Clinical trial: a randomized controlled study on prevention of variceal rebleeding comparing nadolol + ligation vs. hepatic venous pressure gradient-guided pharmacological therapy

Authors

  • C. VILLANUEVA,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD o Ciberehd), Madrid, Spain
    Search for more papers by this author
  • C. ARACIL,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    Search for more papers by this author
  • A. COLOMO,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD o Ciberehd), Madrid, Spain
    Search for more papers by this author
  • J. M. LOPEZ-BALAGUER,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    Search for more papers by this author
  • M. PIQUERAS,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    Search for more papers by this author
  • B. GONZALEZ,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    Search for more papers by this author
  • X. TORRAS,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD o Ciberehd), Madrid, Spain
    Search for more papers by this author
  • C. GUARNER,

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD o Ciberehd), Madrid, Spain
    Search for more papers by this author
  • J. BALANZO

    1. Gastrointestinal Bleeding Unit, Department of Gastroenterology, Hospital de la Santa Creu i Sant Pau, Autonomous University, Barcelona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD o Ciberehd), Madrid, Spain
    Search for more papers by this author

Dr C. Villanueva, Servei de Patologia Digestiva, Hospital de la Santa Creu i Sant Pau, Avgda. Sant Antoni M̃ Claret, 167, 08025 Barcelona, Spain.
E-mail: cvillanueva@santpau.es

Summary

Background  Hepatic venous pressure gradient (HVPG) monitoring of therapy to prevent variceal rebleeding provides strong prognostic information. Treatment of nonresponders to β-blockers ± nitrates has not been clarified.

Aim  To assess the value of HVPG-guided therapy using nadolol + prazosin in nonresponders to nadolol + isosorbide-5-mononitrate (ISMN) compared with a control group treated with nadolol + ligation.

Methods  Cirrhotic patients with variceal bleeding were randomized to HVPG-guided therapy (= 30) or nadolol + ligation (n = 29). A Baseline haemodynamic study was performed and repeated within 1 month. In the guided-therapy group, nonresponders to nadolol + ISMN received nadolol and carefully titrated prazosin and had a third haemodynamic study.

Results  Nadolol + prazosin decreased HVPG in nonresponders to nadolol + ISMN (< 0.001). Finally, 74% of patients were responders in the guided-therapy group vs. 32% in the nadolol + ligation group (P < 0.01). The probability of rebleeding was lower in responders than in nonresponders in the guided therapy group (P < 0.01), but not in the nadolol + ligation group (P = 0.41). In all, 57% of nonresponders rebled in the guided-therapy group and 20% in the nadolol + ligation group (P = 0.05). The incidence of complications was similar.

Conclusions  In patients treated to prevent variceal rebleeding, the association of nadolol and prazosin effectively rescued nonresponders to nadolol and ISMN, improving the haemodynamic response observed in controls receiving nadolol and endoscopic variceal ligation. Our results also suggest that ligation may rescue nonresponders.

Introduction

Recurrent variceal bleeding remains a leading cause of death in patients with cirrhosis.1 Monitoring the hepatic venous pressure gradient (HVPG) provides strong prognostic information that may be valuable for improving therapy to prevent rebleeding.2,3 Haemodynamic response is adequate when the HVPG decreases to <12 mmHg or by >20% of baseline because it provides a reduction in the risk of haemorrhage to below 10%.4,5 Furthermore, responders have a lower risk of developing other complications of portal hypertension such as ascites and also have better survival.6,7 HVPG monitoring may be particularly useful in the high-risk setting for preventing variceal rebleeding, while it may not be cost effective in primary prophylaxis because of the low rate of haemorrhage achieved with medical therapy.2

Nonselective β-blockers are the mainstay long-term pharmacological treatment in primary and secondary prevention of variceal bleeding.8 However, only 30–40% of patients have a long-term haemodynamic response with this treatment.5,9,10 Nonresponders may be rescued adding isosorbide-5-mononitrate (ISMN).11–13 This association achieves a greater reduction in HVPG than β-blockers alone, increasing the rate of responders up to 40–50% of cases11–13 and does not produce any deleterious effect on renal function or sodium handling even in patients with ascites.14,15 At present, it is not clear which rescue therapy may be effective in patients who do not respond to the association of β-blockers and ISMN.16 Combined treatment with β-blockers and prazosin induces a greater reduction in HVPG than achieved with β-blockers and ISMN, increasing the rate of response to 85%, one of the highest reported to date.17 Concerns have been raised about the safety of prazosin because this drug enhances peripheral arterial vasodilation and induces hypotension favouring sodium and water retention,18,19 although these effects are attenuated by the addition of β-blockers.17 It has not been assessed whether a careful titration of prazosin in patients already treated with β-blockers may be effective to rescue nonresponders to β-blockers and ISMN, while simultaneously avoiding adverse effects.16 This randomized study aimed to investigate whether the association of β-blockers and prazosin may rescue nonresponders to β-blockers and ISMN and to assess the value of this HVPG-guided therapy for preventing variceal rebleeding compared to a control group treated with the current recommended therapy combining β-blockers and endoscopic variceal ligation (EVL).

Methods

Selection of patients

From October 2000 to December 2002, 438 patients were admitted to our hospital because of gastrointestinal bleeding. All underwent emergency endoscopy and 83 patients with cirrhosis and haemorrhage from oesophageal varices were eligible for the study. Cirrhosis was diagnosed on the basis of previous liver biopsy or clinical, biochemical and ultrasonographic findings. Variceal bleeding was diagnosed according to consensuated criteria.20

Twenty-four patients were excluded from the trial because of a Child–Pugh score >12 (n = 5), advanced hepatocellular carcinoma (n = 3), previous endoscopic therapy (n = 5) or treatment with β-blockers and ISMN (n = 4), failure of medical therapy in controlling acute bleeding (n = 5) or refused consent (n = 2).

Study protocol, randomization and treatment

On the fifth day of admission, 59 patients were randomly assigned to one of two groups using opaque and sealed envelopes that contained the treatment assigned as derived from computer-generated random numbers. Randomization was stratified according to the severity of liver failure (Child–Pugh class A or B vs. C). One group was treated with nadolol and EVL. The second group received pharmacological therapy tailored according to HVPG monitoring. Written informed consent was obtained from all the patients or their next of kin. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethics committee of our hospital.

A baseline haemodynamic study was performed before randomization and nadolol was started immediately thereafter in all the patients included. Nadolol was given orally at an initial dosage of 80 mg once daily and was then adjusted over 5 days, to reach the maximum tolerated dose without reducing the resting heart rate below 55 beats/min. A second haemodynamic study was conducted 2–4 weeks later to assess response.

In the HVPG-monitoring guided therapy group, oral ISMN was started immediately after nadolol. Over the course of 1 week, the dose was progressively increased up to 40 mg twice a day unless side effects appeared. In patients who had no-haemodynamic response in the second study and who had a Child–Pugh score <11, ISMN was changed to prazosin. The dose of prazosin was increased stepwise at 3-day intervals from 1 mg at bedtime to a maximum of 4 mg twice daily. The dose was not increased if side effects (such as weakness or dizziness) developed or if systolic blood pressure fell to 90 mmHg or less. The doses of ISMN and prazosin were modified in visits to the out-patient clinic. At each follow-up visit, heart rate and blood pressure were measured every 5 min for 30 min, tolerance and signs of fluid retention were investigated and adherence to the regimen was assessed through careful questioning of patients and relatives. In patients receiving prazosin, a third haemodynamic study was performed 1–2 months after the second study to assess the final response.

In the nadolol + EVL group, ligation was performed with the use of a commercial multibanding device. Each varix was ligated at least once. Up to 14 bands were placed within the lower oesophagus in the first session and up to seven bands in the following procedures. Sessions were conducted at the time of index bleeding and every 3–4 weeks until variceal eradication. The varices were considered eradicated when they had either disappeared or could not be grasped and banded by the ligator. Three and six months after eradication, a follow-up endoscopy was performed and additional sessions of ligation were conducted if varices reappeared. This process was repeated every 12 months thereafter.

Haemodynamic studies and definition of endpoints

Haemodynamic studies were performed after an overnight fast. Under local anaesthesia, a catheter introducer was placed in the right internal jugular vein by the Seldinger technique and was used to advance, under fluoroscopy guidance, a 7F balloon-tipped catheter into the right main hepatic vein and a Swan–Ganz catheter into the pulmonary artery. Portal pressure was measured as the HVPG, the difference between wedged and free HVP. All intravascular pressure measurements were performed in triplicate using a previously calibrated highly sensitive transducer, with external zero at the mid-axillary line and a permanent recording of tracings was obtained. The occluded position was checked by the absence of reflux after injection of contrast medium. Haemodynamic response was defined as a decrease in HVPG below 12 mmHg or >20% of baseline value. Cardiopulmonary pressures and cardiac output were also measured. Electrocardiography, mean arterial pressure, heart rate and oxygen saturation were monitored non-invasively throughout the study with an automatic vital signs monitor.

Rebleeding was defined as any episode of haematemesis and/or melaena that occurred during the follow-up and was evaluated by emergency endoscopy. In both groups, the index haemorrhage and rebleeding episodes were treated with somatostatin and emergency sclerotherapy or EVL. Therapeutic failure was defined as the occurrence of a single rebleeding episode that required the transfusion of 2 units of red cells.

Sample size and statistical analysis

The primary endpoint was the achievement of haemodynamic response and this was used to estimate sample size, which was calculated on the assumption of a 40% rate of response in the nadolol + EVL group.9,10 To detect a difference between groups of at least 40% with the use of a two-tailed test17 at alpha and beta levels of 0.05 and 0.15 respectively, 25 patients were required in each therapeutic group.

Categorical variables were compared with the Fisher’s exact test. Continuous variables, reported as mean ± standard deviation, were compared using the Student’s t test for paired data within each group. Comparisons between groups were performed by the unpaired Student’s t-test with Bonferroni correction for multiple comparisons or with the nonparametric Mann–Whitney rank-sum test. Actuarial probabilities were calculated by the Kaplan–Meier method and compared with the log-rank test. Follow-up was censored at the date of death, liver transplantation or last visit. Significance was established at P < 0.05. All P-values were two-tailed. Calculations were performed with the spss 13.0 statistical package (SPSS Inc., Chicago, IL, USA).

Results

Twenty-nine patients were randomized to the nadolol + EVL group and 30 patients to the guided therapy group (Figure 1). Baseline data were similar in the two groups (Tables 1 and 2). The second haemodynamic study was not performed in seven patients in the nadolol + EVL group because of previous failure or death in two, follow-up loss in one and refused consent in the remaining four patients. The second study was not performed in three patients in the HVPG-guided therapy group because of previous failure or death. The mean dose of nadolol was similar in both groups (Table 2). In the nadolol + EVL group, a median of four sessions of ligation were performed (range 1–7). In the guided therapy group, the mean dose of ISMN was 45 ± 23 mg/day and the mean dose of prazosin was 5 ± 2 mg/day. The maximal intended dose of prazosin was achieved in only three of 13 patients (23%). The mean follow-up period was 34 ± 19 months (Table 1).

Figure 1.

 Flow-chart of the study. During the study period, 438 patients were admitted with gastrointestinal bleeding and a total of 83 met the eligibility criteria. Of these, 24 had one or more exclusion criteria and 59 were randomized. Three patients had absolute contraindications to β-blockers in addition to other exclusion criteria. The second haemodynamic study was not performed in seven patients in the nadolol + EVL group (one patient had a rebleeding episode in the 10th day) and in three patients in the HVPG-guided therapy group (one patient had a rebleeding episode in the 23th day). Finally, haemodynamic response was assessed in 22 patients in the nadolol + EVL group and in 27 patients in the HVPG-guided therapy group.

Table 1.   Characteristics of patients of the two groups at admission and during follow-up
 Nadolol + EVL group (n = 29)Guided therapy group (n = 30)
  1. Plus/minus values are means ± s.d. No differences between groups were statistically significant.

  2. EVL, endoscopic variceal ligation.

  3. * Grade 1 denotes varices that were flattened by insufflation, grade 2 varices that were not flattened by insufflation and were separated by areas of normal mucosa and grade 3 confluent varices that were not flattened by insufflation.

  4. † Four patients in the nadolol + EVL group were referred for orthotopic liver transplantation between the 20th and the 33rd month of follow-up. Four patients in the hepatic venous pressure gradient (HVPG)-guided therapy group were referred for liver transplantation between the fifth and the 44th month of follow-up.

  5. ‡ The follow-up was only for 1 month in one patient in the nadolol + EVL group who was lost to follow-up after discharge and in two patients in the HVPG-guided therapy group (both died).

  6. § Three patients in the nadolol + EVL group were lost to follow-up between 1 and 12 months and three patients in the HVPG-guided therapy group were lost to follow-up between 6 and 8 months.

Baseline
 Age (years)64 ± 1262 ± 12
 Gender (men/women)19/10 19/11
 Alcoholic cirrhosis (%)10 (34)13 (43)
 Viral cirrhosis (%)11 (38)9 (30)
 Alcoholic + viral cirrhosis (%)5 (17)3 (10)
 Previous bleeding (%)7 (24)4 (13)
 Child–Pugh class (A/B/C) 1/24/44/20/6
 Active haemorrhage (%)7 (24)6 (20)
 Associated diseases (%)16 (55)13 (43)
 Transfusions during index episode (units of packed red cells)
  Mean3.4 ± 2.43.8 ± 3.5
  Median33
  Range0–110–14
 Ascites (%)22 (76)20 (67)
 Encephalopathy (%)8 (28)9 (30)
 Variceal grade (1/2/3)*0/9/200/9/21
 Biochemical values
 Bilirubin (μmol/L)28 ± 1332 ± 28
 Albumin (g/L)27 ± 630 ± 18
 Prothrombin time (INR)1.4 ± 0.21.4 ± 0.3
 Platelet count (×109/L)105 ± 48112 ± 48
 Haemoglobin (g/L)86 ± 2289 ± 23
Follow-up
 Nadolol dose (mg/day)83 ± 3098 ± 44
 Abstinence from alcohol (%)10/15 (67)12/16 (75)
 Hepatocellular carcinoma (%)7 (24)4 (13)
 Liver transplantation (%)†4 (14)4 (13)
 Duration of follow-up (months)
  Mean35 ± 1934 ± 21
  Median3337
  Range‡1–691–64
 Lost to follow-up (%)§3 (10)3 (10)
Table 2.   Changes in haemodynamic variables and liver and renal functions in the two treatment groups
 Nadolol + EVL groupGuided therapy group
BaselineFollow-up*BaselineFollow-up†
  1. Results expressed as mean ± s.d.

  2. EVL, endoscopic variceal ligation.

  3. * In the nadolol + EVL group, the follow-up study was performed in 22 patients after 40 ± 13 days from the baseline study (median 42, range 15–63).

  4. † In the hepatic venous pressure gradient-guided therapy group, the final follow-up study was performed in 27 patients. In 13 patients treated with nadolol + prazosin, this was the third study performed after 81 ± 14 days from baseline measurements (median 78, range 45–98). In the remaining 14 patients treated with nadolol + isosorbide-5-mononitrate, this was the second study performed after 23 ± 10 days from baseline measurements (median 24, range 10–34).

  5. ‡ P < 0.01 for the comparison with the baseline value.

  6. § P ≤ 0.05 for the comparison between values at follow-up of both groups.

  7. ¶ P < 0.05 for the comparison with the baseline value.

Wedged hepatic venous pressure (mmHg)32 ± 528.2 ± 5‡31.8 ± 628.1 ± 5‡
Free hepatic venous pressure (mmHg)10.4 ± 39.7 ± 411.3 ± 412 ± 4
Hepatic venous pressure gradient (mmHg)21.6 ± 418.5 ± 4‡20.4 ± 516.1 ± 4‡§
Cardiac output (L/min)8.4 ± 27 ± 2‡8.7 ± 26.8 ± 2‡
Mean arterial pressure (mmHg)81.3 ± 1074.6 ± 11¶76.8 ± 970.4 ± 10‡
Heart rate (beats/min)86 ± 1167 ± 12‡89 ± 1265 ± 10‡
Systemic vascular resistance (dyne/s/cm5)787 ± 297843 ± 279714 ± 212809 ± 229¶
Pulmonary artery pressure (mmHg)17 ± 619.1 ± 616.7 ± 520.1 ± 5¶
Pulmonary wedge pressure (mmHg)9.9 ± 411.4 ± 510.3 ± 413.3 ± 5¶
Right atrial pressure (mmHg)5.4 ± 36.8 ± 46.2 ± 38.1 ± 4‡
Body weight (kg)69 ± 1368.4 ± 1471 ± 1171.8 ± 11
Child–Pugh score 7.9 ± 1.36.9 ± 1.8¶8.2 ± 1.76.5 ± 1.8‡
MELD score11.4 ± 410.9 ± 4 10.1 ± 39.8 ± 3
Serum sodium (mEq/L) 136 ± 4137 ± 3137 ± 4137 ± 3
Serum potassium (mEq/L) 3.9 ± 0.54.3 ± 0.4¶4 ± 0.54.2 ± 0.3
Serum urea (mmol/L) 8.3 ± 56.6 ± 2¶7.4 ± 56.3 ± 2
Serum creatinine (μmol/L) 88.7 ± 2294.7 ± 1788.3 ± 2491.4 ± 14

Splanchnic haemodynamic

In the nadolol + EVL group, the HVPG decreased significantly (Table 2), with a mean reduction of 12 ± 21%. In the guided therapy group, the HVPG decreased from 20.4 ± 5 to 17.7 ± 4 mmHg in patients treated with nadolol and ISMN (P < 0.001). The administration of nadolol + prazosin to nonresponders to nadolol and ISMN further reduced the HVPG to 15.2 ± 2 mmHg (P < 0.001; Table 3). In the guided therapy group, the final HVPG was significantly lower than in the nadolol + EVL group (Table 2) and the mean decrease in HVPG was 22 ± 11% (= 0.01 as compared with the nadolol + EVL group).

Table 3.   Changes in haemodynamic variables and renal function in the subgroup of patients treated with nadolol and prazosin
 Guided therapy group: subgroup treated with nadolol and prazosin (n = 13)*
BaselineNadolol + ISMNNadolol + prazosin
  1. Results expressed as mean ± s.d.

  2. ISMN, isosorbide-5-mononitrate.

  3. * In the hepatic venous pressure gradient-guided therapy group, 13 patients with nonresponse to nadolol and ISMN were treated with nadolol and prazosin. This table includes baseline values in these patients as well as values with nadolol and ISMN and values obtained with nadolol and prazosin.

  4. † P < 0.05 for the comparison with the baseline value.

  5. ‡ P < 0.01 for the comparison with the baseline value.

  6. § P < 0.05 for the comparison with the value obtained with nadolol + ISMN.

  7. ¶ P = 0.07 for the comparison with the value obtained with nadolol + ISMN.

Wedged hepatic venous Pressure (mmHg)31.4 ± 528.2 ± 4†27.6 ± 3‡
Free hepatic venous pressure (mmHg)11.3 ± 49.6 ± 312.4 ± 4¶
Hepatic venous pressure gradient (mmHg)20.1 ± 318.6 ± 2 15.2 ± 2‡§
Cardiac output (L/min)9.4 ± 36.6 ± 2‡6.7 ± 2‡
Mean arterial pressure (mmHg)74.2 ± 871.7 ± 1168.6 ± 10†
Heart rate (beats/min)89 ± 1361 ± 7‡64 ± 6‡
Systemic vascular resistance (dyne/s/cm5)637 ± 183862 ± 299‡751 ± 244
Pulmonary artery pressure (mmHg)16 ± 518.1 ± 420.1 ± 5†
Pulmonary wedge pressure (mmHg)10.7 ± 511.7 ± 413.1 ± 5†
Right atrial pressure (mmHg)6.4 ± 46.5 ± 38.7 ± 4‡
Body weight (kg)72.9 ± 874.4 ± 774.7 ± 7
Serum sodium (mEq/L) 138 ± 5137 ± 2138 ± 3
Serum potassium (mEq/L) 4 ± 0.64.3 ± 0.44.1 ± 0.4
Serum urea (mmol/L) 6.6 ± 35.8 ± 2 5.9 ± 2
Serum creatinine (μmol/L) 88.7 ± 1293.4 ± 1596.1 ± 12

In the nadolol + EVL group, seven of the 22 patients (32%) who completed the study were responders. In the guided therapy group, 10 of 27 patients (37%) were responders to nadolol and ISMN. Thirteen of the 17 nonresponders had a Child–Pugh score <11 and were treated with nadolol and prazosin. Response was achieved in 10 of these patients (77%; Figure 1). Finally, 20 of the 27 patients (74%) who completed the study in the guided therapy group were responders, a significantly higher rate than observed in the nadolol + EVL group (= 0.004). Among responders, the final HVPG was <12 mmHg in two patients (9%) in the nadolol + EVL group (both had a decrease >20%) and in six patients (22%) of the guided therapy group (four had a decrease >20%).

Systemic haemodynamic

The degree of β-blockade was similar in the two groups as shown by decreases in heart rate and cardiac output (Table 2). Mean arterial pressure decreased significantly in both groups, with a mean reduction of 7 ± 16% in the nadolol + EVL group and 8 ± 19% in the guided therapy group (= 0.81). Cardiopulmonary pressures and systemic vascular resistance increased significantly in the guided therapy group, although the final value of these parameters was similar to that observed in the nadolol + EVL group (Table 2).

In the guided therapy group, the administration of nadolol and prazosin to nonresponders to nadolol and ISMN induced a further nonsignificant decrease in mean arterial pressure (from 71 ± 11 to 69 ± 11 mmHg, P = 0.21). Similarly, slight nonsignificant changes were observed in the subgroup of patients treated with nadolol and prazosin compared with those in the nadolol + ISMN group when considering cardiopulmonary pressures, cardiac output, systemic vascular resistance, body weight and renal function (Table 3).

Complications and subsidiary outcome measures

The overall rate of complications was similar in the two groups (Table 4). The incidence of serious side effects was higher in the nadolol + EVL group, although the difference was not significant. The global rate of minor side effects was similar (Table 4). However, in the guided therapy group, 13 patients (43%) presented drug-related complications such as weakness, dizziness, headache or impotence vs. five (17%) in the nadolol + EVL group (P = 0.03).

Table 4.   Complications related to therapy
 Nadolol + ligation groupGuided therapy group RR (95% CI)P-value
  1. RR, relative risk; CI, confidence interval.

  2. * In the nadolol + endoscopic variceal ligation (EVL) group, 16 patients had 18 adverse events of which five were serious. One patient in this group had both a major complication in the first month (oesophageal bleeding ulcer) and a minor complication in the fifth month (weakness). In the hepatic venous pressure gradient (HVPG)-guided therapy group, 14 patients had 15 adverse events.

  3. † In the nadolol + EVL group, five patients had serious complications between the first and the 44th month. In the HVPG-guided therapy group, one patient had a major complication within the first month.

  4. ‡ In the nadolol + EVL group, 12 patients had 13 minor complications. In the HVPG-guided therapy group, 13 patients had 14 minor complications.

Overall16 (55%)14 (47%)1.4 (0.5–3.9)0.61
Total no. side effects*1815  
Major side effects† 5 (17%) 1 (3%)6.1 (0.7–42)0.1
 Aspiration pneumonia 1 0  
 Bradyarrhythmia 1 1  
 Oesophageal bleeding ulcer 3 0  
Minor side effects‡12 (41%)13 (43%)0.9 (0.3–2.6) 1.0
 Chest pain 4 0  
 Transient dysphagia 3 0  
 Weakness 3 6  
 Dizziness (hypotension) 2 4  
 Headache 0 2  
 Impotence 1 2  

Kidney function varied in similar fashion in the nadolol + EVL group and in the guided therapy group and laboratory values remained within the normal range (Table 2). Treatment with nadolol and prazosin in nonresponders to nadolol and ISMN did not induce significant changes in renal function (Table 3).

Ascites developed during the follow-up in 19 patients (68%) in the nadolol + EVL group (five of them without previous history) compared with 15 (58%) in the guided therapy group (four without previous history) (P = 0.57). In the subgroup treated with nadolol and prazosin, six patients (46%) had ascites (two de novo).

The Child–Pugh score in the third month of follow-up decreased significantly and similarly in both groups.

Rebleeding and survival

Eight patients (27%) rebled in the guided therapy group compared with six (21%) in the nadolol + EVL group (P = 0.45, log-rank test). Oesophageal varices were the most frequent site of rebleeding (Table 5).

Table 5.   Rebleeding and mortality during follow-up in the study groups
 Nadolol + EVL group (n = 29)Guided therapy group (n = 30)P-value
  1. Plus/minus values are means ± s.d. P-value obtained by log-rank test for the comparison between groups.

  2. EVL, endoscopic variceal ligation; HVPG, hepatic venous pressure gradient.

  3. * One patient in each group rebled within the first month of follow-up, before the second HVPG measurement.

  4. † In the nadolol + EVL group, five patients had seven rebleeding episodes from oesophageal varices. In the HVPG-guided therapy group, seven patients had eight rebleeding episodes from oesophageal varices.

  5. ‡ In the nadolol + EVL group, two patients had two rebleeding episodes from oesophageal ulcers; one of them also had a rebleeding episode from oesophageal varices.

  6. § In the HVPG-guided therapy group, one patient had a rebleeding episode from portal hypertensive gastropathy.

  7. ¶ Treatment failed in five patients in the nadolol + EVL group; three received a TIPS, one was treated with nadolol and isosorbide-5-mononitrate + EVL and the remaining one received no further treatment. Treatment failed in eight patients in the HVPG-guided therapy group; five were then treated with EVL, two received a TIPS (one of them was referred for liver transplantation 4 months later) and the remaining one received no other treatment.

  8. ** In the nadolol + EVL group, five patients without the second HVPG measurement died between the fourth and the 33rd month (death was the reason for missing the second measurement in one patient). In the HVPG-guided therapy group, the three patients without the second HVPG measurement died between the first and the fifth month (death was the reason for missing the second measurement in two patients and in the remaining patient it occurred after a rebleeding episode).

Rebleeding
 Overall (%)6 (21)8 (27)0.45
 In haemodynamic responders1/7 (14)3/20 (15)0.85
 In haemodynamic nonresponders4/15 (27)4/7 (57)0.05
 No second HVPG measurement*1/7 (14) 1/3 (33) 0.13
Total no. rebleeding episodes99 
Site of rebleeding
 Oesophageal varices†5 (17)7 (23)0.43
 Oesophageal ulcer‡2 (7)00.18
 Portal hypertensive gastropathy§01 (3)0.37
Therapeutic failure (%)¶5 (17)8 (27)0.31
Mortality
 Overall (%)18 (62)13 (43)0.19
 In haemodynamic responders3/7 (43)5/20 (25)0.51
 In haemodynamic nonresponders10/15 (67)5/7 (71)0.58
 No second HVPG measurement**5/7 (71) 3/3 (100)0.08

In the HVPG-guided therapy group, the probability of rebleeding was significantly greater in nonresponders than in responders (62% vs. 12% at 3 years, P = 0.006). In the nadolol + EVL group, the probability of rebleeding was also greater in nonresponders, although the difference was not significant (29% vs. 5% at 3 years, P = 0.41). Among responders, the likelihood of rebleeding was similar in the two groups (Table 5). Among nonresponders, the likelihood of rebleeding was significantly greater in the guided therapy group than in the nadolol + EVL group (Figure 2). Among patients without the second haemodynamic study, rebleeding was slightly more common in the guided therapy group (Table 5).

Figure 2.

 Probability of rebleeding in haemodynamic responders (a) and in nonresponders (b) in the two treatment groups. (a) The probability of remaining free from any rebleeding in haemodynamic responders was similar in the two groups (P = 0.85). (b) The probability of remaining free from any rebleeding in nonresponders was significantly higher in the Nadolol + EVL group than in the HVPG-guided therapy group (P = 0.05).

Thirteen patients (43%) died in the guided therapy group compared with 18 patients (62%) in the nadolol + EVL group (P = 0.09, log-rank test). Factors contributing to death were similar in both groups: rebleeding (two patients in the guided therapy group vs. one in the nadolol + EVL group, P = 0.55), liver failure (5 vs. 11, P = 0.12), hepatocellular carcinoma (two vs. five, P = 0.24) and other causes (four vs. one).

Discussion

This study shows that HVPG-guided pharmacological therapy using nadolol and prazosin to rescue nonresponders to nadolol + ISMN significantly improves the haemodynamic response observed in a control group treated with nadolol and EVL. The association of β-blockers and EVL is currently the recommended first-line therapy to prevent recurrent variceal bleeding.20 Controlled trials have shown that it improves the efficacy of EVL alone, achieving rebleeding rates as low as 14–23%.21,22 However, evidence suggests that haemodynamic response may be a relevant therapeutic target, as it is linked with improved efficacy and survival.6,7 Although there has been some controversy regarding the prognostic value of a reduction in HVPG >20% from baseline, this has mainly been based on a single study23 the methodology of which is controversial.24 Furthermore, a systematic review of all available trials assessing HVPG monitoring clearly confirmed the value of this threshold.3

The addition of ISMN to β-blockers enhances the fall in portal pressure achieved using only β-blockers12,13,25 and this combined therapy has shown efficacy in preventing variceal rebleeding.25–28 Responders to β-blockers have no further decrease in HVPG with the addition of vasodilators and the beneficial effects are restricted to nonresponders.11,17 It has been suggested that it may be useful to adapt medical therapy according to HVPG monitoring, adding ISMN to β-blockers in individual nonresponders.29 However, no rescue therapy has yet proven to be efficacious in the high-risk subset of patients who do not respond to β-blockers and ISMN.16 Our results suggest that the addition of prazosin to β-blockers may be beneficial to rescue previous nonresponders to first-line therapy with β-blockers + ISMN. It should be pointed out that this study was conducted in patients treated to prevent variceal rebleeding, a high-risk setting with a greater probability of haemorrhage and a lower likelihood of achieving haemodynamic response compared with primary prophylaxis.30 However, the applicability of HVPG-monitoring is still limited, as it is not widely available and whether or not HVPG-guided therapy may improve clinical practice should be established in future adequately sized studies.

Nonselective β-blockers induce an increase in portocollateral and intrahepatic resistance that may hinder the reduction in portal pressure induced by the decrease in portal venous inflow.31,32 The addition of vasodilators to β-blockers counteracts the increase in portohepatic resistance.12,17 This study suggests that shifting to an α1-adrenoceptor blocker such as prazosin may be effective in nonresponders to NO donors such as ISMN. In the current subset of nonresponders to nadolol + ISMN, the association of nadolol + prazosin induced a further significant decrease in HVPG and markedly increased the rate of response in up to 77% of cases. Such a high rate of haemodynamic response is comparable with that observed in a previous study in which the combination of β-blockers and prazosin was used as first-line therapy in all cirrhotics with varices (with or without previous bleeding) achieving a greater reduction in HVPG than obtained with propranolol and ISMN, although with more side effects.17 This study shows that restricting the association of β-blockers + prazosin to patients with previous variceal bleeding and nonresponse to β-blockers + ISMN may be equally effective and may also avoid side effects in many cases.

A potentially detrimental effect of vasodilators in advanced cirrhosis is the risk of aggravating the already present systemic hypotension and relative hypovolaemia, as this may adversely affect sodium retention and renal function.31 These adverse effects are markedly attenuated by the addition of nonselective β-blockers,17 possibly because of the effect of β-blockade enhancing peripheral resistance and suppressing renin release.33 Untoward consequences on renal function have not been noted with the combination of propranolol and prazosin in cirrhosis.17 However, this combination is still associated with frequent side effects such as asthenia or dizziness. More importantly, it also induces a greater reduction in arterial pressure than propranolol + ISMN and increases fluid retention and body weight.17 To avoid these complications, we introduced prazosin, once β-blockade had been achieved; it was not administered in advanced liver dysfunction and a low dose carefully titrated against blood pressure and tolerance was used. With this approach, we observed a slightly significant decrease in arterial pressure that was not different from that observed with β-blockers + EVL and no significant changes were noted on renal function, ascites management or body weight. It should be noted that such an acceptable safety profile was accompanied by a marked effect on HVPG similar to that observed using a higher dose of prazosin in association with β-blockers.17 However, even when associated with β-blockers, the use of prazosin should be cautious and restricted to controlled trials. Combined therapy with β-blockers and prazosin requires very close monitoring. In addition, the safety profile in cirrhosis is narrow, particularly in advanced disease and may limit its utility. Indeed, we found that drug-related side effects were more frequent with this treatment.

The better haemodynamic response we observed in the HVPG-guided therapy group did not translate into an improvement in preventing recurrent bleeding. However, the sample size was calculated on the probability to reach haemodynamic response and secondary outcomes, such as rebleeding, may be biased by both type I and type II errors. The rebleeding incidence was low in patients treated with nadolol + EVL and the probability of rebleeding among nonresponders was lower in patients treated with nadolol + EVL than in those in the HVPG-guided therapy group suggesting that EVL may be effective in rescuing nonresponders. Opposite results were obtained using this approach in another study.29 Such a discrepancy may be because of the small series of patients treated in both studies.

In conclusion, this study shows that in patients treated to prevent variceal rebleeding, the association of nadolol and prazosin effectively rescues nonresponders to nadolol and ISMN significantly improving the rate of haemodynamic response observed in a control group treated with nadolol and EVL. The incidence of complications and the final mean arterial pressure were similar with both treatments, renal function did not change and ascites development was also similar. Furthermore, our results suggest that EVL may also be useful to rescue nonresponders.

Acknowledgements

Declaration of personal interests: There is no commercial affiliation or consultancy of any author that could be construed as a conflict of interest with respect to the submitted data. Declaration of funding interests: This study has been partially funded by a grant from the Instituto de Salud Carlos III (PI05/1574 and PI02/0398) and AGAUR (2005SGR01085) of Generalitat de Catalunya.

Ancillary