PGE1

on urea synthesis.16 Glucagon is the most potent stimulatorydrive in normal subjects,44 and probably also mediates theeffects of other hormones, namely cortisol and catechol-amines. It possibly stimulates urea synthesis through anincreased hepatocellular amino acid uptake,44 but, in patientswith cirrhosis, such an effect is blunted or absent.33 PGE1

infusion did not change basal preinfusion glucagon levels,and the higher-than-normal alanine-stimulated glucagon con-centrations at the end of infusion might be secondary to ahepatic resistance to glucagon stimulation.

Changes in splanchnic hemodynamics might theoreticallybe involved in the metabolic effects of PGE1. Experimentaldata showed that pharmacologically or mechanically in-creased liver perfusion leads to an improvement in liverfunction.45 In patients with cirrhosis, PG infusion into thesuperior mesenteric artery was shown to increase portal andhepatic vein flow,14 whereas indomethacin administration (aninhibitor of PG synthesis) decreased hepatic blood flow.15

According to the kinetic characteristics of urea formation, itcan be calculated that any changes in splanchnic perfusionmay only affect FHNC by less than 5%.46 In our study,systemic PGE1 infusion significantly affected the systemiccirculation, in keeping with the well-documented activity ofPGE1 at a similar dose,47 but no significant effects onsplanchnic blood flow were detected. It might be argued thatthe splanchnic vasodilatation previously observed followingintra-arterial infusion14 could not be obtained by systemicinfusion, and/or that the hyperdynamic circulatory system ofcirrhosis might prevent any selective effect on splanchnicvessels. More interestingly, the large effect on systemiccirculation (30% increase in femoral artery flow), associatedwith the positive amino acid balance previously demon-strated,6 might be an additional argument supporting thehypothesis that the metabolic effects of PGE1 on nitrogenmetabolism are mediated at the peripheral level.

In humans, PGs have been proposed in solid organtransplantation to reduce the need of intensive care support11

and hospital stay.12 The drug is cost-effective, leading to netsavings of money and hospital charges.48 In other studies,PGs have been reported to improve graft function10 and toreduce primary nonfunction after orthotopic liver transplan-tation49 because of favorable effects on microcirculatorysystem and accelerated liver regeneration processes. Suchregenerative processes might be the results of the nitrogensparing observed in our experiments. Along this line are alsostudies showing that PGEs block hepatic glycogenolysis andglucose production induced by stress hormones in rats,50 inkeeping with an insulino-mimetic activity of PGE1.

The favorable effects on nitrogen economy of patients withcirrhosis presented in this study have potential clinicalrelevance. Data are needed to determine whether the resultsonly apply to cirrhosis, or they are part of a generalphysiological activity also occurring in normal subjects andin disease states. Other studies should also clarify themechanism(s) underlying the nitrogen-sparing effect of PGE1

in cirrhosis, as well as their potential metabolic effects inlong-term clinical studies.51

REFERENCES

1. Baracos V, Rodemann HP, Dinarello CA, Goldberg AL. Stimulation ofmuscle degradation and prostaglandin E2 release by leucocytic pyrogen(interleukin-1). A mechanism for the increased degradation of muscleproteins during fever. N Engl J Med 1983;308:553-558.

2. Goldberg AL, Kettelhut IC, Furuno K, Fagan JM, Baracos V. Activation ofprotein and prostaglandin E2 production in rat skeletal muscle in fever issignaled by a macrophage product distinct from interleukin 1 or otherknown monokines. J Clin Invest 1988;81:1378-1383.

3. Wahren J, Euler C, Eriksson LS. Effect of indomethacin on leg exchangeof amino acids in septic traumatized patients. Clin Nutr 1988;7:203-206.

4. Hasselgren PO, Warner BW, Hummel III RP, James JH, Ogle CK, FisherJE. Further evidence that accelerated muscle protein breakdown duringsepsis is not mediated by prostaglandin E2. Ann Surg 1988;207:399-403.

5. McKinley CJ, Turinsky J. Prostaglandin E2 and muscle proteolysis: effectof burn injury and cycloheximide. Am J Physiol 1986;250:R207-R210.

820 FABBRI ET AL. HEPATOLOGY March 1998

6. Stiegler H, Rett K, Wicklmayr M, Mehnert H. Metabolic effects ofprostaglandin E1 on human skeletal muscle with special regard to theamino acid metabolism. VASA 1989;28(Suppl):14-18.

7. Sakurai Y, Zhang XJ, Wolfe RR. Effect of tumor necrosis factor onsubstrate and amino acid kinetics in conscious dogs. Am J Physiol1994;266:E936-E945.

8. Dutta-Roy A, Gopalswamy N, Trulzsch DT. Prostaglandin E1 binds to Zprotein of rat liver. Eur J Biochem 1987;162:615-619.

9. Sinclair SB, Greig PD, Blendis LM, Abecassis M, Roberts EA, Phillips MJ,Cameron R, et al. Biochemical and clinical response of fulminant viralhepatitis to administration of prostaglandin E. J Clin Invest 1989;84:1063-1069.

10. Tancharoen S, Jones RM, Angus PW, Hardy KJ. Beneficial effect ofprostaglandin E1 on hepatic allograft rejection following orthotopic livertransplantation. Transplant Proc 1993;25:2890-2891.

11. Klein AS, Cofer JB, Pruett TL, Thuluvath PJ, McGory R, Uber L,Stevenson WC, et al. Prostaglandin E1 administration following ortho-topic liver transplantation: a randomized prospective multicentre trial.Gastroenterology 1996;111:710-715.

12. Henley KS, Lucey MR, Normolle DP, Merion RM, McLaren ID, CriderBA, Mackie DS, et al. A double-blind, randomized, placebo-controlledtrial of prostaglandin E1 in liver transplantation. HEPATOLOGY 1995;21:366-372.

13. Quiroga J, Prieto J. Liver cytoprotection by prostaglandins. PharmacolTher 1993;58:67-92.

14. Nishida O, Moriyasu F, Nakamura T, Ban N, Tamada T, Kawasaki T,Uchino H. Relationship between splenic and superior mesenteric venouscirculation. Gastroenterology 1990;98:721-725.

15. Bruix J, Bosch J, Kravetz D, Mastai R, Rodes J. Effects of prostaglandininhibition on systemic and hepatic hemodynamics in patients withcirrhosis of the liver. Gastroenterology 1985;88:430-435.

16. Vilstrup H. On urea synthesis—regulation in vivo. Dan Med Bull1989;36:415-429.

17. Heindorff H, Almdal TP, Vilstrup H. Indomethacin prevents the increasein urea synthesis capacity and the weight loss after hysterectomy in rats.Clin Nutr 1990;9:103-107.

18. Heindorff H, Almdal TP, Tygstrup N, Vilstrup H. Effect of surgical stressand ‘‘stress hormones’’ on urea synthesis in perfused rat livers. Surg ResComm 1993;14:71-78.

19. Tygstrup N. Determination of the hepatic elimination capacity (Lm) ofgalactose by single injection. Scand J Clin Lab Invest 1966;18:118-125.

20. Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R.Transection of the oesophagus for bleeding oesophageal varices. Br JSurg 1973;60:646-649.

21. Bianchi GP, Marchesini G, Vilstrup H, Fabbri A, De Mitri MS, Zoli M, PisiE. Hepatic amino-nitrogen conversion to urea-nitrogen in controlsubjects and in patients with cirrhosis: a simplified method. HEPATOLOGY

1991;13:460-466.22. Cawello W, Leonhardt A, Schweer H, Seyberth HW, Bonn R, Leon Lomeli

A. Dose proportional pharmachokinetics of alprostadil (prostaglandinE1) in healthy volunteers following intravenous infusion. Br J ClinPharmacol 1995;40:273-276.

23. Døssing M, Poulsen HE, Andreasen PB, Tygstrup N. A simple method fordetermination of antipyrine clearance. Clin Pharmacol Ther 1982;32:392-396.

24. Hansen BA, Vilstrup H. Increased intestinal hydrolysis of urea in patientswith alcoholic cirrhosis. Scand J Gastroenterol 1985;20:346-350.

25. Marchesini G, Bua V, Brunori A, Bianchi GP, Pisi P, Fabbri A, Zoli M, et al.Galactose elimination capacity and liver volume in aging man. HEPATOL-OGY 1988;8:1079-1083.

26. Zoli M, Magalotti D, Bianchi GP, Ghigi G, Orlandini C, Grimaldi M,Marchesini G, et al. Functional hepatic flow and Doppler-assessed totalhepatic flow in control subjects and in patients with cirrhosis. J Hepatol1995;23:129-134.

27. Fawcett JK, Scott JE. A rapid precise method for the determination ofurea. J Clin Pathol 1960;13:156-159.

28. Goodwin JF. Spectrophotometric quantitation of plasma and urinaryamino nitrogen with fluorodinitrobenzene. Stand Meth Clin Chem1970;6:89-98.

29. Marchesini G, Bianchi GP, Vilstrup H, Checchia GA, Patrono D, Zoli M.Plasma clearances of branched-chain amino acids in control subjects andin patients with cirrhosis. J Hepatol 1987;4:108-117.

30. Fabbri A, Marchesini G, Bianchi GP, Bugianesi E, Bortoluzzi L, Zoli M,Pisi E. Unresponsiveness of hepatic nitrogen metabolism to glucagoninfusion in patients with cirrhosis: dependence on liver cell failure.HEPATOLOGY 1993;18:28-35.

31. Hamberg O, Vilstrup H. Effects of glucose on hepatic conversion ofaminonitrogen to urea in patients with cirrhosis: relationship to gluca-gon. HEPATOLOGY 1994;19:45-54.

32. Hamberg O, Nielsen K, Vilstrup H. Effects of an increase in proteinintake on hepatic efficacy for urea synthesis in healthy subjects and inpatients with cirrhosis. J Hepatol 1992;14:237-243.

33. Marchesini G, Fabbri A, Bianchi GP, Motta E, Bugianesi E, Urbini D,Pascoli A, et al. Hepatic conversion of amino nitrogen to urea nitrogen inhypothyroid patients and upon L-thyroxine therapy. Metabolism 1993;42:1263-1269.

34. Marchesini G, Fabbri A, Bianchi GP, Motta E, Giovanelli P, Urbini D,Lodi A. Hepatic conversion of amino nitrogen into urea in thyroiddiseases. II. A study in hyperthyroid patients. Metabolism 1994;43:1023-1029.

35. Almdal T, Jensen T, Vilstrup H. Control of non–insulin-dependentdiabetes mellitus partially normalizes the increase in hepatic efficacy forurea synthesis. Metabolism 1994;43:328-332.

36. Weber FL Jr, Banwell JG, Fresard KM, Cummings JH. Nitrogen in fecalbacterial, fibers and soluble fractions of patients with cirrhosis: effect oflactulose plus neomycin. J Lab Clin Med 1987;110:259-263.

37. Arendshorst WJ, Johnston PA, Selkurt EE. Effect of prostaglandin E1 onrenal hemodynamics in nondiuretic and volume-expanded dogs. Am JPhysiol 1974;226:218-225.

38. Fevery J, Van Cutsem E, Nevens F, Van Steenbergen W, Verberckmoes R,De Groote J. Reversal of hepatorenal syndrome in four patients byperoral misoprostol (prostaglandin E1 analogue) and albumin adminis-tration. J Hepatol 1990;11:153-158.

39. Wølthers T, Grøfte T, Jørgensen JOL, Møller M, Vahl N, Christiansen JS,Vilstrup H. Effects of growth hormone (GH) administration on func-tional hepatic nitrogen clearance: studies in normal subjects andGH-deficient patients. J Clin Endocrinol Metab 1994;78:1220-1224.

40. Marchesini G, Fabbri A, Bugianesi E, Bianchi GP. Glucagon and aminoacid disposal in normal man and in patients with cirrhosis. Ital JGastroenterol 1993;25:93-98.

41. Plumley DA, Watkins K, Bode BP, Pacitti AJ, Souba WW. Cyclo-oxygenase blockade abrogates the endotoxin-induced increase in Na1-dependent hepatic amino acid transport. JPEN 1995;19:9-14.

42. Watkins KT, Copeland EM, Souba WW. Induction of hepatic system y(1)-mediated L-arginine transport by prostaglandin E2. Am J Surg1994;167:128-133.

43. Felig P. Amino acid metabolism in man. Annu Rev Biochem 1975;44:933-955.

44. Vilstrup H, Hansen BA, Almdal TP. Glucagon increases hepatic efficacyfor urea synthesis. J Hepatol 1990;10:46-50.

45. Reichen J, Le M. Verapamil favourably influences hepatic microvascularexchange and function in rats with cirrhosis of the liver. J Clin Invest1986;78:448-455.

46. Keiding S. Hepatic clearance and liver blood flow. J Hepatol 1987;4:393-398.

47. Carlson LA, Ekelund L-G, Or L. Circulatory and respiratory effects ofdifferent doses of protaglandin E1 in man. Acta Physiol Scand 1969;75:161-169.

48. Henley K, Smith D. Prostaglandins in liver transplantation: a $50,000bonus. Gastroenterology 1997;112:670-671.

49. Greig PD, Woolf GM, Sinclair SB, Abecassis M, Strasberg SM, Taylor BM,Blendis LM, et al. Treatment of primary liver graft nonfunction withprostaglandin E1. Tranplantation 1989;48:447-453.

50. Brass EP, Garrity MJ. Structural specificity for prostaglandin effects onhepatocyte glycogenolysis. Biochem J 1990;267:59-62.

51. Schafer DF, Sorrell MF. Prostaglandins in liver transplantation: a promiseunfulfilled. Gastroenterology 1996;111:819-820.

HEPATOLOGY Vol. 27, No. 3, 1998 FABBRI ET AL. 821