Kidney International, Vol. 63 (2003), pp. 1417–1425

ION CHANNELS – MEMBRANE TRANSPORT – INTEGRATIVE PHYSIOLOGY

Urinary aquaporin-2 in healthy humans and patients with livercirrhosis and chronic heart failure during baseline conditionsand after acute water load

ROBERT S. PEDERSEN, H. BENTZEN, J.N. BECH, O. NYVAD, and E.B. PEDERSEN

Department of Medicine, Holstebro Hospital, Holstebro; and Department of Medicine, Aarhus University, Aarhus, Denmark

Urinary aquaporin-2 in healthy humans and patients with liver The vasopressin (AVP)-sensitive water channel aqua-cirrhosis and chronic heart failure during baseline conditions porin-2 (AQP-2) mediates water transport across theand after acute water load. apical plasma membrane of the renal collecting ducts [1].Background. Patients with liver cirrhosis and chronic heart

The permeability of the collecting ducts is dependent onfailure (CHF) have a reduced capacity to excrete water. Studiesthe water channels located at the apical membrane. AVPin healthy humans have shown that an acute water load reduces

the excretion of aquaporin-2 in urine (u-AQP-2). We wanted to acts on a short-term basis through the cyclic adenosinetest the hypothesis that an acute water load reduces u-AQP-2 monophosphate (cAMP)/protein kinase A (PKA) signal-less in patients with liver cirrhosis or CHF than in healthy ing pathway translocating AQP-2 from intracellular vesi-humans.

cles to the apical plasma membrane of the collecting ducts.Methods. Fourteen healthy subjects, 14 patients with liverWhen AVP levels decline, AQP-2 is retrieved by endocy-cirrhosis, and 14 patients with CHF were given an oral water

load of 20 mL/kg. Urine was collected every 30 minutes for 4 tosis, and the membrane returns to its resting state ofhours for analysis of u-AQP-2. Blood samples were drawn at lower water permeability [2–4]. AVP also regulates thethe beginning and at the end of the study for analysis of arginine

long-term water permeability of the collecting ducts byvasopressin (AVP). u-AQP-2 was determined by radioimmu-increasing AQP-2 gene expression. The increase in tran-noassay.

Results. During the study period, urinary output was 22.8% scription may be the result of the ability of cAMP tohigher than water intake in the healthy controls and increased activate PKA, resulting in phosphorylation of transcrip-14-fold from baseline, but in patients with liver cirrhosis and tion factors such as cAMP regulatory element bindingCHF urinary output was 14% and 24% less than the intake,

protein (CREB) [5–7]. The long-term regulation increaseswhile urinary output increased 7- and 19-fold from baseline,the total amount of AQP-2 available for the modulationrespectively. u-AQP2 decreased significantly more in patients

with CHF (39%) than in healthy controls (17%) but it was of the apical membrane’s water permeability.unchanged in those with liver cirrhosis. AVP decreased 46% AQP-2 is excreted into the urine [8–9] and can bein patients with CHF, but was unchanged in healthy controls quantified by either radioimmunoassay [10] or quantita-and those with liver cirrhosis. A 24-hour urinary excretion of

tive Western blot analysis [11]. Short-term studies haveAQP-2 was significantly elevated in patients with CHF (me-shown that urinary AQP-2 excretion (u-AQP-2) can bedian, 25.7 nmol/mol creatinine) compared to healthy controls

(15.7 nmol/mol creatinine) and those with liver cirrhosis (17 used as a marker for the action of AVP on the collectingnmol/mol creatinine). ducts [10].

Conclusion. The excretion of AQP-2 in urine is abnormal Chronic heart failure (CHF) is generally associatedboth in liver cirrhosis in which we find less suppression ofwith marked defects in the renal handling of sodium andu-AQP2 by an acute water load and in CHF in which we

find a high baseline level and an exaggerated suppression of water resulting in extracellular fluid expansion and inu-AQP2 by an acute water load. severe CHF with hyponatremia. Studies have shown that

AVP is increased in rats with congestive heart failureand that there is an increased expression of AQP-2mRNA [12]. Furthermore, the total amount of AQP-2is increased, especially at the localization of the apicalKey words: urinary aquaporin-2, water load, healthy humans, water

retention. membrane [13]. Studies in rats with liver cirrhosis withand without ascites have shown different results regard-Received for publication November 5, 2001ing the expression of AQP-2 in the renal collecting ducts.and in revised form March 13, 2002, July 2, 2002, and October 17, 2002

Accepted for publication November 8, 2002 Fujita et al [14] reported that the expression of AQP-2mRNA was increased in decompensated cirrhotic rats 2003 by the International Society of Nephrology

1417

Pedersen et al: Urinary aquaporin-2 during water loading1418

and did not change after an acute water load. However, impairment in renal function defined as a creatinineclearance � 40 mL/min.other studies have shown a decrease in AQP-2 expres-

For the CHF patients, the following inclusion criteriasion in rats with compensated liver cirrhosis [15] and inwere used: (1) men and women, (2) age, 35 to 75 yearsdecompensated liver cirrhosis [16].old; (3) New York Heart Association (NYHA) class IIThe exact role of AQP-2 under conditions of sodiumor III; and (4) left ventricular (LV) systolic dysfunctionand water retention is not fully clarified in humans. Theas evidenced by an LV ejection fraction � 40% docu-effect of an acute oral water load on u-AQP-2 and itsmented by echocardiography within 6 months of studyrelationship to AVP and urinary osmolality (u-osm) inenrollment. Exclusion criteria were (1) a history or clini-healthy humans, in patients with liver cirrhosis, and incal signs of disease of the lungs, liver, or endocrine or-patients with CHF has not been studied by simultaneousgans; (2) neoplastic disease; (3) arterial hypertension;measurement of other important regulatory hormons of(4) alcohol abuse; (5) treatment with cyclooxygenasethe water and sodium homeostasis such as the renin-inhibitors; (6) unwillingness to participate; and (7) sig-angiotensin-aldosterone system and the natriuretic pep-nificant impairment in renal function defined as a creati-tidesnine clearance � 40 mL/min.The purpose of the present study was to measure the

The studies was approved by the local medical ethicseffect of an acute water load on (1) urinary excretioncommittee, and written informed consent was obtainedof AQP-2, (2) plasma concentration of vasopressin, (3)from all participants.urine osmolality, (4) the renin-angiotensin-aldosterone

system, and (5) the natriuretic peptide system. We Experimental procedureswished to test the hypotheses that urinary excretion of

A 24-hour urine collection was performed the dayAQP-2 was reduced less in patients with liver cirrhosisbefore the study. All medication was withdrawn 24 hoursand CHF than in healthy controls after an acute oralbefore the study. The subjects were allowed to drinkwater load, and that baseline levels of u-AQP-2 werefreely until the beginning of the study. Smoking washigher in liver cirrhosis and CHF than in controls.prohibited during the study. Baseline was the periodfrom the end of the 24-hour urine collection to just before

METHODS the water load. An indwelling catheter for blood sam-pling was placed in a forearm vein. After voiding, a waterSubjectsload of 20 mL/kg was given orally for 15 minutes. Urine

For the healthy subjects, the following inclusion crite-was collected every 30 minutes for 4 hours. Voiding took

ria were used: (1) men and women; and (2) age, 35 to place in the standing or sitting position. Blood samples75 years old. Exclusion criteria were (1) a history or were drawn at the beginning and at the end of the study.clinical signs of disease of the heart, lungs, liver, kidneys Plasma was analyzed for AVP, osmolality (osm), reninor endocrine organs; (2) neoplastic disease; (3) arterial concentration (PRC), angiotensin II (Ang II), aldoste-hypertension; (4) alcohol or drug abuse; (5) medical rone, atrial natriuretic peptide (ANP), brain natriuretictreatment except for birth control pills; (6) unwillingness peptide (BNP), sodium, and hemotocrit. Urine was ana-to participate; and (7) abnormal laboratory screening lyzed for AQP-2 concentration, urine osmolality (u-osm),tests (i.e., abnormal blood hemoglobin, white cell count, urine creatinine concentration (u-creatinine), and urineplatelets, p-sodium, p-potassium, p-creatinine, p-albu- sodium concentration.min, b-glucose, s-cholesterol, p-bilirubin and p-alanineaminotransferase, or albuminuria or glucosuria). Measurements

For the liver cirrhosis patients the following inclusion AQP-2 was measured by radioimmunoassay, as pre-criteria were used: (1) men and women; (2) age, 35 to viously described [10]. Urine samples was centrifuged75 years old; and (3) cirrhosis based on liver histology for 5 minutes at 1.6 � 100g (3000 rpm) and 125 to 3000or in the absence of a liver biopsy, anamnestic, clinical, �L of the supernatant (depending on u-osm) was freeze-and laboratory evidence of cirrhosis, including ascites, dried and kept frozen at �20�C until assayed. For radio-esophagogal varices (verified by endoscopy), hypoal- immunoassay rabbit anti-AQP-2 antibody was obtainedbuminemia, and increased p-alanine aminotransferase. from Søren Nielsen (Department of Cell Biology, Insti-Exclusion criteria were (1) a history or clinical signs tute of Anatomy, Aarhus University, Aarhus, Den-of disease of the lungs, heart or endocrine organs; (2) mark). The minimum detection level was 32 pg/tube.neoplastic disease; (3) arterial hypertension; (4) treat- The coefficients of variation were 11.7% (interassay) andment with cyclooxygenase inhibitors; (5) positive hepi- 5.9% (intra-assay).titis B surface antigen (HbsAg), positive hepatitis B en- AVP in plasma was measured by radioimmunoassay,dothelial antigen (HbeAg), or positive antihepititis C which was a modification of the method described pre-

viously [17]. Before the assay procedure C18 Sep-Pakvirus; (6) unwillingness to participate; and (7) significant

Pedersen et al: Urinary aquaporin-2 during water loading 1419

(Water Associates, Milford, MA, USA) extraction was Statisticsperformed. The antibody was a gift from professor Data from 14 healthy subjects, 14 patients with liverJacques Durr (Miami, Florida, USA). The minimum de- cirrhosis, and 14 patients with CHF were included in thetection level was 0.5 pmol/L. The coefficients of variation statistical analyzes. Due to lack of normality or inhomo-were 13% (interassay) and 9% (intra-assay). geneity of the variance we used Friedman’s ANOVA

PRC was measured by a commercial immunoradiome- (analysis of variance) for paired comparisons within thetric assay (Nichols Institute Diagnostics, Geneva, Swit- groups followed by comparisons between baseline andzerland). The coefficients of variation were 2.5% (intra- the other periods as described by Siegel and Castellanassay) and 9.9% (interassay). Minimal detection level [22]. Since baseline values were different between healthywas 1.4 �U/mL. subjects and patients, we compared relative changes

Ang II in plasma was determined by radioimmunoas- from baseline by comparing healthy subjects on one handsay using a modification [18] of the method originally and patients on the other. Between groups comparisonsdescribed by Kappelgaard, Damkjaer-Nielsen, and Giese were performed with Kruskal-Wallis’s test for unpaired[19]. Radioimmunoassay was performed after previous comparisons followed by Mann-Whitney’s rank sum test.extraction of plasma by C18 Sep-Pak cartridges (Water P � 0.05 was considered the limit of significance.Associates). The antibody was obtained from the De-partment of Clinical Physiology, Glostrup Hospital RESULTS(Glostrup, Denmark). Minimal detection level was 2

Demographicspmol/L plasma. The coefficients of variation were 12%Fourteen healthy subjects of mean age of 60 years(interassay) and 8% (intra-assay).

(range, 44 to 64 years old), six men and eight women,Aldosterone in plasma was measured by a modifica-were studied.tion [18] of the method originally described [20]. With

Fourteen patients with liver cirrhosis of mean age 54a rabbit anti-Aldo antibody (Simoco, Denmark) radio-years (range, 42 to 62 years old), eight men and siximmunoassay was performed after extraction of plasmawomen participated in the study. Nine patients had cir-by C18 Sep-Pak cartridges (Water Associates). The mini-rhosis verified by a liver biopsy. The five patients inmum detection level was 42 pmol/L plasma. The coeffi-which a liver biopsy had not been performed had yearscients of variation were 13% (interassay) and 9% (intra-of excessive alcohol abuse. All had ascites, and one hadassay).portal hypertension. Five patients had esophageal vari-ANP in plasma was determined by radioimmunoassay,ces (three ascites patients) and two patients had portalas previously described [18]. ANP was extracted fromhypertension (one ascites patient) verified by ultrasound.plasma by means of C18 Sep-Pak cartridges (Water Asso-Nine patients (six patients with ascites) received diureticciates), using ethanol, acetic acid, and water. For radio-treatment. The diuretic treatment consisted of eitherimmunoassay rabbit anti-ANP antibody was obtainedspironolactone (six patients) or a combination therapyfrom the Department of Clinical Chemistry, Bispebjergof spironolactone and furosemide (three patients). TheHospital (Copenhagen, Denmark). The minimum detec-cirrhotic patients treated with diuretics were in a statetion level was 0.5 pmol/l plasma. The coefficients of varia-of normohydration or with mild-to-moderate leg edemation were 12% (interassay) and 10% (intra-assay).and ascites.BNP in plasma was measured by radioimmunoassay,

Fourteen patients with CHF of mean age 65 yearspreviously described [21]. Immunoreactive BNP was ex- (range, 49 to 74 years old), 11 men and three womentracted from plasma by use of C18 Sep-Pak cartridges were studied. The etiology of CHF was ischemic heart(Water Associates) eluted by 80% ethanol in a 4% acetic disease (93%) and cardiomyopathy (7%). Nine patientsacid solution. Radioimmunoassay was performed using were in NYHA class II [left ventricular ejection fractiona rabbit anti-BNP antibody without cross-reactivity with (LVEF), mean 28%] and five in NYHA class III (LVEF,�-ANP and urodilatin (URO). The minimum detection mean 29%). Medication included digoxin, diuretics, an-level was 0.5 pmol/L plasma. The coefficients of variation giotensin-converting enzyme (ACE) inhibitors, Ang IIwere 11% (interassay) and 6% (intra-assay). inhibitors, aspirin, warfarin, isosorbide mononitrate,

Plasma and urinary concentrations of sodium and po- statins, beta blockers, and amiodarone. Four patientstassium were measured by routine methods at the De- had an ICD pacemaker. All patients were in a clinicallypartment of Clinical chemistry, Holstebro Centralsy- stable condition. Clinical and laboratory data for all par-gehus. Plasma and urinary osmolality was measured by ticipants are given in Table 1.freezing-point depression (Advanced model 3900 multi-

Water balance during the studysample osmometer).Blood pressure was determined by UA-743 digital The water intake was 1605 mL (25th percentile, 1362;

75th percentile, 1669) in healthy subjects, 1562 mL (1253blood pressure meter (A&D Cmpany, Ltd., Japan).

Pedersen et al: Urinary aquaporin-2 during water loading1420

Table 1. Clinical and laboratory data for 14 healthy subjects, 14 patients with liver cirrhosis, and 14 patients with chronic heart failure(CHF). Urine analyses are based on 24-hour urine collections the day before the studies

Healthy Cirrhosis CHF

Number 14 14 14Men/women 6/8 8/6 11/3

Years 60 (44–64) 54 (42–62) 65 (49–74)Body mass index 26.6 (24.3–28.4) 29.2 (23.6–32)a 24.7 (22.4–29.3)a

Systolic blood pressure mm Hg 120 (114–136) 110 (98–116)a 115 (103–129)a

Diastolic blood pressure mm Hg 75 (72–80) 67 (60–72)a 64 (60–74)a

Pulse rate min 65 (62–68) 67 (62–81)a 60 (55–66)a

p-sodium mmol/L 141 (139–142) 136 (135–139)b 138 (136–139)b

p-potassium mmol/L 3.8 (3.7–3.9) 3.9 (3.6–4.3)a 3.9 (3.9–4.3)a

p-creatinine lmol/L 87 (82–100) 78 (69–93)a 100 (83–143)a

s-osmolality mosmol/kg 290 (288–292) 287 (280–291)a 287 (284–292)a

u-sodium excretion mmol/day 195 (160–227) 156 (84–226)a 199 (130–262)a

Median with 25% and 75% percentiles.aNot significant compared to healthy subjectsbP � 0.05 compared to healthy subjects

to 1839) in patients with liver cirrhosis and 1469 mL Plasma AVP(1257 to 1826) in patients with CHF. During the study, The absolute values of AVP in healthy subjects, inthe total urinary output was 1973 mL (1647 to 2024) in patients with liver cirrhosis, and patients with CHF arehealthy subjects, 1343 mL (729 to 1744) in patients with shown in Figure 3. AVP was unchanged during the studyliver cirrhosis and 1111 mL (601 to1390) in patients with in healthy subjects and in patients with liver cirrhosis.CHF. When the total amount of urinary output was AVP was significantly higher in patients with liver cirrho-compared with the amount of water intake the urinary sis compared to healthy subjects at baseline (48%, P �output was 22.8% larger than the water intake in healthy 0.05), but no significant difference in relative values wassubjects, but 14% less in patients with liver cirrhosis and observed between healthy subjects and patients with24% less in patients with CHF. liver cirrhosis during the experiment. AVP decreased by

46% during the study in patients with CHF (P � 0.005).u-AQP-2 AVP was significantly elevated at baseline in CHF pa-

Figure 1 shows u-AQP-2 given as nanomole per mole tients (155%, P � 0.006) compared to healthy subjects,creatinine in healthy subjects, in patients with liver cirrho- and the changes in relative values were significantlysis, and patients with CHF. In healthy subjects, u-AQP-2 greater in CHF patients (43%, P � 0.004) compared towas almost unchanged during the study except after 210 healthy subjects (�13%).minutes where a 17% (P � 0.05) decrease was found.

Osm in urine and urinary outputIn patients with liver cirrhosis, u-AQP-2 was unchangedduring the study and no differences were observed in Figure 4 shows u-osm in healthy subjects, in patientsthe relative changes in u-AQP-2 between patients with with liver cirrhosis, and patients with CHF. In healthyliver cirrhosis and healthy subjects, at baseline or during subjects, u-osm decreased significantly with a maximumthe study (Fig. 2). u-AQP-2 decreased throughout the decrease of 88% after 120 minutes. At the end of thestudy in patients with CHF and was significantly lower study, u-osm did not deviate significantly from baseline inafter 120 minutes with a maximal 39% decrease after healthy subjects. In patients with cirrhosis u-osm decreased210 minutes. u-AQP-2 was significantly greater in CHF significantly after 90 minutes. The maximal decrease waspatients compared to healthy subjects at baseline (54%, seen after 120 minutes and was 72%. Subsequently, u-osmP � 0.03) and the relative decreases in u-AQP-2 was slowly increased but was still significantly lower than base-significantly stronger in patients with CHF compared to line at the end of the study. There was no significanthealthy subjects from 120 minutes (P � 0.007) and until difference between healthy subjects and patients with liverthe end of the study (Fig. 2). cirrhosis at baseline. The relative decrease in u-osm was

In healthy subjects, patients with liver cirrhosis, and significantly less in patients with liver cirrhosis comparedCHF patients the 24-hour u-AQP2 excretion was 15.7 to healthy subjects between 30 and 180 minutes. In CHFnmol/mol creatinine (25th percentile, 13.8; 75th percentile, patients, u-osm decreased significantly after 90 minutes.19.2), 17 nmol/mol creatinine (13.6 to 19.7), and 25.7 nmol/ The maximal decrease was seen after 150 minutes and wasmol creatinine (18.7 to 36.3), respectively. The 24-hour 80%. Subsequently there was a slow increase, but u-osmu-AQP-2 excretion was significantly increased in patients was still significantly lower than baseline at the end of

the study. There was no significant difference betweenwith CHF compared to healthy subjects (P � 0.003).

Pedersen et al: Urinary aquaporin-2 during water loading 1421

increases were significantly greater in CHF patients com-pared to healthy subjects (data not shown).

PRC, Ang II, aldosterone, ANP, and BNP

Table 2 shows PRC, Ang II, aldosterone, ANP, andBNP in healthy subjects, patients with liver cirrhosis,and patients with CHF. PRC and aldosterone were sig-nificantly decreased during the study in healthy subjectsand in CHF patients with only a tendency toward adecrease in patients with liver cirrhosis. The decrease inPRC was significantly greater in CHF patients comparedto healthy subjects. Ang II and ANP were unchangedin all the groups. BNP increased significantly in patientswith CHF with no change in patients with liver cirrhosisand healthy subjects. All the hormones were significantlyincreased at baseline in CHF patients and in patientswith liver cirrhosis compared to healthy subjects.

Blood pressure and heart rate

Mean arterial blood pressure (MAP) and heart ratewere measured every 30 minutes. There was no signifi-cant change in MAP in healthy subjects during the study.MAP was significantly increased in patients with cirrho-sis at 30 minutes and at 120 minutes [baseline, 80 mm Hg(73 to 87); 30 minutes, 90 mm Hg (86 to 98, P � 0.05);120 minutes, 114 mm Hg (104 to 133, P � 0.001)]. Thesame pattern was seen in CHF patients [baseline, 82mm Hg (73 to 91); 30 minutes, 94 mm Hg (87 to 103,P � 0.05); 120 minutes, 120 mm Hg (110 to 142, P �0.001)]. Subsequently MAP decreased to baseline levelduring the rest of the study in all patients. The heartrate showed no change during the study in either of thegroups (data not shown).

DISCUSSION

In the present study of healthy subjects and patientswith liver cirrhosis and CHF, we investigated the effectFig. 1. Urinary excretion of aquaporin-2 (u-AQP2) after an acute wa-

ter load in 14 healthy subjects (A ), 14 patients with liver cirrhosis (B ), of an acute oral water load on u-AQP-2, water excretion,and 14 patients with chronic heart failure (CHF) (C ). Baselines are and hormones in plasma of importance for renal functionfrom the end of the 24-hour urine collection to the beginning of the

and sodium and water homeostasis in 30-minute periodsstudy. Results are means with � SD. *P � 0.05 vs. baseline; †P � 0.05vs. healthy subjects at baseline. from 8:00 a.m. to 12:00 a.m.

In healthy subjects, the total urinary output was greaterthan the total water intake after an acute water load.The acute water load resulted in a small but significanthealthy subjects and patients with CHF at baseline. Thedecrease in u-AQP-2 but no change in AVP. In addition,relative decrease was also significantly smaller in patientsthere was a substantial increase in urinary output and awith CHF compared to healthy subjects after 30 minutesdecrease in u-osm. Patients with liver cirrhosis had aand until 180 minutes.reduced ability to excrete an acute water load. In addi-The maximal increase in urinary output was seen aftertion, they showed only a tendency toward a decrease in120 minutes and was 14-fold higher than at baselineu-AQP-2 and AVP. The relative increases in urinaryin healthy subjects, 7.6-fold higher than at baseline inoutput, and the relative decrease in u-osm was less thanpatients with liver cirrhosis and 19-fold higher than atin healthy subjects. Patients with CHF also had a dimin-baseline in patients with CHF. The relative increasesished capacity to excrete an acute water load with a 24%were significantly lower in patients with liver cirrhosis

compared to healthy subjects. In addition, the relative less urinary output than water intake. In patients with

Pedersen et al: Urinary aquaporin-2 during water loading1422

Fig. 2. Relative changes in urinary aquaporin-2(u-AQP2) after an acute water load in 14 healthysubjects (�), 14 patients with liver cirrhosis( ), and 14 patients with chronic heart failure(�). Baselines are from the end of the 24-hoururine collection to the beginning of the study.Results are means � SD. †P � 0.05 vs. healthysubjects.

CHF, u-AQP2 and AVP decreased throughout thestudy. The decrease in u-AQP2 was larger in patientswith CHF than in healthy subjects. The relative increasesin urinary output were larger in patients with CHF thanin healthy subjects, whereas the decrease in u-osm wassmaller. For ethical reason, we did not find it justifiedto discontinue the medical treatment of the patients formore than 24 hours. It is, therefore, possible that thedrugs still have a small effect on water balance.

The role of AQP-2 in cirrhosis has been studied exten-sively in rats. The results have been conflicting with re-gard to the renal expression of AQP-2. The initial studiesshowed an increased level of AQP-2 protein and AQP-2mRNA in carbon tetrachloride (CCL4)–induced cirrhoticrats with ascites [14, 23]. In addition, an acute water loaddid not decrease AQP-2 mRNA within 1 hour [14]. Livercirrhosis induced by common bile duct ligation (CBL)showed a decreased expression of AQP-2 both in ratswith and without ascites [15, 16]. Recently, Jonassen et al[24] showed no difference in the expression of AQP-2protein in rats with CCL4-induced liver cirrhosis andcontrol rats. The rats had hyponatremia, ascites, andFig. 3. Plasma levels of vasopressin (AVP) at baseline and at 240increased plasma level of AVP and showed an impairedminutes after an acute water load in 14 healthy subjects (�), 14 patients

with liver cirrhosis ( ), and 14 patients with chronic heart failure (CHF) ability to excrete an intravenous water load. We are the(�). Baselines are from the end of the 24-hour urine collection to the first to study the role of AQP-2 in humans with liverbeginning of the study. Results are means � SD. *P � 0.05 vs. baseline;†P � 0.05 vs. healthy subjects; ††P � 0.05 vs. healthy subjects at baseline. cirrhosis, and our results seem to be in agreement with

Pedersen et al: Urinary aquaporin-2 during water loading 1423

Table 2. Plasma levels of renin (PRC), angiotensin II (Ang II),aldosterone, atrial natriuretic peptide (ANP), and brain natriuretic

peptide (BNP) at baseline and 240 minutes after an acute water loadin 14 healthy subjects, 14 patients with liver cirrhosis, and 14

patients with chronic heart failure (CHF)

Baseline 240 min

PRC lU/mLHealthy 14 (12–22) 12 (9–15)a

Liver cirrhosis 29 (13–54)b 23 (11–40)CHF 73 (46–142)b 41 (22–86)a,c

Ang II pmol/LHealthy 8.6 (4.5–11.7) 8.1 (5.5–13.6)Liver cirrhosis 9.1 (3.8–14.8) 7.2 (4.8–14.8)CHF 13.4 (7.4–18.7)b 10.2 (9.6–18.2)

Aldo pmol/LHealthy 61.2 (45.9–105.0) 41.7 (35.4–50.0)a

Liver cirrhosis 122.3 (74.4–182.1)b 70.9 (40.3–207.1)CHF 94.5 (61.1–149.4)b 50.0 (41.7–100.8)a

ANP pmol/LHealthy 6.1 (5.5–6.9) 5.4 (4.4–7.0)Liver cirrhosis 8.0 (3.9–10.5) 7.3 (4.4–9.8)CHF 20.2 (13.3–41.1)b 21.0 (11.5–30.8)

BNP pmol/LHealthy 1.9 (1.1–3.3) 2.0 (1.2–3.9)Liver cirrhosis 4.2 (2.9–9.7)b 4.9 (3.0–11.9)CHF 16.9 (9.1–44.8)b 18.7 (10.2–59.8)a

Median with 25% and 75% percentiles.a P � 0.05 vs. baseline; bP � 0.05 vs. healthy subjects at baseline; cP � 0.05

vs. healthy subjects

It has been demonstrated that AQP-2 protein andmRNA expression is higher in rats with severe CHF[12, 13], but, in the study by Nielsen et al [13], ratswith compensated CHF showed no increase in AQP-2expression. In our study, the patients had hyponatremia,increased basal AVP concentration, and a 24-hour urinesample showed an increased excretion of AQP-2. De-spite a larger increase in urinary output and a decreasein u-AQP-2, the ability to dilute urine was impaired inpatients with CHF. They did not excrete all the ingestedwater and the decrease in u-osm was less than in healthysubjects. Thus, we have demonstrated that u-AQP-2 isincreased in patients with CHF, and that an acute waterload can suppress AVP and decrease u-AQP-2, but water

Fig. 4. Urinary osmolality (u-osm) after an acute water load in 14excretion was not normalized.healthy subjects (A ), 14 patients with liver cirrhosis (B ), and 14 patients

with chronic heart failure (CHF) (C ). Baselines are from the end of The results of the present study show that AQP-2 wasthe 24-hour urine collection to the beginning of the study. Results not up-regulated in patients with liver cirrhosis, despiteare medians with 25th, 75th, 10th and 90th percentiles. *P � 0.05 vs.

an increase in AVP and an inability to excrete an acutebaseline.water load normally. However, the increase in AVP wasonly moderate and only six patients had ascites. Thismight at least partly explain the absence of an increasethe findings in animal experiments by Jonassen et al [24].in u-AQP-2 excretion. Possibly u-AQP-2 is increased inOur patients had hyponatremia, slightly but significantlymore advanced liver cirrhosis. The inability to excreteincreased AVP, and six patients had ascites or edema.an acute water load by the cirrotic patients in our studyThe patients had impaired ability to excrete an acutemust, therefore, be due to a different mechanism, whichoral water load because the increase in urinary outputis unknown at present. One question to be answered iswas less than in healthy subjects, and the cirrhotic pa-why the increase in AVP does not increase u-AQP-2tients did not excrete all the water they were given.excretion. Ecelbarger et al [25] raised the question ofFurthermore, we only saw a tendency toward a decrease

in u-AQP-2. the existence of AVP-independent mechanisms for the

Pedersen et al: Urinary aquaporin-2 during water loading1424

regulation of AQP-2 expression and showed that, despite did not change ANP and BNP, while BNP increased inpatients with CHF. This difference may be attributed toa continuous subcutaneous infusion of 1-desamino-8-D-

AVP (synthetic AVP V2-receptor agonist), AQP-2 was differences in cardiac output, which is reduced in CHFbut often normal in liver cirrhosis.down-regulated after water loading. In addition, thirsting

of rats in the presence of chronic V2-receptor blockade(OPC-31260) increases AQP-2 expression levels [26].

CONCLUSIONThus, vasopressin-independent mechanisms might play

We have demonstrated that 24-hour urinary AQP-2a significant role in the abnormal water balance in liverexcretion was increased in patients with CHF but not incirrhosis.cirrhotic patients compared to healthy subjects. An acuteEarlier studies have shown that AVP decreases inwater load did not change u-AQP-2 in patients withresponse to an acute water load in CHF patients [27, 28].cirrhosis but decreased u-AQP-2 in patients with CHFThe decrease was maximal at 2 hours and AVP wasand even more than in healthy subjects. Our results indi-still significantly lower at 5 hours. AVP was significantlycate that excretion of AQP-2 in urine is abnormal bothhigher in patients with CHF compared to healthy sub-in liver cirrhosis and in CHF.jects throughout the study except after 5 hours. In our

study, we measured AVP after 4 hours, and there wasACKNOWLEDGMENTSno difference between healthy subjects and patients with

CHF. However, the relative change in AVP was larger The study was supported by grants from Ringkjoebing County.Laboratory technicians Lisbeth Mikkelsen, Anne Jaritz-Nielsen, Susanin patients with CHF than in healthy subjects, probablyMilton Rasmussen, Henriette Hedelund Vorup Simonsen, and Anne

because AVP had returned to baseline values in healthy Mette Ravn Torstensen are thanked for skillful technical assistanceand commitment. We thank Dr. Carsten Thordal and Dr. Niels Thors-subjects and not in patients with CHF. It seems likely,gaard for their help in recruiting patients with cirrhosis.therefore, that AVP and AQP-2 may play a role in the

inability to excrete an acute water load in patients with Reprint requests to Dr. Robert Schou Pedersen, Department of Medi-cine, Holstebro Hospital, DK-7500 Holstebro, DenmarkCHF. Recently Martin et al [29] demonstrated the roleE-mail: arsp@ringamt.dkof AQP-2 in the enhanced water reabsorption in CHF,

because u-AQP-2 was significantly decreased by the ad-REFERENCESministration of a selective V2-receptor AVP antagonist

(VPA-985) in CHF patients. Although both AVP and 1. Fushimi K, Uchida S, Hara Y, et al: Cloning and expression ofapical membrane water channel of rat kidney collecting tubule.u-AQP-2 were reduced by an oral water load in CHFNature 361:549–552, 1993

to an even greater extent than in healthy control subjects, 2. Marples D, Knepper MA, Christensen EI, Nielsen S: Redistribu-tion of aquaporin-2 water channels induced by vasopressin in ratour study clearly demonstrated an abnormal water excre-kidney inner medullary collecting duct. Am J Physiol Cell Physioltion in CHF. Thus, the reduced capacity to excrete water269:C655–C664, 1995

cannot be attributed to the short-term AVP and AQP-2 3. Nielsen S, DiGiovanni SR, Christensen EI, et al: Cellular andsubcellular immunolocalization of vasopressin-regulated waterregulation only.channel in rat kidney. Proc Natl Acad Sci USA 90:11663–11667,Both the renin-angiotensin-aldosterone system and1993

the natriuretic peptides are involved in the pathogenesis 4. Yamamoto T, Sasaki S, Fuhimi K, et al: Localization and expressionof a collecting duct water channel, aquaporin, in hydrated andof CHF and liver cirrhosis. Both patients with liver cir-dehydrated rats. Exp Nephrol 3:193–201, 1995rhosis and CHF patients showed an elevated level in

5. Hozawa S, Holtzman EJ: and Ausiello: cAMP motifs regulatingPRC and aldosterone at baseline compared to healthy transcription in the aquaporin-2 gene. Am J Physiol Cell Physiol

270(39):C1695–C1702, 1996subjects. Ang II was only increased in patients with CHF.6. Matsumura Y, Uchida S, Rai T, et al: Transcriptional regulationIn patients with liver cirrhosis, there was no change in

of aquaporin-2 water channel gene by cAMP. J Am Soc Nephrolthe renin-angiotensin-aldosterone system after the acute 8:861–867, 1997

7. Yasui M, Zelenin SM, Celsi G, Aperia A: Adenylate cyclase-water load consistent with a previous study [30]. In pa-coupled vasopressin activates of AQP2 promoter via a dual effecttients with CHF the renin-angiotensin-aldosterone sys-on CRE and AP1 elements. Am J Physol Renal Physiol

tem was suppressed in response to the water load, but 272(41):F443–F450, 19978. Kanno K, Sasaki S, Hirata Y, et al: Urinary excretion of aquap-this did not normalize water excretion.

orin-2 in patients with diabetes insipidus. N Engl J Med 332:1540–It has been shown previously that ANP is normal or1545, 1995

increased and BNP is increased in patients with liver 9. Rai T, Sekine K, Kanno K, et al: Urinary excretion of aquaporin-2 water channel protein in human and rat. J Am Soc Nephrolcirrhosis presumably as a compensatory phenomenon to8:1357–1362, 1997promote sodium excretion. This is consistent with our

10. Pedersen RS, Bentzen H, Bech JN, Pedersen EB: Effect ofresults. In patients with CHF, it is also well known that water deprivation and hypertonic saline infusion on urinary AQP2

excretion in healthy humans. Am J Physiol Renal Physiol 280:F860–both ANP and BNP is increased. ANP is secreted inF867, 2001response to increased atrial stretching and BNP is se-

11. Elliot S, Goldsmith P, Knepper MA, et al: Urinary excretioncreted in response to ventricular dilatation. In healthy of aquaporin-2 in humans: A potential marker of collecting duct

responsiveness to vasopressin. J Am Soc Nephrol 7:403–409, 1996subjects and patients with liver cirrhosis, water loading

Pedersen et al: Urinary aquaporin-2 during water loading 1425

12. Xu D-L, Martin P-Y, Ohara M, et al: Upregulation of aquaporin- tion of human rectum in obese patients treated with intestinalbypass operation. Scand J Gastroenterol 9:417–426, 19742 water channel expression in chronic heart failure rat. J Clin Invest

21. Jensen KT, Carstens J, Ivarsen P, Pedersen EB: A new, fast99(7):1500–1505, 1997and reliable radioimmunoassay for brain natriuretic peptide in13. Nielsen S, Terris J, Andersen D, et al: Congestive heart failurehuman plasma. Reference values in healthy subjects and in patientsin rats is associated with increased expression and targeting ofwith different diseases. Scand J Clin Lab Invest 57:529–540, 1997aquaporin-2 water channel in collecting duct. Proc Natl Acad Sci

22. Sigel S, Castellan NJ: Nonparametric Statistics for the BehavioralUSA 94:5450–5455, 1997Sciences, New York, McGraw-Hill, 1998, pp 174–18414. Fujita N, Ishikawa S-E, Sasaki S, et al: Role of water channel

23. Asahina Y, Izumi N, Enomoto N, et al: Increased gene expressionAQP-CD in water retention in SIADH and cirrhotic rats. Am Jof water channel in cirrhotic rat kidneys. Hepatology 21:169–173,Physiol Renal Fluid Electrolyte Physiol 269:F926–F931, 1995 1995

15. Jonassen T, Nielsen S, Christensen S, Petersen JS: Decreased 24. Jonassen TEN, Christensen S, Kwon T-H, et al: Renal watervasopressin-mediated renal water reabsorption in rats with com- handling in rats with decompensated liver cirrhosis. Am J Physiolpensated liver cirrhosis. Am J Physiol Renal Physiol 275:F216– Renal Physiol 279:F1101–F1109, 2000F225, 1998 25. Ecelbarger CA, Nielsen S, Olson BR, et al: Role of renal aqua-

16. Fernandez-Llama P, Turner R, Dibona G, Knepper MA: Renal porins in escape from vasopressin-induced antidiuresis in rat. Jexpression of aquaporins in liver cirrhosis induced by chronic com- Clin Invest 99:1852–1863, 1997mon bile duct ligation in rats. J Am Soc Nephrol 10:1950–1957, 26. Marples D, Christensen BM, Frøkjær J, et al: Dehydration re-

verses vasopressin antagonist-induced diuresis and aquaporin-21999downregulation in rats. Am J Physiol Renal Physiol 275:F400–F409,17. Pedersen EB, Danielsen H, Spencer ES: Effect of indapamide1998on renal plasma flow, glomerular filtration rate and arginine vaso-

27. Goldsmith SR, Francis GS, Cowley AW: Arginine vasopressinpressin in essential hypertension. Eur J Clin Pharmacol 26:543–547,and the renal response to water loading in congestive heart failure.1984Am J Cardiol 58:295–299, 198618. Pedersen EB, Eiskjær H, Madsen B, et al: Effect of captopril on

28. Pruszczynski W, Vahanian A, Ardaillou R, Acar J: Role ofrenal extraction of renin, angiotensin II, atrial natriuretic peptideantidiuretic hormone in impaired water excretion of patients withand vasopressin, and renal vein ratio in patients with arterial hyper- congestive heart failure. J Clin Endocrinol Metab 58:599–605, 1984

tension and unilateral renal artery disease. Nephrol Dial Transplant 29. Martin PY, Abraham WT, Lieming X, et al: Selective V2-receptor8:1064–1070, 1993 vasopressin antagonism decreases urinary aquaporin-2 excretion in

19. Kappelgaard AM, Damkjær-Nielsen M, Giese J: Measurement patients with chronic heart failure. J Am Soc Nephrol 10:2165–2170,of angiotensin II in human plasma: Technical modification and 1999practical experience. Clin Chim Acta 67:299–306, 1976 30. Castellano G, Solis-Herruzo JA, Morillas JD, et al: Antidiure-

20. Rask-Madsen J, Bruunsgaard A, Munck O, et al: The signifi- tic hormone and renal function after water loading in patients withcirrhosis of the liver. Scand J Gastroenterol 26(1):49–57, 1991cance of bile acids and aldosterone for the electrical hyperpolariza-