Potential conflict of interest: Nothing to report.
The effects of the addition of clonidine to diuretics on the mobilization of ascites in the short term (diuretic response and requirement of diuretics) and the long term (readmissions for tense ascites and requirement of diuretics) were examined in patients with cirrhosis and with increased sympathetic nervous system (SNS) activity. We also studied neurohormonal, hemodynamic effects and side effects of clonidine and diuretics. Patients were randomized to receive placebo (group1, n = 32) or clonidine (0.075 mg) twice daily (group 2, n = 32) for 3 months. After 8 days and for 10 days duration, spironolactone (200 mg/day) was added in both groups. After this period, the dosages of diuretics were individually increased until diuretic response. Responding patients were discharged and followed at the outpatient clinic. During the first hospitalization, the time needed for diuretic response was shorter in group 2 than in group 1. The mean requirement for diuretics was significantly higher in group 1 than in group 2, and the diuretic complications (hyperkalemia and renal impairment) were significantly lower in group 2. Clonidine induced a permanent decrease in SNS activity and delayed decrease in renin/aldosterone levels. During the follow-up, the time to the first readmission for tense ascites was shorter in group 1 than in group 2. Readmissions related to tense ascites or diuretic complications were significantly lower in group 2. The mean requirement for diuretics was significantly higher in group 1 than in group 2. In conclusion, the additional administration of clonidine to diuretics induced an earlier diuretic response associated with fewer diuretic requirements and complications. (HEPATOLOGY 2006;44:844–849.)
The mechanisms leading to renal sodium retention in cirrhosis are not fully understood. Splanchnic arterial vasodilatation associated with portal hypertension induces a reduction in effective blood volume and permanent activation of antinatriuretic and vasoconstrictive mechanisms (stimulation of renin/aldosterone and sympathetic systems, vasopressin secretion).1 The sympathetic nervous system (SNS) contributes to renal hypoperfusion and sodium retention. The activated SNS stimulates renal α1-adrenoreceptors and causes decreases in renal blood flow and glomerular filtration rate. Additionally, norepinephrine increases proximal tubular reabsorption of sodium and enhances renin, aldosterone, and vasopressin secretions.2, 3 We recently showed that in patients with cirrhosis and ascites, the marked activation of SNS and renin/aldosterone axis reduced diuretic response.4 Consequently, inhibitions of SNS activity and renin/aldosterone system would improve diuretic response.
Clonidine, a centrally acting α2-agonist, has been shown to offer sympatholytic activity in patients with arterial hypertension or cirrhosis.5–8 Co-administration of clonidine and spironolactone has been shown to increase natriuresis and body weight loss more efficiently than spironolactone alone in patients with cirrhosis and ascites and activated SNS in 2 pilot studies.9, 10
This study examined the short-term (diuretic response and requirement of diuretics) and the long-term (rehospitalizations for tense ascites and requirement of diuretics) effects of adding clonidine to diuretics (spironolactone and furosemide) in patients with cirrhosis and increased SNS activity. Secondary objectives were to study neurohormonal, hemodynamic parameters and side effects of clonidine and diuretics.
SNS, sympathetic nervous system: MAP, mean arterial pressure; SEM, standard error of the mean.
Patients and Methods
Patients with cirrhosis and ascites were studied. The main inclusion criteria was a plasma norepinephrine level >300 pg/mL (normal value: 185-275 pg/mL) (n = 64). Cirrhosis diagnostic was based on liver biopsy (n = 28) or on clinical, laboratory, and ultrasonographic data (n = 36).
We excluded patients with a serum bilirubin concentration above 4.5 mg/dL, a prothrombin time below 40%, a platelet count below 40 × 109/L, and a serum creatinine concentration above 2 mg/dL. Gastrointestinal hemorrhage due to variceal rupture within the previous 2 months was another cause of rejection. Patients with alcoholic hepatitis, diabetes mellitus, hepatocellular carcinoma, or respiratory or cardiac failures were discarded. Patients suffering from hepatic encephalopathy or bacterial infection were candidates for the investigation after recovery. The Ethical Committee of the Centre Hospitalier Universitaire de Charleroi, Belgium, approved the study and informed consent was obtained.
Patients were studied after a 5-day low-sodium diet (40 mEq/day) without diuretic, vasoactive therapy, or paracentesis. In hyponatremic patients (serum sodium < 130 mEq/L), water ingestion was restricted to 800 mL per day. In the early morning following this 5-day period, before standing up, blood samples were drawn to determine plasma renin, aldosterone, and norepinephrine concentrations. Plasma norepinephrine concentration was immediately determined in order to randomize the patient after a 24-hour urine collection. Mean arterial pressure (MAP) and heart rate were measured before treatment.
Our estimation of the number of patients needed to observe differences in body weight loss and natriuresis between the association of clonidine/diuretics and placebo/diuretics was based on data from our recent pilot study.10 With the α risk of 0.05 and the β risk of 0.2, we estimated that 60 patients would provide the comparison between the 2 treatments. Assuming a drop-out rate of 5% in our experience, the study size should be 64 patients. We planned to enroll 64 patients for a 2-year period.
Patients were randomized using a closed list in 8 blocks of 8 units. Randomization was performed via telephone to a central randomization desk. Patients and investigators were blinded to the treatment assignments. Active drugs and placebos were kept in the hospital pharmacy and dispensed by a pharmacist who was the only caregiver with access to the randomization code. Patient compliance was monitored by a nurse who dispensed medications in each visit.
Sixty-four patients were randomized to receive placebo tablets twice daily (group 1, n = 32) or clonidine 0.075 mg twice daily (group 2, n = 32) for 3 months. This low dosage appeared to be the safer in our previous nonpublished pilot studies, because this drug has potential significant side effects or drug interactions. After 8 days and for 10 days duration, spironolactone (200 mg/day) was added in both groups (Fig. 1). After this period, the daily dosage of spironolactone was adjusted according to individual response. If body weight loss was inferior to 200 g/day during 5 days, it was increased stepwise by 100 mg/day, every 5 days, from 200 to 400 mg. In case of insufficient response to maximal dosage, furosemide 40 mg/day was added to treatment and its daily dosage was increased stepwise by 40 mg/day, every 3 days, from 40 to 160 mg. In case of tense ascites, a paracentesis plus intravenous albumin was performed.
Hemodynamic and neurohormonal parameters were measured after 8 and 18 days. Body weight, urine volume, and urine sodium were measured daily, and liver and renal functions were checked every 3 days in patients of both groups. Patients were discharged after diuretic response (loss of body weight >200 mg/day during 5 days and urinary sodium >50 mEq/day). The follow-up period started at the end of the first hospitalization. The patients were examined in the outpatient clinic weekly during the first month, monthly for the next 2 months, and bimonthly afterward. The schedule was used after readmission to treat episodes of ascites.
The following biological criteria were used to define the diuretic complications. Renal impairment was defined as an increase in serum creatinine concentration of more than 50% of the pretreatment value or up to a level >2 mg/dL. Hyponatremia was defined as serum sodium concentration decreases of >10 mEq/L or down to a level <120 mEq/L. Hyperkalemia was defined as serum potassium increases >1.5 mEq/L or up to a level >6 mEq/L. Rehospitalizations during the follow-up arose from 4 causes: tense ascites, hepatic encephalopathy, severe bacterial infection, and gastrointestinal hemorrhage. The following clinical criteria were arbitrarily used when tense ascites was associated with other complications: (1) Ascites associated with gastrointestinal hemorrhage, severe bacterial infection, grade II to IV (moderate to severe) hepatic encephalopathy, or some other complication requiring emergency treatment. The associated conditions were considered to be the cause of readmission. (2) Ascites associated with grade I (mild) hepatic encephalopathy or another condition not requiring emergency treatment, ascites being considered to be the cause. (3) Hepatic encephalopathy following severe bacterial infection or gastrointestinal hemorrhage was considered to be secondary and not related to diuretics. (4) Gastrointestinal bleeding and bacterial infection, hemorrhage being considered to be the cause of readmission, since bacterial infection frequently follows gastrointestinal bleeding in cirrhosis.
Methods of Measurement.
MAP was calculated according to the formula: MAP = [systolic pressure + (diastolic pressure × 2)]/3. Heart rate was derived from continuous electrocardiographic monitoring. Plasma renin concentration was measured by a 2-site immunoradiometric assay (Nichols Institute Diagnostics, Wychen, The Netherlands).11 This reagent set was endowed with an intra-run imprecision of 5.3% at 34 μU/mL (n = 94), decreasing to 2.7% (n = 92) at 300 μU/mL. Inter-run assay pointed out imprecisions of 7.1% and 5.4%, respectively, for the lowest and the highest concentration (n = 47). Reference values were 7-76 μU/mL.
Plasma aldosterone was routinely measured by a direct radioimmunoassay with antibody-coated tubes (Radim SA, Liège, Belgium).12 Regarding the high concentrations found in this report, values were confirmed by preliminary organic extraction and Sephadex exclusion chromatography separation before submitting the samples to the radioimmunoassay using antibodies raised in sheep to aldosterone-18,21-dihemisuccinate:HSA and provided by MP Biomedicals Inc. (Irvine, CA). Reference values (supine) ranged from 30 to 100 pg/mL. The intra-run and inter-run imprecisions amounted to 6% (n = 92) and 9.5% (n = 46), respectively, at a plasma concentration of 221 pg/mL.
Norepinephrine was determined by electrochemical detection after separation by high-performance liquid chromatography.13 Reference values ranged from 185 to 275 pg/mL. Intra-run imprecision was 4.3 % at 1,700 pg/mL (n = 56). Inter-run assays led to 8.6% imprecision at the same concentration (n = 28). Other measurements were made by standard laboratory screenings.
Analyses of the results were performed using the nonparametric tests of Mann-Whitney and Wilcoxon as appropriate. The χ2 test was used for comparing distributions and frequencies of readmissions and complications. The results are presented as means ± SEM. All P values less than .05 were considered to indicate statistical significance. The results were not evaluated on an intention-to-treat analysis.
From October 2000 to October 2002, of 130 patients with cirrhosis hospitalized for ascites, 64 patients with activated SNS were included in the study. We excluded 66 patients without activated SNS. We randomized 32 patients to group 1 and 32 patients to group 2. No significant initial baseline differences were noticed between the 2 groups, regarding clinical characteristics, cirrhosis etiologies, plasma sodium and potassium concentrations, liver and renal functions, MAP, heart rate, Child-Pugh score, and hormonal concentrations (Table 1). The number of patients with previous episodes of ascites (22 in each group), hepatic encephalopathy (3 in group 1 and 2 in group 2), and gastrointestinal hemorrhage (1 in each group) were also similar. Before the study, 3 patients of each group had refractory ascites and required paracentesis every month for at least 6 months. Two patients of group 1 and 1 of group 2 abandoned the study just after the beginning of the investigation because of lack of compliance.
Table 1. Clinical and Laboratory Data at Baseline in Both Groups
Group 1 (n = 30)
Group 2 (n = 31)
Results are given as mean (SEM). No significant differences were noticed between the 2 groups. Normal values are plasma renin: 7-76 μU/mL; plasma aldosterone: 30-100 pg/mL; plasma norepinephrine: 185-275 pg/mL.
Serum bilirubin (mg/dL)
Prothrombine time (%)
Serum albumin (g/L)
Serum creatinine (mg/L)
Serum sodium (mEq/L)
Serum potassium (mEq/L)
Heart rate (beat/min)
Plasma renin (μU/mL)
Plasma aldosterone (pg/mL)
Plasma norepinephrine (pg/mL)
Results During the First Hospitalization.
No changes in standard liver tests were noticed in both groups. Table 2 shows the evolution of neurohormonal tests, body weight, and systemic hemodynamics in both groups. In group 1 (n = 30) after 8 days, placebo alone did not induce changes in neurohormonal measurements, body weight, or systemic hemodynamics.
Table 2. Evolution of Neurohormonal Tests, Body Weight Loss and Hemodynamics of Both Groups
NOTE. Results are given as mean (SEM).
Significantly different from baseline value (P < .01).
Significantly different from the result obtained in group 1 (P < .01).
Significantly different from the result obtained on day 8 in group 2 (P < .01). Normal values: plasma renin: 7-76 μU/mL; plasma aldosterone: 30-100 pg/mL; plasma norepinephrine: 185-275 pg/mL. Abbreviations: UnaV, urinary sodium excretion; MAP, mean arterial pressure.
In group 1, 10 days after additive treatment with 200 mg spironolactone daily, plasma renin, aldosterone, and norepinephrine concentrations were increased (P < .01). The other parameters were unchanged. In group 2 (n = 31), when administered alone over 8 days, clonidine decreased plasma norepinephrine concentrations (P < .01). No effects on natriuresis or body weight were observed. In this period, clonidine decreased MAP and heart rate (P < .01) and increased the activity of the renin/aldosterone axis (P < .01).
In group 2, after 10 days of clonidine/spironolactone (200 mg daily) additive treatment, hemodynamic changes were not found. Plasma renin and aldosterone levels decreased to values lower than the baseline ones (P < .01) resulting in increased sodium excretion and significant body weight loss. Plasma norepinephrine continued decreasing from day 8 until day 18. At this time, all patients had diuretic response with body weight loss. In group 2, after 8 days and compared to the placebo group, plasma norepinephrine, heart rate, and MAP were lower (P < .01) and plasma renin and aldosterone were higher (P < .01).
In group 2, after 10 days of clonidine/spironolactone treatment and compared to the placebo group, plasma renin, aldosterone, and norepinephrine were lower (P < .01); natriuresis and body weight loss were higher (P < .01). Table 3 shows the mean diuretic dosage requirements, complications, and the total time spent in the first hospitalization for both groups. The mean requirement for diuretics was significantly higher (P < .01) in group 1 (spironolactone 284.09 ± 76.6 mg/day, furosemide 15.3 ± 4.14 mg/day) than in group 2 (spironolactone 200 mg ± 0 mg/day, furosemide 0 mg/day).
Table 3. Mean Diuretic Complications, Dosages and Total Time in the First Hospitalization of Both Groups
NOTE. Results are given as mean (SEM). Significantly different from the result obtained in group 1:
The diuretic complications and, in particular, hyperkalemia and renal impairment were lower (P < .001) in group 2 than in group 1. The number of large paracentesis plus iv albumin procedures were higher (P < .01) in group 1 (5 paracentesis) than in group 2 (0 paracentesis). The total time needed for diuretic response, before patient was discharged, was significantly shorter (P < .01) in group 2 (23.06 ± 0.12 days) than in group 1 (31.97 ± 0.56 days). We could not demonstrate side effects related to clonidine alone or in association with diuretic, in particular, symptomatic hypotension, dry mouth, or depression.
Results During the Follow-up (3 Months).
No discharged patient was lost during the follow-up period. Five patients of group 1 and 4 of group 2 continued drinking alcohol during the follow-up. Table 4 shows the time to the first readmission, the number of readmissions, mean diuretic dosages, and complications of both groups during the follow-up period. The time to the first readmission for tense ascites was significantly shorter (P < .01) in group 1 (24.5 2 days) than in group 2 (41.8 6.9 days). The total number of readmissions and the admissions related to ascites or diuretic complications (hyperkalemia and renal impairment) were significantly lower (P < .01) in group 2 than in group 1. Fourteen patients of group 1 required 25 hospitalizations. Six patients (3 with refractory ascites) were readmitted for tense ascites and needed 10 large paracentesis plus iv albumin. Other reasons for readmissions were bacterial infections and gastrointestinal hemorrhages. In group 2, 4 patients required 6 hospitalizations; 1 patient with refractory ascites was rehospitalized for tense ascites.
Table 4. Time to First Readmissions, Mean Diuretic Dosages, and Complications During Follow-up of Both Groups
NOTE. Results are given as mean (SEM). Significantly different from the result obtained in group 1:
The mean diuretic requirement was significantly higher (P < .001) in group 1 (spironolactone 350 ± 9.28 mg/day, furosemide 25.99 ± 7.31 mg/day) than in group 2 (spironolactone 260.16 ± 11.68 mg/day, furosemide 3.13 ± 2.21 mg/day). No side effects of clonidine were noticed during the follow-up. Two patients in group 2 died; the causes of death were gastrointestinal hemorrhage and bacterial infection. No patients in group 1 died..
This study focused on a group of patients with ascites who presented with increased SNS activity and elevated plasma norepinephrine concentrations resulting in a low urinary sodium excretion (Table 1). The results show that clonidine induced a prolonged sympathetic inhibition as indicated by the continuous decrease in plasma norepinephrine (Table 2).
The effects of clonidine on systemic hemodynamics and on renin/aldosterone axis were divided into 2 phases. In the first phase, after 8 days, clonidine decreased heart rate and MAP. Arterial pressure decreased because the reduction in cardiac output was not counterbalanced by reflex vasoconstriction. The concomitant increases in plasma renin and aldosterone can be seen as a partial compensatory response to the reduction in effective volemia, which took place despite the known lowering effect of clonidine on renin secretion.14 At this time, clonidine did not increase natriuresis because clonidine sympathetic inhibition of sodium reabsorption especially occurs in proximal tubules. Consequently, endogenous antinatriuretics like aldosterone induced distal reabsorption of sodium, counterbalancing the natriuretic effect of clonidine.
During the second phase, after an 18-day treatment with clonidine and 10 days after introduction of spironolactone, MAP and heart rate returned to baseline. The reason why clonidine did not induce chronic hemodynamic modifications cannot be clearly explained. Roulot et al. showed that clonidine pharmacokinetics were similar in subjects with cirrhosis and in normal subjects.8 They noticed that MAP and heart rate returned to baseline after 1 week. The differences with our results might be explained by the single oral dose of 0.150 mg daily in Roulot's study and also by different patient characteristics such as 3 patients with cirrhosis with refractory ascites in our study and none in the French study. Our results suggest that compensatory mechanisms might protect patients with cirrhosis against arteriolar hypotensive effects of prolonged sympathetic inhibition. A compensatory increase in cardiac output might contribute to this, but we did not measure it. In addition, studies dealing with hypertension show that the hypotensive long-term effects of clonidine disappeared in some patients because of activation of vascular peripheral alpha-receptors.14, 15
Studies in arterial hypertension show that clonidine depresses renin secretion in the renal juxta-glomerular apparatus.14, 16 In the first phase, the lowering effect of clonidine on renin/aldosterone secretion was concealed by the decrease in arterial pressure. During the second phase, the return of arterial pressure to baseline values reveals the action of clonidine on the renin/aldosterone axis. These results suggested also that clonidine was able to inhibit the activity of renin/aldosterone enhanced by diuretic drugs. Decreased plasma renin results in decreased plasma aldosterone. The effect of direct sympathetic inhibition on adrenal glands or on adrenocorticotropic hormone secretion should also be taken into account.15, 17
Clonidine would improve the effect of spironolactone by decreasing the activity of the renin–aldosterone axis. Indeed, some studies show that clonidine induces a reduction in norepinephrine concentrations associated with a fall in afferent arterial tone and elevation in glomerular filtration rate.7, 9, 10 Increasing the glomerular filtration rate and decreasing the proximal reabsorption of sodium, clonidine would raise the delivery of sodium to the distal nephron. Spironolactone, initially ineffective, would increase natriuresis by impairing distal reabsorption of sodium. As a result, clonidine/spironolactone association increases mobilization of ascites and body weight loss in all patients, even in patients with refractory ascites.18 Studies show that continuous clonidine administration decreases the portosystemic gradient.7, 8 This mechanism might contribute to mobilization of ascites.
In conclusion, in patients with cirrhosis and high plasma norepinephrine concentration, the association of clonidine/diuretics was more effective than diuretics alone in the first treatment of ascites as indicated by the shortened first hospitalization, the lower requirement for diuretics or large paracentesis, and the lower diuretics side effects (hyperkalemia and renal impairment) (Table 3). In the long-term control, clonidine/diuretics was more effective also, as indicated by longer delay for the first readmission, significantly lower readmissions rate, requirement for diuretics or paracentesis, and diuretics side effects (Table 4).