Article Text


Somatostatin plus isosorbide 5-mononitrate versus somatostatin in the control of acute gastro-oesophageal variceal bleeding: a double blind, randomised, placebo controlled clinical trial
  1. F Junqueraa,
  2. J C López-Talaverab,
  3. F Mearina,
  4. E Saperasa,
  5. S Videlaa,
  6. J-R Armengola,
  7. R Estebanb,
  8. J-R Malageladaa
  1. aDigestive System Research Unit, Hospital General Universitari Vall d'Hebron, Autonomous University of Barcelona, Spain, bLiver Unit
  1. Dr F Mearin, Digestive System Research Unit, Hospital General Vall d'Hebron, Autonomous University of Barcelona, Pº Valle de Hebrón 119-129, 08035 Barcelona, Spain.


BACKGROUND Variceal bleeding is a severe complication of portal hypertension. Somatostatin reduces portal pressure by decreasing splanchnic blood flow, and nitrates by diminishing intrahepatic resistance. Experimental studies have shown that the combination of somatostatin and nitrates has an additive effect in decreasing portal pressure.

AIM To compare the therapeutic efficacy of either intravenous infusion of somatostatin plus oral isosorbide 5-mononitrate or somatostatin alone in gastro-oesophageal variceal bleeding associated with liver cirrhosis.

METHODS A unicentre, double blind, placebo controlled, clinical trial was conducted. Sixty patients bleeding from oesophageal or gastric varices were randomised to receive intravenous infusion of somatostatin (250 μg/hour) plus oral isosorbide 5-mononitrate (40 mg/12 hours) (group I) or somatostatin infusion plus placebo (group II) for 72 hours.

RESULTS The two groups of patients had similar clinical, endoscopic, and haematological characteristics. Control of bleeding was achieved in 18 out of 30 patients (60%) in group I and 26 out of 30 patients (87%) in group II (p<0.05). There was no significant difference in mean transfusion requirements between the two groups: 2.6 (2.2)v 1.8 (1.6) respectively; means (SD). Mortality and side effects were similar in the two groups, but development of ascites was higher in group I (30%) than in group II (7%) (p<0.05).

CONCLUSION In cirrhotic patients with acute gastro-oesophageal variceal bleeding, addition of isosorbide 5-mononitrate to somatostatin does not improve therapeutic efficacy, induces more adverse effects, and should not be used.

  • gastro-oesophageal bleeding
  • haemorrhage
  • portal hypertension
  • clinical trial
  • isosorbide 5-mononitrate
  • somatostatin

Statistics from

Variceal bleeding is a major complication of portal hypertension in patients with liver cirrhosis.1 Vasoactive drugs, such as vasopressin and somatostatin (SMS), are the initial treatment of choice for acute variceal bleeding because of effectiveness, availability, and low complication rate.2Most of these drugs decrease portal pressure by reducing portal blood flow through splanchnic vasoconstriction.3 SMS has been shown to be a safe and effective drug in controlling variceal bleeding. Its major advantage is that it avoids the major side effects of vasopressin.4 5 Most trials, which have compared SMS with vasopressin, vasopressin plus nitroglycerine, balloon tamponade, or emergency sclerotherapy, have produced equivalent results.6-11 A different approach to reducing portal pressure is to diminish splanchnic vascular resistance using venous vasodilators such as nitroglycerine or isosorbide 5-mononitrate (Is-5-Mn). Pharmacokinetic characteristics of Is-5-Mn make this drug preferable to other organic nitrates in cirrhotic patients; thus oral administration of Is-5-Mn reduces portal pressure in a dose-dependent, acute, and sustained manner in patients with cirrhosis.12Experimental data have shown that the association of SMS and nitrates has an additive effect in decreasing portal pressure in dogs with portal hypertension.13 Therefore we proposed that addition of Is-5-Mn to SMS would enhance the lowering effect of SMS on portal pressure and hence improve its control of haemorrhage. To test this hypothesis, we conducted a double blind randomised trial to compare the efficacy of SMS plus oral Is-5-Mn with SMS plus placebo in the control of acute variceal haemorrhage and in preventing early rebleeding in cirrhotic patients.



Between September 1995 and February 1997, all cirrhotic patients referred to our institution with the suspicion of gastrointestinal bleeding from oesophageal or gastric varices underwent emergency endoscopy to determine the site of the bleeding and were admitted to our Gastrointestinal Bleeding Unit. Emergency endoscopy was performed during the first eight hours after admission. Endoscopic criteria defining variceal bleeding were as follows: active bleeding from varices, recent stigmata of haemorrhage on varices, or presence of varices and blood in the stomach without other potential source of gastrointestinal bleeding. Exclusion criteria were hepatocarcinoma or any other previously diagnosed malignant neoplasm, endoscopic treatment in the previous four weeks, prior participation in the study, history of side effects to the drugs used in this study, and pregnancy. Patients who fulfilled the inclusion criteria were stratified into two groups according to the Child-Pugh classification: AB and C. Afterwards, each patient was randomly assigned to receive in a double blind manner either SMS plus Is-5-Mn (group I) or SMS plus placebo (group II). Randomisation of patients was carried out in blocks of four participants with the help of the EPISTAT program.


Patients in group I received an initial bolus of SMS (250 μg intravenously) followed by infusion of 250 μg/hour plus Is-5-Mn (40 mg/12 hours orally). Group II patients received an initial bolus of SMS (250 μg intravenously) followed by an infusion of 250 μg/hour plus placebo (one tablet/12 hours orally). The administration of Is-5-Mn or placebo was begun when the patient was haemodynamically stable and systolic blood pressure had been ⩾ 100 mm Hg for at least one hour. Tablets of Is-5-Mn and placebo, which were not distinguishable and had the same organoleptic characteristics, were kindly packed, labelled (protocol code), and donated by Boehringer-Mannheim, Barcelona, Spain. All patients, apart from the medication used in the protocol, were equivalently managed according to the guidelines of our Gastrointestinal Bleeding Unit: norfloxacin 400 mg, one tablet/24 hours orally as prophylaxis against bacterial infection14; ranitidine 300 mg/24 hours intravenously for prevention of acute gastroduodenal ulcers15; lactitol 10 mg/8 hours orally and lactitol enema every 12 hours over 72 hours as prophylaxis for hepatic encephalopathy.16


All patients were monitored continuously for blood pressure, heart rate, temperature, and diuresis; packed cell volume and glycaemia were determined every six hours; transfusion requirements, electrocardiographic changes, and side effects, with special attention to hepatic encephalopathy, ascites, infectious disease such as pneumonia, urinary tract infection, spontaneous bacterial peritonitis, and bacteriaemia, were carefully evaluated throughout treatment. Transfusion requirements were pre-established according to emergency packed cell volume values. Thus 4, 3, 2, or 0 blood units were given when packed cell volume was less than 20, between 20 and 25, between 25 and 30, or more than 30 respectively. Bleeding control was assessed clinically, analytically, and by nasogastric tube aspiration. Endoscopy was performed whenever failure to control bleeding was suspected. Bleeding control was evaluated at the end of 12 hours, 24 hours, 48 hours, and 72 hours. Arrest of bleeding for 72 hours was considered the main end point of the study, and control of bleeding was evaluated at that time. Additional end points were persistence of bleeding, rebleeding, new bleeding episodes, complications, and mortality during the study period. After 72 hours of SMS infusion, and provided that haemodynamic stability was achieved, SMS infusion was stopped and β blocker treatment was initiated. Treatment failure was managed by sclerotherapy and/or balloon tamponade if required. Sclerotherapy was also performed in patients with cardiovascular and respiratory failure and insulin dependent diabetes mellitus in whom β blocker treatment was contraindicated. Patients stayed in hospital for at least one week, and, after discharge, were followed as outpatients at two and four weeks after the bleeding episode.

Definitions for active variceal bleeding, control of bleeding, and rebleeding were established according to the Second Baveno's Consensus Meeting on Portal Hypertension.17 Persistence of bleeding was defined as any bleeding episode within the 48 hour interval from time zero (admission). Evidence of any bleeding after 48 hours (with no evidence of clinically significant bleeding between 24 and 48 hours) was considered to be a new bleeding episode.


As there were no previous studies in the literature using the combination of SMS and Is-5-Mn for the control of gastro-oesophageal variceal bleeding, this clinical trial was planned as a pilot study. Therefore, given its exploratory nature, it was not compulsory to calculate the necessary sample size to confirm any previous hypothesis.18 However, on the basis of collective experience with SMS,19 we assumed a bleeding control rate of 60%. As we had to compare it with the combination with Is-5-Mn and, a priori, this association should represent a major number of adverse effects, we considered that an increased efficacy of the combined treatment of 30% would be clinically acceptable and would compensate for the theoretical increased risk of side effects. With these assumptions, the sample size was calculated to be 50 patients per group, with an overall type I error (α) of 0.05, adjusted for one interim analysis by Pocock's method, and a type II error (β) of 0.1.

As this trial is the first study using the combination SMS plus Is-5-Mn to combat variceal bleeding associated with liver cirrhosis, and, for ethical reasons, it was planned to perform an interim analysis. This intermediate analysis was planned to be performed when half of the total sample was recruited, and the Pocock method20 was used to adjust the study overall significance level. An external group was in charge of performing the interim analysis, preserving the blinding of the investigators, and issuing a recommendation on the continuation of the study based on efficacy and safety criteria.

When the interim analysis had been performed with the first 50 patients, the results showed a large, although not statistically significant, trend against the initial hypothesis. Given these interim results, the “conditional power”—that is, the probability of the plausibility of the initial hypothesis (SMS efficacy 60% and SMS plus Is-5-Mn efficacy 90%) if the study were completed with 100 patients—was then calculated. A total of 100 000 simulations were performed by the stochastic restriction method.21 22 In fact, the probability of reversing the results if the trial had been completed was 3.6%. Based on this probability, the external monitoring group informed the investigators that this trend would be extremely unlikely to be reversed in favour of the initial hypothesis by completing the study. Safety considerations prevailed and it was decided to stop the study at the end of February 1997 for obvious ethical reasons. However, it was decided to complete the investigation on the 10 patients already included in the study while the interim analysis was running, and to reanalyse the data including them in the final report of the study.


The statistical analysis was carried out according to an “intention to treat” analysis (all the patients included in the study). Differences between groups were evaluated by the χ2 test, Fischer's exact test for qualitative variables, and Student's t test for quantitative variables following a normal distribution, or by the Mann-Whitney rank sum test for those who failed the normality test. The corresponding 95% confidence intervals were calculated for the differences between groups. The statistical analysis was performed using Windows SigmaStat software, version 1.0.


The protocol was approved by the hospital ethics committee, and written informed consent was obtained from all patients (or patient's legal representative when shock or encephalopathy was present) enrolled in the study. We had no external financial sources, including pharmaceutical funding, supporting this work.



During the study period, 100 cirrhotic patients bleeding from oesophageal or gastric varices were admitted to our Gastrointestinal Bleeding Unit. Forty patients were excluded from the study: eight because of a previously diagnosed hepatocarcinoma, 10 because initially we could not obtain haemodynamic stability, six were referred from other hospitals to our centre with balloon tamponade, six had already participated in the study, and 10 refused to participate. Thus a total of 60 patients were included in the study. The two treatment groups were similar with regard to age, sex distribution, aetiology, Child-Pugh classification, and associated diseases. Likewise, clinical presentation of gastrointestinal bleeding, packed cell volume, period of bleeding until treatment, and systolic, diastolic, and mean arterial pressure were similar in both groups (table 1). No endoscopic differences were observed with regard to grade and location of varices, presence and type of stigmata of haemorrhage, and portal gastropathy between the groups (table 2). Haematological and biochemical profiles of patients, including packed cell volume, mean corpuscular volume, prothrombin time, platelets, glucose, urea, creatinine, sodium, potassium, calcium, direct bilirubin, total proteins, and albumin (obtained in the emergency room), as well as alkaline phosphatase, γ-glutamyltransferase and γ-globulins (obtained 12 hours after admission), were similar in the two groups. The only exceptions were the aminotransferases, which were significantly higher in the SMS group (aspartate aminotransferase 86 (70) U/l and alanine aminotransferase 71 (70) U/l) than in the SMS + Is-5-Mn group (aspartate aminotransferase 59 (43) U/l and alanine aminotransferase 43 (36) U/l) (p<0.05), and urinary sodium excretion at 72 hours, which was lower in the SMS + Is-5-Mn group than in the SMS group (36 (11)v 93 (20); p<0.05) (all results means (SD)).

Table 1

Demographic and clinical features at entry

Table 2

Endoscopic characteristics of the two groups of patients


Arrest of bleeding for 72 consecutive hours after the onset of treatment was achieved in 18 (60%) patients in the SMS + Is-5-Mn group and in 26 (87%) patients in the SMS alone group (p = 0.039). Rebleeding was similar in SMS + Is-5-Mn group and in the other group (three patients v two patients respectively). Likewise, new bleeding episodes were similar in the first 72 hours (two v one) and during the first month after index variceal bleeding (seven in both groups). Mean transfusion requirements (2.6 (2.19) v 1.8 (1.64) packed red cells) and total fluid replacement (excluding packed red cells) during 72 hours (7851 (909) v7818 (1424) ml respectively) were not different between the two groups. Mortality related to bleeding varices was identical in the SMS + Is-5-Mn group with that in the SMS alone group (two patients). Mortality in three patients was directly related to a bleeding episode, whereas in another patient it was caused by complicating bronchopneumonia (table 3).

Table 3

Therapeutic efficacy of both treatments


Tables 4 and 5 describe in detail the side effects and complications that occurred during the study period. The total number of complications was higher in the SMS + Is-5-Mn group than in the SMS group (43 v 33) but the difference was not statistical. However, its temporal distribution was different. The number of complications was significantly higher in the SMS + Is-5-Mn group during the first week after gastrointestinal bleeding than in the SMS alone group (p<0.05). However, in the follow up period, the number of side effects was greater in the latter group than in the SMS + Is-5-Mn group (p = 0.01). In the SMS + Is-5-Mn group, one patient developed atrial fibrillation, and another respiratory distress during the first week of treatment. Likewise heart failure and bronchospasm, both related to β blockers, were present in two patients of the SMS group during the follow up period. Complications including hepatic encephalopathy, pneumonia, urinary tract infection, spontaneous bacterial peritonitis, bacteraemia, and hepatorenal syndrome were similar in the two groups. However, the appearance of ascites during the total study period was significantly more common in the SMS + Is-5-Mn group (9/30, 30%) than in the SMS alone group (2/30, 7%) (p = 0.04).

Table 4

Side effects and complications occurring in the first week after gastrointestinal bleeding

Table 5

Side effects and complications occurring in weeks 2–4 after gastrointestinal bleeding


The results of this study show that combining SMS and Is-5-Mn produces inferior results to those using SMS alone in the clinical management of acute variceal bleeding in cirrhotic patients.

Treatment of oesophageal-gastric variceal bleeding must aim to both control the presenting haemorrhage and prevent early rebleeding. Collective evidence suggests that patients with high portal pressure are likely to continue to bleed or rebleed early,23 24implying that intensive drug treatment aimed at decreasing portal pressure may be beneficial. Pharmacological treatment is the initial treatment of choice for acute variceal bleeding because of its effectiveness, availability, and low complication rate. SMS is a safe and highly effective drug in the control of variceal haemorrhage. Meta-analysis of the randomised-controlled trials of SMS determined that control of variceal bleeding was significantly greater with SMS than with placebo.25 Moreover, different studies have shown that SMS is as effective as sclerotherapy in controlling bleeding, preventing early rebleeding, and patient survival.11 26 27 Endoscopic variceal banding28 as well as transjugular portosystemic shunting29 may be useful in bleeding control, mainly when pharmacological treatment has failed.

Nitrovasodilators (nitroglycerine, isosorbide mononitrate and dinitrate) decreased hepatic venous pressure gradient and portal pressure by reducing intrahepatic or collateral vascular resistance.12 This effect is mediated by generation of nitric oxide in vascular smooth muscle cells.30 Is-5-Mn has been shown to be as effective as propranolol in preventing initial variceal bleeding31 and reduces the incidence of rebleeding when administered alone during sclerotherapy.32These beneficial effects of Is-5-Mn on portal pressure were not associated with adverse effects on liver function.12 The combination of nitrovasodilators with other vasoactive drugs would be predicted to exert a synergistic effect on reduction of portal pressure. Several studies have shown that the addition of nitrates to vasopressin enhances the reduction in portal pressure while counteracting the adverse systemic haemodynamic effects of vasopressin.33 The addition of Is-5-Mn to β blockers further reduces hepatic venous pressure gradient, and it has been noted that about one third of non-responders to β blockers became responders after the addition of Is-5-Mn.34-37

To test the hypothesis that the addition of Is-5-Mn to SMS enhances the portal pressure lowering effect of SMS and improves its control of haemorrhage, we conducted a double blind randomised trial comparing the efficacy of SMS plus Is-5-Mn with SMS plus placebo in the control of acute variceal haemorrhage and in preventing early rebleeding in cirrhotic patients.

Unexpectedly, control of variceal bleeding was less effective in patients treated with SMS + Is-5-Mn than in those treated with SMS alone. Moreover, patients who received Is-5-Mn developed more complications during the first week, including larger ascites. Only during the follow up after the first week did patients who received Is-5-Mn have fewer complications than those who received placebo. The two groups of patients showed no significant differences with regard to rebleeding incidence, new bleeding episodes, transfusion requirements, and mortality.

The reasons why Is-5-Mn reduces the efficacy of SMS in the control of acute variceal bleeding are unclear. Although SMS has been shown to induce splanchnic vasoconstriction in an in vitro perfused mesenteric artery preparation from normal and portal hypertensive rats, it did not increase splanchnic vascular resistance.38 These data indicate that SMS is not a direct vasoconstrictor. This agent may increase vascular tone by inhibiting secretion of gut derived vasodilatory peptides, such as glucagon, vasoactive intestinal peptide, substance P, and calcitonin gene related peptide.39Nitrates reduce collateral vascular resistance because of their vasodilatory effect on splanchnic circulation. We postulate that this effect may counteract the vasoconstrictor effect of SMS on the splanchnic vascular bed. On the other hand, acute administration of Is-5-Mn may impair systemic haemodynamics and renal function in cirrhotic patients, in particular those with ascites.40Long term administration of this drug, although apparently well tolerated by compensated cirrhotics in whom it does not alter renal plasma flow or glomerular filtration rate, can induce hypotension and sodium retention in cirrhotics with ascites. Vorobioffet al 35 showed that more than 50% of patients with previous or present ascites receiving isosorbide dinitrate presented an impairment in renal sodium metabolism, as reflected by the development or worsening of ascites and increasing diuretic requirements.

In agreement with the data presented above, other investigators found that administration of Is-5-Mn for three months did not impair renal function,41 but potentiated the state of peripheral vasodilatation in cirrhosis,41 42 and promoted sodium and water retention. These changes resulted in expansion of plasma volume and increased ascites.40-43

In our study, despite similar fluid load, patients receiving Is-5-Mn showed a statistically significant reduction in urinary sodium excretion during the acute phase of treatment (⩽ three days), as shown previously,40 42 implying greater sodium retention. Taking into account that our data suggest not only absence of therapeutic efficacy but even suggest the possibility of a deleterious effect of the combined SMS + Is-5-Mn treatment, we speculate that the increased sodium retention and plasma volume secondary to Is-5-Mn administration decreased the favourable effects of SMS in patients with portal hypertension.

The remarkably high rate of therapeutic success with SMS alone merits additional consideration (SMS group 87%, SMS + Is-5-Mn group 60%). Ten patients were not included because we could not obtain haemodynamic stability after one hour, perhaps eliminating the most resistant patient subgroup. We also had a relatively low death rate (7%) compared with data in the literature (30%). This could be due to the low number of Child C patients (n = 8, 13%) included in our study. This contrasts with the higher percentage of patients in functional class Child C participating in pharmacological trials of acute variceal bleeding (range 8–88 patients, mean 41.5 (4)%).2

On the other hand, as this trial is the first study using the combination SMS plus Is-5-Mn in variceal bleeding associated with liver cirrhosis and considering ethical reasons, it was planned to perform an interim analysis. This analysis showed a large, although not statistically significant, trend against the initial hypothesis. The information on this analysis was discussed and evaluated meticulously. On one hand, there were scientific reasons not to stop the trial until a formal conclusion on the efficacy of the treatments could be reached, but, on the other, to carry on with the study would mean exposing new patients to a presumed but important increased risk of bleeding with the new treatment.

The difference in arresting bleeding between treatments in the interim analysis was 18.3% (95% CI, −3.9 to 40.4) (p = 0.175, Fischer's exact test) in favour of SMS alone. These results showed a strong trend against the initial hypothesis, and it would have been extremely unlikely to find a significant difference in favour of the initial hypothesis by completing the study. In fact, the probability of reversing the results if the trial had been completed was 3.6%. The difference in arresting bleeding was higher when the results for the 10 patients who were in the trial at the time of the interim analysis were included in the analysis: 26.7% (95% CI, +5.3 to +48.0) (p = 0.039, Fischer's exact test). Therefore the trial was stopped for ethical reasons.

In summary, our study shows that the combination of Is-5-Mn with SMS is detrimental in patients with cirrhosis during the acute phase of variceal haemorrhage. Is-5-Mn significantly diminishes the effects of SMS in controlling the initial bleeding episode, and we observed that patients treated with Is-5-Mn developed significantly more ascites. We hypothesise that during the acute phase of treatment of variceal bleeding, Is-5-Mn increases sodium and water retention and expands plasma volume, and these effects worsen the vasodilator state of these patients, decreasing the effects on variceal bleeding and favouring the development of ascites.

We conclude that in acute gastro-oesophageal variceal bleeding in cirrhotics, addition of Is-5-Mn to SMS does not improve the therapeutic efficacy of SMS alone, induces more side effects, and therefore should not be used.


We are very grateful to Dr R Jodar, Dr X Carné, and Vall d'Hebron's gastroenterology, medical, and nursery personnel for their clinical contribution and to Dr M Sust (external biostatistician) and Dr F Torres (Lab Bioestadística y Epidemiología, UAB) for methodological advice. Part of this study was presented at the Meeting of the American Gastroenterological Association, May 1998, New Orleans, Louisiana, USA.


View Abstract


  • Abbreviations used in this paper:
    isosorbide 5-mononitrate

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.